Handbook of Quality of Life in the Enlarged European Union
Recent enlargement to the east made the European Union a mo...
50 downloads
797 Views
3MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Handbook of Quality of Life in the Enlarged European Union
Recent enlargement to the east made the European Union a more diverse social space and brought it into more direct contact with the social and cultural aftermath of communism. Sound empirical knowledge on heterogeneity and homogeneity in European societies after the EU enlargement is lacking. By bringing together a collection of informative analyses of key domains of social life in the new member states and candidate countries, viewed in comparison both to each other and to the ‘old’ EU-15, this handbook will help social scientists, policy-makers and other observers cope with the unfamiliarity of this new world. In particular, it examines the implications of the new member states’ membership for the future course of EU integration. This substantial text contains 17 chapters with a focus on social conditions, such as: • • • •
poverty and living conditions; social inclusion and life satisfaction; work and labour markets; family and housing.
Making use of a range of data, this handbook will be an essential resource for undergraduate and postgraduate students and researchers of Sociology, Social Policy and Welfare, European Studies and European Union Policy. Jens Alber is Professor of Sociology at the Free University Berlin and Director of the Unit ‘Inequality and social integration’ at the Social Science Center, Berlin, which chaired the international consortium that analyzed the European Quality of Life Survey. He is the author of numerous books and articles in the field of comparative social policy and political sociology. Tony Fahey is Professor of Social Policy at University College Dublin. He has published extensively on the family, religion, demography, the elderly, housing and various aspects of social policy. Chiara Saraceno is Professor of Sociology at the University of Turin, Italy, and Research Professor at the Social Science Center, Berlin. She has written extensively on family changes and family policies, on poverty and social policies, and on gender and women’s issues.
Praise for Handbook of Quality of Life in the Enlarged European Union A superb collection that delivers accessible, balanced and penetrating analyses of social conditions among the old and new member states – Alber, Fahey and Saraceno have created an indispensable reference for social scientists, policymakers and students concerned about the quality of life in the enlarged European Union. Neil Gilbert is Chernin Professor of Social Welfare at the University of California, Berkeley. This Handbook will prove invaluable to all who want to understand the social realities of the enlarged European Union. It is an excellent source of data and analysis for those who want to compare and contrast the social situation in the Member States that joined in 2004 and 2007 as against the much more comprehensively researched EU-15. It will prove a “must read” for all scholars and students of social trends in European societies and is essential background for anyone wanting to understand the depth of the future challenges of integration and cohesion in the EU-27. Roger Liddle worked for seven years as European adviser to Tony Blair and recently, as a prinicipal adviser in the think tank of European Commission President, Jose Manuel Barroso. He is now Vice Chair of the progressive international think tank, Policy Network, and a Visiting Fellow at the London School of Economics.
Handbook of Quality of Life in the Enlarged European Union
Edited by Jens Alber, Tony Fahey and Chiara Saraceno
First published 2008 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Simultaneously published in the USA and Canada by Routledge 270 Madison Avenue, New York, NY 10016 Routledge is an imprint of the Taylor & Francis Group This edition published in the Taylor & Francis e-Library, 2007. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” © 2008 Jens Alber, Tony Fahey and Chiara Saraceno Typeset in Sabon by Keyword Group Ltd Printed and bound in Great Britain by TJ International Ltd, Padstow, Cornwall All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data Handbook of quality of life in the enlarged European Union / edited by Jens Alber, Tony Fahey, and Chiara Saraceno. p. cm. 1. Quality of life—European Union countries. 2. Social indicators— European Union countries. 3. European Union countries—Social conditions—21st century. 4. European Union countries—Economic conditions—21st century. 5. European Union. I. Alber, Jens. II. Fahey, Tony. III. Saraceno, Chiara. HN374.H35 2007 306.3072′04—dc22 2007017233 ISBN 0-203-93630-2 Master e-book ISBN
ISBN 10: 0–415–42467–4 (hbk) ISBN 10: 0–203–93630–2 (ebk) ISBN 13: 978–0–415–42467–7 (hbk) ISBN 13: 978–0–203–93630–6 (ebk)
Contents
Figures Tables Contributors Preface
vii xi xiv xv
ROBERT ANDERSON
Introduction: EU enlargement and quality of life: the context and purpose of this book
1
J E N S A L B E R , T O N Y FA H E Y A N D C H I A R A S A R A C E N O
PART I
Fertility, families and households
25
1
27
Fertility patterns and aspirations in Europe T O N Y FA H E Y
2
Patterns of family living in the enlarged EU
47
CHIARA SARACENO
3
Is there a generational cleavage in Europe? Age-specific perceptions of elderly care and of the pension system
73
WOLFGANG KECK AND AGNES BLOME
4
Family policy patterns in the enlarged EU
100
THOMAS BAHLE
PART II
Employment and working conditions
127
5
129
Employment patterns in the enlarged EU JENS ALBER
6
Working conditions and quality of work: a comparison of Eastern and Western Europe C L A I R E WA L L A C E A N D F L O R I A N P I C H L E R
162
vi 7
Contents Extension through dilution? European integration, enlargement and labour institutions
175
JELLE VISSER
PART III
Material living conditions 8
Poverty, deprivation and economic vulnerability in the enlarged EU
199 201
C H R I S T O P H E R T. W H E L A N A N D B E RT R A N D M A Iˆ T R E
9
Minimum income policies in old and new member states
218
B E A C A N T I L L O N , N ATA S C H A VA N M E C H E L E N A N D B E R N D S C H U LT E
10
Housing conditions
235
´ SKI H E N RY K D O M A N
11
Institutional drivers of housing inequalities in the enlarged EU
254
MICHELLE NORRIS
PART IV
Social capital and social cohesion
277
12
279
Patterns of sociability in the enlarged EU M A N U E L A O L A G N E R O , PA O L A T O R R I O N I A N D C H I A R A S A R A C E N O
13
Feeling left out: patterns of social integration and exclusion
304
PETRA BÖHNKE
14
The perception of group conflicts: different challenges for social cohesion in new and old member states
328
JAN DELHEY AND WOLFGANG KECK
PART V
Processes of Europeanisation
353
15
355
Migration and mobility culture: an analysis of mobility intentions H U B E RT K R I E G E R
16
Where we stand in Europe: citizen perceptions of European country rankings and their influence on subjective well-being
385
JAN DELHEY AND ULRICH KOHLER
17
Assessing the quality of European surveys: towards an open method of coordination for survey data
405
ULRICH KOHLER
Index
425
Figures
1.1 2.1 2.2 2.3 2.4 2.5 3.1
3.2 3.3 3.4 3.5 3.6 3.7 3.8 4.1 4.2 4.3 4.4 4.5 4.6 4.7 5.1 5.2
Relationship between actual number of children and personal ideal number 38 of children among women with completed fertility (ages 40–64) Household status of young men (18–34): three distinct patterns 57 Incidence of unemployment among European young singles living 58 with their parents, by country Households of elderly people: differences between men and 60 women (65 and over) in the three clusters Households of 35- to 64-year-olds: differences between the 35–49 63 and the 50–64 age groups, by gender and country cluster Distribution of working-age couples, by working status in Europe 66 Share of individuals who care for elderly persons within the 78 household and preference of working-age family members as care givers for own parents Share of individuals in different age groups who opt for family 79 support as the preferred care solution, by country Difference between men and women according to their preference 81 for family care Family orientation in care issues and public care efforts, by country 82 Preferences on care financing, by country 84 Trust in pension system, by country 89 Trust in pension system, by country and age groups 90 Perception of tensions between young and old, by level of trust in 94 public pension system Synopsis 1: Conceptual map of family policy forms in Europe 103 Structure of social expenditure on family benefits, Europe 2004 108 Variations in family allowances, Europe 2004 111 Maternity and parenting benefits, Europe 2004 112 Childcare services and parenting policies, Europe around 2000 115 Synopsis 2: Variations in family policies – a ‘cultural map’ 116 of Europe Family policy problem constellation, around 2000 118 Development of economic output during the Great Depression 131 and post-communist transformation Employment rates of the population aged 15–59 in post-communist 131 member states and in Eastern European countries outside the EU, 1989–2004
viii 5.3
Figures
Employment rates in consumption-related and production-related services, 2003 5.4 Development of real GDP and employment rate in post-communist NMS-8 5.5 Economic growth rates and change in employment rate in EU-25, 2000–2004 5.6 Development of male and female employment rates in Europe and the United States, 1979–2003 (% of population aged 15–64) 5.7 Employment rates of older (55–64) and younger (15–24) age cohorts in EU-25, 2004 5.8 Employment rates of people with high and low educational attainment in EU-25, 2004 (population aged 25–64) 5.9 Social expenditure ratio and employment levels in EU-25, 2004 5.10 Level of social assistance benefits and low-skill employment in EU countries, 2004 5.11 Level of unemployment insurance benefits and employment rate, 2004 5.12 Burden of social security contributions and low-skill employment in EU countries, 2004 6.1 Perceptions of pay: gender and occupational differences across Eastern Europe 6.2 Perceived job security: gender and occupational differences across Eastern Europe 6.3 Perceived intrinsic rewards: gender and occupational differences across Eastern Europe 6.4 Job satisfaction: gender and occupational differences across Eastern Europe 7.1 Bargaining coverage and union density 7.2 Earnings inequality and bargaining coverage 7.3 Union presence, 2003–2004 8.1 Median monthly household equivalised incomes in purchasing power standards, by country 8.2 Median monthly household income, by income quartile across countries (in purchasing power standards) 8.3 Mean deprivation (10 items) across countries 8.4 Mean deprivation (10 items), by income quartile, by country 8.5 Share of persons in poverty at 60% median equivalised household income, 2003 8.6 Vulnerable class size, by economic clusters 8.7 Conditional probabilities for latent class models for income deprivation and economic stress 9.1 Expenditure in % of GDP, social exclusion (NEC), means-tested cash benefits, 2003 9.2 Monthly net disposable income of social assistance recipients 224, (working age), 2004 9.3 Cross-country correlations between net incomes of various household types on social assistance, 2004
143 145 146 148 148 149 150 152 153 155 166 168 168 171 181 183 191 204 204 205 206 207 212 212 221 225 227
Figures
ix
Net disposable income of social assistance recipients (working age, average of four family types), 2004 10.1 Tenure status according to accommodation, by country grouping 10.2 Relationship between GDP per capita and % of ownership with mortgage 10.3 Relationship between GDP per capita and index of housing quality 11.1 Number of dwellings and housing output per 1,000 inhabitants in European countries, 2000 11.2 Ratio of outstanding mortgage debt to GDP in European countries and annual growth to mortgage debt, 2004 11.3 % of dwellings in high-rise buildings in European countries, 2004 12.1 Distribution of countries along two dimensions of the secondary and tertiary sphere of sociability 12.2 Patterns of sociability in Europe (country clusters) 12.3 Distribution of sociability clusters with regard to selected structural indicators 13.1 The prevalence of belonging and marginalisation in the EU (mean index values) 13.2 Mean index value of belonging in three population groups 13.3 Perceptions of belonging and the availability of social network support (mean index values) 13.4a Social support outside the household as a buffer? 13.4b Family back-up as a buffer? 13.5 The impact of long-term poverty on perceived social exclusion across countries according to their welfare level 14.1 Impact of perceived tensions between rich and poor people on social trust 14.2 Intensity of group tensions as perceived by the population 14.3 Intensity of perceived vertical and ethnic tensions (country clusters) 14.4 Cross-national correlates of vertical tension perception 338, 14.5 Cross-national correlates of ethnic tension perception 341, 15.1–15.4 Stock of macro-indicators and change in intended 375, migration, 2001–2005, for EU-25 15.5–15.8 Change in macro-indicators and change in intended 377, migration, 2001–2005, for EU-25 15.9–15.10 Stock of subjective indicators, 2001, and change in intended migration, 2001–2005, for EU-25 16.1 Country–EU comparisons (share of missing answers, by survey country) 16.2 Outcomes of country–EU comparisons 16.3 Cross-check between reality and perceptions (economic situation) 16.4 Cross-check between reality and perceptions (employment situation) 16.5 Cross-check between reality and perceptions (overall quality of life) 16.6 Impact of country–EU comparison on life satisfaction 17.1 Components of the sampling process, by country and survey programme
228
9.4
236 238 246 262 263 267 289 291 293 308 312 316 318 318 324 331 333 335 339 342 376 378 380 390 393 395 396 397 400 412
x
Figures
17.2 17.3
Proportions of women (differences between survey and official sources) Proportions of women among gender heterogeneous couples
415 419
Tables
1.1 The ideal–actual fertility gap in three broad age groups of women in European countries, 2001–2002 1.2 Mean personal ideal number of children and actual number of children by school-leaving age among women with completed fertility (ages 40–64) in the EU-15, NMs and CC-3 1.3 Fertility ideal attainment and school-leaving age among women with completed fertility (ages 40–64) 1.4 Mean general ideal family size 1.5 Fulfilment of general ideal number of children among women with completed fertility (ages 45–64), pooled data for 8 European societies, 1981, 1990, 2001 A1.1 Sample Ns for female samples by countries and age groups, combined Eurobarometer dataset, 2001–2002 2.1 Household patterns in an enlarged Europe 2.2 Household statuses of the young in Europe (ages 18–34) 2.3 Elderly men and women (65 and over) living with their children, by patterns of family formation by the young 2.4 Use of time among working Europeans 3.1 Age-group differences in perceived financing responsibility for care 3.2 Logistic regression on tension perception by age and trust in pension scheme 4.1 Social expenditure on family benefits, Europe 2004 4.2 Eligibility for family allowances and variations in benefit rates, Europe 2004 4.3 Childcare coverage rates in Europe, around 2000 5.1 Indicators of employment development in the post-communist new member states 5.2 Recent employment records in EU-25, employment levels, 2004, and changes, 2000–2004 5.3 Measures of labour market inclusion and exclusion, 2004 5.4 Sector-specific employment rates in the enlarged EU A5.1 Possible determinants of employment development in EU-25 6.1 Working conditions in Eastern Europe I (working hours, unemployment, wages, job security, pay, intrinsic rewards and job satisfaction, means and standard deviations)
34, 35 39
40 41 42
44 51 54, 55 61 68 86 95 107 110 114 132 134 136 141 156 165
xii 6.2
Tables
Working conditions in Eastern Europe II (workload, tight deadlines, dangerous and unhealthy work environment, autonomy and influence, future prospects of promotion and career and job satisfaction, means and standard deviations) 7.1 Key changes in employment, unemployment and contracts, 1998–2004 7.2 Collective bargaining (private sector) and wage setting, recent years 7.3 Statutory minimum wages in Europe 9.1 Statutory adjustment mechanisms relating to social assistance benefit standards in EU countries, 2004 10.1 Housing tenure 10.2 Persons living in own homes, by age, income, area of residence and occupational status 10.3 Average floor space 10.4 Number of persons per room, by ownership status, income, region and age 10.5 Households that reported deficits in accommodation 10.6 Households that do not have an indoor toilet, by age of respondent and area of residence 10.7 Good housing (having at least one room per person and perceiving none of five housing deficits 10.8 Complaints about environment 10.9 Relationships between satisfaction with accommodation and general satisfaction with life and objective characteristics of housing 10.10 Relationships between trust and objective characteristics of housing 11.1 Occupied dwellings, by tenure, in European countries, 1980, 1990, 2000 11.2 Sources of funding for new construction in seven longstanding EU member states, 1994–1995 11.3 Age distribution of the housing stock in European countries, various years 11.4 Institutional drivers of housing inequalities in an enlarged Europe 12.1 Dimensions of sociability, by country clusters 12.2 The secondary sphere of sociability (type of support and 283, contacts), by country 12.3 The tertiary sphere of sociability (activity in political and 286, associative domains) 12.4a Involvement in public sociability, by cross-cluster 12.4b Involvement in private sociability, by country cluster 13.1 Perceived social exclusion, mean index value and distribution of 309, single indicators per country (agree and strongly agree summarised) 13.2 What determines perceived marginalisation and belonging? 313, 13.3 Socio-economic precariousness and social support as 320, determinants of belonging 13.4 The independent effect of country characteristics on the individual experience of social exclusion 14.1 Vertical tension perception in different social groups
170
177 180 185 223 237 239 240 242 243 244 245 248 250 251 256 261 265 272 282 284 287 294 295 310
314 321 322 343
Tables
xiii
14.2 Ethnic tension perception in different social groups 345 A14.1 Documentation of indicators 348 15.1 Mobility intentions in EU member states, 2001 and 2005 359 (population aged 18–64) 15.2 Mobility intentions (regression models for 2005) 362–364 15.3 Mobility intentions (regression models for 2001) 367–369 15.4 Comparison of the effect of macro-economic and subjective 381 conditions on the dynamic intention to migrate between EU-25 and EU-15 16.1 Explaining difficulties of rating one’s own country (coefficients 392 of logistic regressions of missing answer vs. non-missing answer) 16.2 Explaining personal life satisfaction: the influence of economic 400, 401 country–EU comparisons with various control variables 17.1 Target population, number of countries and number of 407 observations, by survey programme 17.2 Minimum, average and maximum response rate, by survey 409 programme 17.3 Number of countries for which complete information about the 411 specified component of the sampling process is available, by survey programme 17.4 Fraction of countries with best available sampling process, 413 by survey programme 17.5 Mean difference from the true value and fraction of countries 417 with values outside the confidence bounds, by survey programme 17.6 Mean difference from the true value and fraction of countries with values outside the confidence bounds among gender heterogeneous couples, by survey programme 420 17.7 Overall sampling quality of survey programmes 421
Contributors
Jens Alber, Social Science Research Center Berlin, Germany Robert Anderson, European Foundation for the Improvement of Living and Working Conditions, Dublin Thomas Bahle, University of Mannheim, Germany Agnes Blome, Social Science Research Center Berlin, Germany Petra Böhnke, Social Science Research Center Berlin, Germany Bea Cantillon, University of Antwerpen, Belgium Jan Delhey, Jacobs University Bremen, Germany Henryk Doman´ski, University of Warsaw and Polish Academy of Sciences, Poland Tony Fahey, University College Dublin, Republic of Ireland Wolfgang Keck, Social Science Research Center Berlin, Germany Ulrich Kohler, Social Science Research Center Berlin, Germany Hubert Krieger, European Foundation for the Improvement of Living and Working Conditions, Dublin Bertrand Maître, Economic and Social Research Institute, Dublin Michelle Norris, University College Dublin, Republic of Ireland Manuela Olagnero, University of Turin, Italy Florian Pichler, University of Aberdeen, UK Chiara Saraceno, University of Turin, Italy, and Social Science Research Center Berlin, Germany Bernd Schulte, Max Planck Institute for Foreign and International Social Law, Germany Paola Torrioni, University of Turin, Italy Natascha Van Mechelen, University of Antwerpen, Belgium Jelle Visser, University of Amsterdam, Netherlands Claire Wallace, University of Aberdeen, UK Christopher T. Whelan, Economic and Social Research Institute, Dublin
Preface
For the 2007 Spring European Council the European Commission presented a first stocktaking of social reality in an enlarged Europe. This report argued that there is a lack of consensus on the common social challenges facing Europeans, and that there is a need for better analysis and understanding of the social situation. To support debate on the social issues and challenges facing Europe, the Bureau of European Policy Advisers (a Directorate General of the European Commission) has issued a consultation paper, which begins: How can the social well-being of all Europe’s citizens be best advanced within a globalising world? This question should be at the heart of everything the EU and its Member States do. Public policy imperatives, such as “Growth and Jobs”, the Lisbon strategy, and the drive for greater competitiveness are not ends in themselves – but means to an end – the well-being of European citizens. (Liddle and Lerais 2007) The challenges arising from social exclusion, an ageing population, changing family structures and gender roles, and now enlargement, have pushed quality of life issues to the fore in the EU policy debate. Their impact is direct on people’s everyday lives, families, communities and society. The recent work of the European Foundation for the Improvement of Living and Working Conditions has thus focused on monitoring trends and changes in living conditions and quality of life across the EU and in candidate countries. ‘Living conditions’ clearly embraces a very wide area of policy interest, with a particular need to map and understand disparities associated with age, gender, health, ethnicity and region. The Foundation’s four-year programme points to the need to link the assessment of living conditions to the changing nature of employment, work organisation, and working conditions on the one hand, and to the modernisation of social protection and social welfare services on the other. Quality of life for Europe’s population is increasingly at the centre of the Foundation’s work. The enlargement of the EU to incorporate twelve new member states over the past four years has increased not only size and population, but the diversity of people, lifestyles and culture in Europe. This diversity is undoubtedly enriching daily lives but, as with other developments in economy and employment, not for all European citizens. The flipside of diversity is inequality, which is evident between member states, but often as much or more within countries and regions. The European institutions have a range of policies and programmes that impact on key ‘quality of life’ issues: employment conditions; health and safety; social inclusion;
xvi
Preface
mobility; equal opportunities. The elaboration of policy responses to established and emerging social challenges will depend upon information, but more so on insight and understanding of the living conditions and experiences of people in the EU. Appropriate measures will demand intelligence not only on objective conditions or the social situation, but regarding how people feel about these conditions, their concerns and priorities. Monitoring and analysis should cover key domains that influence quality of life – income, health, family – but should also examine how people assess the quality of the society in which they live, including conflicts and tensions in society, the quality of the environment and their satisfaction with services of general interest. This volume takes forward the analysis and interpretation of the complex social challenges in an enlarged Europe. I welcome its contribution to the debate on quality of life in Europe and believe it provides an informed basis for both further research and the identification of priorities for public policy. Robert Anderson European Foundation for the Improvement of Living and Working Conditions
Reference Liddle, R. and Lerais, F. (for the European Commission) (2007) ‘Europe’s social reality: a consultation paper from the Bureau of European Policy Advisers’. Online. Available (accessed March 2007).
Introduction: EU enlargement and quality of life The context and purpose of this book Jens Alber, Tony Fahey and Chiara Saraceno Introduction The fall of the Berlin wall in 1989 and the subsequent collapse of communism were a major advance in healing the divisions of Europe left after World War II. The continent was no longer at war with itself, either in the form of the hot war that ended in 1945 or the cold war that endured ten times longer. The recent eastern enlargements of the European Union decisively advanced this process of reconnection. Eight former communist states joined the Union in 2004 (along with the small Mediterranean states of Cyprus and Malta) and two more – Bulgaria and Romania – joined in 2007. This development consolidated the move to democracy and the freemarket economy in a whole swathe of the continent where these possibilities had long been snuffed out. It also enhanced the long-term prospect of peace within Europe and strengthened the EU’s capacity to hold its place in the world. In view of these gains, the eastern enlargement could be ranked among the EU’s greatest achievement so far. It is also notable for having been brought about by means of an inter-state machinery, embodied in the institutions of the EU, that has proved surprisingly effective and has consistently defied predictions that it would collapse under the weight of its own ungainliness. While the EU cannot claim the credit for bringing communism to an end, it can claim to be a central mechanism by means of which European states have been enabled to come to terms with both the possibilities and the stresses of the post-communist era. Great though these gains have been, their very ambition has stretched the political and institutional capacities of the EU to their limits, as is sketched out later in this chapter. How well the Union will cope with these strains is partly a matter of macrolevel factors such as the evolution of relationships between member states, the adaptation of EU institutions and the impact of macro-economic forces. However, it also depends on the micro-level impact of the EU on the daily lives of its people. The geo-political significance of EU integration will not of itself ‘sell’ the project to ordinary citizens. It must also deliver improvements in their day-to-day well-being and quality of life – or at least not be seen as a threat to what the various national populations have already achieved in these areas. The eastern enlargement has a particular significance in this regard. For people in the new member states (NMS), the hope of improvements in economic opportunities and standards of living was a major motivation for joining the EU in the first place. For people in the incumbent states, the very poverty of the new members and the fear that they would drag down living conditions and social cohesion in the richer parts
2
Jens Alber, Tony Fahey and Chiara Saraceno
of Europe, through either low-cost competition or floods of migrants, gave rise to concerns that enlargement may have gone too far, too fast. The candidate countries that have been listed as possible members in the future – of which Turkey is the most significant – accentuate these contrasts further, as they promise to add further to the already high level of diversity and inequality across countries in the EU. The hopes of the new members and the worries of the incumbents thus place quality of life at the centre of debate about the future of the European project. The ‘quality of life’ perspective on people’s living conditions, as we describe further below, aims to broaden the scope of what is considered beyond the economic to include a range of social domains of life, such as family life, the social dimensions of work, neighbourhood and community, housing and so on. It also, crucially, is not just concerned with objective indicators but also with how people themselves view their circumstances and evaluate the quality of their lives. The usefulness of this broader view is that it approximates more closely to how people experience and react to their own situation. It thus taps into what are likely to be core concerns of the EU as it seeks to deliver benefits that will establish its appeal and legitimacy among citizens. The concern that motivates this book is that, important though these issues may be for the future of the EU project, knowledge of quality of life in the European Union is seriously incomplete, particularly with regard to the situation in the NMS and candidate countries. Social reporting has a long tradition in Western Europe and has been built upon by the EU to develop wide-ranging data and analyses on social conditions (for example, through the European Community Household Panel survey, which ran from 1994 to 2002). However, this work has largely been confined to the EU-15 or even to sub-sets of older member states within the 15. The NMS and candidate countries lack a similar tradition, particularly at the level of a cross-national picture that would enable one to view these countries alongside each other rather than in isolation. As the eastern enlargement took place, therefore, assessment of living conditions in the NMS and candidate countries relied on a narrow range of economic and demographic indicators, such as GDP per capita, labour market measures and mortality data. Valuable though these are, they do not amount to the multidimensional approach required to capture broader quality of life. A further relevant gap in knowledge concerns the institutional context within which various domains of social life function. These contexts are a particular concern in former communist countries since despite the transition to democracy and the free-market economy, an institutional legacy of the old regime survives in many areas of life. Thus, for example, housing conditions in these countries are not only of distinctive quality (usually in the direction of being quite poor) but also are the product of distinctive housing systems, which in turn are the outcome of institutional histories unlike anything experienced in the west. Quite often, in areas such as these, the details of day-to-day living conditions are difficult to make sense of unless one also understands something of government policies, state regulation, market mechanisms and civil society institutions that together make up the institutional framework within which they function. Most, if not all, welfare state typologies, which are used to understand national differences in living conditions and behaviours, have been developed looking at Western Europe. We still lack a similar comprehensive exercise including also the formerly socialist countries. The purpose of this book, which consists of a series of chapters written by authors who are experts in their field, is to try to fill these gaps in knowledge on quality of
Introduction
3
life in the EU, with particular reference to the NMS and candidate countries, and to draw some implications on the future of the EU project. In later sections of this chapter, we set out in more detail what this purpose entails and how it is fulfilled in the book. We first set the scene by considering a number of macro-level consequences of the enlargement process for the EU.
1. Background: an enlargement crisis? The accession of 12 new member states has profoundly transformed the nature of the European Union. The Union’s official motto as specified in Article I-8 of the Treaty establishing a Constitution for Europe – ‘United in diversity’ – has since 2004 certainly become more true with respect to diversity. But how much unity still remains after enlargement? Will Europe grow together or will a lasting cleavage between old and new member states put a stop to further integration? For a number of reasons, many policy pundits expect the enlargement process to lead to a lasting chasm and to the hitherto most serious challenge to the unity and cohesion of the European Union. First of all, there are huge differences in economic wealth as well as in population size. Adding roughly 74 million people, the 2004 enlargement augmented the number of EU member states by 66 per cent, increased the population size of the Union by roughly 20 per cent, while raising total economic output as measured by the Union’s Gross Domestic Product by a mere 5 per cent. Even when measured in purchasing power parities, the GDP per head in the ten new member states at the time of accession was less than one half of the EU-15 average. The entry of Bulgaria and Romania in 2007 added another 30 million citizens (+ 6 per cent), but an increase of total economic output of only less than 1 per cent (Berié and Kobert 2006: 575). Among the new member states, only four – Poland, Romania, Hungary, and the Czech Republic – have double-digit population sizes. Bulgaria and Slovakia are the only other countries to exceed a population size of 5 million. The other six newcomers are small with population sizes between 400,000 (Malta) and 3.4 million (Lithuania). The addition of several countries that combine demographic smallness with economic poverty is bound to exacerbate the always difficult task of the European Union to strike a balance between the equality of all sovereign member states on the one hand and their different economic and demographic weight on the other. Second, the accession process itself is likely to leave lasting bad memories. During the accession negotiations, several representatives of the aspiring countries complained about a marked power asymmetry in the accession process (Heidenreich 2006). Experiencing the accession criteria as an encroachment upon their national sovereignty, they felt they were being asked to meet expectations that not all of the established members of the club fulfilled themselves. Such concerns were repeatedly and perhaps most forcefully voiced by the Czech president and former Prime Minister Vaclav Klaus, but also by Polish politicians. Even after accession, the new member states were not put on a par with the old members. Due to various transitional arrangements, basic rights and freedoms will only be fully extended to NMS citizens after the end of various transitional periods (e.g. 2016 for the free movement of capital, 2011 for the free movement of persons, 2008 for the free movement of goods). The farmers of the new member states were initially entitled to only 25 per cent of the direct payments granted to their peers in old member states, and they will reach
4
Jens Alber, Tony Fahey and Chiara Saraceno
equal status only gradually up to 2013 (von Baratta 2003: 119). Whereas the farmers of France (21 per cent), Spain (14 per cent), Germany (13.5 per cent), and Italy (11 per cent) received roughly 60 per cent of the payments made under the Common Agricultural Policy (CAP) in 2004, the agricultural sector in the ten new member states initially had to settle with a combined 4 per cent of total expenditure (Berié and Kobert 2005: 553). Third, after many decades under the shadow of the Soviet empire, the post-communist countries seem to look more to the US than to the EU as a social and political model. This is so for a number of reasons. First, the entry of the former communist countries into the market economy and the restructuring of their social security systems have been strongly shaped by the requirements – and ideology – of the IMF and the World Bank and of the so-called Washington consensus, with its priority for rapid privatisation and lack of attention to building social and organisational capital (Stiglitz 1999a, 1999b). Second, the US welcomed these countries as members of NATO long before the European Union opened up its ranks (Ash 2004). During the Iraq war, American newspapers and government officials such as Donald Rumsfeld, then Secretary of Defence, frequently made a polemical distinction between ‘old and new Europe’. Originally coined to distinguish between proponents and opponents of the war against Iraq, the concept was soon extended to distinguish a group of supposedly dynamic countries actively facing the future from a passive and nostalgic group clinging to the past and supposedly coping only reluctantly with the modern globalised world.1 For all these reasons, many observers see the European Union in a profound crisis after enlargement (e.g. Bach 2006 has a particularly gloomy vision). It is claimed that due to the failure of the constitution, the decision-making structures which were originally designed for a small number of fairly homogenous nations do not suit a much bigger and more heterogeneous Union. For sheer reasons of size, in this view, finding consensus and making decisions in the Commission or in the European Council has become a far more difficult task now that there are 27 rather than 15 members with differing interests. In addition, the poverty and economic backwardness of the new members will strain EU Structural Funds, which were never designed to cope with this degree of regional inequality or this big an agricultural sector. Consequently, conflicts of interest between the net contributors to the EU budget – above all the Netherlands, Sweden and Germany – and the net beneficiaries – especially the new member states and Greece, Portugal and Spain – are likely to grow. In 2005, policymakers in the old member states such as the former Swedish Prime Minister Goran Persson or the former German chancellor Gerhard Schröder complained that the EU’s big net contributors were effectively subsidizing the low tax rates of their competitors in the new member states, who could afford to cut taxes because hefty EU subsidies compensated for any lost revenue (http://euractiv.com/en/taxation/ flat-tax/article-136190, accessed 31 January 2007). Of course, there are also more positive assessments of the enlargement process. First of all, the promise made to the aspiring countries that joining the European Union would lead to freedom, democracy and economic development has basically been fulfilled. In international comparisons of democratic progress or the level of corruption as measured by Freedom House (2006) or Transparency International (2005), the new member states usually score much better than other post-communist countries. Economically, the new member states have recently begun to catch up,
Introduction
5
experiencing higher growth rates than the old members. The three Baltic countries even had double-digit growth rates in some years, thus coming close to the growth rate of the Chinese economy. While the post-communist transformation put the transition countries into a much deeper economic slump than western countries had experienced during the Great Depression, all new member states have by now exceeded the material standard of living (as measured by GDP per capita) which they had reached under the communist regime. The idea that the EU institutions could not cope with the post-enlargement situation unless fundamentally overhauled proved too pessimistic. For many years the Union has already proved capable of practising a pattern of differentiated integration where selected groups of countries rather than all members opt for closer cooperation in specific policy fields. Thus, only five countries originally signed the Schengen agreement on the elimination of border controls, from which Ireland and the United Kingdom still abstain, thus leaving not only the new member states outside the scope of the agreement. The single currency was adopted by only eleven of the old member states (absent: Denmark, Greece, Sweden, the United Kingdom), and is now valid in 13 countries including Greece and Slovenia. Thus, the Union has been capable of living with different degrees of integration in different policy realms, relying on a gradual or stepwise expansion of cooperation to new participants. This underlines the notion that the European Union resembles a tricycle that can halt the process of deepening at any moment rather than a bicycle that has to keep moving forward in order not to falter or fall (Moravcsik 2004). Yet, even if one adopts a more pragmatic and less gloomy outlook, there are two problems which, having not yet been solved successfully, will require some innovative solutions, because they are exacerbated by enlargement. The first one concerns growing disunity among political elites, the second one declining compliance among taxpayers and voters. Given the divergent political interests of net contributors and net beneficiaries and vast differences in economic development, consensus building in budget negotiations has already become visibly more difficult. As pointed out most forcefully by the British Prime Minister Tony Blair (2005) during the British Presidency, the EU budget is predominantly geared to provide support for declining sectors of the economy such as the coal and steel industry at the very beginning and agriculture today, instead of focusing on investments in education, science, and research that would help to bolster European productivity and competitiveness. Indeed more than 40 per cent of the EU budget is still spent on various supports for agriculture and more than another third is spent on structural funds supporting the development of poor regions whose GDP per head is less than 75 per cent (in the case of so-called objective 1 areas entitled to subsidies), or less than 90 per cent of the EU average (in the case of the so-called cohesion fund). Even though they are only gradually being incorporated into this system of support, the demand of the new member states is particularly high, because 92 per cent of their population live in poor objective 1 regions. Overall, 36 regions in new member states and 32 regions in old member states belong to this group (Süddeutsche Zeitung 17.03.05). In order to cope with the growing demand, the European Commission had proposed to raise the EU budget to 1.24 per cent of the Union’s Gross National Income in the period 2007–2013. But the camp of net contributing countries – led by Germany, France, the United Kingdom, the Netherlands, Austria, and Sweden – insisted in the 2005 budget negotiations on limiting it to 1 per cent. A compromise
6
Jens Alber, Tony Fahey and Chiara Saraceno
was reached after long and tough negotiations in December 2005 by setting the budget ceiling at 1.045 per cent of EU output. The compromise also included a 20 per cent decrease of the UK rebate which Prime Minister Blair had agreed upon in exchange for France’s agreement to endorse a full and wide-ranging review of the budget for farm subsidies in 2008–2009. Of the total budget of E864.3 billion for the period 2007–2013 which was formally endorsed in May 2006, E157 billion or roughly 18 per cent will go into aid for the new member states (http://news.bbc.co.uk/ 2hi/Europe/4538100.stm, accessed 31 January 2007; Berié and Kobert 2006: 577). The fact that more than one-third of the EU budget goes into funds for regional policies is usually justified as an indispensable means for fostering cohesion in the European Union. In the Commission’s reports on economic and social cohesion, cohesion is understood as a synonym for the similarity or equality of standards of living across European countries and regions. From this perspective, the more similar countries and regions become with respect to the standard of living, the more social cohesion there will be in the European Union.2 In short, EU regional policy sees convergence as resulting in cohesion, and the structural funds serve to promote cohesion by redistributing resources in a way that helps to make regions more similar. Defining cohesion as a synonym for equality poses a few problems, however. From a theoretical perspective, sociologists in the tradition of Emile Durkheim tend to see growing equality also as a platform for increased competition. They relate the origins of solidarity under modern conditions to the division of labour, which makes unequal individuals mutually dependent on each other. From this perspective, cohesion is not a synonym for equality, but a concept capturing the strength of social bonds, the degree of connectedness and felt solidarity among individuals and groups in society, and the degree to which the members of a society develop shared values and a common identity. If we keep the concepts of equality and cohesion conceptually apart, it will be possible to examine the extent to which people in equal or unequal positions actually engage in social relations, share a common sense of identity and similar norms of appropriateness. The question whether equality of living conditions fosters social cohesion then becomes a matter of empirical analysis rather than a matter of definition. Once we accept that convergence of living standards is not the same as cohesion, the relevant question concerning social cohesion on the European level is to what extent Europeans have moved beyond the nation state when it comes to defining the territorial unit to which they attribute a sense of belonging, of collective identity and of shared responsibilities. In a recent book and a series of articles based on comparative data from surveys, the sociologist Jürgen Gerhards (2005; 2006) has studied these issues. His result is that there is a marked discrepancy between the official institutionalisation of a European citizenship on the one side and the ideas of ordinary Europeans on the other. The latter continue to frame citizenship rights in national terms and hesitate to extend notions of solidarity beyond the boundaries of the nation state. In this sense, there is a widening gap between the policy visions of top-level Eurocrats and ordinary citizens. One of the specific findings is that the idea of equal access to national labour markets for all Europeans anywhere in the Union has not caught on with the citizens of Europe, least of all in new member states. According to the European Values Survey, there are only four EU member states in which a majority of respondents support equal treatment for foreigners in the labour market when jobs for co-nationals are
Introduction
7
scarce (Sweden, Netherlands, Denmark and Luxembourg). In ten of the old member states large majorities of two thirds or more advocate making a distinction between their co-nationals and foreigners. In the new member states, these majorities are even larger, varying between 80 and 96 per cent in all countries bar Estonia3 (see the data in Gerhards 2006). If this is the case, redistributive attempts to produce more equality across Europe will not automatically foster more social cohesion among Europeans, but may instead invoke problems of legitimation and citizen backlash. There are several additional signs of a widening gap between the supranational structure of the European Union and the predominantly national mindsets of Europeans. First of all, voter turnout at European elections has always been much lower than in national elections. This so-called ‘Euro-gap’ (Rose 2004) is even higher – about one-third – in new member states (29 points) than in old ones (22 points Wessels 2006). In the eight post-communist new member states only 31 per cent of the citizens – compared to 53 per cent in the old EU-15 – participated in the 2004 elections to the European Parliament. Over time, average turnout at European Parliament elections has shrunk from 66 per cent in 1979 to 48 per cent in 2004 (Wessels 2006). Second, Eurobarometer time-series data on perceptions of Europe among Europeans show that identification with Europe was highest in the early 1990s, but has since been declining. The percentage of Europeans considering the EU membership ‘a good thing’ decreased by almost 20 percentage points from a peak of 72 per cent at the beginning of the 1990s to 53 per cent in autumn 2006 (European Commission 2002: 19; European Commission 2006a: 6). There is also much scepticism regarding further enlargement. In Eurobarometer 66 of autumn 2006, the percentage of respondents opposing further enlargement (42 per cent) was almost as high as the percentage of proponents (46 per cent). In five of the old member states, less than 40 per cent expressed themselves in support of further enlargements (with a variation from 36 per cent in the UK to 30 per cent in Germany, and with France, Luxembourg and Austria in between - European Commission 2006a: 29). All this reflects the fact that European integration has primarily been an elite project, driven by the war experience of political leaders and technocratic elites who sought to overcome the tradition of European Civil Wars (as Göran Therborn, 2000, characterised the two world wars of the twentieth century). Protracted postwar prosperity and security concerns during the Cold War helped the project to gain citizen consent even though effective chances for citizen participation remained very limited, as politics were designed in rather non-transparent networks of experts which largely left the people out. The common threat posed by the Soviet Union has now ceased, and the long years of prosperity and welfare state expansion have given way to sluggish growth, welfare state restructuring and increased international competition in a globalised world. As a result, further integration will presumably need actively to win citizens consent by convincingly demonstrating that concrete welfare gains are associated with the European project. As Mark Kleinman described the new political climate concisely even before the time of enlargement: ‘The time of permissive consensus to European integration among European citizens is over’ (Kleinman 2002: 214). The failure of the French and Dutch referendums on the European constitution is convincing proof of this thesis. The European Commission acknowledged the problem of declining citizen consent when it noted already in a 2001 White Paper on European Governance that ‘people
8
Jens Alber, Tony Fahey and Chiara Saraceno
increasingly distrust institutions and politics or are simply not interested in them’ and identified these problems as ‘particularly acute at the level of the European Union’. This was followed by the self-critical diagnosis that: ‘Many people are losing confidence in a poorly understood and complex system to deliver the politics that they want. The Union is often seen as remote and at the same time too intrusive’ (Commission of the European Communities 2001: 3). One of the Commission’s proposals for change was to communicate more actively with the general public on European issues and to reach out to involve regional and local stakeholders as well as civil society actors more actively into the development of policy proposals. The Commission report on European governance (2003) reiterated the importance of these objectives and described strategies for implementing the White Paper. In a Communication published in 2005 the Commission also responded to the negative votes on the European Constitution in France and the Netherlands by declaring a ‘period of reflection’ and by publishing its ‘Plan-D for Democracy, Dialogue and Debate’ (Commission of the European Communities 2005). In order to find out how Europeans conceive of the future of the European Union, a special Eurobarometer was launched in 2006. Among its many complex results, four findings are particularly worth mentioning: (1) A majority of Europeans is in favour of shifting more decision-making to the European level, particularly so with respect to the fight against terrorism (European Commission 2006b: 40). (2) The proportion who consider that things are moving in the right direction in the European Union is only 39 per cent, compared to 27 per cent who feel that things are going in the wrong direction, and 34 per cent who are undecided (2006b: 17). (3) In most policy areas, the interviewees have a critical assessment of the European Union’s performance; particularly poor grades are given for the fight against unemployment, the protection of social rights, and ensuring economic growth (ibid: 35). (4) When asked what they would consider to be most helpful for the future, most respondents declare comparable living standards as the key element for the future of Europe. Particular urgency is given to this point in the new member states, where 74 per cent of the respondents hold this opinion, compared to 47 per cent in the old member states (2006b: 37). The proponents of European integration and EU enlargement have so far often settled with rather idealistic celebrations of diffuse general progress rather than demonstrating concrete welfare gains linked to the European project. Thus, EU official documents usually refer rather vaguely to ‘productivity gains’ which the European Social Model supposedly implies, while Sandra Kalniete, the former Latvian Foreign Minister and EU Commissioner, hailed the 2004 enlargement as ‘Europe’s triumph over the twentieth century’ (Verheugen 2005). In similarly diffuse terms the Italian banker Padoa-Schioppa praised the EMU for restoring the ancestral monetary unity which Europe had once enjoyed (Kleinman 2002). Winning the future consent and support of sceptical voters who demand comparable living standards will probably require much more concrete forms of ‘output-legitimation’ (Scharpf 1999), which demonstrate that the EU is indeed a government by the people and for the people and promotes the common welfare of Europeans. This is where the social monitoring of the development of economic and social conditions in Europe will play an important role. If serious attempts at social accounting show that the further development of the EU is not only associated with continuing or increasing economic growth, but also with measurable progress in the quality of
Introduction
9
life of Europeans, then the process of further enlargement and integration is likely to gain more active and widespread support from European citizens. So far the European project has integrated economies and to some extent also political systems, but it has done fairly little to integrate societies (Bach 2006). The Copenhagen criteria required the fulfilment of economic (functioning market economy), political (democracy with rule of law and protection of minorities) and administrative criteria (adoption of the acquis communautaire). They did not imply any requirements in terms of social policy, social structure or living conditions of the population. As a consequence, even comparative knowledge about living conditions and quality of life in the enlarged EU remained in scarce supply. This book is one of the first attempts to fill the gap based on empirical research.
2. Objectives and approach of book As just indicated, the overall purpose of this book is to present a picture of key aspects of quality of life in the EU, with reference especially to the NMS and candidate countries viewed in comparison with the EU-15. It seeks to examine both people’s experience of selected domains of social life in EU states and the institutional contexts in which those experiences arise. It is interested in particular in the degree to which the eastern enlargement has increased heterogeneity and inequality in quality of life in the EU and the implications that may arise for EU integration. Most chapters have a large descriptive component, as they try to present a comparative picture of social conditions and institutional contexts across a large number of states in Europe. There is also an analytical focus on the relationship between daily conditions and institutional contexts on the one hand and implications for integration on the other. The book owes its origin to a comparative research project initiated in 2001 by the European Foundation for the Improvement of Living and Working Conditions with the purpose of comparing living conditions and quality of life in old and new member states of the European Union. A centrepiece of this project was the first European Quality of Life Survey (EQLS), which was carried out in 2003 on behalf of the European Foundation in 28 European countries – the EU-15, the 10 new member states (NMS) and what were then the 3 candidate countries (Bulgaria, Romania, and Turkey – CC-3). The national samples covered people aged 18 and above. The sample sizes in each country were approximately 1,000 persons, apart from the smaller countries (Luxembourg, Cyprus, Malta, Slovenia and Estonia) where they were approximately 600 (see Kohler’s methodological appendix in this volume (ch. 17) for further details and methodological assessment of this data source). Other datasets were also assembled for the project, including Eurobarometer data covering more or less the same countries from 2001 and 2002 (for an overview of results from the latter data, see Alber and Fahey 2004). Over the years since the project was launched, a range of senior sociologists from various European countries has been engaged in the analysis of these data-sets, most of whom are among the authors in this book. From this body of work, six main domains of quality of life issues have been selected as the focus of this book on the basis (a) that they are important issues that figure prominently in current policy debates, (b) that relevant comparative data are available for all, or nearly all, of the EU-27 countries and Turkey, and (c) that high-level experts are available to provide analysis and comment on the domain. The six domains are: (1) fertility,
10
Jens Alber, Tony Fahey and Chiara Saraceno
family and household structures; (2) employment and working conditions; (3) income and living standards; (4) housing; (5) sociability and social cohesion; (6) patterns of Europeanisation. A seventh technical domain has also been included, dealing with the quality of EU-wide data on social conditions. All of the six substantive domains are covered by chapters dealing with objective conditions and their subjective perception, while the first four also have parallel chapters that examine the institutional context and policies. 2.1 Fertility, families and households The chapters in Part I, Fertility, families and households, address from different perspectives the role of family relationships and of patterns of family formation in shaping quality of life. The degree to which people succeed in having the number of children they want is not often spoken of as a quality-of-life issue. Yet it clearly relates to fundamental human aspirations and also, in view of the crisis of fertility that many observers now ascribe to Europe, it has a strong bearing on the future vitality of Europe at the collective level. The relationship between aspirations and outcomes as far as numbers of children are concerned thus warrants the treatment Tony Fahey gives it in his chapter (1). It has often been observed that Europeans now generally have fewer children than they say they want (among the countries examined here, it is only in Turkey that the opposite holds true – women in Turkey on average have more children than they say they want). Fahey’s concern is to explore this gap in detail and assess the common claim that it provides a promising basis for future pronatalist policies – that is, policies designed to remove or counterbalance the obstacles that inhibit people from attaining the family sizes they want. He finds that while under-attainment of family size ideals is the norm on average in EU states, there is a spread around that average such that for some women the gap between ideal and actual is substantial, for others there is no gap at all (women have the number of children they want) and for yet others the gap is in the direction of over- rather underattainment – they have more children than they want. Furthermore, he finds that patterns of over-attainment and under-attainment are linked to patterns of social advantage and disadvantage: over-attainment is concentrated among less educated women while under-attainment is the dominant experience of the well educated. This points to a policy dilemma for those who might wish to introduce pro-natalist supports aimed at raising actual fertility closer to preferred levels. Insofar as they would be likely to work at all, they would be most effective if targeted on better educated women, since it is they who have the widest gap between actual and ideal fertility. But such upwards targeting would run counter to the traditional focus of social and family policy on less well-off groups. Patterns of family formation in Europe are increasingly diverse within each country as well as cross-nationally. After a long period in which they seemed to converge, family forms at the turn of the century have ‘recovered their complexity’ (as Göran Therborn once observed). In her chapter (2), Chiara Saraceno argues that these differences occur at three levels: in the institutional forms of family life, in the shape and the timing of life course transitions, and in everyday living arrangements. Her analysis focuses mainly on the second and third aspects. With regard to lifecourse transitions, given the lack of comparative longitudinal data, she focuses on the household
Introduction
11
arrangements of three broad age groups: the young, the elderly and those at midlife. The age of exiting the parental household shows high variation across Europe, thus producing the most striking differences with respect to the young, though these have an impact on the other two age groups also. For the elderly, the main differences are gender-related, as European men and women differ with respect to their life expectancies and the age at marriage. Some of the cross-country differences in patterns of family formation have their roots in old anthropological and historical divides, while others are new and are currently reshaping the European map ‘of families of families’. In several respects, the Southern European countries stand most clearly apart from other EU nations, as a high prevalence of conventional marriage and legitimate births coincides with very low fertility and late exit out of the parental household. The new member states do not form a distinct pattern. Some joint properties emerge, however, with regard to everyday living arrangements and particularly with respect to the high incidence of dual earner households. Here all former communist countries except Poland resemble the Nordic and Continental countries and stand apart from the Southern European ones. The time structures and gender tensions that underlie these apparently common patterns are different, however, especially because dual earner households in the former communist countries stand out for their very heavy overall workload. Families are not only about households, but also about kinship and intergenerational ties. Chapter 3 by Agnes Blome and Wolfgang Keck explores this issue from a limited, but crucial, perspective: that of relationships of solidarity and obligations between generations in ageing societies. The authors empirically test a widespread notion that informs much of public discourse, namely, that there is an inevitable conflict between generations over the redistribution of resources and responsibilities. Using comparative survey data (Eurobarometer and EQLS), they test this hypothesis with respect to three specific policy issues: care arrangements for the frail elderly, financing responsibilities for care, and the assessment of pension systems. Their findings show the persistent strength of intergenerational solidarity at the kinship level, but also at the societal level. In the field of care there is widespread recognition across all age-groups that adult children have a responsibility to look after their elderly parents when the parents can no longer manage to live on their own. Younger and middle-aged adults also stress the collective responsibility to pay for non-family care out of the public budget, while the elderly more often argue for taking the financial responsibility themselves, in order to unburden the younger generations. Perceptions of public pensions are more stratified by age, as younger age groups profess less trust in pension schemes than older generations. Low levels of trust in pensions are associated with a higher awareness of tensions between the generations, suggesting that the perception of intergenerational conflict is strongly mediated by the degree of trust in pension systems. Comparing old and new member states, the authors find no consistent country differences that would correspond to this distinction. Trust in the pension systems tends to be lower in new member states, but nation-specific differences within the old EU-15 are much bigger than the difference between the two country-group averages. In the field of care, countries with a stronger tradition of public care services have lower proportions of citizens who feel that care for a frail elderly is mainly a family responsibility. With their strong reliance on family provision, the new member states here resemble the Southern countries more than the Nordic ones.
12
Jens Alber, Tony Fahey and Chiara Saraceno
Interacting strongly with patterns of everyday organisation within households and families, public policies and welfare state arrangements also shape perceptions of family responsibilities. Thomas Bahle’s chapter (4) deals specifically with family policies. Usually ignored in mainstream comparative analysis and typologies, family policy probably constitutes the most differentiated and differentiating area of social policies in the EU. It is one where convergence is totally left to self-initiated processes of mutual learning, as the EU level lacks even an open method of coordination in this field. On the other hand, the maternity leave directive, the target concerning childcare coverage for the under three agreed upon at the Barcelona summit, and the various directives and recommendations concerning equal opportunities and work-life reconciliation might be defined as a nucleus of EU family policy. This joint nucleus interacts, however, with national policies that vary widely with respect to aims, means, and levels of generosity. Bahle shows that the stark differences are rooted in different ideas concerning a state’s legitimacy to develop a family policy, which in turn are based on the interaction of two phenomena: national-historical specific patterns of family formation and different church–state relationships. Having developed different ways of dealing with state–church competition, Protestant, Catholic and Orthodox countries formed different ‘families of family policy’. In the former communist countries one must add the impact of communism with its idea of the ‘working family’ and of the respective division of labour between state and family. The complex historical past thus produced four ‘families of family policy’ in Europe which the author identifies by looking at child allowances, childcare services, maternity and parental leaves. He applies three criteria to generate this typology: the degree of universality, the generosity of benefits and services, and the definition of family and state responsibilities. Bahle distinguishes four resulting types which he terms work compatibility, subsidiarity, mixed, and no family policy. Rather than developing a specific model of their own, the post-communist Central and Eastern European countries have joined Western European countries on the basis of their pre-communist religious culture and family history. Most of them joined the Central European ‘subsidiarity model’. This model is currently changing, however, as family policies all over Europe, except in the south, are moving closer to the Scandinavian model. From the perspective of family policies, enlargement thus has not increased European diversity to any sizeable degree. 2.2 Employment and working conditions The chapters in Part II on employment and working conditions deal with a central dimension of quality of life. Constituting the main access to income and social status in contemporary societies, employment crucially affects individual well-being as well as social integration. Having a job structures everyday time organisation, while work conditions affect the everyday living conditions of individuals and families. The question then is to what extent the enlarged European Union reaches the goal of highquality jobs for all. Based on official statistics, the chapter on employment (5) by Jens Alber describes and analyses the development of employment patterns in the enlarged European Union. Even though the new member states tend to have lower levels of employment on average, recent labour market developments cut across the East–West divide, as the groups of good and poor labour market performers consist of old and new member states alike. Whilst group-specific risks of labour market exclusion tend
Introduction
13
to be similarly concentrated on vulnerable groups across Europe, the new member states stand out for two peculiarities: several of them have very high levels of youth unemployment, and low-skilled groups face particularly steep impediments. The structures of employment still show much national diversity. Whilst industrial employment has declined to similarly low levels in Western and Eastern European countries, the new member states resemble the Southern European old members in having higher levels of agricultural employment and lower levels of employment in the service sector, where there is most future growth potential. After having gone through a protracted phase of jobless growth, the Central and Eastern European new member states have translated their above-average economic growth rates into sizeable employment increases, especially in the Baltic nations. The chapter concludes with an analysis of the impact of welfare policies on employment. Even though data are not available for all new member states, this analysis suggests two preliminary conclusions. (1) The new member states did not pursue social dumping policies, but stand out for particularly high burdens of social insurance contributions which are associated with low levels of employment in the low-skill sector. (2) The group of successful labour market performers includes nations with generous and stingy welfare benefits alike, suggesting that there are two paths to full employment in Europe, i.e. the liberal model of the Anglo-Saxon countries and the more state-centred model of the Scandinavian countries where public services function as the engine of employment growth in the service sector. Claire Wallace and Florian Pichler shift attention from the overall structure and functioning of labour markets to the workers’ experience of their jobs (Chapter 6). They assess that experience from a quality-of-life point of view, focusing especially on the contrast between Eastern and Western Europe. The communist era in Eastern Europe was in theory a worker’s utopia, with guaranteed full employment and a range of services, such as health, education, holidays, pensions and extensive childcare, provided in return for work effort. The reality of work and its rewards was less rosy, as captured in the adage from the communist era quoted by Wallace and Pichler – ‘we pretend to work and they pretend to pay us’. The transition to the market economy shattered old work relations and the benefit systems that went with them. Today, the problems experienced by workers in Eastern Europe are many, and while being of a similar kind as in Western Europe, are much more severe in the new member states. Eastern Europeans in general work longer hours (partly because they often have secondary jobs), they are badly paid, their employment is insecure and their jobs are often dull, boring and located in unpleasant working environments. However, there is considerable variation around the average, both between and within countries. The accounts of labour market developments and of the quality of jobs are complemented by Jelle Visser’s Chapter 7 on changes in labour market institutions. Visser’s account centres on the effects of enlargement on regulatory structures and on the relationship between unions and employers. Visser considers a future convergence towards what used to be the mainstream Western European model of labour relations based on sectoral bargaining and upward harmonisation as highly unlikely, because the extension of a common market with the free movement of workers and services has constraining effects on unions and regulations. On both sides of the old EU-15 border, more flexible regulatory approaches allow exit from standards that can no longer be sustained in the face of increased international competition. Under the
14
Jens Alber, Tony Fahey and Chiara Saraceno
motto of ‘flexibilisation’, contractual arrangements at variance with standard employment contracts have proliferated across the enlarged EU. With independent unions and autonomous employer organisations absent, and larger shares of employment located in small firms, the new member states lack decisive preconditions for the implementation of the Western European model of collective bargaining. Bargaining at the enterprise level has become more important almost everywhere. Minimum wages which exist in 18 of the 25 member states and in all former socialist countries counteract the trend towards rising wage inequality only partly, because minimum wage adjustments lagged behind average wage growth in many member states. Whereas several EU-15 countries retained or reintroduced social pacts as instruments of macro-economic coordination, Slovenia is the only new member state where social pacts played a role. However, all CEE countries have founded tripartite councils in an attempt to comply with the European idea of social dialogue. Yet this dialogue remains particularly underdeveloped on the transnational level, as key sectors of the European economy are without transnational employer associations. Visser does not expect framework directives on the European level to have strong standardising effects, because their impact depends on how the legislation integrates with national structures. Within weakly developed national institutions for the collective representation of employees at the firm or national level, the harmonising effect of EU legislation will remain limited. 2.3 Material living conditions The four chapters in Part III on material living conditions focus on two crucial dimensions that affect quality of life: material deprivation and housing conditions. These chapters offer an integrated overview of objective circumstances and of policies in these two fields. Although there is no consensus on what should be included in measuring quality of life, all agree that living standards are central – it is difficult to live a good life if one is beset by poverty, deprivation and economic vulnerability. However, in taking up these topics in Chapter 8, Christopher T. Whelan and Bertrand Maître are faced with the difficulty of deciding who should be considered as suffering from inadequate living standards in the enlarged EU. The problem here is that member states differ so widely in their level of economic development that those who are conventionally defined as poor in some member states countries are better off in absolute terms than those who are considered well-off in others – the classic difficulty arising from cross-national comparisons of relative notions of poverty. Whelan and Maître show that this difficulty is not only a matter of income but is evident also in direct measures of material hardship: in many of the NMS, even middle-income groups experience shortfalls in basic consumption resources that are unknown amongst most of the ‘poor’ in the richer member states. Whelan and Maître’s contribution is to recommend a novel way of coping with this contradiction. They reject the view that a pan-European approach to poverty measurement would help illuminate this issue and present instead a complex, multi-dimensional measure of economic vulnerability that combines relative and absolute indicators using latent class analysis. Their key result, similar to the findings in the chapter by Petra Böhnke, is that the economically vulnerable do indeed make up a larger share of the population in the poorer member states than in the richer, but that they are more excluded – more cut off from the mainstream – in the richer countries.
Introduction
15
In order to be effective, the fight against income poverty, deprivation, and social exclusion requires more than a single policy measure. Employment and local development policies, support for working women, adequate family transfers and old age pensions are among the most effective preventive measures in this field. Most European countries also provide an entitlement to minimum income support in the case of poverty as a fundamental element of their social rights package. On the European level, a 1992 resolution promoted a minimum resources guarantee. Its actual implementation was originally left to the single member states, but later incorporated into the ‘soft’ framework regulation of the Open Method of Coordination. Bea Cantillon, Natascha Van Mechelen and Bernd Schulte in their Chapter 9 examine the effect that the EU resolution and the monitoring process had on the persistent differences in implementation of minimum income policies. Differences persist because of the complexities of national social security systems (old age pensions, invalidity benefits, unemployment benefits, family transfers, tax allowances and so forth), but also because there are implicit and explicit assumptions in national welfare cultures concerning the deserving and undeserving poor, the role of family solidarity, or the risks of welfare dependency. Minimum income support measures, therefore, differ not only with regard to eligibility rules and generosity, but also with respect to administrative styles and the application of discretion rules. Most countries increasingly link income support to some kind of ‘activation’ goal, but the way this goal is implemented and the rights and obligations attached to it add a further element of diversity among the systems. Once again, the largest differences are found not between old and new member states, but across them. The only two countries that do not have a general system of income support for the poor are two ‘old EU’ members: Italy and Greece. In all former communist countries, however, the issue of poverty surfaced dramatically in the transition period in an institutional framework that traditionally did not include any specific poverty policies. They have started slowly to set them in place only recently. The broad-ranging treatment of living standards and income maintenance issues contained in Chapters 8 and 9 is followed in Chapters 10 and 11 with a focus on a specific but highly important element of material living conditions, namely, housing. In Chapter 10, Henryk Domanski ´ describes and assesses the housing conditions of the populations of the enlarged EU. The starting point here is the widespread perception that this is one area where the communist legacy in Eastern Europe has been unfortunate in that it has left people badly housed. Domanski’s ´ account confirms that this perception is largely accurate: housing conditions are generally poor in the eastern countries, as evidenced by a range of indicators from over-crowding to lack of indoor plumbing. Furthermore, these differences are reflected in how people in the worse-off countries feel about their situation – knowing that the majority of one’s compatriots are in the same situation does not really make it much easier to live in cold, cramped and badly built apartments. However, Domanski ´ also highlights some interesting shades of variation in this broad picture. One is that the east–west divide in housing standards is far from clear-cut: some parts of Southern Europe have as many, or more, problems in housing conditions as do the better-off of the former communist countries. Another is the extraordinary situation with regard to housing tenure that suddenly emerged in Eastern Europe with the collapse of communism. The huge stock of state-owned housing that had been a feature of social provision in the communist era was quickly handed over to the ownership of its occupants as the
16
Jens Alber, Tony Fahey and Chiara Saraceno
communist system fell apart, largely because the new state administrations lacked the capacity to manage or maintain the masses of shoddy buildings previously under state control. The consequence is that the NMS have home ownership rates generally exceeding 70 per cent – but in a situation where home ownership has none of the connotations of accumulated wealth and success in life that often attach to it in the older member states. Rather, for most newly minted home owners in the NMS, ownership of their dwellings has simply brought onto themselves the massive task of upgrading decrepit buildings that formerly they might have hoped would be undertaken by the state. In Chapter 11, Michelle Norris takes the analysis of the housing situation in the NMS a step further by examining the institutional context within which housing is provided in these countries, viewed in contrast with the housing systems of the ‘old’ EU. Her main focus is on the institutional legacy left in the wake of the ‘East European housing model’ created in the communist era, particularly as it affects housing inequalities within and between countries. She identifies four key strands to that legacy – housing tenure policy, housing finance and subsidy systems, housing construction systems, and governance. For each strand, she provides a brief sketch of the communist-era background, the present-day situation in the NMS and the contrasts offered by the rest of EU – contrasts which differ broadly between the northern and southern states of the EU-15. The key historical features of the East European housing model can be summarised as: (1) high levels of state ownership and an aspiration for universal social provision of housing that could not really be sustained; (2) inadequate investment in housing as available capital was concentrated on the build-up of heavy industry; (3) large monopolistic state construction companies that were hugely inefficient and unresponsive to consumer needs; (4) poor governance at local level, resulting in inadequate management and maintenance of dwellings and poor policies on such things as rent setting and housing allocations. The rush to the market that occurred after the fall of communism transformed the surface character of some of these features – as in the massive privatisation of housing referred to in Domanski’s ´ chapter and the opening up of housing markets to private finance. However, these surface changes did not solve the underlying difficulties. Privatisation did nothing to rehabilitate dwellings and private finance made scarcely any difference to most housing since mortgage markets, the key channel of private financing for housing in other countries, remained underdeveloped. Not all of these problems were confined to Eastern Europe since some of them, such as the lack of ready access to private finance and poor management structures, were found in other parts of Europe, especially in the Southern European states. 2.4 Social capital and social cohesion The chapters in Part IV on Social capital and social cohesion address the issue of subjective perceptions of one’s own quality of life and of the quality of society including the perception of tensions which may threaten social cohesion. Chapter 12 by Manuela Olagnero, Paola Torrioni and Chiara Saraceno focuses on the quality of social networks, with particular regard to the balance of integration in family and non-family networks, including investments into the civic and public sphere. They find distinct patterns of sociability that identify six country clusters. These overlap only partly with known welfare state typologies such as Esping-Andersen’s
Introduction
17
(Western Europe based) distinction between three types of welfare regimes. Being distributed across several clusters, the former communist countries are as internally differentiated as the old EU-15 member states. The findings furthermore suggest that patterns of public solidarity and social cohesion can explain (or be explained by) patterns of private sociability and feelings of social belonging only partly. High standards of living, high rates of employment, and high levels of technological literacy are drivers of a multi-dimensional kind of sociability, whereas high rates of unemployment and of poverty restrict the range of social networks. There is no indication that a high involvement in family networks crowds out involvement in non-family ones or that involvement in the private sphere limits involvement in public life. There is, however, indication that disadvantageous circumstances which are more frequent in some Southern European countries and most of the new member states restrict the range and kind of social networks to close family and kin. Focusing on individuals’ own perceptions, the chapter by Petra Böhnke (13) analyses how widespread feelings of marginalisation are in the European Union and how differences between nations in this regard can be explained. Given the heterogeneity of living conditions in the enlarged EU, the author’s central concern is to find out if similar disadvantages bring about identical perceptions of marginalisation even if the macro-contexts are different. On average, Böhnke finds the prevalence of feelings of belonging to be lower in the new member states. However, the rank-order of member states does not fully correspond to the distinction between old and new members, as self-assessed marginalisation is found to be least prevalent in Scandinavia, Cyprus and Slovenia and most widespread in the Baltic countries, the Czech Republic and Slovakia, and in Bulgaria and Romania (as well as in Turkey). Across Europe perceived social exclusion is strongly related to unemployment and poverty. The lower the income the more likely it is that people feel at the margins of society, especially if they do not have a job. Yet unemployment or economic strains are not the only determinants of perceived marginalisation. The absence of family back-up and social network support are similarly important drivers. Thus, social back-up helps to considerably soften the consequences of material deprivation. This suggests that quality of life requires a share in economic prosperity as well as access to meaningful social relationships. Beyond individually experienced disadvantages, it also matters in which country or social context people experience social or economic hardship. Adverse conditions are most likely to give rise to feelings of marginalisation in the more prosperous countries where deprivation is least common. Rich countries thus have lower levels of perceived social exclusion, but higher degrees of polarisation in the perception of exclusion between marginalised and privileged groups, because vulnerable groups are likely to feel more stigmatised and marginalised in countries where the overall level of well-being is high. Dealing with the perception of group conflicts in Europe, the chapter by Jan Delhey and Wolfgang Keck (14) focuses on an aspect of social cohesion which is central to the concept, yet frequently overlooked: the degree to which different groups share common feelings of belonging and have cooperative rather than hostile relations. Based on the 2003 EQLS, the authors examine to what extent Europeans perceive strong tensions between various kinds of social groups. Focusing on vertical tensions such as rich vs. poor or management vs. workers on the one side, and on ethnic tensions on the other, they show that old and new member states do not differ much in the overall level of perceived social conflicts and that throughout Europe perceptions of
18
Jens Alber, Tony Fahey and Chiara Saraceno
tensions are weakly stratified, as privileged and disadvantaged groups within single nations show a similar awareness of strains. However, the citizens in old and new member states are preoccupied with different kinds of conflicts. Whereas people in the new member states worry most about distributional conflicts, the citizens in EU-15 nations are most concerned about ethnic tensions. Hence the authors conclude that the Eastern enlargements will shift the conflict agenda within the European Union towards distributional issues which have been toned down in the more affluent old member states. This means that successful cohesion policies in the enlarged European Union will have to be two-pronged. In the post-communist new member states – and to a similar extent also in France and Greece – the priority must be to smooth vertical conflicts about the distribution of material living conditions. In most of the old member states, and also in Hungary and the Czech Republic, the focus must be on mediating ethnic tensions which have become a major citizen concern in these countries. 2.5 Processes of Europeanisation Part V on Processes of Europeanisation deals with the extent to which Europeans are moving beyond the nation state and beginning to think and act in transnational European terms. The possibility of such a shift arises especially in connection with migration. The image of an EU where workers move freely between member states as they seek to advance their careers is dear to the hearts of EU policy makers and is part of the core vision of the EU project. The problem is that EU citizens do not share in this vision, not only because they dislike having migrants move into their areas but also because, as Hubert Krieger points out in his Chapter 15 on migration intentions, they are slow to move themselves. Krieger sets out to examine the latter aspect of Europe’s lack of ‘mobility culture’. He focuses on people’s stated intentions to move across regions or to other countries in the EU as a means of probing mobility culture. This focus on intentions rather than actual migration is adopted partly for the pragmatic reason that data on mobility intentions are available for the whole of the EU (something which is not true for actual migration), but also because they reveal something of people’s perceptions of the desirability of moving, even if obstacles of various kinds may inhibit them from acting on those perceptions. He uses data on mobility intentions for all the present EU-27 states drawn from 2001 and 2005. A key result is that there is an upward shift in people’s intention to move in the future that occurred between 2001 and 2005. Krieger interprets this as evidence of a growing mobility culture in Europe, but he also points out that the shift is so variable across countries that it is hard to view it as constituting a single consistent underlying trend. In looking at the influences on mobility intentions, he finds a range of socio-demographic factors to have reasonably stable effects over time. Subjective quality of life perceptions such as the general life satisfaction are even stronger predictors of cross-border migration intentions than objective macroeconomic indicators. Not surprisingly, the analysis also reveals the strong disincentive effects which language barriers exert on cross-border migration intentions in Europe. Chapter 16 by Jan Delhey and Ulrich Kohler probes into transnational aspects of perceived quality of life by examining how citizens judge the quality of life in their own country relative to the EU as a whole. Departing from sociological reference group theory and based on 2004 Eurobarometer survey data for the old and new
Introduction
19
member states, the authors take up three questions: Do citizens see their own country as offering better or worse life chances than the EU average? To what extent do their beliefs match reality? Which impact do perceptions of the European ranking of their country have on quality of life as measured by individual life satisfaction? The authors show that with respect to the material standard of living, the employment situation, and the overall quality of life, Europeans have a fairly precise idea of where their country stands in the league table of member states. They also demonstrate that perceptions of collective deprivation relative to the EU average impact negatively on personal life satisfaction, even if other determinants are statistically controlled. Their results thus suggest two major conclusions: (1) As also shown in Chapter 13, macro-conditions have an effect on well-being independent of individual life circumstances. (2) Having become a relevant yardstick for ordinary citizens who evaluate the social conditions they live in, the EU increasingly constitutes an integrated social space that provides a common cognitive frame for European citizens. As citizens are no longer only nationally oriented, but develop European aspiration levels, economic growth and improvements of material living conditions in single nations cannot be expected to produce similar improvements in subjective well-being as long as the relative position of the country within the European Union remains constant. Hence if Europeans are to become similarly satisfied across the Union, some convergence mitigating regional disparities and improving the lot in poorer countries will be required. Finally, Chapter 17 by Ulrich Kohler, gives a methodological quality assessment of different European Surveys including the European Quality of Life Survey on which most articles in this book are based. Examining the ‘European Value Study 1999’, the ‘European Social Survey 2002’, the ‘International Social Survey Programme 2002’, the ‘European Quality of Life Survey 2003’, and the ‘Eurobarometer 62.1’, he discusses to what extent certain methodological quality criteria were met and proposes some guidelines for future rounds of comparative European surveys.
3. Conclusions From the wide range of analyses presented in the chapters just outlined, no simple picture of quality of life in the enlarged EU emerges. Yet some brief overall conclusions can be drawn. The first concerns the value of the quality of life perspective itself as a way of evaluating the condition of societies in Europe. There are many instances where the multi-dimensional assessment that characterises the quality of life approach does no more than confirm what we already knew from narrower old-style indicators such as GDP per head. It is not just that conventional measures of economic output are good predictors of those aspects of quality of life that have a largely economic character, such as living standards and the risk of material hardship, or housing conditions, or working conditions. In all these areas, not surprisingly, there is a rough correspondence between how well societies in Europe measure up and how advanced their economies are. However, a similar relationship also seems to hold in connection with some important social aspects of quality of life, such as feelings of marginalisation and to some degree also patterns of social networks. It might be argued, of course, that in European states measures based on GDP are not just about the economy but capture also the whole gamut of civic, social and political characteristics that are particularly favourable in the old, deeply rooted
20
Jens Alber, Tony Fahey and Chiara Saraceno
democracies of north-west Europe and that weaken somewhat as one goes south (Portugal, Spain and Greece, after all, escaped dictatorship only in the 1970s), and weaken further as one enters the former communist new member states. This gradient in the age of democracy across Europe is similar to that in GDP per head. It would be a difficult matter to disentangle how the elements of this correlation fit together and which of them are most important as influences on quality of life. In any event, as a general rule in present-day Europe, it might justifiably be argued on the basis of evidence presented here that strong economies and good quality societies tend to go hand in hand and together provide a strong under-pinning for a high quality of life for individuals. All that being said, however, there are many ways in which the chapters in this book take us beyond a simple identity between the level of economic development and the quality of life, and it is here that the value of the broader multi-dimensional assessment shows its worth. For one thing, there are some central underpinnings of people’s well-being that are only weakly influenced by economic conditions – or at least where variation across Europe does not run in tandem with level of GDP. The most important of these is family life with its different kinds of gender and intergenerational relations. Differences across the member states do not vary along rich–poor lines, nor indeed between east and west. Although there are some indications of a distinctive Southern European pattern, the similarities and differences between countries in this area are little associated with current economic conditions, or indeed with any other contemporary societal features, so that one might speculate on a possible link with older historically rooted patterns of cultural difference. In any case, these different patterns of family formation suggest that there might be different understandings across Europe about what quality of life is about, at least in the domain of private, family relationships. They also suggest that cross-country differences in resources available to pursue one’s idea of a good life – e.g. achieving autonomy when young, leaving a bad marriage, balancing family and work obligations – might differ not only because of GDP levels but because of the overall way society is organised through its welfare state and gender and intergenerational arrangements. As indicated in the chapter on family policies, these arrangements differ in ways that cannot at all be explained by GDP levels. The analysis of the gap between actual and ideal family size also reveals complex differences between countries on similar levels of economic development, though – standing back a little – one might also be struck by how uniform European countries are in this area, with the near-universal pattern of small actual family sizes and larger ideal family sizes, especially in contrast to Turkey. The important cleavage in this respect is not between countries but between educational categories within countries, with the well educated across all countries generally having as much or more in common with each other than with the less educated in their own countries. Complexities along somewhat similar lines arise in connection with patterns of poverty and social marginalistion, an area where, as already mentioned, the richer EU states have less severe problems in absolute terms than the poorer ones. For the former communist countries, the experience of mass poverty has been one of the unforeseen results of the transition process, radically modifying people’s expectations and conditions of living. Yet, in richer countries the superior absolute level of wellbeing goes hand in hand with relative problems that are worse. Those on the lowest rungs of the ladder of advantage and social inclusion in the richer states are located
Introduction
21
on ground that in certain senses is quite elevated, since it places them as high as those well up the ladder in the poorer countries. But they are further away from the mainstream in their own societies and so, relatively speaking, are worse off and more marginalised in their own national contexts than the lower rungs in the poorer countries. Here again, subtleties are revealed that are important to be aware of, even if it is not at all clear how they should be reckoned up in trying to arrive at overall judgements of quality of life. As well as examining people’s direct experience of major domains of quality of life, part of the purpose of this book is to provide an idea of the institutional context within which individual experiences arise. This purpose was adopted with particular reference to the new member states, where key institutions are likely to reflect a mix of legacies from the communist era, of importation from the western democracies and of improvisation arising from the admixture of the two. Furthermore, it is sometimes neither the old nor the new regime that is the crucial influence on people’s well-being but the manner of the transition between the two. With respect to employment, the shock of transition from the communist to the market economy brought about a period of turbulence in the NMS that was of a similar order of severity as the great depression of the 1930s in the west. Since labour market institutions are weak overall in the former communist countries, workers have less means than in most old EU countries to negotiate their position in increasingly flexible labour markets. But also within the EU, labour market conditions and institutional mechanisms remain quite differentiated, and the most successful cases in terms of employment development represent two almost opposite social models (the Anglo-Saxon and the Nordic approach to full employment). Institutional transformation in the NMS was and is to a great deal shaped also by the accession process itself. Originally, under the impact of the Copenhagen criteria, emphasis was placed almost exclusively on establishing a market economy and democratic forms of governance. Later, following the Lisbon summit, there was also a new emphasis on the European social model with developed social policies. Yet, since this European model is to a large degree a kaleidoscope of nationally specific arrangements and welfare regimes, which the EU promotes mostly through means of ‘soft law’ while leaving its concrete implementation to national governments, it is not surprising that the NMS also contribute to this kaleidoscope, rather than forming a homogeneous group of their own. More than in employment policies, this is apparent in the way anti-poverty/social inclusion policies are developed across the old and enlarged EU. The fact that the former communist countries had no policy legacy to build on in this field would have provided the most favourable condition for the development of largely similar policies. Yet, each country pursued different approaches with regard to aspects that are important for a ‘social model’: generosity, duration, activation. These differences partly overlap and partly add to those already existing in the old EU, where two countries (Greece and Italy) even still lack such an income support policy for the poor. None of these findings point to a clear answer to the question posed earlier as to the likely impact of the eastern enlargement on the future cohesion and viability of the EU. Viewed from a quality of life perspective, there are many signs of convergence and of successful re-integration of the former communist states into the west European model of democratic, market-economy states. But there are also many signs of deep, persistent heterogeneity and of gulfs in quality of life between east and west.
22
Jens Alber, Tony Fahey and Chiara Saraceno
And there are persistent differences also within the West both in everyday arrangements (patterns of family formation, gender arrangements, the balances between autonomy and interdependence of generations) and in institutions. Judged on the basis of the evidence presented in the chapters that follow, whether or in what way the EU will be able to hold its now very different parts together and move towards closer union remains an open question. Interestingly, as suggested in Chapter 16, European citizens are beginning to become European in their framework of expectations and of comparative evaluations. This may be an important resource in the construction of Europe. But it may on the contrary fire back on this very construction if the conditions of living and the national patterns of social citizenship remain too far apart. In preparing this volume, the editors had valuable help from several staff members at the Social Science Center Berlin, especially Florian Fliegner, Bettina Mertel, and Marion Obermaier to whom we wish to express our gratitude. Our special thanks are due, however, to Martina Sander-Blanck, also from the Social Science Center, who transcribed all manuscripts into the template form for the publishers with admirable care, diligence, and speed, as well as with never-failing good humour.
Notes 1 The problem, of course, was that this distinction never fitted the situation even with respect to the war in Iraq. The ‘coalition of the willing’ who rallied behind the United States in their ‘letter of eight’ written in 2002 included five old member states (Denmark, Italy, Portugal, Spain and the United Kingdom), as well as three acceding countries (Czech Republic, Hungary, and Poland). Only later did the ten Central and Eastern European countries who were applying to join NATO unite as the ‘Vilnius Ten’ to sign a letter expressing even more explicit support for the United States (Ash 2004). 2 One of the implications of this approach is that, judged by this standard, the United States stands out as far more cohesive than the European Union, because regional differences are less pronounced. 3 The fact that almost one-half of the respondents in Estonia welcomed openness may perhaps be related to the mixed Estonian/Russian ethnic composition of the country.
References Alber, J. and Fahey, T. (2004) Perceptions of Living Conditions in an Enlarged Europe, Luxembourg: Office for Official Publications of the European Communities. Ash, T.G. (2004) Free World, London: Penguin. Bach, M. (2006) ‘The enlargement crisis of the European Union: from political integration to social disintegration?’, pp. 11–28, in M. Bach, C. Lahusen and G. Vobruba (eds), Europe in Motion: Social Dynamics and Political Institutions in an Enlarging Europe, Berlin: edition sigma. Baratta, M. von (ed.) (2003) Der Fischer Weltalmanach 2004, Frankfurt a. M.: Fischer Taschenbuch Verlag. Berié, E. and Kobert, H. (eds) (2005) Der Fischer Weltalmanach 2006, Frankfurt a. M.: Fischer Taschenbuch Verlag. Berié, E. and Kobert, H. (eds) (2006) Der Fischer Weltalmanach 2007, Frankfurt a. M.: Fischer Taschenbuch Verlag.
Introduction
23
Blair, T. (2005) ‘Vision for the UK presidency of the EU’, speech given by Tony Blair to the European Parliament, 23 June. Online. Available http: (accessed February 2006). Blair, T. (2006) ‘Europe is falling behind’, Newsweek, special edition (issues 2006): 26–27. Brown, G. (2005) Global Europe: Full-employment Europe, London: HM Treasury. Commission of the European Communities (2001) European Governance: A White Paper, Brussels, COM (2001) 428 final. Commission of the European Communities (2003) Report from the Commission on European governance, Luxembourg: Office for Official Publications of the European Communities. Commission of the European Communities (2005) The Commission’s Contribution to the Period of Reflection and Beyond: Plan-D for Democracy, Dialogue and Debate. Communication from the Commission to the Council, the European Parliament, the European Economic and Social Committee and the Committee of the Regions, Brussels, Com (2005) 494 final. European Commission (2002) Standard Eurobarometer 56: Full Report, Brussels. European Commission (2006a) Eurobarometer 66: First Results, Brussels. European Commission (2006b) The Future of Europe: Special Eurobarometer 251, Brussels. Freedom House (2006) Nations in Transit 2006: Democratization from Central Europe to Eurasia. Online. Available http: http://www.freedomhouse.hu/nitransit/2006/ corruption2006.pdf (accessed 23.10.2006). Gerhards, J. (2006) ‘Europäische versus nationale Gleichheit. Die Akzeptanz der Freizügigkeitsregel für Arbeitskräfte in den Mitglieds- und Beitrittsländern der Europäischen Union’, pp. 253–278, in M. Heidenreich (ed.), Die Europäisierung sozialer Ungleichheit zwischen nationaler Solidarität, europäischer Koordinierung und globalem Wettbewerb, Frankfurt a. M.: Campus. Gerhards, J. (with M. Hölscher) (2005) Kulturelle Unterschiede in der Europäischen Union. Ein Vergleich zwischen Mitgliedsländern, Beitrittskandidaten und der Türkei, Wiesbaden: Verlag für Sozialwissenschaften. Giddens, A. (2007) Europe in a Global Age, Cambridge: Polity Press. Heidenreich, M. (2006) ‘The decision-making capacity of the European Union after the fifth enlargement’, pp. 29–57, in M. Bach, C. Lahusen and G. Vobruba (eds), Europe in Motion. Social Dynamics and Political Institutions in an Enlarging Europe, Berlin: edition sigma. Kleinman, M. (2002) A European Welfare State? European Union Social Policy in Context, Houndmills: Palgrave Macmillan. Moravcsik, A. (2004) ‘The myth of a European “leadership crisis”’. Online. Available http: (accessed February 2006). Official Journal of the European Union (2004) ‘Treaty establishing a Constitution for Europe’, C310, Vol. 47, 16 December. Online. Available http: (accessed February 2006). Rose, R. (2004) ‘Voter turnout in the European Union member countries’, pp. 17–24, in International Institute for Democracy and Electoral Assistance (ed.), Voter Turnout in Western Europe, Stockholm: Publications Office, International IDEA. Scharpf, F. (1999) Regieren in Europa. Effektiv und demokratisch?, Frankfurt a. M., New York: Campus. Schmitt, H. and Scholz, E. (2005) The Mannheim Eurobarometer Trend File 1970–2002, Data Set Edition 2.00 Codebook on Unweighted Frequency Distributions, Mannheim, Cologne: MZES, ZUMA and ZA. Online. Available HTTP: (accessed February 2006). Stiglitz, J.E. (1999a) ‘Whither reform? Ten years of the transition’, in B. Plesovic and J.E. Stiglitz (eds), World Bank Annual Conference on Development Economics, Washington, DC: World Bank.
24
Jens Alber, Tony Fahey and Chiara Saraceno
Stiglitz, J.E. (1999b) ‘The World Bank at the millennium’, Economic Journal, 109, 459: F577–97. Therborn, G. (2000) Die Gesellschaften Europas 1945–2000. Ein soziologischer Vergleich, Frankfurt a. M.: Campus. Transparency International (2005) Corruption Perception Index (CPI). Online. Available http: (accessed 25.10.2006). Verheugen, G. (2005) Europa in der Krise. Für eine Neubegründung der europäischen Idee, Köln: Kiepenheuer & Witsch. Wessels, B. (2006) ‘“Founding elections” der erweiterten EU und europäische Integration’, pp. 231–251, in J. Alber and W. Merkel (eds), Europas Osterweiterung: das Ende der Vertiefung?, WZB-Jahrbuch 2005, Berlin: edition sigma.
Part I
Fertility, families and households
1
Fertility patterns and aspirations in Europe Tony Fahey
Introduction The European Commission’s Green Paper on demographic change expressed the view that the EU no longer has a ‘demographic motor’: the Union now has nearly as many deaths as births per year and without inward migration would soon be headed towards population decline (European Commission 2005). The implications for the future of Europe are hard to predict but are unlikely to be good. It is difficult to see how the EU can become the most dynamic and competitive economy in the world, as is the aspiration of the Lisbon agenda, while its population is greying and its workforce shrinking. The key problem is the very low birth rates now found in Europe. At present, the total fertility rate (TFR) in the EU-25 is at about three-quarters of the level needed to replace the population (the TFR is estimated at 1.52 for 2005, while the replacement TFR is conventionally defined as 2.1 – New Cronos 2005). This is one area where the divide between new and old member states is not that significant a part of the EU picture. The ten member states that joined in 2004 have an even weaker reproductive performance than the EU-25 average, with a TFR in that year of 1.27 (Bulgaria and Romania are at a similar level). But a number of EU-15 states also have an equally low TFR so that in this area diversity within the ‘old’ member states is as great as any gap between the old and the new. While many aspects of low fertility in Europe have been examined by researchers (for recent overviews, see Billari 2005; d’Addio and Mira d’Ercole 2005; United Nations 2003), the feature focused on in this chapter is the gap that has emerged between actual and preferred fertility: the number of children people have is, on average, less than the number they would like to have (Bernardi 2005; d’Addio and Mira d’Ercole 2005: 41–44; Goldstein et al. 2003; Bongaarts 2002: 426–427; van Peer 2002; van de Kaa 1998). Chesnais (1998), for example, points out that while women in Europe on average say they want an average family size of 2.2, the actual total fertility rate is only 1.45. This shortfall, which emerged historically in the course of the transition to low fertility, is the reverse of what is found in high fertility countries, where women typically have more births than they say they want (United Nations Populations Division 1995: 59–67; Bongaarts 1998: 8–11). The gap between actual and preferred fertility in low fertility countries is not usually included among the quality of life issues studied by researchers who work within the quality of life approach. However, it is known that the family context is one of the strongest social influences on the quality of people’s lives (see Chapter 13: Böhnke, this volume), and having or not having children relates to quite profound aspects of
28
Tony Fahey
family life. The actual–preferred fertility gap could thus be thought of as having fundamental significance for quality of life. It has also been read by some as having considerable policy significance in the light of the increasing interest in revitalising the EU’s demographic performance. It suggests that governments might be successful in getting Europe’s demographic motor moving again if they adopted appropriate pronatalist policies. For Chesnais (1998), for example, the actual–preferred fertility gap reflects a ‘latent demand for family support’; while for Sleebos (2003: 30) it ‘provides a window of opportunity for policies aimed to increase fertility and to bring it into line with individual preferences’. The European Commission’s Green Paper asserts that ‘if appropriate mechanisms existed to allow couples to have the number of children they want, the fertility could rise overall’ (European Commission 2005: 5). Others may doubt that great weight should be attached to people’s stated preferences for children. Conventional economics assumes that we all want more than we have of all good things, and the compromises we settle for through our behaviour are a better guide to the mix we really want than are our stated preferences. Nevertheless, stated preferences are worth taking into account in the present instance since they indicate that most people still regard children as a good they want more of. This is quite a significant fact, since it indicates that people feel some unease or dissatisfaction with how few children they have had. This is good news for governments concerned about low fertility, since, at the very least, it suggests that family policies aimed at supporting birth rates are likely to be swimming with rather than against a tide of popular preferences. It is another matter whether such policies are likely to be effective, since research has shown that state supports for child-rearing at best have only modest effects on fertility outcomes and are less important than macroeconomic influences, of which buoyant demand for female labour seems to be the most important (Sleebos 2003; d’Addio and Mira d’Ercole 2005). Nevertheless, the actualpreferred fertility gap is worth taking some note of, since it is a part of the context within which the problem of Europe’s demographic weakness might be addressed. In spite of the interest in the gap between preferred and actual fertility, it has been little analysed from the point of view of its significance for present or future pronatalist policy in low fertility countries. Demographers have regularly included measures of desired fertility in fertility surveys since the 1950s but they have been preoccupied with their value for predicting the future fertility behaviour of women still in their childbearing years rather by the final gap between preferred and actual number of children among those with completed families (from a large literature, see e.g. Freedman et al. 1980, Thomson and Brandreth 1995, Thomson 1997, Schoen et al. 1999; for a rare example of a focus on the gap between preferred and actual family size as an object of interest in its own right, see van Peer 2000). It is this final gap between outcomes and preferences, rather than the predictive value of the preferences, that becomes a matter of central interest when, as here, the concern is with the specific issue of very low fertility rather than with broader explanations of reproductive behaviour. This chapter first provides some contextual information on fertility trends and patterns in Europe, focusing especially on the question of whether there are broad regional distinctions to be found in these patterns with regard either to west–east distinctions or to differences between ‘families of nations’. The chapter then turns to the gap between actual and preferred family size in Europe. It examines evidence of the extent of the gap and how it has changed over recent decades, and assesses the
Fertility patterns and aspirations
29
relevance of that gap from a family policy point of view, with particular reference to its significance for potential pronatalist policy.
1. Context: general patterns While below-replacement fertility is now the norm in the developed world, and is increasingly common in poorer countries, there is considerable diversity in how far fertility has fallen below replacement levels. In Europe, Billari (2005) classifies countries into those with ‘lowest low’ fertility (a total fertility rate below 1.3), ‘very low’ fertility (below 1.5), and ‘low’ fertility (below 2). This categorisation roughly coincides with a spatial gradient from north-west to south-east Europe (Council of Europe 2005). ‘Low’ fertility is found in a band of countries running roughly along the north-west edge of the continent – the Scandinavian countries, Britain, Ireland and France – while ‘lowest low’ is found along the south and east of Europe – the Mediterranean countries and some of the former communist countries of Central and Eastern Europe. The countries lying geographically between these two extremes are in the intermediate ‘very low’ fertility band, with Turkey in the south east forming an exceptional case of relatively high fertility. The family formation patterns from which these fertility patterns emerge also differ across Europe and indicate that there are different ways of arriving at similar low fertility outcomes. For example, delayed childbearing is often cited as a contributor to very low fertility (Billari 2005) but in eastern Europe countries, where fertility is at the bottom of the range, women give birth at a younger age than they do in northern European countries such as Sweden, Finland, Ireland, Britain and France where fertility is higher (United Nations 2003: 51). A rise in the incidence of childlessness is an important contributor to the emergence of low fertility in some countries (such as Germany, where over 25 per cent of the 1960 birth cohort remained childless compared to less than 10 per cent of birth cohorts of the late 1930s) but not in others (such as France, where the birth cohorts around 1960 had less childlessness than those of the 1930s) (United Nations 2003: 66). One reasonably pronounced feature of the pattern of lowest low fertility found in the former communist new member states is that it is relatively recent (Billari 2005: 59). In 1990, the total fertility rate in countries like Poland (2.20), the Czech Republic (1.90) and Hungary (1.87) was more or less similar to that of the UK (1.83) and France (1.78) but by 2003, the Czech Republic had the lowest fertility rate of the present EU countries (1.18), while Poland (1.22) and Hungary (1.28) were not much higher. Those countries thus made the transition from low to ‘lowest low’ fertility quickly and recently, reflecting an impact on family formation associated with the transition from communism. Italy and Spain, by contrast, countries with similarly low fertility today, were already close to their current position by 1990 (with total fertility rates in that year of 1.33 and 1.36 respectively).
2. Data The key indicators focused on in this chapter consist of responses to the following questions in surveys of the adult population in Europe (these questions produce the variables as labelled here in brackets): •
What is the ideal number of children for a family? (general ideal number of children)
30 • •
Tony Fahey What is the ideal number of children for you personally (personal ideal number of children) Have you had any children? If yes, how many? (actual number of children)
These three questions were asked in Eurobarometer 56.2 carried out in EU member states (EU-15) in 2001 and the Candidate Country Eurobarometer carried out in 2002 in what were then the 10 accession states (now the new member states – NMS) and the three candidate countries, Bulgaria, Romania and Turkey (CC-3). The common variables from these two surveys were compiled into the single 28-country dataset used here by the Social Science Research Centre, Berlin (WZB), as part of a project on living conditions in Europe carried out for the European Foundation for the Improvement of Living and Working Conditions in 2003 (Alber and Fahey 2004). This combined dataset is the principal data source used in this chapter. The main focus is on the second and third of the three variables listed above (personal ideal number of children and actual number of children), since they allow us to examine the gap between people’s personal family size preferences and their actual family size for 28 European countries. The chapter also draws on two further data sources – the European Values Study (EVS) surveys carried out in 1981 and 1990. Eight countries are common to both of these surveys and to Eurobarometer 56.2 mentioned earlier. For these countries, the first variable listed above (general ideal number of children) is measured at each of the three time points. This variable is of limited interest for our purposes because it taps into general social norms about family size rather than personal preferences. However, it has value because it provides a rare trend measure of family size ideals for the two-decade period between 1981 and 2001–2002. It is examined briefly below on that account. 2.1 Classification by age In examining the gap between ideal and actual fertility across adults of all ages in cross-sectional data, it is necessary to distinguish between what might be termed the ‘final’ gap between ideal and actual fertility among those whose childbearing is completed and the interim gap that arises among younger adults who may yet have more children. The final gap is more important than the interim gap for long-term demographic outcomes and therefore will occupy most of our attention here. For this reason also, the present analysis concentrates on women, since in their case it is possible to identify an age at which, in biological terms, childbearing can be said to be complete. Men’s fertility is not age limited in the same way and so it is more difficult to speak of completed fertility in their case. In much of the analysis below, we concentrate on the youngest possible segment of women who could be said to have completed fertility, while at the same maintaining reasonable sample size at the country level in the data at our disposal. We therefore frequently focus on women with completed fertility, defined as those who are aged 40–64 and who, if they are aged under 50, have said they plan to have no more children. Even though the age-range of this group is narrowed down as far as sample size considerations will allow, the childbearing time span of the women involved is quite wide. In the case of the 2001 data, for example, the oldest women in the age group 40–64 would have entered their childbearing years in the late 1950s, while the
Fertility patterns and aspirations
31
youngest would be arriving at the end of their childbearing years at around the time of the survey. The childbearing time span represented by these women, therefore, amounts to most of the second half of the twentieth century. This indicates the difficulty of linking cross-sectional data of this kind to short-term temporal patterns of fertility, and so drawing conclusions about time trends that might be of interest to policy.
3. The role of education The effect of educational level on the ideal-actual fertility gap is viewed here as a useful means of elucidating what that gap signifies. The measure of education level used is respondents’ age when they left full-time education. While this is a crude measure, it provides a serviceable basis for classifying education in a comparable way across countries. For the present analysis, we use a three-fold classification of the age at which people left full-time education: under age 16 years, between 16 and 19 years, and 20 years or over. The focus on education is adopted here partly for methodological reasons, reflecting the technical character of education as a proxy measure of socio-economic status that is stable over the adult life course and independent of fertility. In cross-sectional data such as are used here, education can be used as a measure of socio-economic background not only at the point of observation but also prior to or during a person’s childbearing years: for the most part, people complete their education before they begin childbearing, and their educational level does not change as their family formation proceeds. There may be exceptions to this rule (e.g. young women who drop out of education to have children) but in the fertility regimes found in Europe in the latter part of the twentieth century, these exceptions are unlikely to be common enough to render invalid the assumption that education is independent of fertility. Other possible measures of socio-economic background, such as current employment status, occupation and income, are not independent of fertility in the same way since they are subject to change through the family cycle and in particular may be affected by the number or timing of children that a person might have. In addition to the methodological significance of education, it is also important because, as Cleland (2003: 187) states, ‘education of adults consistently emerges as the single most powerful predictor of their demographic behaviour’. In less developed societies, women’s education first causes a short-term rise in fertility, because of increased fecundity, reduced risk of foetal death, and the decline of traditional practices such as prolonged breastfeeding and postpartum abstinence (United Nations 1995: 23). As societies begin to develop, education causes fertility to fall both because of individual-level effects on resources and incentives and community-level effects on cultural norms regarding family size (Caldwell 1980; Castro Martin 1995). As societies make the transition to very low fertility, fertility differentials by education get narrower at the individual level (United Nations 1995: 23; Cleland 2003), while at the aggregate level the relationship changes direction – societies with higher levels of female education come to have marginally higher fertility rates than those with lower levels of education (Sleebos 2003: 19-20). The latter pattern echoes the recent reversal of the female employment–fertility relationship found in industrialised societies. Where low female employment was associated with relatively high fertility in industrialised societies in the 1960s and 1970s, the relationship reversed during the
32
Tony Fahey
1980s and low female employment came to be associated with very low fertility. Today, the industrialised countries with the lowest fertility now generally also have low rates of female labour force participation (Brewster and Rindfuss 2000; Sleebos 2003: 20; Ahn and Mira 2002). There is a paradox here because at the individual level, women with higher education and with stronger attachment to the labour force continue to have smaller families (though this effect is not strong in advanced industrial societies – Cleland 2003). In other words, when we look at the situation within countries, we find that women with weaker labour market and educational profiles have somewhat larger families, but when we compare developed countries with each other we find that those that enhance the overall labour market and educational profiles of the female population have the higher birth rates (Castles 2003). Our concern here is with the effect of education on the ideal–actual fertility gap. This effect can be broken down into two components: the effect on the ideals themselves (whether the better educated have larger or smaller ideal family sizes than the less educated) and on the degree to which those preferences are fulfilled (whether the better educated are more or less likely to attain their ideal family size). Lower family sizes among better educated women could be the result of either of these two components: better educated women might prefer smaller families or they might be less likely to attain whatever family size they might prefer. An effect that operates through a preference for smaller families would suggest that the cultural influence of education on fertility ideals and aspirations is the important causal mechanism, while an effect that operates through reduced attainment would point to opportunity costs and the problems of reconciling work with motherhood as the key influences. In the latter instance, better educated women might have more resources in an absolute sense, but the scope for careers, income and other uses of time open to them could be experienced as an obstacle to motherhood and a constraint not faced to the same degree by less educated women. The paradox here is that the range of options open to better educated women might confront them with choices that less educated women do not face. As a result they might experience these options as sources of strain and tension that might make motherhood seem less rather than more possible.
4. The ideal–actual gap We first consider the gap between personal ideal family size and actual family size among women in 28 European countries in 2001–2002. As well as looking at women with completed fertility as defined above (i.e. those aged 40–64), we also consider in this section women still in their childbearing years (aged 18–39) and older women (those aged 65 and over) as an indirect means of assessing trends over time (see the appendix (Table A1.1) for details of sample numbers in each age category). In the case of the two older age groups of women (those aged 40–64 and 65 and over), childbearing can be assumed to be complete and family size is measured by women’s own reports of their actual number of children. In the case of the women aged 18–39, among whom further childbearing is likely to occur, the relevant comparable variable is eventual completed family size rather than current family size. As this variable is not directly measurable, we use the total fertility rate (TFR) in 2000 as a proxy (the TFR is the number of births a woman would have during her reproductive life if she
Fertility patterns and aspirations
33
had children at the same rate as the various childbearing age groups in a particular year). The TFR is unlikely to match exactly the eventual completed family size of the age groups on which it is calculated but it provides the best available approximation to a final outcome measure and is used here for that reason. Table 1.1 sets out the data. The countries are ordered according to the ‘families of nations’ categorisation described in the introduction to this volume and are also clustered into the EU-15, the new member states (NMS) and the candidate countries (CC-3). 4.1 Ideal number of children We can look first at the differences in ideal number of children across the three age groups of women. In all countries except France and Turkey, ideal number of children declines consistently with age, with the differential on average between the oldest and the youngest of the three age groups being less than half a child. This is an indication of a definite, though not large, decrease in ideal family size over time. In France, one of the exceptional cases, there is no real differential. The mean ideal family size for women aged 65 and over in France is 2.52. This falls to 2.48 for women aged 40–64 but rises to 2.54 for women aged 18-39. In Turkey, the other exception (which, as we shall see further below, differs from the other countries in more ways than just this), the youngest age group has a larger ideal number of children than the oldest (2.24 versus 2.07) but taking the possibility of sampling error into account, the difference is not very great. It is worth noting that among the youngest age-group, ideal number of children exceeds 2.1 (the conventional definition of the replacement fertility rate) in 20 of the 28 countries, even though the actual fertility rate as measured by the TFR in 2000 is below 2.1 in all countries bar Turkey. Although sub-replacement fertility outcomes are now found in all of the countries bar Turkey, sub-replacement fertility ideals among younger women have not become the norm, contrary to what Goldstein et al. (2003) suggest may be happening. In many of the countries with sub-replacement ideals, those ideals are only barely below the 2.1 threshold, though in Germany and Austria, where ideal number of children among 18–39-year-old women is 1.75 and 1.7 respectively, they are quite far below. Romania, at 1.91, is also quite low. (The low figure for Malta – 1.91 – has to be viewed with caution as the sample N on which it is based is small – see Table A1.1 in the Appendix.) Further breakdowns of these data presented in Fahey and Spéder (2004: 30) show that younger women in Germany and Austria have larger-than-normal minorities who choose ‘none’ as their personal ideal number of children (17 per cent and 13 per cent of women aged 1839 in Germany and Austria respectively, compared to a weighted average of 5.5 per cent for the EU-15). A personal preference for having no children is thus held by a distinctively large minority of young women in Germany and Austria. The main contrast with Germany and Austria is provided by the Scandinavian countries, Ireland, the UK and France, in all of which ideal number of children among women aged 1839 exceeds 2.4. In this instance, there is as much contrast within the EU-15 as there is between the EU-15 and the NMS, thus reinforcing the point made earlier that contrasts between east and west are not a dominant feature of the fertility-related landscape of the current EU.
Scandinavian Denmark Finland Sweden Western Ireland UK Continental Austria Belgium France Germany Luxembourg Netherlands Mediterranean Greece Italy Portugal Spain
2.43 2.42 2.44 2.43 2.43 2.61 2.42 2.09 1.7 2.2 2.54 1.75 2.16 2.1 2.16 2.28 2.14 2.1 2.16
1.65 1.76 1.73 1.54 1.65 1.89 1.64 1.57 1.32 1.65 1.89 1.34 1.78 1.72 1.27 1.3 1.25 1.54 1.22
0.78 0.66 0.71 0.89 0.77 0.72 0.78 0.5 0.38 0.55 0.65 0.41 0.38 0.38 0.89 0.98 0.89 0.56 0.94
2.51 2.45 2.53 2.52 2.49 3.02 2.42 2.23 2.19 2.29 2.48 2.04 2.28 2.34 2.42 2.84 2.33 2.52 2.4
2.01 1.94 2.14 2.03 2.37 3.07 2.33 1.99 1.91 1.91 2.31 1.85 1.91 2.1 2.04 2 1.86 2.61 2.25
Actual number of children (mean)
Personal ideal number of children (mean)
Ideal– TFR
Personal ideal number of children (mean)
Total fertility rate, 2000
Women aged 40–64
Women aged 18–39
0.5 0.51 0.39 0.49 0.12 −0.05 0.09 0.24 0.28 0.38 0.17 0.19 0.37 0.24 0.38 0.84 0.47 −0.09 0.15
Ideal– actual
2.63 2.58 2.95 2.51 2.88 3.84 2.83 2.37 2.41 2.4 2.52 2.2 2.43 2.74 2.74 3.04 2.75 2.66 2.65
Personal ideal number of children (mean)
Women aged 65+
Table 1.1 The ideal–actual fertility gap in three broad age groups of women in European countries, 2001–2002
2.36 2.26 2.72 2.22 2.73 3.67 2.67 2.53 2.59 2.77 3.05 2.11 1.86 2.83 2.6 1.87 2.54 2.86 2.83
Actual number of children (mean)
0.27 0.32 0.23 0.29 0.15 0.17 0.16 −0.16 −0.18 −0.37 −0.53 0.09 0.57 −0.09 0.14 1.17 0.21 −0.2 −0.18
Ideal– actual
2.16 2.65 1.91 2.05 2.11 2 2.07 2.16 2.1 2.09 2.18 2.06 2.08 2.17 2.09 1.91 2.24 2.11 2.08 2.17
0.72 0.82 0.01 0.8 0.82 0.86 0.74 0.82 0.9 0.78 0.79 0.82 0.75 −0.02 0.84 0.61 −0.26 0.62 0.78 0.56
0.496* (excl. Turkey: 0.588**)
1.45 1.83 1.9 1.25 1.29 1.14 1.33 1.34 1.2 1.31 1.39 1.24 1.33 2.08 1.25 1.3 2.5 1.49 1.30 1.61
2.48 3.2 1.98 2.31 2.28 2.14 2.21 2.36 2.32 2.26 2.21 2.23 2.27 2.24 2.1 2.07 2.38 2.33 2.29 2.32
0.54 0.59 0.18 0.55 0.19 0.18 0.22 0.13 0.29 0.37 0.37 0.41 0.32 −0.29 0.18 0.18 −0.75 0.26 0.22 0.19
0.633* (excl. Turkey: 0.762**)
1.94 2.61 1.8 1.76 2.09 1.96 1.99 2.23 2.03 1.89 1.84 1.82 1.95 2.53 1.92 1.89 3.13 2.07 2.07 2.13
Note: For sample Ns, see Table A1.1 in the Appendix.
Sources: Eurobarometer 56.2 (2001); CC Eurobarometer 2002 data files; total fertility rate 2000: Eurostat
Correlation of ideal and actual
West NMS Cyprus Malta Slovenia Visegrad Czech Rep. Hungary Poland Slovakia Baltic Estonia Latvia Lithuania CC-3 Bulgaria Romania Turkey EU-15 NMS All countries
2.81 4.1 2.9 2.47 2.56 2.48 2.29 2.75 2.4 2.37 2.12 2.44 2.46 2.3 2.1 2.52 2.07 2.58 2.55 2.57
0.52 0.49 −0.12 0.6 0.08 0.28 0.26 −0.02 −0.28 0.45 0.43 0.65 0.34 −0.38 0.1 0.35 −2.01 0 0.12 0.22 0.492* (excl. Turkey: 0.734**)
2.29 3.61 3.02 1.87 2.48 2.2 2.03 2.77 2.68 1.92 1.69 1.79 2.12 2.68 2 2.17 4.08 2.58 2.43 2.55
36
Tony Fahey
4.2 The ideal–actual gap We now turn to the gap between ideal and actual fertility, looking again at the three age groups in Table 1.1. Among women aged 65 and over, actual number of children exceeded ideal number of children in 10 of the 28 countries, suggesting a widespread lack of adequate control over their own fertility by women in these countries. This excess was by far the widest in Turkey, where older women had quite large families on average (4.08 children) but had low ideal family sizes (2.07 children). It is striking to note that there was no similar excess of actual over ideal number of children in either Ireland or Cyprus, the two countries that were closest to Turkey in levels of actual family size among older women. In both these countries, women’s expressed preference was for even larger families than the relatively large families they actually had. Thus, where older women in Turkey experienced ‘traditional’ fertility outcomes while holding ‘modern’ family size ideals, their counterparts in Ireland and Cyprus were traditional in both outcome and ideal. Apart from these countries, however, the most common experience even among older women was that actual number of children fell short of the ideal. In most cases, the shortfall was quite small, but the very fact that it occurred so widely indicates that the present generation of women of childbearing age is not the first to fail to attain its ideal number of children. Moving on to women aged 40–64, we find that ideal number of children was almost universally lower than among those aged over 65 and the excess of actual over ideal number of children had disappeared in all but three countries (Ireland, Portugal and Turkey). This suggests that effective contraceptive practice had become more widely established among women of this generation in a larger number of countries. We will examine the ideal–actual gap for this group of women in more detail further below. The gap between ideal and actual fertility among the youngest group of women – those aged 18–39 – strictly speaking cannot be compared that of the two older age groups since, as mentioned earlier, their childbearing is not yet complete and we therefore use a different measures of actual fertility in their case (the total fertility rate in 2000). The gap between ideal family size among 18–39-year-old women and the TFR in 2000 is wider than the ideal-actual gap among 40–64-year-olds in all countries bar Malta and Turkey. Although we cannot be sure whether this indicates a real increase in the extent of unfulfilled fertility aspirations among younger women or is an artefact of differences in measurement, the most plausible interpretation is that a real widening of the gap between ideal and actuality has occurred.
5. Composition of the ideal–actual gap The overall gap between ideal and actual fertility for any category of the population has to be interpreted carefully since it is a composite arrived at by averaging quite different outcomes. These outcomes can be classified as under-attainment (having ‘too few’ children relative to one’s ideals), over-attainment (having ‘too many’ children) and ‘just right’ outcomes (where the ideal and actual number of children match). Even in a country where there was no gap on average between ideal and actual numbers of children, substantial proportions of women could have family sizes that either exceeded or fell short of their ideals.
Fertility patterns and aspirations
37
Figure 1.1 summarises the distributions across these three categories for women with completed fertility in the age group 40–64. In the majority of countries, the ‘just right’ outcome was achieved by between half and two-thirds of women aged 40–64, which means that the women who had other outcomes amounted to significant minorities. Turkey is an exception, in that the ‘just right’ outcome was achieved only by a minority (33 per cent) of Turkish women in this age group. In most cases, as we would expect from findings already presented, under-attainment of family size was the most common other outcome: in the EU-15 and NMS, onethird of women on average had ‘too few’ children relative to their ideal. Greece had particularly high levels of such under-attainment (56 per cent), though the sample on which this estimate for Greece is based is somewhat small (see Table A1.1 in the Appendix). Turkey had particularly low levels of under-attainment (15 per cent). In some countries with especially low fertility (such as Spain), the proportion of women whose number of children fell below their ideal was not especially large. Alongside such under-attainment, there were smaller but significant levels of overattainment (‘too many’ children relative to the ideal). In the EU-15 and NMS, 10 to 11 per cent of women aged 40–64 had excess fertility in this sense. Again, Turkey was an exception: here excess fertility was the most common situation and was present among half of women. The level of excess fertility found in Turkey was three to four times greater than that found in most other countries. Although excess fertility was a dominant feature only in Turkey, its widespread if less prominent presence in other countries is notable. It indicates that even in countries where on average actual family size falls short of the ideal, there can be significant minorities whose family size exceeds the ideal.
6. Influence of education We now turn to differences in the ideal–actual fertility gap by educational level as a means to shed further light on what that gap signifies. Table 1.2 examines the relationship between ideal and actual family size for women with completed fertility, classified by educational level. If we look first at ideal family size, we find that differences by educational level are absent or slight in both the EU-15 and the NMS and become substantial only in the CC-3. In the EU-15, for example, the lowest and highest educational categories have the same ideal number of children (2.46), while in the NMS a difference between the two is present but slight (2.39 among the less educated compared to 2.20 among the best educated). Substantial differences in ideal number of children by educational level are evident only in the CC-3 (2.33 among the less educated compared to 1.93 among the best educated). Separate analysis not shown in the table indicates that the distinctiveness of the CC-3 in this regard is more or less wholly driven by Turkey, where the gap in ideal family size between the least and best educated is wide (2.41 compared to 1.61). Apart from the Turkish case, therefore, Table 1.2 suggests that the effect of education on ideal family size is weak. This is a finding of some interest, as it suggests that whatever influence education might have on women’s fertility behaviour, it does not operate primarily through an effect on family size preferences. We get an indication of where education does make a difference when we look at the column for actual family size in Table 1.2. Here it is clear that differences by educational level are substantial and are present in each of the three regions: the higher
38
Tony Fahey Greece
56
Cyprus
43
Sweden
43
Italy
39
Denmark
38
Slovenia
41
3
43
14 53
5
51
10
56
38
5
57
5
Latvia
36
57
7
Estonia
36
57
7
Belgium
35
France
35
53
12
49
16
Luxembourg
33
50
17
Slovakia
33
51
16
31
Hungary
58
11
Poland
29
57
14
Finland
28
60
11
Lithuania
28
Czech Rep.
28
Romania
28
Portugal
26
Netherlands
25
Malta
25
Spain
25
Austria
25
Great Britain
24
Ireland
22
Bulgaria
22
8 15
61
11
56
18
63
12
50
25 62
13
66
10
60
24
Germany
Turkey
64 57
16
64
12
56
22 67
16
11
33
51
EU-15
32
56
12
NMS
32
57
11
CC-3
22 Too few
52 Just right
26 Too many
Figure 1.1 Relationship between actual number of children and personal ideal number of children among women with completed fertility (ages 40–64) Sources: Eurobarometer 56.1 (2001), CC Eurobarometer 2002 Note: Countries ordered in descending importance of ‘Actual children less than ideal’.
Fertility patterns and aspirations
39
the educational level, the smaller the actual number of children. In the EU-15, for example, women with completed families who finished their education at 15 years or younger had 2.34 children on average, while women who finished their education at 20 years or older had 1.81 children on average, a differential of 0.53 children. The difference in family size by education level is somewhat wider in the NMS, where the less educated have 2.5 children and the better educated 1.68 children (a differential of 0.82). It is wider still in the CC-3 (at 1.34 children). This lower level of childbearing among better educated women, when taken in conjunction with the small or absent differences in family size ideals by educational level noted earlier, means that the better educated have a larger gap between actual and ideal family size than do the less educated. In the EU-15, for example, the shortfall of the actual from the ideal number of children is five times greater among the best educated compared to the least educated (0.65 compared to 0.12). This ratio is somewhat greater in the NMS and greater again in the CC-3. As before, we need to be conscious that the gaps between ideal and actual fertility reported in Table 2 are composites of under-attainment (‘too few’), over-attainment (‘too many’) and ‘just right’ outcomes. Table 1.3 shows the distribution of these outcomes by age at which respondents left school in the EU-15, the NMS and the CC-3. There is a striking similarity between the better educated in all three regions in the degree to which they are likely to have ‘too few’ rather than ‘too many’ children: 41–2 per cent of women in this educational category have ‘too few’ children, while only 5–8 per cent (depending on the region) have ‘too many’. The pattern is quite different, and is less uniform across regions, for those in the lowest educational category. In the EU-15, the less educated are less likely to have ‘too few’ children than the better educated (26 per cent versus 41 per cent) and they are more likely to have ‘too many’ (16 per cent versus 8 per cent). Nevertheless, in this region the ‘too few’ outcome is more common than the ‘too many’ outcome even for the less educated. In the NMS, by contrast, the balance among the less educated women between those with too few and too many is reversed: those with too many (22 per cent) slightly exceed those with too few (20 per cent). This reversal is even more pronounced in the less educated category in the CC-3, where women with ‘too many’ are more than twice as numerous as those with ‘too few’ (38 per cent vs. 17 per cent). Thus, the likelihood that those with low education will over-attain their fertility ideals increases as we move from the EU-15 to the NMS and to the CC-3. Further breakdowns of the data Table 1.2 Mean personal ideal number of children and actual number of children by schoolleaving age among women with completed fertility (ages 40–64) in the EU-15, NMS and CC-3 EU-15
NMS
CC-3
School leaving age
Ideal Actual Gap: I-A Ideal Actual Gap: I-A Ideal Actual Gap: I-A
Up to 15 years 16-19 years 20+ years All
2.46 2.20 2.46 2.34
2.34 1.98 1.81 2.07
0.12 0.22 0.65 0.27
2.39 2.26 2.20 2.29
2.50 1.97 1.68 2.07
Sources: Eurobarometer 56.1 (2001); CC Eurobarometer 2002
−0.11 0.29 0.52 0.22
2.33 2.05 1.93 2.20
2.73 1.93 1.39 2.32
−0.4 0.12 0.54 −0.12
40
Tony Fahey
Table 1.3 Fertility ideal attainment and school-leaving age among women with completed fertility (ages 40–64) Fertility ideal attainment (%) School leaving age
Too few
Just right
Too many
26 28 41
58 59 52
16 12 8
100 100 100
20 31 42
58 58 53
22 11 6
100 100 100
17 24 41
45 64 55
38 12 5
100 100 100
20 12 7
56 60 51
24 29 42
100 100 100
EU-15 Up to 15 years 16-19 years 20+ years NMS Up to 15 years 1619 years 20+ years CC-3 Up to 15 years 16-19 years 20+ years All Up to 15 years 16-19 years 20+ years
Total (%)
Sources: CC Eurobarometer 2002; Standard Eurobarometer 56.1 (2001)
not shown here, however, indicate that these regional contrasts can be overstated, in that over-attainment among less educated women varies considerably within regions as well as between regions (in particular, in the case of less educated women with completed fertility, over-attainment is below 5 per cent in the Scandinavian countries but is in the region of 25 per cent in Britain and Ireland). In sum, these patterns show that under-attainment of fertility ideals is not evenly distributed within the female populations in European countries but is particularly characteristic of better-educated women. It occurs among the less educated also, but less frequently and is counter-balanced to some degree among the less educated by substantial levels of over-attainment.
7. Trends since 1981 As a check on the patterns just looked at, we can briefly examine some trend data on ideal and actual family size since 1981 for a sub-set of European countries. As indicated earlier, the measure of ideal family size that is available across time relates to ‘general’ rather than ‘personal’ ideals (that is, to what people think is desirable for families in general rather than for themselves personally). The countries for which this measure is available are eight in number. The data are presented first for all adults for the eight countries in Table 1.4. These data broadly confirm the picture derived from the age comparisons for 2001–2002 set out earlier, keeping in mind that the latter relate to personal rather than general ideal number of children. There is a modest decline in general ideal family size between 1981 and 2001, with a fall in most countries from around 2.5 to something in the region of 2.2 or 2.3. Ireland again is an outlier: it had an ideal family size of 4 in 1981, and this fell to 2.8 in 2001.
Fertility patterns and aspirations
41
1
Table 1.4 Mean general ideal family size 1981
1990
2001
2.6 2.5 2.2 2.4 2.6 2.5 2.6 3.5
2.4 2.3 2.0 2.3 2.2 2.1 2.3 2.8
1990 (35–44) 2.6 2.3 2.0 2.3 2.4 2.4 2.5 3.2
2001 (45–54) 2.4 2.4 2.0 2.3 2.4 2.1 2.4 2.9
All adults France Britain W. Germany Italy Netherlands Belgium Spain Rep. Ireland
France Britain W. Germany Italy Netherlands Belgium Spain Rep. Ireland
2.6 2.4 2.3 2.3 2.5 2.4 2.9 4.0 Age cohort of women 1981 (25–34) 2.6 2.3 2.1 2.2 2.4 2.4 2.6 3.7
Sources: 1981 EVS; 1990 EVS; Eurobarometer 56.1 (2001) Note: 1 Ideal number of children for a family.
West Germany is the only one of the eight countries which moves to a sub-replacement family size ideal, again echoing earlier findings concerning the emergence of subreplacement fertility ideals in Germany. The lower panel in Table 1.2 presents the trend data for an approximate age cohort of women – those aged 25–34 in 1981, 35–44 in 1990 and 45–54 in 2001. The cohort data show that the sharp decline in family size ideals found in Ireland occurred within this age cohort across time and thus is not solely a consequence of changing attitudes from one cohort to the next. In the other countries, the change in ideals across time were generally slight and do not indicate a great deal of within-cohort change. Table 1.5 takes the pooled data for the eight countries shown in Table 1.4 and examines the incidence of under-attainment, over-attainment and ‘just right’ outcomes by educational level among women with completed fertility for 1981, 1990 and 2001. Keeping in mind that these outcomes are measured relative to general rather than personal family size ideals and relate only to eight countries, their significance should not be overstated. Nevertheless, it is of interest to note that they show no increase over time in the proportions of women at any educational level who have either ‘too few’ children or ‘too many’ children. Otherwise, the patterns confirm earlier findings: less educated women are more likely to have ‘too many’ children and less likely to have ‘too few’ children than better educated women.
8. Conclusions and implications The key empirical finding of this chapter is that under-attainment of family size preferences is a function of social advantage rather than disadvantage in Europe: well
42
Tony Fahey
Table 1.5 Fulfilment of general ideal number of children among women with completed fertility (ages 45–64), pooled data for 8 European societies, 1981, 1990, 2001 School leaving age (%)
1981 Too few Just right Too many Total 1990 Too few Just right Too many Total 2001 Too few Just right Too many Total
15 yrs or earlier
16-19
20+
All
No. of cases
37 39 24 100
41 39 20 100
44 46 10 100
39 39 21 100
704 533 311 1548
33 42 25 100
40 41 19 100
46 40 14 100
36 41 22 100
730 853 446 2029
25 52 24 100
33 44 23 100
43 47 10 100
32 47 20 100
341 662 172 1175
Sources: European Values Study 1981, 1990; Eurobarometer 56.1 (2001)
educated women are more likely to fall short of the number of children they would want than are less educated women. There is a slight effect of education on the family size ideals themselves in some countries (better educated women prefer slightly fewer children than less educated women) but the main effect of education operates on the attainment of those ideals rather than on the ideals themselves: better educated women have substantially fewer children than less educated women, even where their ideals are broadly similar. A less substantial though nonetheless notable pattern is that small but significant minorities of less educated women have too many children: their family size outcomes exceed their preferences, even in societies where on average women fall short of their preferences. These patterns do not divide along east– west lines in the enlarged EU. Differences within the EU-15 are as great as those between the EU-15 and the NMS. Turkey is sharply differentiated from all the other countries in that Turkish women still typically have more children than they would prefer. But apart from that, consistent regional variations in patterns of under-attainment and over-attainment are hard to identify in the EU. These empirical findings have a number of implications for policies that European governments might contemplate if they wish to boost fertility rates. First, they confirm that such policies might usefully focus on the constraints that prevent women from attaining their ideals rather than on the ideals themselves. Women in Europe do not need to be persuaded of the desirability of having more children – apart, perhaps, from countries such as Germany and Austria where women’s preferred family size has fallen to very low levels. Rather, they need to be facilitated in having the number of children they desire. Our findings do not tell us what kind of interventions would be most effective in achieving that end – nor even whether macro-economic measures designed to raise employment rates would be better than family-oriented measures such as child
Fertility patterns and aspirations
43
support payments, parental leave, improved childcare provision, or more flexible work practices. But they do tell us what categories of women should be the primary target of whatever interventions are chosen – namely, better educated women. These are the women most likely to fall short of their ideal number of children, presumably because the opportunity cost of childbearing is greatest in their case. While we have not tested the likely effectiveness of pronatalist interventions here, a targeting of such interventions on better educated women would probably require that the resources deployed be large, since if interventions were to be effective, their cost would have to be in proportion to the earning power and value of time (that is, to the opportunity costs of childbearing) of the segment of the population they would seek to influence. This would be so no matter what specific form the interventions would take (as between cash payments, maternity leave, etc.). It costs more for a highly educated, highly skilled woman to be given maternity leave, or to have flexible working hours, or to be provided with cash supports amounting to a meaningful share of household income, than for a less educated woman to be given supports with a similar relative value. There is an additional reason for arguing that the efficiency of pronatalist measures, evaluated in strict pronatalist terms, would be enhanced by targeting them on better educated women. This is that while less educated women also often fail to attain their ideal number of children, there are many countries where the mismatch in their case is as likely to arise in the form of having too many children as having too few. If resources were used to enable less educated women to attain their family size ideals, therefore, the result would consist of downward as well as upward movement in fertility. In many countries the downward movement would be large enough fully to counterbalance upward movement. While downward movement for women with more children than they want might be desirable for their welfare and that of their families, it would do nothing to raise fertility rates and would therefore fail to serve pronatalist objectives. The problem with pronatalist policies targeted on educated women is that, while they might work in raising fertility, they would also be socially regressive: they would give most to those who already have most. They would therefore be difficult to reconcile with the traditional socially progressive logic of family support, where the aim is to enhance child welfare and living standards among less well-off families. There is a less oppositional way of posing the alternatives: pronatalist policy could be viewed in terms of horizontal equity, where the aim is to transfer resources from those without children to those with children, rather than in terms of vertical equity, where the aim is to transfer resources from the well-off to the less well-off. (Income tax reliefs for families with children are one possible mechanism of horizontal distribution, especially in systems where income taxes in general are progressive. In those instances, income tax reliefs for children would provide the greatest benefits to high income earners with children, would impose the highest costs on high income earners without children, and would provide little or no costs or benefits on low income earners, whether with children or without.) Nevertheless, even if pronatalist measures were designed along horizontal equity lines, they would reduce the overall package of resources available for downward distribution (for example, income tax reliefs for high earners with children, while not imposing direct costs on low income families with children, would reduce nevertheless reduce the resources available for distribution to such families). It would therefore be difficult to design effective pronatalist measures that would not weaken the possibilities for socially progressive distribution, even if they did not obviate them entirely.
44
Tony Fahey
Table A1.1 Sample Ns for female samples by countries and age groups, combined Eurobarometer dataset, 2001–2* Female sample
Denmark Finland Sweden Ireland Great Britain Austria Belgium France Germany Luxembourg Netherlands Greece Italy Portugal Spain Cyprus Malta Slovenia Czech Rep. Hungary Poland Slovakia Estonia Latvia Lithuania Bulgaria Romania Turkey Total
Total sample
18-39
40-64
65+
All women
1000 1003 1000 1001 1312 999 1007 1005 2007 604 999 1002 999 1001 1000 500 500 1002 1000 1020 2000 1067 1010 1000 1015 1000 1049 2000
205 227 204 267 311 228 219 248 394 143 255 215 216 222 241 99 55 246 237 195 465 252 218 235 211 173 218 747
189 198 175 166 250 199 185 175 421 109 186 198 189 188 180 138 84 227 223 255 497 267 211 220 221 268 236 228
89 121 124 91 129 105 113 86 262 34 62 88 120 143 94 35 50 122 85 163 139 113 117 62 77 165 120 30
483 546 503 524 690 532 517 509 1077 286 503 501 525 553 515 272 189 595 545 613 1101 632 546 517 509 606 574 1005
30102
6946
6083
2939
15968
*Sources: Combined from Eurobarometer 56.1 (2001); CC Eurobarometer 2002
References Ahn, N. and Mira, P. (2002) ‘A note on the changing relationship between fertility and female employment rates in developed countries’, Journal of Population Economics, 15: 667–682. Alber, J. and Fahey, T. (2004) Perceptions of Living Conditions in an Enlarged Europe, Luxembourg: Office for Official Publications of the European Communities. Online. Available Http: (accessed February 2007). Bernardi, F. (2005) ‘Public policy and low fertility: rationales for intervention and a diagnosis for the Spanish case’, Journal of European Social Policy, 15, 2: 123–139. Billari, F. (2005) ‘Europe and its fertility: from low to lowest low’, National Institute Economic Review, 194, 1: 56–73. Bongaarts, J. (1998) ‘Fertility and reproductive preferences in post–transitional societies,’ Policy Research Division Working Paper, no. 114, New York: Population Council. Bongaarts, J. (2002) ‘The end of the fertility transition in the developed world’, Population and Development Review, 28, 3: 419–443.
Fertility patterns and aspirations
45
Brewster, K.L. and Rindfuss, R.R. (2000) ‘Fertility and women’s employment in industrialized nations’, Annual Review of Sociology, 26: 271–296. Caldwell, J.C. (1980) ‘Mass education as a determinant of the timing of the fertility decline’, Population and Development Review, 6, 2: 225–255. Caldwell, J.C. (2002) ‘Policy responses to low fertility and its consequences: a global survey’, Journal of Population Research, May 2002. Online. Available Http: (accessed February 2007). Castles, F.G. (2003) ‘The World turned upside down: below replacement fertility, changing preferences and family-friendly public policies in 21, OECD countries’, Journal of European Social Policy, 13, 3: 209–227. Castro Martin, T. (1995) ‘Women’s education and fertility: results from 26 demographic and health surveys’, Studies in Family Planning, 26, 4: 187–202. Chesnais, J.-C. (1998) ‘Below-replacement fertility in the European Union (EU-15): facts and policies, 1960-1997’, Review of Population and Social Policy, 7: 83–101. Cleland, J. (2003) ‘Education and future fertility trends with special reference to mid-transitional countries’, pp. 187–202, in United Nations Population Division of the Department of Economic and Social Affairs (ed.), Completing the Fertility Transition, New York: United Nations. Online. Available Http: (accessed February 2007). Council of Europe (2005) Recent Demographic Developments in Europe 2004, Strasbourg: Council of Europe Publications. d’Addio, A.C. and Mira d’Ercole, M. (2005) Trends and Determinants of Fertility Rates: The Role of Policies, OECD Social, Employment and Migration Working Papers, Paris: OECD. European Commision (2005) Confronting Demographic Charge: A New Solidarty between Generations, Green Paper, Communication from the Comission Com (2005) 94, final, Brussels: Commision of the European Communities. Eurostat (2002) European Social Statistics: Demography, Luxembourg, Office for Official Publications of the European Communities. Fahey, T. and Spéder, Z. (2004) Fertility and Family Issues in an Enlarged Europe, Luxembourg: Office for Official Publications of the European Communities. Online. Available Http: (accessed February 2007). Freedman, R., Freedman, D.S. and Thornton, A. (1980) ‘Changes in fertility expectations and preferences between 1962 and 1977: their relation to final parity’, Demography, 17, 4: 365–378. Gauthier, A.H. and Hatzius, J. (1997) ‘Family benefits and fertility: an econometric analysis’, Population Studies, 51: 295–306. Goldstein, J., Lutz, W. and Testa, M.R. (2003) ‘The emergence of sub-replacement family size ideals in Europe’, European Demographic Research Papers 2003, no. 2, Vienna Institute of Demography in collaboration with the European Observatory on the Social Situation, Demography and the Family. McDonald, P. (2000) ‘Gender equity, social institutions and the future of fertility’, Journal of Population Research, 17, 1: 1–16. New Cronos (2005) Eurostat on-line database. Online. Available Http: (accessed February 2007). Pearse, D. (1999) ‘Changes in fertility and family size in Europe’, Population Trends, no. 95, London: Office for National Statistics. Schoen, R., Ashtone, N.M., Kim, Y.J., Nathanson, C.A. and Fields J.M. (1999) ‘Do fertility intentions affect fertility behaviour?’, Journal of Marriage and the Family, 61: 790–799. Sleebos, J.E. (2003) Low Fertility Rates in OECD Countries: Facts and Policy Responses, OECD Social, Employment and Migration Working Paper no. 15, Paris: OECD.
46
Tony Fahey
Thomson, E. and Brandreth, Y. (1995) ‘Measuring fertility demand’, Demography 32, 1: 81–96. Thomson, Elizabeth (1997) ‘Couple childbearing desires, intentions, and births’, Demography, 34, 3: 343–354. United Nations (2003) Partnership and Reproductive Behaviour in Low-fertility Countries, ESA/P/WP.177, New York: United Nations. United Nations Population Division (1995) Women’s Education and Fertility Behaviour: Recent Evidence from the Demographic and Health Survey Series, New York: United Nations. Van de Kaa, D. (1998) ‘Postmodern fertility preferences: from changing value orientation to new behaviour’, Working Papers in Demography, no. 74, Australian National University, Canberra. Van Peer, C. (2002) ‘Desired and realized fertility in selected FFS countries’, pp. 117–142, in M. Macura and G. Beets (eds), Dynamics of Fertility and Partnership in Europe: Insights and Lessons from Comparative Research, Vol. 1, New York: United Nations.
2
Patterns of family living in the enlarged EU Chiara Saraceno
Introduction: A history of diversity Patterns of family formation and family living have a long history of diversity throughout Europe (e.g. Reher 1998; Therborn 2004). Diversity in marriage rates, age at marriage, family structures, fertility, gender relations and intergenerational power relations continues to exist. These historical differences shape patterns of change and the impact of external pressures on existing family arrangements. To some degree, they also inform the way in which needs are understood and how responsibility for these needs is allocated between individuals, families and other social actors, including the welfare state. As a consequence, difference, rather than convergence, seems to be the persistent feature of family arrangements even within the limited space of Europe. An analysis performed for the Council of Europe in 2001, for instance, concluded that ‘The results of this analysis show that some demographic patterns are diverging (fertility quantum, fertility and marriage timing); others are geographically unchanged (divorce and out-of-wedlock births) even if their levels are different, and there is no tendency towards convergence’ (Pinnelli et al. 2001: 16). In their overview of changes in family patterns across Europe since the sixteenth century, Barbagli and Kertzer (2003) argue that at the beginning of the twentieth century there remained enormous differences in patterns of family formation among the various regions of the European continent. According to their reading, the twentieth century, differently from the preceding ones, was actually marked primarily by convergence. Yet, at the end of the twentieth century, although differences had diminished, substantial differences remained in the ways in which ‘families are formed, transformed and divided, and in the relations of family members and of more distant kin’ (Barbagli and Kertzer 2003: xxxviii). Differences in particular involve the role of marriage and its relationship to both sexuality and fertility, fertility rates, patterns of intergenerational exchange, gender relations and arrangements. These differences underline what Therborn (2004: 11), looking at the world as a whole, defined as ‘geo-cultures’: ‘To view family systems as geo-cultures means to treat them as institutions or structures taking their colouring from customs and traditions, from the history of a particular area, a cultural wrapping which may remain after structural, institutional change, leaving imprints in the new institution.’ Thus, for instance, the delay in leaving the parental home in Italy is certainly a contemporary device used by the young and their families to deal with a weak welfare state and a rigid housing market. But this behaviour has its roots in a historical pattern of
48
Chiara Saraceno
family formation according to which many Italian men in the centre-north regions until the first decades of the twentieth century never exited the parental household. If and when they married, they brought their wife into the parental household (Barbagli et al. 2003). Many authors have offered their own reading of this diversity. In a famous essay, John Hajnal (1982) identified two broad areas within Europe, on the basis of three criteria: prevalence of the nuclear, rather than the complex, family; degree of universality of marriage; and age at marriage for women. East of the conceptual line that runs from St Petersburg to Trieste, in what Hajnal called the Eurasian marriage pattern, the incidence of complex households has been substantial for many centuries. In these countries marriage also was virtually universal, and the age at marriage for women was lower than in western European countries. The complex household was almost absent in western European countries, particularly in Scandinavia, the Netherlands, England and northern France. It was present – mostly in the form of the stem family, as observed by Frédéric Le Play (1875), in southern Austria, some parts of Germany and in the rural areas of northern and central Italy, but not in southern Italy. More recently, David Reher (1998) has pointed to another aspect of diversity: the relevance of, and embeddedness in kinship ties. In the Mediterranean countries, irrespective of the household structure (nuclear, stem or extended), households traditionally have been embedded in and dominated by a dense kinship network to a degree very different from that found in northern European households, even in the past. According to Reher, within Europe at least three general patterns of family formation and arrangement may be found up to the twentieth century. The first was prevalent in the central-western and Nordic countries. Here, individuals married late, at a comparatively lower rate, and had fewer children; moreover, couples were relatively unstable and had loose ties with their kinship network. The second was prevalent in some parts of southern Europe. Here, households also were nuclear, but people married earlier and at a relatively higher rate, households were embedded in dense kinship networks. The third pattern was prevalent in eastern European countries and the Balkans, as well as in parts of southern Europe (e.g. rural central and northern Italy). Households were complex, and age at marriage for women often was comparatively low. As a result, gender asymmetry was sharp, and the fertility rate was high. Hajnal’s and Reher’s typologies partly overlap, but also partly differ, designing different boundaries in traditional patterns of family formation across Europe, depending on the specific dimension of family arrangements on which they focus. They do, however, share the idea that family arrangements belong to the longue durée and involve some kind of cultural path dependence. Past differences, therefore, shape also patterns of change. To stress the relevance of long-standing arrangements in family formation and cultures does not mean that one ignores history and the relevance of changing social, economic and political conditions. Rather, the approach suggests that the impact of social, economic and political conditions on family arrangements is to some degree dependent on these arrangements themselves. This (inter)dependence helps explain the substantial differences that even today can be found across Europe in family patterns – including gender and intergenerational arrangements. It may also partly explain differences in welfare arrangements (see also Bahle, this volume Ch. 4). One might argue that long-standing family cultures offer different resources to deal with new problems or opportunities.
Patterns of family living
49
These differences are not obliterated by the construction of the EU as a political and social entity. Rather, they enter, more or less explicitly, into the formal and informal negotiations and confrontations through which Europe as a supra-national body is constructed, shaping country-specific understandings of common goals and priorities, as in the case of what civil and social rights are about, or of gender equality. They are relevant not only for demography but for understanding patterns of everyday organisation, such as dealing with needs of income and care, for gender and intergenerational relations, and for the division of labour and responsibilities. This chapter is largely based on a new comparative source: the European Quality of Life Survey (EQLS), performed in 2003 on the 15 old EU member states, the 10 new member states and three countries that have applied for EU membership: Bulgaria, Romania and Turkey. This dataset has its limitations and does not allow for a full exploration of the range of possible variations in patterns of family formation and living.1 Yet the EQLS data do offer a rich, cross-sectional picture of what kinds of households Europeans live in and of how they organise their everyday life in dealing with their needs for income and care. The following analysis starts with the political and institutional divide that stems from the different timing of access to the EU – the EU-15, the new member states, and the three countries which have applied for membership – and then moves towards a deeper exploration of differences in patterns of family formation. In the last section, the analysis addresses differences in gender arrangements and, more specifically, in the gendered division of labour – a crucial contemporary difference in family arrangements.
1. Patterns of family formation in the European Union: an overview Within the ‘old’ 15 European Union member states (hereafter, ‘EU-15’), the main divide is between the northern European pattern, including the United Kingdom, and the southern European one, with the central European countries somewhere in between. The first group of countries for a long time had a marriage pattern that can be characterised as late, with a comparatively high percentage of the population who never marries. Until the 1970s, it also exhibited the lowest fertility rates in Europe; it now has the highest, together with Ireland and France. It also is the first group of countries where cohabitation as a pattern of first couple-formation emerged in the late 1970s and then spread, to become the most common pattern. Women’s labour force participation rates already were comparatively high in the 1960s. In comparison, in southern European countries the young, in particular young men, always have exited the parental household latest, and mostly only to marry. The gender differentials in age at marriage and the marriage rates have traditionally been higher here than in the northern countries, although in recent years the latter has been declining. Moreover, after having been the countries with the highest fertility rates, they now compete with eastern European countries (and Japan) for lowest fertility. Still, in these countries, marriage instability rates are comparatively low, as are labour force participation rates among women. EU enlargement is progressively including within the EU boundaries such countries as Estonia, which clearly lie west of Hajnal’s divide; countries such as Hungary, Poland, Latvia and Lithuania, which are on the border of this divide; and still others, such as Slovenia, Cyprus and, even more clearly, Bulgaria, Romania and Turkey,
50
Chiara Saraceno
which lie east of that divide.2 Whatever their positions with regard to this divide, most of these countries share some of the features of the so-called eastern European model: nearly universal marriage, a low age at marriage for women, and high fertility (Council of Europe 2001: 55–76). In 1985, the marriage rates in Communist Eastern (excluding Russia) and Central Europe were, respectively, 0.88 and 0.90, compared to 0.83 in Greece, 0.79 in Portugal and 0.70 in Austria and West Germany. Fertility, including non-marital fertility, also was higher in these countries than in Western Europe. The pattern of universal marriage and a comparatively low age at marriage for women persisted, particularly in Bulgaria, the Czech Republic and Hungary, until the generation born after 1955 (see also Tomka 2002). Due to widespread impoverishment since the end of communism, marriage rates have declined sharply, as have fertility rates. According to some analyses (e.g. Therborn 2004), however, this phenomenon reflects a conjuncture of events: marriage rates will pick up again, remaining on average higher than in Western Europe. In addition to fertility, marriage rates and age at marriage, another aspect showing variation in the past, but particularly at present, is a mother’s partnership situation at first birth. According to the data of the United Nations Fertility surveys in the 1990s (United Nations Economic Commission for Europe 2002), substantial differences exist across Europe, reflecting in part historical differences. In the Netherlands, Belgium, Estonia, the Czech Republic, Hungary, Poland, Spain, Italy and Greece, between 89 per cent and 95 per cent of first-time mothers were married. In another group of countries (Finland, France, western Germany, Latvia, Lithuania, Slovenia), the percentage was between 70 per cent and 85 per cent. In Norway it was 62 per cent, in Austria 56 per cent. In Denmark and Sweden it was well below half of all mothers: 26 per cent and 32 per cent, respectively. These two countries had the highest percentage of first-time mothers in an unmarried partnership (50 per cent and 58 per cent, respectively), followed at a distance by Norway (23 per cent), France (20 per cent), Austria (18 per cent), Finland (16 per cent) and Slovenia (15 per cent). Denmark also had the highest percentage of single mothers (24 per cent), followed by western Germany (19 per cent), Norway (15 per cent), Slovenia (15 per cent) and Poland (15 per cent). Poland, Greece and Lithuania had the lowest percentages of first-time mothers in a cohabitant unmarried partnership (1–2 per cent), followed by Spain, Italy, the Czech Republic and Belgium (3–5 per cent). With the exception of Denmark and Sweden, with Norway following at a distance, no clear Nordic pattern can be found in these data. Within the continental countries, a similarity emerges in that a substantial minority of first-time mothers are in a cohabitant unmarried partnership. But the incidence of both single and married first-time motherhood varies. The Mediterranean pattern of the prevalence of first-time motherhood within marriages is clearer. The pattern in most central-eastern European countries resembles the Mediterranean one with regard to the dominance of married first-time motherhood and the very low incidence of first-time mothers in cohabitant unmarried partnerships. But the incidence of single motherhood is higher. Furthermore, by 2002, Estonia had the second-highest extramarital fertility rate in Europe. Latvia and Bulgaria also were among the ten European countries with the highest fertility rate for unmarried mothers (e.g. higher than that in the United Kingdom; Council of Europe 2004). The risk of a marriage breaking up also varies substantially across Europe, ranging from one in ten (in the Mediterranean countries) to one in two (in the
Patterns of family living
51
Scandinavian countries). In 2002, Sweden, Belgium, Luxembourg, Finland, Estonia, the Czech Republic and Austria were among the ten countries with the highest rate of divorce, whereas Turkey, Spain, Italy, Croatia and Spain had the lowest divorce rate in Europe (Council of Europe 2004). Given these differences, the incidence of specific household patterns at a given point in time varies across countries. Not all countries use the same definition of the household in their statistics, despite the guidelines both of the United Nations and of Eurostat. Therefore comparisons are not always precise. The main differences concern whether a household is defined primarily as a dwelling unit or as a housekeeping unit. Other differences pertain to the definition of family units within households, and whether there is an age limit to define the relationship between parents and children. This variation in definitions renders not fully comparable the official statistical data, including the census data (see, for example, De Vos and Sandefur 2002; Hantrais 2004). Differences in definition affect in particular the comparability of the data on unmarried cohabitant couples, lone parents, and extended and multiple households. In the light of these methodological and conceptual difficulties, the EQLS data have the virtue of having been generated within the same conceptual framework. Households are primarily defined as dwelling units. Respondents were asked with whom they lived. Household types were reconstructed on the basis of these individual answers, focusing on whether or not a couple, children and/or other relatives were present in the household. There is no age limit in defining the parent–child relationship. Within the EU-25 (the 25 EU member states as of 2004), only 35 per cent of European households include a couple with children of any age, compared to 54 per cent in the three other countries; 30 per cent include a couple only.3 Lone parents comprise 8 per cent of all households. One-person households make up a quarter of all households, whereas extended households, in which (usually older) relatives live with a couple and their children, constitute a very small group (see Table 2.1).
Table 2.1 Household patterns in an enlarged Europe (%) Country cluster EU-15 NMS EU-25 CC-3
Living alone 26 15 25 9
Couple without children
Couple with children1
Lone parent2
Extended or multiple household3
32 19 30 19
33 46 35 54
7 11 8 8
2.5 10 3 10
Source: EQLS 2003; layer % Question HH3 (household grid): Now think about other members of your household, starting with the oldest: What is this person’s relationship to you? Notes: 1 Includes unmarried children of all ages. 2 Includes all adults living with their children, irrespective of the children’s age, but without a partner or any other person. 3 Extended households include one or more relative, but there is only one conjugal couple. Multiple households include more than one couple.
52
Chiara Saraceno
However, the incidence of the latter household pattern reaches 10 per cent in the new member states, compared to just 2 per cent in the old EU-15, confirming indirectly the persistence of Hajnal’s long-standing geographical divide.4 Conversely, singleperson households and childless couples are fewer in number in the ten new member states (hereafter, ‘NMS’) than in the EU-15. They are even less common in the three other countries (hereafter, ‘CC-3’), which show a distribution of household patterns more similar to that of NMS than that of EU-15. Households in the NMS and CC-3 are also more likely to include unmarried children than are those in the EU-15. These distributions suggest that different mechanisms of family formation are in place. The data do not, in fact, mean that most couples in Europe never have a child, nor that one fourth of the EU-25 population live alone throughout their entire adult life. Rather, the combination of increased life expectancy, low fertility rates, the widespread but diversified popularity of the nuclear household (in which each couple sets up its own household) and increasing marital instability opens up new household and individual life phases and/or renders old ones (e.g. the empty-nest phase) more prolonged and widespread. For instance, it is no longer widowhood alone that transforms a surviving spouse into a single person or a lone parent. The transformation also may occur through divorce or separation. Individual ‘household careers’ and the life courses of households have become more diversified, if not fragmented. For this reason, the kinds of household that people live in during different phases of the life course is a more meaningful indicator of patterns of household formation than is the simple distribution of household types within and across countries. Although the EQLS data are not longitudinal, they do offer the possibility of comparing across countries, as well as across gender and economic circumstances, the kinds of household situations that individuals experience and the household statuses that they have at various ages. The household unit – whatever its definition – of course does not fully coincide with ‘living arrangements’. Some household units may simply represent an address and possibly a sleeping place, whereas the individuals inhabiting them may spend all or most of their day in other households. This is the case of elderly individuals who spend their day and eat their meals in one of their children’s households. In some cases, young adults may have their own dwelling, but are supported financially by their parents, often eating their meals with them and having their laundry done by their mothers. More generally, ‘living arrangements’ may involve more or less frequent exchanges and relationships within the kinship network. As recent research has shown, kinship – particularly intergenerational ties – is a crucial practical as well as emotional resource throughout Europe (e.g. Kohli et al. 2005; Ogg and Renaut 2005). The EQLS data also testify that contacts with kin are frequent. Moreover, kin is the main resource for expected or received practical support in all countries, irrespective of household patterns and of welfare regime type or political history (Saraceno et al. 2005). ‘Households’, therefore, do not fully coincide with ‘families’ in the practices and understanding of most people. Yet households define some kind of boundaries within kinship networks. Specific household patterns, therefore, define those boundaries, and the need and the possibilities to cross them, in different ways. One difference concerns the experience of living alone or with others. The highest percentage of adult individuals living with someone else is found in CC-3 (91 per cent), the lowest in EU-15 (75 per cent), with NMS in between (85 per cent). Within the
Patterns of family living
53
EU-15, however, the Mediterranean countries are more similar to the NMS than to the central and Nordic European countries. Not surprisingly, whereas the households of those in the middle age-groups are more similar across countries, the biggest differences across countries and country clusters are found among the young and the elderly. These differences throw a light on distinct patterns of family formation and household experience along the entire lifecycle. When children leave the parental household late or, in some cases, bring their spouse or partner into it, adults in their mature and older years may spend many years living in the same house as their adult children. Some will become old and die while still living with them. Many lone mothers in Mediterranean countries, for instance, are neither unmarried single mothers nor separated or divorced ones. They are widows still living with their unmarried adult children. On the contrary, when children exit the parental household early, the adults in their mature years are likely to live many years as a childless couple or, particularly when old, as a single person. The kinship network is affected by these patterns, for in some cases the parents of both partners live in different households. In other cases, one set of parents (or one parent only), and sometimes other relatives, live with the couple and even with grandchildren. Although the kin network is larger than any kind of extended or multiple household and nuclear households and even single individuals may be embedded in dense kinship networks, demands and forms of support, patterns of giving and receiving across kinship networks are affected by patterns of family formation and household structure in the different phases of life.
2. Patterns of family formation when young Throughout Europe, the formation of a married or unmarried cohabitant couple and the beginning of parenthood – events traditionally marking completion of the transition to adulthood – normally follow the individual’s exit from the parental home. Yet for a small number of the young, the couple is formed within the parental household of one of the partners. This phenomenon is more evident, as one might expect, in some of the eastern European countries, such as Hungary, Bulgaria, Romania and Cyprus, but also in Ireland. Thus, for a proportion of the young in these countries, family support in ‘forming a new family’ occurs in the form of including a child’s partner in the household. On average, in the EU-15 the young become parents later than in the NMS and the candidate countries. They do, however, leave the parental household at a younger age. Young persons in the CC-3 leave the parental household later than in the EU-15, but earlier than in the NMS, and they also become parents earlier. The largest crosscountry differences among the young concern the percentages of those who still live as children in the parental household. Within the EU-15, differences are substantial for both genders, but particularly for men. The incidence of men under 35 years of age still living in the parental household, without a partner or children, ranges from 12 per cent in Sweden to almost 6 times as much, 67 per cent, in Italy. The other Mediterranean countries score 20 percentage points less than Italy. Greece is more similar to France and Luxembourg than to Italy. Only Malta, in the new member states, has a percentage similar to the Italian one. The other new member states range from 36 per cent for Estonia to 57 per cent for Slovakia. The CC-3 are similar to the higher end of the NMS spectrum (excluding Malta) (see Table 2.2).
Austria Belgium Denmark Finland France Germany Greece Ireland Italy Luxembourg Netherlands Portugal Spain Sweden United Kingdom
Men 29 28 17 13 30 21 39 25 67 35 36 48 45 12 19
Women 13 28 8 11 18 14 19 19 60 25 22 33 34 10 12
Living with parents
Household Status
Men Women 35 23 28 11 33 28 39 29 36 24 40 26 33 24 10 7 11 6 12 5 27 20 7 5 5 5 44 31 33 14
Living alone
Men Women 14 26 20 14 28 28 26 25 16 23 11 27 8 14 14 15 8 12 13 13 19 22 8 10 22 16 17 24 22 36
Living as childless couple
Table 2.2 Household statuses of the young in Europe (ages 18–34) (%)
Men 13 20 20 18 15 20 9 15 11 25 12 28 13 22 8
Women 18 33 21 26 23 23 33 18 20 39 25 30 26 23 19
Living as a couple with children
Men 4 1 0 1 1 1 2 12 0 4 3 4 5 1 4
Women 3 1 2 0 1 1 3 13 1 5 1 11 4 2 1
Living as a couple with or without children in an extended household Men 3 2 0 0 1 2 1 6 0 4 1 0 1 0 1
Women 14 11 9 6 7 5 1 15 2 7 10 10 4 5 14
Lone parent in nuclear or in extended household
Men 2 1 1 3 2 6 8 18 2 7 2 5 9 5 12
Women 4 2 5 3 3 4 6 13 0 6 1 2 10 5 4
Other kind of household
43 48 36 47 39 41 67 55 57 53 50 51 49 36 52 49
29 28 29 32 22 18 55 45 41 43 23 19 29 27 39 26
16 9 17 13 14 9 7 2 4 10 11 6 6 26 6 6
6 8 19 5 13 8 4 3 1 6 3 11 1 16 4 4
9 17 26 10 13 8 9 4 4 10 3 12 7 15 8 8
20 7 23 14 20 14 14 4 4 11 13 18 9 21 8 11
19 20 10 16 21 34 9 22 21 17 23 11 20 14 21 18
30 34 17 25 24 35 11 23 31 26 25 20 40 23 26 35
9 2 5 11 9 8 5 11 12 5 10 16 7 3 10 10
8 9 2 16 7 9 8 15 13 9 27 16 12 2 14 15
2 0 2 1 2 1 1 2 1 1 0 1 2 1 1 2
3 11 10 4 11 13 1 8 7 3 6 4 4 7 8 4
3 4 3 2 3 0 2 4 2 5 4 4 8 5 3 7
5 3 1 5 4 4 7 2 2 3 3 12 4 4 3 6
Notes: The category ‘Living with parents’ includes all respondents without a partner and childless, who are living with parents with or without other aggregate members. The category ‘Other kind of household’ includes the respondents who live with friends, brothers or sisters (without parents) or other aggregate members
HH3c: ‘Now think about other members of your household, starting with the oldest: What is this person’s relationship to you?’
Source: EQLS 2003; layer %
Cyprus Czech Republic Estonia Hungary Latvia Lithuania Malta Poland Slovakia Slovenia Bulgaria Romania Turkey EU-15 NMS CC-3
56
Chiara Saraceno
Turkey has the highest, Malta the lowest percentage of young women who are mothers: respectively, 48 per cent (including those living with a partner and children in an extended household) and 12 per cent. Ireland, the United Kingdom, Austria and Lithuania have the highest percentages of lone mothers (15–13 per cent; living alone or in an extended household) in this age group; Greece and Malta have the lowest, followed by Italy. Sweden has a percentage close to that of Spain and much lower than that of Denmark. The former socialist countries have comparatively high percentages. Thus, the EQLS data indirectly confirm findings based on demographic sources: the rise in natural births is accompanied by a rise in lone motherhood in some countries, such as the former socialist ones and some western European ones as diverse as Ireland, the United Kingdom, Portugal, the Netherlands and Denmark (though not others, such as Sweden, where natural births tend to occur within a cohabitant couple relationship).5 If one examines two of the dimensions which in the literature are considered to be crucial dimensions both for entering adulthood and for forming a family – exiting the parental household and forming a partnership – three patterns may be distinguished (see Figure 2.1 for these patterns among young men).6 They confirm the patterns described by Alessandro Cavalli and Olivier Galland (1996) for the EU-15 only (see also Billari and Wilson 2001, Schizzerotto and Lucchini 2004). The patterns are clearer among men than among women. In the first pattern, prevalent in the Nordic countries, Germany, Austria, Belgium, France and the United Kingdom, as well as in Greece,7 a substantial number of young people is outside the parental household by the age of 34. They either live alone as singles, particularly if they are males, or live with a partner, with or (more often) without children. Among young Swedish and German men 44 per cent and 40 per cent live by themselves, respectively, compared to 12 per cent and 21 per cent who still live at home with their parents. In the second pattern, prevalent in the southern European countries and in the new member states, around half, and sometimes more, of those under 35 years old are still in the parental household without a cohabiting partner, particularly if they are male. For example, 67 per cent of Italian and Maltese young men, 57 per cent of Slovakian, 55 per cent of Polish and over 40 per cent of Portuguese and Spanish young men are still living in the parental home at the age of 34, without a partner. These households usually are nuclear, but, particularly in the NMS and in CC-3, for a small number they are extended; that is, other relatives (usually one or more grandparents or a sibling’s spouse) also are present.8 If they are outside the parental household, the young are more likely than in the former pattern to be already in a partnership and, particularly if female, to have one or more children. The third pattern concerns a smaller group of the young in all countries; it is particularly concentrated in the NMS and the CC-3, but also plays a significant role in Ireland. In this case, when the young form a partnership, they remain in the parental household or enter the partner’s parental household, which sometimes includes other relatives as well. In all three patterns, women generally enter a partnership and become parents earlier than do men. In comparison with men, they are less likely to live as childless singles, but more likely to live as a lone parent, either alone with their child or children, or together with their parents in an extended household. As Cavalli and Galland (1996) have indicated, the three patterns of family formation found among young Europeans involve different kinds of exchange and forms of
Patterns of family living
57
60 56
50
48
40 35 % 30 26
19
20 16
16 14 11
10
7
7 5 2
5
9
10 7
4 2
1
0 Pattern of ʻearly exitʼ
Pattern of ʻlate exitʼ
Living with parents Living as childless couple Living as a couple with or without children in extended household Other kind of household
Pattern of ʻpartnering in a parental householdʼ
Living alone Living as a couple with children Lone parent in nuclear or extended household
Figure 2.1 Household status of young men (18–34): three distinct patterns (%) Source: EQLS 2003 Question HH3c: ‘Now think about other members of your household, starting with the oldest: What is this person’s relationship to you?’ Notes: The category ‘Living with parents’ includes all respondents without a partner and childless, who are living with parents with or without other aggregate members. The category ‘Other kind of household’ includes the respondents who live with friends, brothers or sisters (without parents) or other aggregate members. The ‘early exit’ pattern includes Austria, Belgium, Denmark, Finland, France, Germany, Greece, the Netherlands, Sweden and the United Kingdom. The ‘late exit’ pattern includes Estonia, Hungary, Italy, Malta, Poland, Portugal, Slovakia, Slovenia and Spain. The ‘partnering in a parental household’ pattern includes Bulgaria, Cyprus, the Czech Republic, Ireland, Latvia, Lithuania, Luxembourg, Romania and Turkey.
58
Chiara Saraceno
support across families and kin, as well as different options available for the young. An individual’s ability to leave the parental household early may be supported by cultural values, but also by favourable labour and housing markets as well as by welfare state provisions. If, however, the family is the main financial resource for the young and the housing market is tight, it is more difficult for a young person to leave the parental household when he or she is not yet established in the labour market. Moreover, it may be more costly for parents to help children live on their own. And, finally, if the family culture does not support the extended household pattern, or if parents are unwilling to accept a child’s spouse into their household, the young must wait not only to leave the parental household, but also to marry. The concentration of young couples living in an extended household in the countries east of Hajnal’s line, that is, in countries where the extended or multiple family pattern has a long tradition, confirms the substantial influence of historical, geo-culturally bounded family patterns. Data on the occupational status of the young in different household circumstances and in different countries support the hypothesis that decisions about forming a new household are certainly influenced, but not dictated, by labour market and welfare Cluster EU-15 NMS CC-3
TR
% of the young neither in work nor in education
40 BG UK
30 PL
20
RO
EL
IR LT
DE BE
FI
FR
PT
LV
ES
NL LU
IT EE HU
10
SW DK
10
AT
SI
CZ CY
20
SK
MT
30 40 50 60 % of the young living with parents
70
Figure 2.2 Incidence of unemployment among European young singles living with their parents, by country Source: EQLS 2003, own calculations
Patterns of family living
59
state conditions. Actually, with the partial exception of Poland, the incidence of those who are neither in work nor in education is higher among the relatively few young persons who live without a partner and childless in the parental household in some of the ‘early exit countries’ (e.g. the United Kingdom) than among the more numerous ones in the ‘late exit’ countries (see Figure 2.2). Those who live with a partner, either with or without children, in an extended household are also more likely to be in paid work: 90 per cent in the NMS, slightly less in the EU-15 and CC-3. Thus, remaining in the household of one’s parents is not necessarily a means to buffer unemployment. It might be a means to integrate otherwise low wages, to accumulate capital for establishing a new family, or to support exploratory strategies in the labour market.9 Gender differences are evident, however: young women who are not in education and are still living in the parental household are generally more likely than men not to be working. In sum, these cross-country differences suggest that paths into adulthood are shaped not only by available resources, such as publicly provided income support in the event of unemployment, scholarships for education, or access to housing. Cultural expectations about proper behaviour and the proper sequence of events are also involved.
3. Households of elderly persons By around 2020, the main increase among elderly persons in the EU will occur within the over-80 group. This group will increase by about 50 per cent (European Commission 2003). The situation looks more balanced in the new member states, as their population is on average younger, but the trend is quite similar (Fahey and Spéder 2004). An increasing number of households will comprise only elderly people. Kinship networks will have an age and intergenerational balance skewed towards elderly people – with frail elderly persons constituting a crucial, if relatively small, proportion of them, who will have special needs and demands for care. In which kind of household elderly persons live depends not only on their own past choices concerning marriage and/or cohabitation with a partner, divorce, having children and how many children they have had, at what age, and so forth. It depends also on patterns of household formation by the young: whether they bring their partner into the parental household or they form a new household, at what age they exit their parental household, and so forth. It depends, moreover, on gender, given women’s higher life expectancy on average, which, coupled with their younger age at marriage on average, makes it more likely for them than for men to outlive their partners into old age (see also Iacovou 2000, based on European Community Household Panel data). Gender differences are systematic across countries. Throughout Europe, older men live with their partner to a much greater degree than older women. Conversely, older women live alone to a much greater degree than older men (see Figure 2.3). Thus, in old age there is a partial reversal of the pattern found among the young, among whom more men than women live alone. For instance, in Sweden and Denmark, over 70 per cent of older women live by themselves, compared to over 40 per cent of older men. In Portugal and Greece, over 55 per cent of older women live alone, compared, respectively, to 28 per cent and 10 per cent of older men. In Poland, the percentages
60
Chiara Saraceno 40 30 20 10
%
0 EU-15
NMS
CC-3
−10 −20 −30 −40 Living alone
Living as a childless couple
Living as a couple with children
Living as a couple in a next ended household
Living without partner, with children or relatives
Other kinds of households
Figure 2.3 Households of elderly people: differences between men and women (65 and over) in the three clusters (%) Source: EQLS 2003 Notes: The figure shows the relative differences between the household status of elderly men and that of elderly women (calculated as the difference between the % of elderly men who live in a given household status and the correspondent women’s % in the same household status). Negative differences show how much less men are in that status compared to women. Positive differences indicate how much less women are in that status compared to men. The category ‘Other kind of household’ includes those who live with relatives or other people, but no partner or children.
for women and men are, respectively, 39 per cent and 17 per cent; in Turkey, 23 per cent and 3 per cent. Fewer women than men still live with both their partner and their children. On the other hand, more women than men live with their children, whether or not these children have a partner, when they lose their own partner. This latter gender difference is particularly clear in NMS and CC-3, given the higher incidence of extended households in these countries. Cross-country differences are as important as gender differences. They do not, however, neatly overlap with any standard clustering based on tripartite EU membership, on Hajnal’s divide, or on patterns of family formation by the young (see Table 2.3). Certainly, the experience of living with one’s own children when old is nearly absent in such ‘early exit’ countries as Denmark, Sweden, the Netherlands, France and Germany. But in Austria and particularly in Greece, which share the same pattern, this experience is relatively widespread. As one might expect, the incidence of this
Patterns of family living
61
1
Table 2.3 Elderly men and women (65 and over) living with their children; by patterns of family formation by the young2 (country %) Patterns of family formation by the young ‘Early exit’ Austria Belgium Denmark Finland France Germany Greece Netherlands Sweden United Kingdom
10.8 7.5 2.23 6.7 3.63 1.63 19.8 3.83 2.13 7.8
‘Late exit’ Estonia Hungary Italy Malta Poland Portugal Slovakia Slovenia Spain
14.3 20.9 27.8 31.0 35.0 11.3 18.3 21.6 23.2
‘Partnering in a parental household’ Bulgaria Cyprus Czech Republic Ireland Latvia Lithuania Luxembourg Romania Turkey
22.0 21.4 12.9 15.0 16.7 14.2 9.33 19.1 44.8
Source: EQLS 2003 Notes: 1 All elderly persons living as a couple with children (in a nuclear or extended household) and lone parents living in a nuclear or extended household. 2 See Figure 2.1. 3 N < 15.
household situation is highest in the ‘late exit’ and the ‘partnering in a parental household’ countries, but with notable internal differences. Within the former, Poland, Malta and Italy have substantially higher percentages than do Portugal and Estonia. Slovenia, Slovakia, Spain and Hungary lie in between. Differences are even greater within the third group of countries, ranging from almost half of all elderly persons in Turkey living with their children – whether or not the children have a partner – to 13 per cent in the Czech Republic. This ‘fuzziness’ of clusters is probably due to the fact that household patterns in old age are the outcome both of patterns of household formation by the current younger cohort and of their own cohort’s specific socio-cultural and demographic
62
Chiara Saraceno
histories: marriage rates, age differentials at marriage, fertility, mortality and migration. The interplay of all these phenomena may offer different household ‘options’ when old, even within otherwise similar geo-cultural patterns of family formation. Furthermore, life expectancy varies significantly across Europe. Sweden, Italy, Cyprus, Greece and Norway are among the ten countries with the highest life expectancy for men in Europe (at 76–79 years). Romania, Lithuania, Turkey, Latvia, Estonia and Ukraine are among the ten with the lowest life expectancy, ranging from 63 to 67 years. For women, Sweden and Italy are still in the first group, which also includes Spain, France, Finland and Austria (at 83–84 years). Bulgaria, Romania, Ukraine and Turkey are still together in the lowest expectancy group, at 71 to 76 years (Council of Europe 2004). Given these differences in household patterns, a different proportion of elderly people in each country, as well as of women in all countries, may find themselves living alone when they become frail. This situation puts a different kind of pressure both on family and kinship solidarity and on welfare arrangements. In the EQLS, this issue has not been dealt with in detail. There was one question asking how often the respondent took care of an elderly or disabled relative (with no distinction made between co-resident relatives and relatives living elsewhere). Provision of care for elderly persons is higher in the NMS and CC-3 than in the EU-15, and also involves to a greater degree adults of all ages, not just those over 54 years old (Saraceno and Olagnero 2004). Of the entire sample, 5 per cent assumes such an obligation daily (with more women than men taking on this responsibility), another 5 per cent at least once a week, and the remainder less often. But over half of the sample never takes care of an elderly or disabled relative. This last figure is higher than that found in Eurobarometer data (Alber and Kohler 2004).10 In her study based on European Community Household Panel (ECHP) data, Maria Iacovou (2000) found that a large proportion of elderly persons living with their children receives care within the household, particularly in the Mediterranean countries. She also found that this care is to a large degree reciprocal, particularly in the case of women: the ‘younger old’ among them, in fact, provide childcare almost to the same extent as care is provided by the household to the ‘older old’. This reciprocity does not hold for men, however. Recent research on a smaller number of countries with a specific focus on the issue of intergenerational exchanges (Kohli et al. 2005) indicates that between 40 per cent and 65 per cent of those over 80 receive some kind of help (care, but also help with shopping or with bureaucratic work) from their co-resident or non-co-resident children. Contrary to many assumptions implicit in welfare regime typologies, elderly people living in Mediterranean countries do not seem to receive more help from their family than do those living in Nordic and central European countries. Welfare state regimes possibly make a difference not so much in availability of family support as in the degree to which these exchanges and forms of support are the only, or main, available option, or even a necessity.
4. Household patterns in middle adulthood (35–64) By age 35, about 90 per cent of all Europeans live outside their parental household. The large majority lives with a partner, with or without children. There are, however, noticeable differences with respect to gender and age (see Figure 2.4). Up to 49 years of age, more men than women live alone, either because they have not yet formed a
Patterns of family living
63
30
20
10
0
−10
−20
−30
EU-15: difference between 50–64and 35–49-year-old men
EU-15: difference between 50–64and 35–49-year-old women
NMS: difference between 50–64and 35–49-year-old men
NMS: difference between 50–64- and 35–49-year-old women
CC-3: difference between 50–64and 35–49-year-old men
CC-3: difference between 50–64and 35–49-year-old women
−40 Living with parents
Living alone
Living as childless couple
Living as a couple with children
Living as a couple with or without children in extended household
Lone parent in nuclear or extended household
Other kinds of households
Figure 2.4 Households of 35– to 64-year-olds: differences between the 35–49 and the 50–64 age groups, by gender and country clusters (%) Source: EQLS 2003, own calculations Notes: The figure represents the relative differences between the household statuses of, respectively, (a) men in the 50–64 and 35–49 age groups and (b) women in the 50–64 and 35–49 age groups. The difference has been calculated as that between the % of 50–64-yearold men who have a given household status and the correspondent % in the 35–49 age group; the same calculation has been carried out for the women’s sample. Negative differences indicate how much less 50–64-year-old men (or women) are in that status compared to 35–49-year-old men (or women). Conversely, positive differences indicate how much less 35–49-year-old men (or women) are in that status compared to 50–64-year-old men (or women). The category ‘Other kind of household’ includes the respondents who live with friends, brothers or sisters (without parents) or other aggregate members.
partnership or because they have left it. More women than men are lone parents.11 Actually, these two household statuses are somewhat symmetrical, suggesting that to some degree they are gender-specific outcomes of the process of partnership dissolution, particularly when children are present. Women are more likely to remain with the children. Men are more likely to leave the household to form a new one as a single person household, at least temporarily. Compared to younger women, 35- to 49-year-old lone mothers are less likely to ‘go back’ to, or remain in, their parents’ household.
64
Chiara Saraceno
There are fewer lone mothers living in extended households in this age group. These findings are similar to those from a study on the EU-15, which was based on ECHP data and used a more restricted definition of lone parenthood (i.e. the lone parent lives only with her or his dependent children; see Lehmann and Wirtz 2004). According to this study, in 2001 lone-parent households made up 9 per cent of all households with dependent children in the EU-15, ranging from 22 per cent in Sweden and 17 per cent in the United Kingdom to 4 per cent in Italy, Portugal and Greece and 3 per cent in Spain. Over 90 per cent of lone parents were women. Only in Sweden were 26 per cent of lone parents men. Of the lone parents in the EU-15, 86 per cent were between the ages of 25 and 49. Only in Greece and Portugal was there a relatively high proportion of lone parents in the age group 50–64 years: 23 per cent and 22 per cent, respectively, compared to an average of 11 per cent. These differences in the incidence of lone parenthood imply differences in the risk that European children have of spending part of their childhood and/or adolescence with only one cohabitant parent. Furthermore, some national research reported by Linda Hantrais (2004: 59) suggests that children are likely to spend a longer period of time in a lone-parent household if their parents divorce than if one parent dies or a cohabitant unmarried partnership breaks up. In the EQLS sample, 14.2 per cent of underage children lived in a lone parent household in EU-15, 11.9 per cent in NMS and 8.4 per cent in CC-3. The highest percentages were found in the UK (26.5 per cent) and Estonia, the lowest in Malta. But the numbers are too small to allow any analysis. By the age of 50, the effect of gender differences in life expectancy and in age at partnering starts reversing the differences in household statuses found among the 35- to 49-year-olds. The pattern that is so clear among elderly people starts to emerge: now women have a slightly higher likelihood than men to live alone. They continue to have a higher likelihood to be lone parents, but their children are now likely to be young adults. Thus, although they spend much of their adult life together, women and men shift their differences over the lifecourse. Or rather, their differences change sign: men are more likely to live alone when young; women are more likely to live alone when old. This is also the age at which an increasing percentage of people, particularly women, is involved in the care of a frail elderly person, cohabitant or not. In no country cluster is there a substantial percentage of 35- to 64-year-olds who live in an extended, three-generation household. This circumstance may be a lifecourse effect. As individuals age, it is less likely that their parents are still alive, while their cohabiting children may not have yet formed a partnership. Yet, given the present high life expectancy, this explanation is at best partial. In fact, although among the 50- to 64-year-olds the percentage of those who still have a child in the household decreases dramatically, the percentage of those who are in an extended household increases only slightly. Rather, this finding seems to suggest that, even in the countries where it once was widespread, the extended household pattern has been weakened as a model among the cohorts who are at present in the middle age groups. It remains, however, as a buffer or a necessity, in two phases of household formation: when the young couple does not have enough resources to set up its own household, and when elderly persons become frail and remain alone. Moreover, although co-residence between more than two generations may not be the prevalent pattern of family organisation, exchanges and support across households within the family network may be a crucial organisational as well as emotional resource. Other comparative research has indicated that it is particularly this age group, the so-called pivot generation that
Patterns of family living
65
is most likely to give financial support and care both to the older and to the younger generations (Attias-Donfut et al. 2005; Kohli et al. 2005).
5. One-worker and two-worker couples: another intra-European differentiation in patterns of household organisation? Households are not simply a matter of who lives with whom when. Because households have to deal with needs of income and care, decisions concerning household organisation involve also decisions and arrangements concerning the allocation of paid and unpaid work, particularly within the couple (if one is present). These decisions, particularly as they affect women’s and mothers’ participation in paid work, increasingly differentiate households’ organisational patterns across Europe. As findings of the European Values Survey show (Halman 2001), the decisions certainly are influenced by cultural patterns regarding what is appropriate, particularly when there are small children in the family. But they also are contingent on welfare state and labour market arrangements. As European employment surveys indicate, there are meaningful differences across Europe not only in women’s activity rates, but also in the degree to which motherhood has a negative impact on women’s labour force participation. The negative impact of having a child under six ranges from nearly 40 percentage points in the Czech Republic and Hungary to 3 percentage points in Denmark and minus 8 percentage points in Slovenia (see Plantega and Siegel 2004). EQLS data indicate the considerable variation across Europe in the distribution of three possible paid-work arrangements within couples of working age – one worker only, two workers and no workers. Differences in this case seem to depend less on patterns of family formation than on welfare regime patterns (see Figure 2.5) and on gender models. The social democratic countries and the former socialist countries show a remarkably similar distribution. In these countries, in the large majority of working-age couples both partners work. The outlier is Poland, both because of the higher incidence of couples who do not work and because of the lower percentage of two-worker couples. In the continental countries, the majority of couples also consists of two workers, but the proportion of one-worker couples is higher than in the first two clusters. The Mediterranean countries appear to be quite differentiated. Cyprus and Portugal are similar to the continental countries, and in Spain, too, slightly over half of the couples have both partners in employment. In Italy and Greece, two-worker couples make up less than half of all couples of working age. Finally, in Malta, they represent a clear minority. Among the two countries usually identified with the liberal cluster, the United Kingdom is similar to Belgium and France, whereas Ireland is similar to Italy. With respect to the CC-3, Turkey has by far the lowest percentage of two-worker couples in all 28 European countries, as well as a relatively high percentage of couples with no workers. Bulgaria’s and Romania’s percentages of two-worker couples are similar to those of the continental countries. On the other hand, Romania has a higher percentage of couples with no workers than Poland does. Two-worker couples may, however, combine quite different time schedules. European labour force surveys indicate that across Europe, on average, women work shorter days than men do, and that part-time work is largely women’s work, either
66
Chiara Saraceno CZ 2
19
79
DK
6
17
FI
5
19
76
20
75
77
SK
4
SI
4
23
SE
3
25
EE
3
72
28
AT 3
70
29
HU
6
68 27
FR 2
67
33
LV PT
73
5
64 31
3
64
33
LT
9
64 28
64
UK
7
30
63
BE
7
31
63
CY 1
38
BG
6
LU 2
60
38
DE NL
61 33
60
9
34
5
58
37
ES 2
58 47
RO
51
18
32
IT
5
48
IE
6
48
PL EL
CC-3 0%
47 42
4
45
53
42
61
TR
NMS
47
13
MT 1
EU-15
50
38
17
70
5
13
37 9
58 33
58
16
59
20%
40% No workers
25
60% Single worker
80%
100%
Two workers
Figure 2.5 Distribution of working–age couples by working status in Europe Source: EQLS 2003, own calculations
as an option during a certain life stage (e.g. in the early years of motherhood) or as a permanent condition. However, given the different incidence and availability of parttime work across Europe, couples with ‘one and one-half workers’ (Crompton 2006, Lewis 2003, Pfau-Effinger 1998, 2004) predominate in the continental countries and the United Kingdom, particularly when they have young children (Bielenski et al. 2002, Franco and Winqvist 2002, Plantega and Siegel 2004). They also make up a substantial proportion in the Nordic countries. Two full-time workers is the prevalent pattern in most of the former socialist countries. It is the prevailing pattern in the Mediterranean countries as well – when the couple has two earners – even though there is little support from the welfare state. This lack of support explains the relatively low, though increasing, proportion of two-earner households in the Mediterranean countries. Household organisation is affected not only by the number of paid workers in the couple and by their paid working-time schedule, but also by the needs for care and by the gender division of unpaid family work (Saraceno 2005). With regard to the former, in most countries the presence of children, particularly when they are very
Patterns of family living
67
young, has a negative impact both on women’s activity rates and on their working time, but to a quite different degree across countries (e.g. Plantega and Siegel 2004). Parenthood impacts negatively on women’s participation far more in new member states than the EU-15: the Czech Republic, Hungary, and Slovakia are the three countries where women are most affected by parenthood. On the contrary the impact is non-existent in Sweden and minimal in Denmark. With regard to the gender division of labour, comparative time-use data on a number of European countries (Eurostat 2004; Sabbadini and Romano 2006) indicate that unpaid family work remains mainly women’s responsibility (see Table 2.4). When women are in paid work, however, they reduce their time devoted to family work, whereas their partners increase their share of it. In most countries studied, the overall (paid and unpaid) workingtime load is higher for working women than for working men, although men on average work more hours for pay than women do. This is not the case, however, in Sweden, Norway and the United Kingdom, although the balance between paid and unpaid work may differ between men and women. Estonian, Slovenian and Italian working women have the longest working day. Italian men have the longest paid working day and are the least collaborative in unpaid family work. The EQLS data are less precise. But they give information on a larger number of countries. Generally, they confirm the picture presented above. In particular, they suggest, on the one hand, that both the former socialist and the Nordic countries are the most gender equal with regard to the overall workload, particularly in two-worker couples. But whereas in the Nordic countries the overall workload for both men and women is among the shortest in Europe, in the former socialist countries it results in a very long working day for both. The Mediterranean countries appear to be the most unbalanced with respect to gender, both when households are based on the male breadwinner model and when they have two earners. Even when women work full time, the gender division of labour in family work remains skewed (Saraceno et al. 2005).
6. Conclusions Households in Europe are differentiated in terms of who lives with whom when, timing and sequence of events (e.g. cohabitation, marriage, fertility, exit from the parental household and so forth). During the twentieth century there has been a noticeable convergence with regard to family structures, in so far as neo-locality prevails at present in the formation of households throughout Europe to a much greater extent than in the past. In addition, the widespread downward fertility trend has homogenised countries and social groups which previously had quite distinct patterns. Yet the timing of exit from the parental household and the sequencing of cohabitation, marriage and parenthood distinguish different clusters of countries that to some degree overlap with the north/south and east/west divides. This, in turn, has an impact on the average household size (greater in southern and eastern Europe) and on the likelihood that a substantial proportion of elderly people (particularly elderly women) live alone. At the same time, significant cross-country differences in marital instability rates and birthrates both out of wedlock and out of cohabitation shape different household and life course patterns for children. There are also noticeable differences in the patterns of households’ time organisation and of dealing with needs for income and care. One of the main changes in
8:01 2:35 5:03 2:15 1:43 4:23 24:00
Men Sleep Meals and self-care Paid work and/or study Unpaid family work Travel Free time/unspecified time Total 8:22 2:11 5:05 2:20 1:20 4:47 24:00
8:23 2:36 4:13 4:04 1:15 3:51 24:00
Estonia
8:12 1:55 5:32 1:59 1:17 5:06 24:00
8:22 2:02 4:20 3:21 1:16 4:38 24:00
Finland
8:24 2:58 5:44 1:53 1:10 3:51 24:00
8:38 2:57 4:32 3:40 1:05 3:08 24:00
France
8:00 2:58 5:05 1:52 1:31 3:51 24:00
8:11 2:57 4:32 3:40 1:05 3:08 24:00
Germany
8:08 2:30 5:25 2:09 1:10 4:37 24:00
8:18 2:21 4:43 3:54 1:02 3:43 24:00
Hungary
7:58 2:52 6:13 1:10 1:40 4:06 24:00
8:00 2:44 4:39 3:51 1:28 3:18 24:00
Italy
7:53 1:58 4:56 2:12 1:23 5:37 24:00
8:07 2:02 3:46 3:26 1:17 5:22 24:00
Norway
8:06 2:07 5:20 2:24 1:14 4:52 24:00
8:12 2:02 4:23 4:24 1:09 3:51 24:00
7:52 2:05 5:17 2:23 1:32 4:51 24:00
8:05 2:23 4:05 3:32 1:28 4:27 24:00
Slovenia Sweden
Sources: Eurostat 2004; for Italy: Sabbadini and Romano 2006. The years when the data were collected differ. For example, for France, they refer to 1998; for Italy, to 2003
8:16 2:36 3:53 3:52 1:30 3:51 24:00
Women Sleep Meals and self-care Paid work and/or study Unpaid family work Travel Free time/unspecified time Total
Belgium
Table 2.4 Use of time among working Europeans (hours and minutes)
8:11 1:55 5:42 1:54 1:36 4:41 24:00
8:25 2:07 3:46 3:26 1:17 4:21 24:00
UK
Patterns of family living
69
family relationships over the past half century – women’s emancipation through education and labour force participation – appears to have a differential impact on household organisation across Europe depending both on previous arrangements and cultural models and on the ability of a society to adapt to this change. In the Nordic countries, the two-worker arrangement predominates, but when children are very young a good share of women chooses part time. Couples with one and one-half workers prevail in the continental countries and the United Kingdom, particularly when they have young children. Two full-time workers is the prevalent pattern in most of the former socialist countries as well as in the Mediterranean countries – when the couple has two earners. In the Mediterranean countries, however, this arrangement still pertains to just a minority of all working-age couples; moreover, the skewed gender division of labour remains unfavourable for women. Particularly in the Mediterranean and the former socialist countries, households headed by adults in the middle age-groups may find themselves under considerable time pressure due to long working days and responsibilities to provide care that are little supported by the welfare state. Extended family solidarity may buffer these strains. But the declining fertility rate is creating a demographic context in which there will be an increasing care deficit for the older generations to come. In the former socialist countries, this care deficit might even be intensified by emigration. Furthermore, somewhat paradoxically, eastern European women as migrant care-workers in the EU-15, particularly in Mediterranean countries, are becoming the private market solution in welfare regimes that deal inadequately with the care deficit emerging from the interplay between population ageing and the increase in women’s labour force participation (Da Roit and Sabatinelli 2005). Yet the women filling these positions may open up this same deficit in their own countries. The ‘global care chain’, as it has been defined (Anderson 2000), might produce unbalances not only in individual and family lives, but also in demographic and social structures. In any case, the brief story that has been told in these pages suggests that in family arrangements both differences and similarities do not always overlap with political and institutional divisions, even if these may crystallise them. It also suggests that in this field convergence is a somewhat illusionary (or moving) target. Families and societies interact with each other. The way the latter have adapted to challenges posed by changes in the former has been and continues to be as diverse across Europe as family forms themselves.
Notes 1 For instance, some crucial questions (e.g. whether a couple was married or cohabitant; whether a lone mother previously had been married) were not included in the questionnaire. 2 Exactly which country belongs to which part of the divide is controversial. For instance, Hajnal did not include Hungary in the eastern European model, but, according to Tomka (2002), Hungarian family arrangements bear a great similarity to aspects of this model, particularly with regard to age at marriage. According to De Vos and Sandefur (2002), the same holds for Slovakia, whereas Estonia is more similar to the Nordic pattern and the Czech Republic to the west-central European one. 3 Only 30 per cent of all households across Europe include at least an underage child. 4 In Bulgaria 32 per cent of all underage children live in an extended household, compared to slightly over 1 per cent in Denmark or the UK. In Hungary, Slovakia, Slovenia and Poland the figure is over 15 per cent.
70
Chiara Saraceno
5 A study on lone parents in the EU-15, which was based on ECHP data (Lehmann and Wirtz 2004), found that Ireland was the country with the highest incidence of lone mothers in the youngest age bracket (16–24): 11 per cent compared to the average of 2 per cent. 6 In order to identify these patterns, an agglomerative hierarchical cluster analysis was used. With this procedure, homogenous cases are grouped in order to form clusters. The ‘complete linkage’ method was implemented. In this method, the distance between two clusters is calculated between their two farthest points (Lance and Williams 1967a, 1967b). In the sample of young women, these patterns are not clearly identifiable. 7 The data on Greece are somewhat surprising. It is the only case in which the EQLS data differ substantially from other sources that usually put Greece in the same group as the other Mediterranean countries. 8 Given the small number of these situations, for the purpose of this analysis I have grouped together all childless children without a partner who are living in the parental household, without distinguishing between those living in nuclear and those living in extended households. 9 In Italy, for instance, children of better-off households leave the parental household later than children of low-income households. So, too, do only children in comparison with children with siblings. 10 Cross-survey differences in findings on this issue may depend on different factors: age composition of the sample, wording of the question, and so forth. The issue of comparability in this field has been addressed in a review of European studies by Jacobs et al. (2005). 11 For the purpose of this analysis of household status, ‘lone parent’ includes all adults living with at least one child and no partner, irrespective of the child’s age.
References Alber, J. and Kohler, U. (2004) Health and Care in an Enlarged Europe, European Foundation for the Improvement of Living and Working Conditions, Luxembourg, Office for Official Publications of the European Communities. Anderson, B. (2000) Doing the Dirty Work? The Global Politics of Domestic Labour, London, New York: Zed Books. Attias-Donfut, C., Ogg, J. and Wolff, F.-C. (2005) ‘European patterns of intergenerational and time transfers’, European Journal on Aging, 2, 3: 161-173. Barbagli, M. and Kertzer, D. (eds) (2003) ‘Introduction’, pp. xi–xliv, in M. Barbagli, and D. Kertzer (eds), Family Life in the Twentieth Century, New Haven: Yale University Press. Barbagli, M., Castiglioni, M. and Della Zuanna, G. (2003) Fare famiglia in Italia, Bologna: il Mulino. Bielenski, H., Bosch, G. and Wagner, A. (2002) Working Time Preferences in Sixteen European Countries, Luxembourg: Office for Official Publications of the European Communities. Billari, F. and Wilson, C. (2001) Convergence towards Diversity? Cohort Dynamics in the Transition to Adulthood in Contemporary Western Europe, MPIDR Working Paper WP 39, December 2001, Rostock: Max Planck Institute for Demographic Research. Cavalli, A. and Galland, O. (1996) Youth in Europe, London: Pinter. Council of Europe (2001) Recent Demographic Developments in Europe 2001, Strasbourg: Council of Europe Publications. Council of Europe (2004), Recent Demographic Developments in Europe 2004, Strasbourg: Council of Europe Publishing. Crompton, R. (2006) Employment and the Family, Cambridge: Cambridge University Press. Da Roit, B. and Sabatinelli, S. (2005) ‘Il modello mediterraneo di welfare tra famiglia e mercato’, Stato e mercato, 74, 2: 267-290. De Vos, S. and Sandefur, G. (2002) ‘Elderly living arrangements in Bulgaria, the Czech Republic, Estonia, Finland and Romania’, European Journal of Population, 18: 21-38.
Patterns of family living
71
European Commission (2003) The Social Situation in the European Union, Luxembourg: Office for Official Publications of the European Communities. Eurostat (2004) How Europeans Spend Their Time: Everyday Life of Men and Women. Data 1998–2002, Luxembourg: Office for Official Publications of the European Communities. Fahey, T. and Spéder, Z. (2004) Fertility and Family Issues in an Enlarged Europe, Report to the European Foundation for the Improvement of Living and Working Conditions, Luxembourg: Office for Official Publications of the European Communities. Franco, A. and Winqvist, K. (2002) ‘Women and men reconciling work and family life’, Statistics in Focus: Population and Social Conditions, Theme 3-9/2002. Hajnal, J. (1982) ‘Two kinds of preindustrial Household Formation System’, Population and Development Review, 8, 3: 449–494. Halman, L. (2001) The European Value Study: A Third Wave. Sourcebook, EVS/WORK, Tilburg: Tilburg University. Hantrais, L. (2004) Family Policy Matters, Bristol: Policy Press. Iacovou, M. (2000) Living Arrangements of Elderly Europeans, Working Paper, Institute for Social and Economic Research, Colchester: University of Essex. Jacobs, T., Lodewijckx, E., Craeynest, K., De Koker, B. and Vanbrabant, A. (2005) ‘Mesurer l’aide informelle: synthèse des pratiques européennes et nouvelle proposition’, retraite et société, 46: 60–89. Kohli, M., Künemund, H. and Vogel, C. (2005) ‘Intergenerational transfers in Europe: a comparative overview’, paper presented at the ESA Conference, September 2005, Torun. Lance, G.N. and Williams, W.T. (1967a) ‘A general theory of classificatory sorting strategies I: hierarchical systems’, Computer Journal, 9, 4: 373-380. Lance, G.N. and Williams, W.T. (1967b) ‘A general theory of classificatory sorting strategies II: clustering systems’, Computer Journal, 10, 3: 271-277. Lehmann, P. and Wirtz, C. (2004) ‘Household formation in the EU: lone parents’, Statistics in Focus, Population and Social Conditions, Theme 3, 5/2004. Le Play, F. (1875) L’Organisation de la famille: selon le vrai modèle signalé par l’histoire de toute les races et de tous les temps, by M.F. Le Play, 2nd edn, revd and corrected, Tours: Alfred Mame, libraires-éditeurs, Dentu, Libraire. Lewis, J. (2003) Should We Worry about Family Change?, Toronto: University of Toronto Press. Ogg, J. and Renaut, S. (2005) ‘Le soutien familial intergénérationnel dans l’Europe élargie’, retraite et société, 46: 30–59. Pfau-Effinger, B. (1998) ‘Arbeitsmarkt- und Familiendynamik in Europa. Theoretische Grundlagen der vergleichenden Analyse’, pp. 177–194, in B. Geissler, F. Maier and B. PfauEffinger (eds), Frauen ArbeitsMarkt. Der Beitrag der Frauenforschung zur sozioökonomischen Theorieentwicklung, Berlin: edition sigma. Pfau Effinger, B. (2004) Development of Culture, Welfare States and Women’s Employment in Europe, Aldershot, Burlington: Ashgate. Pinnelli, A., Hoffman-Nowotny, H.J. and B. Fux (2001) Fertility and New Types of Household Formation in Europe, Population Studies, 35, Strasbourg: Council of Europe Publishing. Plantenga, J. and Siegel, M. (2004) ‘Childcare in a changing world’, Position paper, part I, European Childcare strategies, conference ‘Childcare in a changing world’, Groningen, 23–24 October. Online: Available http: (accessed 11 December 2006). Reher, D. (1998) ‘Family ties in Western Europe: persistent contrasts’, Population and Development Review, 24: 203–234. Sabbadini, L.L. and Romano, C. (2006) ‘Principali trasformazioni dell’uso del tempo in Italia’, paper presented at the conference ‘Andare a tempo’, Turin, 20–21, January.
72
Chiara Saraceno
Saraceno, C. (2005) ‘Family work systems in Europe’, pp. 57–84, in J. Alber and W. Merkel (eds), Europas Osterweiterung: Das Ende der Vertiefung? WZB Jahrbuch 2005, Berlin: edition sigma. Saraceno, C. and Olagnero, M. (2004) ‘Households and families’, pp. 33–44, in European Foundation for the Improvement of Living and Working Conditions, The Quality of Life in Europe, Luxembourg: Office for Official Publications of the European Communities. Saraceno, C., Olagnero, M. and Torrioni, P. (2005) First European Quality of Life Survey: Families, Work and Social Networks, European Foundation for the Improvement of Living and Working Conditions, Luxembourg: Office for Official Publication of the European Communities. Schizzerotto, A. and Lucchini, M. (2004) ‘Transitions to Adulthood’, pp. 46–68, in R. Berthoud and M. Iacovou (eds), Social Europe, Cheltenham: Edward Elgar. Therborn, G. (2004) Between Sex and Power, London: Routledge. Tomka, B. (2002) ‘Demographic Diversity and Convergence in Europe, 1918–1990: The Hungarian case’, Demographic Research, 6, 2. Online. Available Http: (accessed 11 December 2006). United Nations Economic Commission for Europe, Family and Fertility Surveys (2002) Economic Survey of Europe 2002: No. 1, New York and Geneva: United Nations.
3
Is there a generational cleavage in Europe? Age-specific perceptions of elderly care and of the pension system Wolfgang Keck and Agnes Blome
Introduction1 All societies have to solve two basic problems concerning old age: how to provide incomes for elderly people who no longer draw earnings from work, and how to provide care for older people who are disabled and no longer self-sufficient. During the post-war period of prosperity European societies could for many years rely on a rather well-established division of labour between the generations. The population at working age financed fairly generous pay-as-you-go pension schemes, as a combination of full employment, high birthrates, and sizeable productivity gains helped to ease the burden of financing and allowed the raising of pensions to levels which considerably alleviated the problem of poverty in old age.2 Care needs were traditionally covered within the extended family system where a fairly large generation of adult and usually female care workers provided for a relatively small generation of elderly parents. A number of social transformations have put the viability of these arrangements increasingly into question. Shrinking fertility and extended unemployment lead to stagnant or even decreasing numbers of economically active contributors to social security systems despite growing female labour force participation, whilst the growing share of the elderly population and the prolonged life expectancy combine to boost the number of people drawing social benefits. Hence social security systems are faced with growing imbalances between the numbers of contributors and beneficiaries. The demographic changes are particularly incisive. Once the net reproduction rate shrinks below factor 1, each generation of contribution payers in pension schemes will – other things being equal – be smaller than the generation of pension recipients (Breyer 1990). Under these conditions a steady level of pensions can only be bought at the price of successive increases in contribution rates. Alternatively, the contribution rate may remain constant only at the price that each successive generation of pensioners will draw less generous pensions. Various models have been proposed to distribute the burden more fairly between the groups of contributors and
74
Wolfgang Keck and Agnes Blome
beneficiaries,3 but the problem remains that under conditions of shrinking populations existing combinations of pensions and contributions will not be sustainable. The problem is even exacerbated as shrinking levels of fertility combine with increasing life-expectancy. Pensions which were originally designed to cover merely a few years after retirement from work now have to be paid for close to two decades.4 The demographic changes also affect the caring relationship. Dependency upon care increases steeply at very advanced stages in the lifecycle, i.e. beyond ages 75, and especially beyond age 85.5 The proportion of these ‘very old’ age cohorts grows even faster than the proportion of the population aged 65+.6 Hence there will be a growing demand for care. This growing demand meets with a shrinking supply of carers in private households, because growing female labour force participation – and in the long run decreasing fertility – lead to a shrinking supply of adult daughters who traditionally used to shoulder the care work in private households.7 If the growing care needs of people at very advanced ages are to be met, the shrinking caretaker potential in private households calls for a public provision of care services, but this would further increase the tax squeeze for the population at working age. The new macro-constellation has put traditional patterns of generational relations under strain, and there has been growing concern that under the new conditions European societies might face growing generational tensions if not an outright generational conflict (see Arber and Attias-Donfut 2000 for a good overview of respective debates). From a social policy perspective, several authors have suggested considering welfare states as stratifying systems which divide the population into groups with different obligations and entitlements (Lepsius 1979; Alber 1984; Esping-Andersen 1990), and Peter Baldwin (1990) proposed the term ‘risk-categories’ to distinguish between groups for which the welfare state provides different cost/benefit ratios. From this perspective, those who finance public transfers and services and those who benefit from them form two distinct risk categories with discrepant interests. The more welfare states are targeted on needy persons, the less overlap there is between the payers and recipients of public benefits, and the more clashes of interest between these two groups may be expected. In the case of generations, however, each generation appears on both sides of the structural divide during the life-course, paying contributions and rendering care services in an earlier phase of the life-cycle, but receiving benefits and care at a later stage. In this sense, the positions in various riskcategories are not fixed but fluid. Nevertheless, different generations will probably end up with different generation-specific cost–benefit ratios as currently younger age cohorts have to expect lower benefits relative to their contributions than the current generation of pensioners (Auerbach et al. 1998, European Commission 1999). However, the fact that each person will hold positions on either side of the structural divide should considerably lower the potential for generational conflicts. The potential for generational conflicts is further lowered by the fact that family ties cut across the structural cleavage. In addition to knitting emotional bonds between parents and children, families also serve as systems of mutual support where considerable resources are transferred between the generations.8 Hence the amount of tensions between the generations is structurally under-determined and must be empirically researched rather than theoretically deducted. The purpose of this chapter is to investigate the degree to which there is an agespecific polarisation in the perception of public benefit systems in European societies. The basic interest is to describe how much polarisation between age groups there is
Is there a generational cleavage?
75
in the assessment of elderly care and of pension systems, and to analyse to what extent such age-specific discrepancies reflect country-specific welfare state arrangements.
1. Hypotheses and data The analyses depart from a number of hypotheses concerning the age-specific distribution of perceptions. The basic assumption is that younger and older age-cohorts form two distinct risk-categories with different positions concerning the distribution of the costs and benefits of public schemes: younger persons are assumed to answer questions regarding old-age risks from the perspective of the financers of social security provisions, older persons are expected to represent the views of the beneficiaries. With respect to the provision of care, one would expect that younger persons advocate shifting the burden of care from families – i.e. from themselves – to the state or to another third party. In particular, individuals in the medium age group who have elderly parents and therefore have to deal actually or at least potentially with the care issue should opt for public care responsibilities. In contrast, one would expect that older people tend to be more in favour of family solidarity. With respect to financing responsibilities, the basic assumption is that each group will have a preference for shifting the burden of financing to others or at least to public authorities. Since women are the predominant providers of care, they should advocate state schemes – which would help to unburden them – more frequently than men. The position in a risk-category on either side of the care relationship might be mediated by two intervening macro-conditions, i.e. welfare state arrangements and the strength of cultural norms regarding family care (Milar and Warman 1996). We expect age-specific preferences regarding care responsibilities to come more to the fore the less developed public services are – because under these conditions the age-specific division into givers and recipients of care is felt most incisively on the personal level. However, cultural norms stressing family solidarity may cushion the effect of interests tied to a specific risk-category, because the emphasis on family obligations has a dual effect. On the one side, countries following the subsidiarity principle and holding families responsible for the delivery of care, impede the growth of public provisions and thus burden families with care obligations. On the other side, to the extent that such policy is rooted in religious values, they also provide a moral underpinning of family responsibilities which strengthen the normative commitment to family care. This would imply that countries with a strong normative emphasis on family responsibilities will be characterised by lower degrees of age-specific polarisation in the assessment of care responsibilities due to the diffusion of respective norms. With respect to the assessment of pension schemes, older people can be expected to have a positive assessment of current pension arrangements because they have a more favourable balance of benefits and contributions than the younger generation. On the other hand older people might also develop a more negative image of pension schemes because, given that their standard of living crucially depends on pensions they are most painfully affected by cutbacks. In contrast younger generations are less immediately affected by reforms and might hence perceive the pension system in a less negative vein. On the macro-level we would expect countries with more universal, less fragmented, and more sustainable pension schemes to produce smaller degrees of age-specific polarisation in the assessment of public pensions. Finally, we assume that
76
Wolfgang Keck and Agnes Blome
negative assessments of pension arrangements provide a basis for the perception of a generational cleavage. This implies that age-specific differences in the perception of pension schemes should translate into corresponding differences in the perception of generational cleavages, and, secondly, that countries with more negative assessments of their pension schemes should prove to have higher potentials for generational tensions. The data we use to test these hypotheses stem from two different data sources: a compiled version of several Eurobarometer surveys9 which we refer to here as the harmonised candidate countries Eurobarometer 1998–2002 and the European quality of-life survey (for survey and sample details see Kohler, this volume, Ch. 17). In the Eurobarometer respondents were asked what kind of care arrangement they would prefer and who should pay for care provisions. In addition, the data provide information whether people actually provide care, and if so to whom and where. This allows us to compare professed preferences with actual behaviour. The European Quality of Life Survey (EQLS) contained two questions where respondents were asked to evaluate their national pension scheme. The first one inquired whether people tend to trust the state pension system. The second one asked to rate the quality of the state pension system on a ten-point scale from low to high quality. Moreover, the EQLS inquired to what extent individuals think that there are tensions between young and old people in society. This allows us to analyse the impact which assessments of the social security system have on perceived generational conflict.10 Our analyses are based on the distinction of three age groups. The youngest age group ranges from 15 years (18 for the EQLS) up to 34 years. This group represents a broad transition phase from education to the labour market, from the parental home to the formation of an own household as well as towards family formation. The second age group ranges from 35 to 59 years. This age category includes people exposed to the most intensive demands required both by family formation processes and by professional obligations. The older among them are likely to have elderly parents. In the post-communist countries, however, this group has also been affected most strongly by the transformation after the fall of the iron curtain, finding it difficult to keep or get jobs in the labour market. Thus, the living conditions for this age group might differ considerably between old and new member states. Including people aged 60 and over, the third group consists mostly of retirees, because age 60 is more or less the average retirement age in most European countries. The analysis proceeds as follows. First, we will analyse to what extent preferences regarding elderly care differ by age and gender. Here we first look at the preferred care arrangements (section 3.1) and then at policy choices regarding the financing of care (section 3.2). In a second step we look at age-specific similarities and differences in the assessment of national pension schemes (section 4). Finally we examine to what extent age-specific differences in the perception of pension schemes translate into the perception of generational cleavages.
2. Age-specific care preferences As pointed out above, the preferences which people profess with respect to care are presumably a function of their values or normative commitments as well as of their interests. A rational choice perspective would suggest that people express preferences which are above all shaped by their self-interests. Since persons at different stages in
Is there a generational cleavage?
77
the life course are likely to be in different positions as care givers or care receivers we expect age-specific preferences to reflect these different positions. Finding themselves frequently in the position of care recipients, members of the older generation should have a predominant interest in getting high quality care at a low price which is typically rendered by family members.11 Younger generations, on the other side, who are usually in the position of care providers, should have an interest in being relieved from caring obligations and in distributing the cost of care widely among all generations so that the cost does not exclusively impinge upon them. This logic should apply above all to those in the middle age brackets who already have elderly parents so that we expect this group to be most in favour of public care solutions which promise to unburden them. Younger people may similarly be expected to prefer public care solutions which are predominantly financed by the elderly themselves. Apart from different interests reflecting the position in care arrangements, care preferences are presumably also shaped by the interplay between cultural norms regarding family life and by the prevailing welfare state arrangements. Cultural norms may stipulate the family – i.e. women – as ‘natural’ providers for care and support, and an absence of public care services may constrain family members to fulfil otherwise orphaned care obligations. With respect to the impact of cultural norms, we expect cross-national differences in care preferences especially between countries with an emphasis on family obligations and an absence of state-provided services on the one hand, and countries in which the state takes over family responsibilities on the other. Here we assume a gradient between Southern and Northern European countries. We would expect generational differences in care preferences to be most pronounced in countries which neither have well-developed public care services nor a strong cultural emphasis on family obligations. This group should include continental welfare states as well as new member states. The following sections analyse to what extent care preferences differ between age groups and whether cultural norms and welfare state arrangements have an impact on professed attitudes. 2.1 Preferred care arrangements The CC-Eurobarometer asked respondents what arrangement they would prefer in case their elderly parents could not manage to live on their own. Five options were offered: (1) the respondent or one of his siblings should invite the parent to live with them; (2) one of the children should move in with the parent; (3) the parent and children should move closer together; (4) the parent should move into a residential care facility; (5) the parent should stay at home assisted by domestic services. We classified these answers according to whether the respondents preferred a family solution. A preference for a family solution was assumed if respondents advocated any of the first three possibilities. The answers to such questions reveal attitudes or general preferences rather than disclosing what respondents actually do in case a frail elderly relative needs care. The harmonised CC-Eurobarometer dataset, does, however, provide an opportunity to analyse the relationship between professed care preferences and actual caring behaviour, since questions regarding both aspects were asked in the same survey. The result is that people actively involved in care giving to elderly people have positive rather than negative thoughts about extended family responsibilities for care (Alber and Kohler 2004: 71). Since the respective questions were asked in different
78
Wolfgang Keck and Agnes Blome
% preferring family care for own parents
100 90
GR
80
ES RO
PT
70 IT
60
IE
50
LU
40
PL LV
HU
SI
EE
BG
LT
DE
SK AT
UK
TR
R2 = 0.6429
CZ
FR CY MT
BE
30 20
DK
NL
FI
SE
10 0 0
2
4 6 8 10 % giving care to a person over 60 inside household
12
14
Figure 3.1 Share of individuals who care for elderly persons within the household and preference of family members as care givers for own parents Source: Harmonised CC Eurobarometer data set 1998–2002
Eurobarometer surveys in the EU-15, such a micro-level analysis is limited to the new member states. A macro-level comparison which is possible for all nations shows that countries with higher proportions of active care givers within private households also have higher proportions of respondents who advocate family solutions for care (Figure 3.1). Hence on each level there is evidence that people’s professed care preferences are not merely a lip service to a general ideal but tend to reflect the actual care behaviour.12 This implies that the professed preferences are a reliable indicator of the kind of solution for care problems which Europeans from different societies consider appropriate. Figure 3.2 shows how widespread support for family care for the elderly is in each European country as well as in the various country group aggregates. National means are given for the three age groups. Comparing the country group aggregates first, the pro-family attitude is most prevalent in the candidate countries13 and least often reported in the old EU member states. In the NMS-10, which are in between those two groups, the division between the former socialist countries and Cyprus and Malta stands out. More than 50 per cent of the respondents in the post-communist transformation countries, but only around 35–40 per cent in the two Southern islands opt for the family solution. The differences between the various country aggregates are rather small compared to the large variation within the EU-15. Here, support for the family solution ranges from 16 per cent in Sweden to 89 per cent in Greece. The contrast between the Nordic states and the Netherlands with low fractions of family support on the one side and the Southern European countries with very widespread advocacy of family obligations is particularly striking and much more conspicuous than differences between old and new EU member states.
Is there a generational cleavage?
79
Sweden (1) Denmark (1) Netherlands (1) Finland France Belgium Luxembourg UK
Austria (1, 2) Germany (1, 2) Ireland Italy
Portugal Spain Greece (1) Cyprus (1) Malta Slovenia Czech Rep. Estonia Slovakia Lithuania Hungary Latvia Poland Bulgaria Turkey Romania (2) EU-15 NMS CC-3 EU-25 EU-28 0
10
20
30
40
50
60
70
80
90
100
% opt for family solution Age:
15–34 years
35–59 years
60 years and older
Figure 3.2 Share of individuals in different age groups who opt for family support as the preferred care solution by country Source: Harmonised CC-EB 1999–2002 (CC-EB 2002.1, Q20 EB 50.1 1999. Q36): ‘Let’s suppose you had an elderly father or mother who lived alone. What do you think would be best if this parent could no longer manage to live on his/her own?’ Notes: 1 Youngest age group differ significantly from the oldest age group (p<0.05) controlling for sex, income, household size. 2 Medium age group differ significantly from the oldest age group (p<0.05) controlling for sex, income, household size.
80
Wolfgang Keck and Agnes Blome
With respect to age-specific preferences we expected younger people to prefer public care services and the elderly to be in favour of family solutions. Figure 3.2 shows, however, that there are only four countries where elderly persons are significantly more in favour of family solutions (Austria, Germany, Greece and Romania). In all other countries there is either no generational cleavage at all, or the younger age-cohorts express a preference for family solutions even more frequently. In Denmark, Sweden and the Netherlands as well as in Cyprus the youngest age cohort clearly prefers family solidarity for care. Moreover, the middle age group does not stand out. Only in Austria, Germany and Romania do people aged 35 to 59 advocate family care less often than elderly people, and in less than half of the countries we find the middle age group to be less family centred than the youngest one. Hence, contrary to our expectations there is no evidence of a generational cleavage which reflects self-interested care preferences. In general younger people accept the idea of family obligations for elderly parents in need of care just as much as the older generation does. Country-specific differences run also counter to our expectations. We had hypothesised that cultural norms in favour of family obligations would mitigate the generational cleavage. The evidence shows, however, that countries with the strongest preference for family care solutions (Spain, Portugal and Greece) are also marked by more visible divisions between older and younger people. We had furthermore expected that generational differences in the advocacy of family solutions would be least marked in countries with well established public care services. In contrast to this expectation, Denmark, Sweden and the Netherlands which are known for their welldeveloped public care services stand out for a certain backlash among the younger generation against the prevailing model of public care. In all three countries younger people advocate family solutions for the problem of care even more frequently than the older generation while still being less family oriented than their age peers in other countries.14 There are several possible reasons why the empirical results are not in line with our theoretical expectations. One is that age groups can only serve as crude approximations for the roles of care providers and care recipients in which we are theoretically interested. Since the burden of care work is predominantly borne by women, and since due to their longer life expectancy women cannot expect to receive the same family care support as their male partners, a comparison by gender may serve as another approximation of different caring roles. Here we would expect that women as the potential providers of care advocate public solutions which help to unburden them more frequently, whereas men as the likely recipients of care should be more in favour of family solutions. Once again the data do not bear out the idea that care preferences predominantly reflect self-interested dispositions on the part of groups with different caring roles. Even though female partners and daughters carry the main burden of care responsibilities, women are more frequently in favour of a family care solution for dependent elderly parents than men (Figure 3.3). The gender differences are not very distinct. Only in Turkey and Estonia differences between men and women turn out to be significant if other factors are controlled. In addition, gender differences do not cluster according to any particular families of nations typology, but in old and new member states alike they tend to be in the same direction. In 20 of the 28 countries female respondents express a preference for family care more frequently than men. Only in
Is there a generational cleavage? Malta
6.5
Turkey*
6.2
Bulgaria
5.3 0.6
Sweden
0.5
Hungary United Kingdom
0.1 0.1
Netherlands
0.0
Latvia
−0.3
Poland
−0.8
Ireland
−1.4
Romania
−1.5
Italy
−1.6
Spain
−1.7
Portugal
−2.7
Slovenia
−2.9
Denmark
−4.0
Greece
−5.1
Czech Republic
−5.5
Austria
−6.1
Germany
−6.3
France
−6.8
Belgium
−7.3
Cyprus
−7.5
Luxembourg
−7.6
Slovakia
−8.5
Finland
−9.3
Estonia*
−9.5
Lithuania −3.7
EU-15
−3.3
NMS CC-3
4.0 −3.8
EU-25 −2.6
−10.0 −8.0
81
−6.0
−4.0
EU-28 −2.0
0.0
2.0
4.0
6.0
8.0
10.0
Figure 3.3 Difference between men and women according to their preference for family care (in % points) Source: Harmonised CC-EB 1999-2002 (CC-EB 2002.1, Q20 EB 50.1 1999. Q36): ‘Let’s suppose you had an elderly father or mother who lived alone. What do you think would be best if this parent could no longer manage to live on his/her own?’ Note: *Women differ significantly from men (p < 0.05) controlling for age, income, and household size.
82
Wolfgang Keck and Agnes Blome
% mentioned care should be provided by family members
the three less modernised countries which did not become members in 2004 (CC-3) do we find women to opt more frequently for public care support than men, but with more than 70 per cent the overall level of those who prefer family care in these countries is still very high. Our data thus indicate a remarkable degree of altruism – or acceptance of selfresponsibility – in care preferences rather than a strong element of self-interested response behaviour. Both groups of potential care providers, younger people as well as women, do not reject care obligations for family members more frequently than elderly people who are the prime recipients of care. In contrast, in several countries the young are even more in favour of moving closer together to solve the care problem of elderly parents, and women tend to prefer family care more often than men. Concerning the influence of welfare state regimes we have argued that countries with well-developed care facilities should have a weaker adhesion to family care solutions, because public services allow elderly people to manage their lives independently of family support. The data on social expenditures published by the European Data Service allow us to construct a proxy measure of the expansion of public services for the elderly, i.e. the per capita benefits in kind spent for the elderly population (expressed as per cent of per capita GDP).15 By this measure, Sweden, Ireland, Denmark, and France have the most extended public services for the elderly in Europe (in that order).16 As Figure 3.4 shows, countries with higher expenditure for
100 90
GR ES
80
PL PT
LV
70
LT
IT
HU
AT CZ
EE
50
LU
UK SI
BE
40
IR
SK
DE
60
FR
MT
2 R = 0.3826
30 FI
NL
20
DK SE
10 0 0
10
20
30
40
50
60
70
Public expenditure for benefits in kind (old-age, invalidity, health) per capita of 65+ population as a share of GDP per capita
Figure 3.4 Family orientation in care issues and public care efforts, by country Source: Harmonised CC-EB 1999-2002 (CC-EB 2002.1, Q20 EB 50.1 1999. Q36), European Data Service 2005 Note: Bulgaria, Romania, Cyprus and Turkey excluded because of missing data.
Is there a generational cleavage?
83
elderly care tend to have fewer proponents of family care solutions. This underlines that citizens’ preferences tend to reflect actual conditions and that people do perceive extended public services as effective functional equivalents of family responsibilities (Zetterberg 1986). In countries where public social services are not available, a professed preference for family care may simply reflect the absence of viable alternatives rather than a normative preference, a pattern for which Land and Rose (1985) have coined the term ‘compulsory altruism’. 2.2 Age-specific perceptions of the responsibility to pay for care Irrespective of the preferred care arrangements, respondents were also asked who should mainly pay for the care of elderly parents. The choice was between four options: (1) the elderly parents themselves, (2) their children, (3) the state or other public authorities, (4) everyone equally. As the answers indicate different grades of private or public responsibilities, we expected largely similar patterns as in the case of preferred care arrangements. This would mean a higher degree of family orientation among the elderly, as well as in countries with a strong normative emphasis on family solidarity and a weak development of public care facilities. Within the family oriented solution we would furthermore expect a tendency for each generation to externalise the burden of financing to the other so that elderly persons should be more in favour of children paying the bill, while younger cohorts should prefer that the elderly pay themselves. Figure 3.5 reports the fractions of respondents who opt for the respective financing system. Overall, the majority of Europeans are clearly in favour of a public responsibility for the financing for care. Fifty-one per cent of all European Union citizens prefer this solution. If we add the category ‘everyone equally’ which indicates that public authorities should at least take over some parts of the care costs, the preference for public solidarity becomes even more dominant. Only in Turkey and Romania as well as in Austria do we find a majority of respondents in favour of a private financing of care. The preferences for private solutions are fairly evenly divided among the three different possibilities. As expected, family-oriented financing preferences where the elderly themselves or their children are held responsible, are most widespread in the southern countries of the EU-15, as well as in the three countries we grouped as candidates. The lowest degree of family orientation is once again found in the three Nordic countries and in the Netherlands. Once again we find differences within the old EU-15 to be much more marked than the difference between old and new member states. The basic similarity of the group averages for EU-15 and NMS-10 with respect to the perception of family responsibilities is in fact the result of a peculiar composition effect of different patterns within the EU-15. In the Nordic and continental European countries (except Austria) the idea that children should pay the care bill is clearly the most unpopular solution. In these countries citizens tend to take the challenge of the younger generation to raise children into account and hence accept the idea that the elderly themselves should be held responsible. In stark contrast, citizens in Southern European countries and the NMS-10 as well as in the CC-3 tend to prefer an intergenerational family solidarity with an emphasis on child obligations rather than on parental self-responsibility. The South-eastern pattern stressing child obligations is also found in Austria and Ireland, whereas Poland and Malta are the only new member states which do not share it.
84
Wolfgang Keck and Agnes Blome
Finland Netherlands
66
10
Malta
65
12
Slovakia
58
Latvia
57
Cyprus
54
Estonia
53
51
14
Germany
51
Italy Slovenia
43
19
41
22
Poland
41
22
Austria
23
Romania
23
20
17
25
15 32 9
5 44
7
46
Public authorities
Everyone equally
50%
60%
17
12
22
12
18
48 40%
19
49 18
30%
17
11
23 7
51
20%
13
15
14
EU-28
30
11
46
EU-25
34 58
24
30
18
8
54
10%
22 28
13
21
NMS
0%
15
35
EU-15
CC-3
21
9
24
30
Hungary
18
6
4
31
Turkey
18 24
38
34
Bulgaria
6 12
15
34
Czech Republic
10
30
19
40 38
Spain
16 17
20
43
Greece
11
17
24 17
13 21
8
47
Luxembourg
2 4 22
12
43
Lithuania
18
20
48
10
7 30
France
Ireland
12 9
21 15
9
11
16
51
Portugal
7
14
21
Belgium
4
14
13
62
United Kingdom
3
12
15
68
2
9
7
81
Sweden
5
2
89
Denmark
70%
Elderly themselves
80%
Their children
Figure 3.5 Preferences on care financing, by country Source: Harmonised CC Eurobarometer dataset, 1998–2002, Q 20
90% Other
100%
Is there a generational cleavage?
85
The analysis of age-specific preferences shows some striking deviations from the hypothesis that age groups reflect different risk categories with an interest to shift the cost of care to others (Table 3.1).17 In stark contrast to our respective expectations, older people are even more likely than younger people to hold elderly parents themselves responsible for paying the care bill. In all countries except Luxembourg and Lithuania elderly people more often advocate themselves as financiers than younger age cohorts. In this sense, elderly people in Europe profess a remarkable degree of commitment to the idea of self-responsibility. In Estonia, Germany, Poland, Slovenia, Sweden and the United Kingdom attitudinal differences between old and young age groups are particularly high. Even though the younger generation does not want to be burdened themselves, they are also rather hesitant to make elderly parents foot the bill. The younger age cohorts – and the youngest even more than those in mid-age – have a strong preference to shift the care costs from the family to public authorities. This pattern is particularly strong in the old EU member states in continental Europe. In summary, we have found indications of some age-specific differences in care preferences, but no signs of a generational cleavage, because most of the differences are in the opposite direction of what the idea of a generational conflict based on agespecific calculations of interest would suggest. What we find is acceptance of responsibilities rather than a marked interest in externalising the burden of care to others. However, it is just as noteworthy that even in countries where majorities opt for family care solutions, at least relative majorities assign the responsibility for the financing of care to public authorities. Contrary to our expectations, young and middle-aged cohorts demonstrate in several countries an even higher developed sense of family responsibility than elderly people. These, in turn, are more in favour of self-responsibility, especially when it comes to financing care. These results clearly run counter to the prominent argument that selfish generations seek to accomplish the best possible living standard for themselves. There are several reasons why the problem of care – even though bound to aggravate in the future – will not provide a basis for generational conflicts: 1
2
Rational choice in favour of alleviating the own care burden is not the only and presumably not even the overriding factor shaping care preferences. Since the family operates as the ultimate safety net, rendering help within the family also serves an insuring function that one will receive similar support oneself in times of need. Being less integrated into the labour market and into connected social security schemes, women are most dependent upon family networks. This suggests that their role as kin-keeper may also be motivated by the interest to strengthen and maintain family networks as a resource in hard times (Rossi 1993). Unlike other dimensions of social inequality, age is not a fixed or permanent characteristic, since everyone will occupy different age positions during the life course. The anticipation of younger people that they will grow old and the retrospective view of the elderly people that they once were young create empathy and mutual understanding (Baltes and Mittelstraβ 1992). Giving care accentuates this kind of personal bonding, because it strengthens the intimate und personal relationship between family members while fostering the young generation’s understanding of what ageing will eventually mean for them.
86
Wolfgang Keck and Agnes Blome
Table 3.1 Age-group differences in perceived financing responsibility for care (% point difference) Who should pay for care? Elderly themselves
Denmark Finland Sweden Ireland United Kingdom Netherlands Austria Belgium France Germany Luxembourg Greece Italy Portugal Spain Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania Bulgaria Romania Turkey EU-15 NMS CC-3 EU-25
Their children
Public authorities
Old – young
Old – medium
Old – young
Old – medium
Old – young
3 5 10* 5 13* 5 0 8 6* 16* 8 2 4 3 2 3 0 14* 1 3 8* 2* 5* 3 1 4 1 4 7 4 3 6
4 1 5* 6 8* 1 0 9* 0 12* −5 4 3 2 6 2 1 9 1 4 3 1* 5* 0 −1 2 0 3 4 2 2 4
−2 1 0 0 −4 −9* 21* −1 8 2 2 −5 −3 0 6 −13* 1 6 5 0 2 −4 4 −2 6* 12* 5 2 1 1 −2 1
0 2 −1 0 −1 −1 15* −3 7 3 5 4 4* 4* 6* −1 7 1 4 −5 2 0 0 −1 9* 4 0 0 3 2 −3 2
−5 −7 −9* 5 −8* 0 −7* −19* −15* −14* −14 2 1 −1 −7* 16* 6 −21* −12* −6 −9* −3 −9 0 −7 −10* 2 −10* −7 −7 −6 −7
Old – medium −7 2 −3* 2 −9* −8 −5* −16* −14* −10* −7* −8* −3* 0 −6 0 −1 −5 −7 −3 −2 −5 −2 1 −3 −2 6 −4 −7 −3 0 −5
Source: CC-EB 1998-2002 (CC-EB 2002.1, Q21, EB 50.1 1999. Q37): ‘Who do you think should mainly pay for taking care of elderly parents?’ Note: *Younger age group differ significantly from elderly people age 60 and over (p < 0.05) controlling for sex, income, and household size.
3
Intergenerational relations within the family are the glue which keeps different age groups together (Kohli 2004; Arber and Attias-Donfut 2000). Norms of reciprocity between generations are established in a twofold way: First, as dependency changes over the life course, adult children compensate the support they have received in their childhood from their parents. This is a kind of direct but delayed reciprocity. Second, adult children who support their parents act as a role model strengthening the likelihood that their children will make similar efforts for them. This rationale represents an indirect reciprocity via intergenerational
Is there a generational cleavage?
87
transmission (Antonucci and Jackson 1990). Moreover, the tight emotional relationship between family members strengthens the ties between generations without applying the idea of quid pro quo (Walker 1996). A third aspect is that intrinsic norms as well as social control mechanisms support family care decisions. Assigning elderly parents to residential care is in some contexts regarded as a shame to the family, while rendering care to family members provides status and social recognition (Millar and Warman 1996). In the last instance, giving support thus contributes to higher self-esteem, a mechanism which Andreoni (1989) describes as ‘impure altruism’. The high degree of variation with respect to a preference for family care in Europe indicates that the motives and the circumstances which influence professed preferences differ widely between the countries. We find a low family orientation in care provisions to prevail in the Nordic countries and in the Netherlands, whereas a high family orientation is found in the Mediterranean old member states and in the CC-3. The Western and continental European nations are in between these poles, forming a very heterogeneous group of countries which comprises old as well as new EU member states. National differences in care preferences thus cut across the distinction between old and new member states. In general, there is a tendency for countries with better developed care services and higher expenditure for elderly care to have fewer advocates of family care solutions. This suggests that citizens do perceive extended public services as effective functional equivalents of family care. The new member states do not cluster closely together, but are found in different groups representing weak, intermediate, and strong degrees of familialism in care preferences. Given the heavy reliance on public services and the high female labour force participation in the former communist world, the rather strong persistence of family orientations in the post-communist countries is noteworthy and in line with findings from previous research showing a strong adhesion to family values (Gerhards and Hölscher 2003; Malnar 1999). Our results on care preferences suggest different degrees of familialism with respect to care giving on the one hand and care financing on the other. While most people favour family care, they also think that it should be financed by the state. However, our data do not allow specifying what this means precisely in terms of specific care arrangements. Perhaps the most plausible interpretation is that people prefer a care mix where the parttime provision of family care is supported by public services (Kröger 2001; Leira 1994). The degree of age polarisation is higher with respect to the financing of care than with respect to the provision of care. This is an indication that differences between the generations are more marked with respect to aspects which do not involve tight personal relationships whereas monetary aspects mobilise more rational interest-driven orientations. If this is true, attitudes towards the pension system should be even more age-differentiated, because intergenerational relations within the family have a minor impact whereas financial and fiscal considerations come to the fore more strongly.
3. Age-specific perceptions of the state pension systems 3.1 The level of trust in pension schemes There are two questions in the European Quality of Life Survey (EQLS) asking respondents to evaluate their national pension scheme: one on trust in state pension
88
Wolfgang Keck and Agnes Blome
system and one on the perceived quality of the pension system.18 For conceptual and empirical reasons we here focus on the question relating to trust. Conceptually, the question on quality is rather diffuse leaving it open, whether respondents predominantly think of the generosity of benefits, of the level of the contributions or of the long-term sustainability of present pension arrangements. Empirically, answers to both questions are positively correlated.19 The question on trust presumably invites respondents more to focus on the problem of sustainability, making it more likely that contributors and beneficiaries think of the same issues when answering the question. Considering that demographic change as well as the usually delayed impact of recent pension reforms will affect the pension entitlements of younger people much more than those of elderly people today, we expect younger people to express lower trust than older people. The question then is how much trust European citizens have in their national pension schemes and how much variation of trust we find among different age groups. Figure 3.6 shows the general level of trust in public pension systems. On average, citizens in the old member states have more trust in their pension schemes than citizens in the NMS-10 or in the CC-3 (56 per cent compared to 45 per cent and 49 per cent). In both groups, however, we find large variation around the means. In the NMS-10 there is a range from below 40 per cent in Slovakia, Latvia, and Lithuania to highs above 60 per cent in Poland, Cyprus and Malta. In the old EU-15 the range is from below 40 per cent in Germany, France and Italy to highs above 80 per cent in Finland and Luxembourg. On average, one out of two citizens in the enlarged European Union does not have trust in the national pension system. Once more we do not find countries to cluster together according to any known families of nations logic. Within the EU-15, there is a general north–south gradient, with the Nordic countries standing out for particularly high levels of trust, the Southern European and (some of the) continental countries having particularly low levels, and the Anglo-Saxon countries in between these extremes. It is noteworthy, however, that within each of these groups of countries which are frequently said to form different welfare regimes there is wide variation. In each group we find at least one high-trust (h) and one low-trust (l) country. Thus there are huge contrasts between high-trusting Finland and low-trusting Sweden, between Ireland (h) and the UK (l), between Belgium or Luxembourg (h) and Germany (l), and between Portugal (h) and Italy (l). Within the group of post-communist new member states there is somewhat more similarity, as trust rates below 50 per cent prevail in all countries except Poland. Given the rather low level of confidence in national pension schemes in most European countries, it is interesting to find out to what extent distrust is widespread throughout all social groups, to what extent there are specific groups which function as the carriers of distrust, and to what extent there are age-specific differences which confirm our expectation that the younger generation stands out with a particular lack of confidence. Figure 3.7 shows that younger people have indeed less trust in their national pension schemes than the elderly. The general tendency is borne out in 24 out of 28 European countries and in 18 countries the differences between the youngest and the oldest age group are statistically significant. In addition, in 15 countries we find considerable deviations between the medium age group and elderly people. The degree of age-polarisation is much higher in the old member states of the EU than in the NMS-10 or CC-3. In Austria, Denmark, France, Portugal and the
Is there a generational cleavage? Slovakia
89
30
Lithuania
32
Latvia
33
Germany
34
Italy
38
France
38
Bulgaria
39
Spain
40
Austria
40
Czech Republic
41 41
Romania
43
Slovenia
45
Hungary
47
Greece
48
Estonia
48
United Kingdom
49
Sweden
58
Poland
61
Cyprus
61
Denmark
61
Turkey
65
Portugal
67
Malta
68
Belgium
69
Ireland
70
Netherlands
79
Luxembourg
82
Finland
87
EU-28
52
EU-25
52
CC-3
49 45
NMS
56
EU-15 0
10
20
30
40
50
60
70
80
90
100
% tend to trust
Figure 3.6 Trust in pension system, by country Source: EQLS 2003: Q27a: ‘How much trust do you have in the ability of the following to systems to deliver when you need it? (a: State pension system): (1) A great deal of trust, (2) Some trust, (3) Hardly any trust, (4) No trust at all. Categories 1 and 2 are combined to ‘tend to trust’, categories 3 and 4 are combined to ‘tend to distrust’
90
Wolfgang Keck and Agnes Blome Germany (1, 2) Italy (1, 2) France (1, 2) Spain (1, 2) Austria (1, 2) Greece UK (1, 2) Sweden (1, 2) Denmark (1, 2) Portugal (1, 2) Belgium (2) Ireland Netherlands (1, 2) Luxembourg Finland Slovakia (1) Lithuania (1, 2) Latvia Czech Rep. (1) Slovenia Hungary (1, 2) Estonia (1, 2) Poland (2) Cyprus Malta (1) Bulgaria (1) Romania Turkey EU-15 NMS CC-3 EU-25 EU-28 0
10
20
30
40
50
60
70
80
90
100
% tend to trust in state pension system Age:
15–34 years
35–59 years
60 years and older
Figure 3.7 Trust in pension system, by country and age groups Source: EQLS 2003: Q27a: ‘How much trust do you have in the ability of the following to systems to deliver when you need it? (a: State pension system): (1) A great deal of trust, (2) Some trust, (3) Hardly any trust, (4) No trust at all’ Notes: 1 Significant negative effect (p < 0.05) for younger people (18–34) compared to people (60+) on perception of trust in state pension system controlled for sex, income position, household class status in country specific models. 2 Significant negative effect (p < 0.05) for mid-age people (35–59) compared to people (60+) on perception of trust in state pension system controlled for sex, income position, household class status in country specific models.
elderly region, elderly region,
Is there a generational cleavage?
91
United Kingdom there is more than a 30 percentage point gap between the youngest and oldest age group. Among the ten new member states, Malta is the only country with a similar degree of polarisation. Thus, even though the NMS-10 tend to have lower levels of trust on average, they also have less age-specific polarisation. This suggests that distrust in pension schemes has disseminated more throughout all agegroups than in the old member states, where the older generation has a much more favourable impression of the trustworthiness of the national pension system than the young. One reason for the low age-polarisation in the post-communist countries is that pensioners were struck particularly hard by the transformation process in the 1990s, ending up with meagre pension benefits which made it difficult to make ends meet. Once again, there is no clustering of countries that would clearly distinguish different families of nations. For example in Denmark and Sweden we find highly overproportionate levels of distrust among the younger generation, whereas in Finland younger people express even more trust in the pension scheme than the older generation. Statistically, the degree of age-specific polarisation varies independently of a country’s general level of trust so that we find similar levels of generational cleavage in countries with low and high trust. This means that a high degree of age-polarisation does not necessarily indicate a particularly low level of trust among the young. In the Netherlands, for example, younger people have much less confidence in the pension scheme than the elderly, but in a comparative perspective their level of trust nevertheless exceeds the levels found among the older generation in low-trust countries such as Germany or Italy. In conclusion, younger people distrust the pension system much more than elderly people. The degree of age polarisation is higher in the EU-15 than in the NMS-10 or CC-3. But at the same time the general level of trust is higher in the EU-15 than in the other country aggregates. Considering both, general level of trust and the age polarisation, a generational cleavage appears to be most severe in countries with a low trust level and high age polarisation. This situation is found in Austria, France and Germany as well as in Lithuania, Hungary, and Slovenia. 3.2 What structures trust in pension schemes? The issue of sustainable pensions is more pressing in some societies than in others, because there are differences in socio-demographic structure – such as fertility rates and life expectancy – on the one hand, and differences in the institutional adaptation of pension schemes to the new challenges on the other. With respect to sociodemographic challenges, we would expect strains on pension schemes to be higher, the higher the population share of elderly people is forecast to be in the future. Even though individual citizens may not be fully aware of the demographic situation, it will be reflected in national policy debates which in turn will have an impact on people’s perceptions of the pension system.20 In countries with higher shares of elderly people in the future we expect more intensive debates and hence also more concern among citizens which we expect to translate into lower levels of trust and higher degrees of age-polarisation. The overall correlation between the expected demographic burden of ageing and the degree of age-specific polarisation in the levels of trust in pension schemes is low (R2 = 0.14). However, there is a tendency for countries with the most youthful population
92
Wolfgang Keck and Agnes Blome
structure – Turkey, Cyprus, Romania, but also Ireland and Luxembourg – to have lower degrees of age polarisation, whereas countries with the highest projected share of elderly people are predominantly found in the group with a higher generational cleavage in the perception of pension systems (Italy, Austria, Germany). Demographic changes will of course affect the sustainability of state pension systems very differently according to the institutional structure of the pension systems and to the intensity of reforms which have already been implemented. We expect two institutional characteristics to have an impact on the age-specific perception of pension schemes. First, the more generous benefits in favour of the current generation of pensioners are, the more incisive future reforms will probably have to be. This implies that the younger generation has to expect sizeable curtailments in entitlements, and this should translate into an over-proportionate distrust in the pension system on the part of the young. Second, pension systems which are more universal, less institutionally fragmented and less designed to have earnings-related entitlements reflecting the work career should foster more unity among the generations than more graded and less universal systems. Institutional characteristics of pension schemes are hard to capture in a small set of indicators. Our respective attempts21 failed to produce any stable association between national characteristics of pension schemes and age-specific patterns of trust in state pensions. This is presumably no coincidence, because citizens usually lack any specific knowledge of social policy institutions and they do not tend to compare their national social security systems with others.22 Hence the perception of pension systems may differ from reality, and prevailing perceptions probably reflect general moods and the character of current policy debates rather than the actual nature of the public schemes.23 3.3 The impact of pension assessments on the perception of inter-generational tensions Beyond questions concerning the perception of public services the European Quality of Life Survey also contained a question on the perception of generational tensions.24 This allows us to examine to what extent distrust in the public pension scheme translates into a perception of generational cleavage. On average around 18 per cent of Europeans notice strong tensions between old people and young people. With the exception of tensions between men and women other kinds of tensions – between management and workers, between poor and rich people, and between different racial and ethnic groups – are much more prevalent in the opinion of Europeans (Delhey and Keck in this volume, Ch. 14). In their rare perception of a generational cleavage old and new member states are fairly united, as in both contexts less than one in five respondents see this tension as strong. Only in the CC-3 is there a higher awareness of generational tensions (27 per cent).25 In the NMS-10 the range is from 9 per cent in Cyprus and 14 per cent in Slovakia to a high of 21 per cent in Slovenia. In the old EU-15 there is once again more diversity with a range from 3 per cent in Denmark to 27 per cent in Greece. On the macro-level of countries, one would expect low levels of trust in state pension systems to be accompanied by more frequent perceptions of tensions between the generations. Contrary to this expectation, however, we do not find any association between level of trust and perceived tensions between young and old
Is there a generational cleavage?
93
people (R = 0.03). There are high trust countries with a considerable level of perceived age group tensions, like Belgium, Malta, the Netherlands and Turkey. And on the opposite side there are countries with low average trust in state pension system and with a rather low generational cleavage, such as Germany, Italy, Slovakia and Spain. A micro-analysis on the level of individuals reveals, however, that lack of trust in the pension scheme does translate into a higher propensity to perceive generational tensions. As shown in Figure 3.8, those who have no trust in pension schemes perceive tensions between the generations more frequently than people who have confidence in their pension system. This result is found in all but two countries (Poland and Malta). Multivariate regression analysis reveals that age itself has no significant effect on the propensity to perceive generational tensions, whereas distrust in the state pension system increases the probability to perceive such tensions significantly, even if other background variables including age are held constant (Table 3.2). The analyses on trust in state pension system reveal that younger people differ strongly in their perception of the pension system compared to elderly people. In most countries the young have substantially less trust than people aged 60 and over. Yet younger people do not see a generational cleavage more frequently than older people. It is not an individual’s age per se which determines whether conflicts between old and young are perceived, but rather the concern about the sustainability of state pension systems. Those who distrust the pension scheme are also more likely to be aware of generational tensions. Subjective evaluations of the pension system, however, do not seem to be based on detailed knowledge about demographic change or the structure of pensions systems, but rather represent a vague insight that something is going wrong and that a lot of effort is being put into reforming the systems, which will finally lead to a deterioration of state social security in old age. 2
4. Conclusion By and large our findings underline that the notion of a generational conflict should be considered a myth. Most Europeans do not perceive strong tensions between old and young people. The perception of such tensions is also not age-polarised to any noteworthy degree; in other words, older and younger people largely agree that there are no strong generational tensions. Secondly, we find no big or consistent differences between old and new member states. Country-specific differences within the old EU-15 are much bigger than the difference between the two country-group averages. This means that affinities and differences between European countries with respect to old age policies cut across the divide between old and new member states. With respect to generational tensions concerning policies, it is important to distinguish between the fields of care and of pension policies. In the field of care we find a remarkable vitality of family solidarity across all age cohorts. Not only is one out of five Europeans actively involved in rendering informal care to family members (Alber and Kohler 2004), but there is also widespread recognition that adult children have a responsibility to look after their elderly parents if they can no longer manage to live on their own. Younger people underline family responsibilities just as frequently as older people, and there is no indication of an age-related cleavage in the perception of care responsibilities. Remarkably, an active involvement in care giving fosters even more positive thoughts about extended family responsibilities in the future.
Denmark Italy Finland* Sweden Portugal Germany* Spain Ireland Luxembourg UK Netherlands Belgium* Austria* France Greece* Cyprus Slovakia* Czech Rep.* Poland Estonia Hungary* Malta Latvia Lithuania* Slovenia Bulgaria* Romania Turkey* EU-15 NMS CC-3 EU-25 EU-28 0
10
20
30
40
50
% perceived strong tensions between young and old distrust pension system
country mean
trust pension system
Figure 3.8 Perception of tensions between young and old, by level of trust in public pension system Source: EQLS 2003: Q27a: ‘How much trust do you have in the ability of the following to systems to deliver when you need it? (a: State pension system); Q29d: In your opinion, how much tension is there between each of the following groups? (d: Old people and young people)’ Note: *Significant positive effect (p < 0.05) of distrust on perception of tension between younger and older people controlled for sex, age, region, income position, employment status in country specific models’.
Is there a generational cleavage?
95
1
Table 3.2 Logistic regression on tension perception by age and trust in pension scheme Perceive strong tensions between young and old people, 1= yes Independent variables Trust in pension system 1 = a great deal of trust, 4 = no trust at all Age in years
Coefficient
Significance
0.22
0.000
0.00
0.432
Source: EQLS 2003: Q27a: ‘How much trust do you have in the ability of the following to systems to deliver when you need it? (a: State pension system), Q29d: In your opinion, how much tension is there between each of the following groups? (d: Old people and young people)’ Note: 1 Controlled for sex, region, employment status, income position, and country dummies. N: 18241, McFaddens R2 = 0.06.
Even though Europeans advocate and practise family solidarity, they are also clearly aware of the families’ limits. Thus with respect to the financing of care, public payment is clearly the most popular solution, whereas the idea that children should pay the bill receives least support. There is no indication of a generational conflict, as age-specific perceptions of financing responsibilities run counter to the idea of an agerelated cleavage. Thus older people advocate shifting the cost of care to the elderly themselves even more frequently than the younger generation who in turn accept the idea that children should pay for care usually just as frequently as older age cohorts. The idea that care should be paid by the public purse is more popular among younger than among older people. Thus the overriding result with respect to care arrangements is that there is a remarkable sense of intergenerational compassion and of self-responsibility in the younger and older generation alike. Noteworthy nation-specific differences appear on the macro-level between countries with different care traditions, but here the decisive divide is not between, but within the groups of old and new member states. In the old EU-15, the Nordic countries and the Netherlands stand clearly apart for their emphasis on public services and for their concomitant reluctance to advocate family responsibilities. In the new member states a similar divide is found between the Mediterranean islands and the post-communist transformation countries where the emphasis on family responsibilities tends to be stronger and thus more akin to Southern European and continental countries than to the Nordic example. In contrast to the field of care arrangements, traces of a generational cleavage can be found in the field of pension policies. As one in two Europeans tends to distrust the state pension scheme, the general confidence in public pension programmes is remarkably low. On average, it is lower in new member states than in the old EU-15, but again there is much nation-specific diversity within the two groups. Within single countries, however, there is a rather general pattern: distrust is distributed unevenly among different age-groups, as younger people distrust the pension schemes more than older people. This also has an impact on the perception of tensions between the generations. Whereas the perception of such tensions is distributed fairly evenly among the generations and hence statistically unrelated to age, distrust in pension schemes does increase the propensity to perceive a generational cleavage. In other words, where we find signs of a generational conflict this is not related to age per se, but mediated by distrust in the public pension system.
96
Wolfgang Keck and Agnes Blome
A conceptual distinction made by the sociologist Lutz Leisering (1992) may help to combine the apparent contradiction between our findings on care and on pensions. Leisering distinguishes between institutionally constructed generational relations (Generationenverhältnis) and personal affiliations between members of different generations (Generationenbeziehungen). In this perspective, the fields of care and of pensions are policy fields with very different degrees of state penetration. Whereas in most European countries care is still predominantly left to (and perceived as belonging to) family provision, pensions are predominantly state-provided. Hence perceptions of the field of care reflect intergenerational affiliations within families – which are basically intact with a strong sense of family solidarity – whereas perceptions of pension policies are much less shaped by personal affiliations, thus reflecting the relationship of citizens to the far more remote institutions of the state which have recently become the topic of much public debate. In other words: personal affiliations between members of different generations will always cut across generational relations shaped by the state, this will weaken the potential for generational conflicts, and this mediating effect will be stronger in policy fields where public action is less prominent.
Notes 1 We would like to thank Jens Alber, Ulrich Kohler, Mary Daly and Chiara Saraceno for their support and helpful comments on earlier versions of this chapter. 2 Poverty statistics show, however, that in 2003 elderly people still had above average poverty rates in 11 out of 14 old EU member states and in 4 out of 9 new member states for which there are data (European Data Service 2005). 3 One model, known as the fixed relative position (FRP), was advocated by Musgrave and proposes to hold constant the ratio of per capita earnings (net of contributions) of those in the working population to the per capita benefits (net of taxes) in the pensioner generation. Once the ratio is fixed, the contribution rate rises as the population ages, but benefits also fall so that the income of both groups shrinks at the same rate (see Myles 2002: 141). 4 In Germany, the average duration of drawing an old-age pension in the white-collar scheme more than doubled for female pensioners between 1960 and 2004 and increased from 9 to 15.5 years for men: (Deutsche Rentenversicherung 2005). 5 In Germany, dependency upon domestic care jumps from 4 per cent in the age group 70–75 to almost 40 per cent among those aged 85+ (Garg 1995). In the European member states the share of very old people grows over-proportionally with the consequence of drastically increasing numbers of persons dependent upon care (McGlone and Cronin 1994). 6 In EU countries the percentage of people above age 75 is expected to grow from roughly 7 per cent in 2004 to 12 per cent in 2030 and to above 17 per cent in 2050 (European Data Service 2005). 7 In Germany, there were 5 women in the potential caring age group 45-69 per each person above age 75 in 1960, but only 2.2 in 2000, and following projections only 1.5 in 2030 (Alber and Schölkopf 1999: 304). 8 The typical pattern is that those in the older generation receive public pensions, but give considerable financial support to their children (Kohli 2004; Attias-Donfut and Wolff 2000). 9 The harmonised aggregated data file combines questions from surveys held at different points in time in various countries. The information for the old EU member states was collected in two surveys held in 1998 and 1999 (EB 50.1, EB 51), whereas the data from the new EU member states and the candidate countries all come from a 2002 survey (CC-EB 2002.1). Micro-level analyses are only possible in cases where all variables needed for analysis come from the same survey. 10 Since the questions on the perception of care on the one side and pensions on the other come from different surveys, we cannot analyse the relationship between attitudes on care and on pensions.
Is there a generational cleavage?
97
11 It should be noted, however, that a sizeable share of care work is done by elderly partners, siblings or children who are themselves beyond the age of retirement, so that the age group of elderly people above 60 includes both, receivers and givers of care. Yet we assume that even elderly care givers – mostly women – also are likely to consider their future situation as care receivers because of their advanced age. 12 The correlation between care giving outside one’s own household and the advocacy of family responsibilities is much weaker, but also positive. 13 The term ‘candidate countries’ is used here to refer to the three countries which did not belong to the European Union after the first eastern enlargement of 2004. Of course, Bulgaria and Romania have now also joined the Union, so that Turkey is the only remaining candidate country from the 2003 survey. Aggregates for Bulgaria, Romania and Turkey will here be referred to as ‘CC-3’. 14 A further expectation was that countries with less developed care facilities should have a more visible generational cleavage, especially if cultural norms in favour of family care are weak. However, the Baltic states, which seem to fit this categorisation most closely, show no indication of a particularly marked generational cleavage. Only the relatively strong generational differences in Germany and Luxembourg are somewhat in line with this hypothesis. 15 Eurostat classifies benefits in kind for the population above age 65 as part of the old-age function. This may, however, include or exclude benefits in kind spent on health and invalidity functions. In order to improve comparability, we here include these benefits as well, because the combined total represents the benefits in kind for the elderly more comprehensively (see the methodological annotations in European Commission 1996). No data are available for Cyprus. 16 Ireland is a somewhat unanticipated member of this group. It has an extraordinarily high share of healthcare expenditure in its comprehensively defined benefit expenditure for old age as used here. 17 As in the section on preferred care support, we examined gender differences in care financing as well. However, the situation for women and men is not the same as in the case of care provision. Given that they have lower incomes on average, women might be expected to prefer financial support from public authorities. However, we did not find large gender differences in care financing. preferences which is probably due to the fact that men and women share their resources when living in a joint household. The results are not presented here. 18 Unfortunately the questions on pensions are not analogous to the care preference items. Here we only know the evaluation of the state pension system but we do not have any information with regard to alternative options or reform preferences. 19 Comparing nations on the macro-level we find a positive correlation of 0.52. On the micro-level of individuals trust in pension system and quality of pension system is similarly correlated (0.55). 20 For an account of how politicians play a crucial role in the framing of issues which in turn influences beliefs about the welfare state see Kitschelt (2001). 21 We have worked with the following operationalisations of institutional characteristics: present gross (and net) earnings-replacement ratios; projected change in replacement levels (as indicator of the intensity of adaptive reforms); degree of universalism in pension schemes (measured as citizenship-based entitlement plus low institutional fragmentation). 22 In a German survey a majority of people believed Germany to have a funded state pension system, even though the country has one of the oldest pay-as-you-go pension systems in the world (Börsch-Supan 2000). 23 Based on surveys conducted in Germany and Italy in the autumn of 2001, Boeri et al. (2002) show that the German Riester reform which was legislated previously in the same year had a discernible impact on public opinion. In contrast to Italy where the last reform had been passed in 1997, the Germans held more pessimistic views concerning future pension benefits. The authors concluded that the Riester reform ‘obviously succeeded in conveying the message that there is an end to pension generosity’ (Boeri et al. 2002: 400). 24 The question read as follows: ‘In all countries there sometimes exist tensions between social groups. In your opinion, how much tension is there between each of the following groups? ‘Old and young people’ were mentioned as one among five possible antagonistic groups
98
Wolfgang Keck and Agnes Blome
(for further analysis of the perception of tensions in European societies see Delhey and Keck, this volume, Ch. 14). 25 This is the unweighted average for the three countries; weighted for the different population sizes, the aggregate mean would be 31 per cent.
References Alber, J. (1984), ‘Versorgungsklassen im Wohlfahrtsstaat: Überlegungen und Daten zur Situation in der Bundesrepublik’, Kölner Zeitschrift für Soziologie und Sozialpsychologie, 36: 225–251. Alber, J. and Kohler, U. (2004) Health and Care in an Enlarged Europe, Dublin: European Foundation for the Improvement of Living and Working Conditions. Alber, J. and Schölkopf, M. (1999) Seniorenpolitik: Die soziale Lage älterer Menschen in Deutschland und Europa, Amsterdam: Fakultas. Andreoni, J. (1989) ‘Giving with Impure Altruims: Applications to Charity and Ricardian Equivalence’, Journal of Political Economy, 97: 1447–1458. Antonucci, T.C. and Jackson, J.S. (1990) ‘The Role of Reciprocity in Social Support’, pp. 173–198, in B.R. Sarason, I.G. Sarason and G.R. Pierce (eds), Social Support: An Interactional View, New York: Wiley. Arber, S. and Attias–Donfut, C. (ed.) (2000) The Myth of Generational Conflict, London, New York: Routledge. Attias-Donfut, C. and Wolff, F.-C. (2000) ‘The redistributive effects of generational transfers’, pp. 22–46, in S. Arber and C. Attias-Donfut (eds), The Myth of Generational Conflict, London, New York: Routledge. Auerbach, A. J., Kotlikoff, L. J. and Leibfritz, W. (1998) Generational Accounting around the World, Tokyo: Bank of Japan. Baldwin, P. (1990) The Politics of Social Solidarity: Class Bases of the European Welfare State 1875–1975, Cambridge: Cambridge University Press. Baltes, P.B. and Mittelstraβ, J. (1992) Die Zukunft des Alterns, Berlin: De Gruyter. Boeri, T., Börsch-Supan, A. and Tabellini, B. (2002) ‘Pension reform and the opinion of European citizens’, The American Economic Review, 92, 2: 396–401. Börsch-Supan, A. (2000) Rentenreform und die Bereitschaft zur Eigenvorsorge: Umfrageergebnisse in Deutschland, Beiträge zur angewandten Wirtschaftsforschung, Mannheim: Institut für Volkswirtschaftslehre und Statistik. Breyer, F. (1990) Ökonomische Theorie der Alterssicherung, Munich: Vahlen Franz GmbH. Deutsche Rentenversicherung (2005) Statistiken – Rente – Zeitreihen. Online. Available http: (accessed 4 January 2007). Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism, Princeton, NJ: Princeton University Press. European Commission (1996) ESSPROS Manual, Brussels. European Commission (1999) Generational Accounting in Europe, Brussels: European Commission. European Data Service (2005) Eurostat Online Datenbank. Online. Available http: (accessed 4 January 2007). Garg, H. (1995) Pflegebedürftigkeit als Gegenstand ökonomischer Sicherungspolitik, Frankfurt a. M., New York: Peter Lang. Gerhards, J. and Hölscher M. (2003) ‘Kulturelle Unterschiede zwischen Mitglieds- und Beitrittsländern der EU. Das Beispiel Familien- und Gleichberechtigungsvorstellungen’, Zeitschrift für Soziologie, 32, 2: 206–225.
Is there a generational cleavage?
99
Kitschelt, H. (2001) ‘Partisan competition and welfare state retrenchment: when do politicians choose unpopular policies?’, pp. 265–304, in P. Pierson (ed.) The New Politics of the Welfare State, Oxford, New York: Oxford University Press. Kohli, M. (2004) ‘Intergenerational transfers and inheritance: a comparative view’, pp. 266–289, in M. Silverstein (ed.), Intergenerational Relations across Time and Place, New York: Springer. Kröger, T. (2001) Comparative Research on Social Care: The State of the Art, Brussels: European Commission. Land, H. and Rose, H. (1985) ‘Compulsory altruism for some or an altruistic society for all?’, pp. 74–96, in P. Bean, J. Ferris and D. Whynes (eds) In Defence of Welfare, London: Tavistock. Leira, A. (1994) ‘Concepts of caring, loving, thinking, and doing’, Social Service Review, 68, 2: 185–201. Leisering, L. (1992) Sozialstaat und demografischer Wandel, Frankfurt, New York: Campus. Lepsius, R.M. (1979) ‘Soziale Ungleichheit und Klassenstrukturen in der Bundesrepublik Deutschland’, pp. 166–209, in H.-U. Wehler (ed.), Klassen in der europäischen Sozialgeschichte, Göttingen: Vandenhoeck & Ruprecht. Malnar, B. (1999) ‘Images of modern family’, pp. 217–240, in: N. Tos, P.Ph. Mohler and B. Malnar (eds), Modern Society and Values, Mannheim: Faculty of Social Sciences, Ljubljana and ZUMA, Centre for Survey Research and Methodology. McGlone, F. and Cronin, N. (1994) A Crisis in Care? The Future of Family and State Care for Older People in the European Union, Occasional Paper no. 19, London: Family Policy Studies Centre. Millar, J. and Warman, A. (1996) Family Obligations in Europe, London: Family Policy Studies Centre. Myles, J. (2002) ‘A new social contract for the elderly?’, pp. 130–172, in: G. Esping-Andersen (ed.), Why We Need a New Welfare State, Oxford: Oxford University Press. Rossi, A.S. (1993) ‘Intergenerational Relations: Gender, Norms, and Behavior’, pp. 191–212, in V. L. Bengtson and A. W. Achenbaum (eds), The Changing Contract across Generations, New York: Aldine. Walker, A. (1996) ‘Intergenerational relations and the provision of welfare’, pp. 37–55, in A. Walker (ed.), The New Generational Contract: Intergenerational Relations, Old Age and Welfare, London: University College London Press. Zetterberg, H. (1986) ‘The rational humanitarians’, pp. 79–96, in S. Graubard (ed.), Norden: The Passion for Equality, Oslo: Norwegian University Press.
4
Family policy patterns in the enlarged EU Thomas Bahle
Introduction1 In all European countries, the family is changing in similar ways, but family policies widely vary (Hantrais 2004). The European nation states have intervened into the family to varying degrees and in different forms (Kaufmann 1994). This pattern of diversity was hardly changed by the EU which has limited competency in family policy.2 Chances for a common ‘model’ have probably further decreased with the entrance of ten new member states into the EU, among them eight former Socialist countries. The likely result is growing diversity in European family policies. This chapter aims at a comparative analysis of family policy within the enlarged European Union. It uses the ‘families of nations’ approach for studying similarities and differences between EU member states.3 The chapter raises two questions: which ‘families of nations’ of family policy can be distinguished and how can they be explained? Do the new member states form a separate group or join other ‘families of nations’? Family policies are defined as policies by which the state interferes into relationships between the family and society and/or relationships among family members. The analysis of ‘external’ relations focuses on functional aspects of the family. The family’s internal relationships are structured by three dimensions: parenthood, partnership and generational bonds. This chapter focuses on parenthood and thus on (nuclear) families with children, excluding partnership and generational relations.4 It does not distinguish between ‘explicit’ and ‘implicit’ policies, because the focus is on institutional variations. In this perspective, family policy is the institutionalised pattern of public policy-makers’ actions on the family (Bahle 1995; Gauthier 1996). The first section of the chapter develops a conceptual framework for identifying and explaining the various families of nations of family policy in Europe. This framework is taken as hypothetical reference for the empirical analysis of family policies following in section two. The third section compares these findings with the postulated ‘families of nations’ and speculates about convergence or divergence in European family policies.
1. Families of nations of family policy A family of nation shows a typical pattern of institutionalised policies that was formed in a specific historical-cultural context. The chapter starts from the assumption that in this context a ‘guiding image’ for state action on the family had developed long before
Family policy patterns
101
family policies became literally enacted. This image ‘framed’ the legitimacy of state intervention into the family and the preferred modes of action. I assume that two elements were crucial in forming this image: family systems and state-church relations. The first primarily concerns the action mode, the second the legitimacy of state interference. Variations in these two dimensions may explain the different families of nations of family policy in Europe. 1.1 Family systems and state–church relationships For the analysis of family forms I refer to the work of Hajnal (1965), Goody (1994), Laslett et al. (1983), Mitterauer (2003) and Todd (1983; 1990). Hajnal (1965) discovered a historical West European marriage pattern characterised by late marriage and high celibacy rates. This pattern distinguished Western Europe from the East and probably from the rest of the world. Geographically, the dividing line runs approximately from Trieste to St Petersburg. Laslett et al. (1983) largely confirmed this finding on the basis of historical household studies. Hajnal’s division partly overlaps with two major historical borderlines between West and East: one between Orthodoxy and the Roman Catholic Church (Goody 1994) and another between the medieval western system of seigneurial domination (feudalism) and eastern serfdom (Mitterauer 2003). 5 Todd (1983; 1990) identifies four historical family forms in Europe on the basis of relations of authority or liberty between parents and (adult) children and relations of equality or inequality among siblings. Main indicators are inheritance rules and household formation patterns (Todd 1990: 29–67). The stem family (famille souche) is characterised by authority and inequality. It is primarily found in central Europe (Germany, Austria, Switzerland, Belgium), Sweden, southern France and the north of the Iberian peninsula. The egalitarian nuclear family (famille nucléaire égalitaire) is defined by liberty and equality. It is primarily found in northern France, southern Spain, southern Italy, Poland and Greece. The ‘absolute’ nuclear family (famille nucléaire absolue) is characterised by liberty and ‘neutrality’ (concerning relations among siblings). It is mainly found in England, the Netherlands, Denmark and Norway. The communitarian family (famille communautaire) is defined by authority and equality. This pattern is found in South-eastern Europe (particularly Serbia) and Russia, but also in the centre of Italy and France (see Saraceno, this volume, Ch. 2). Todd does not take into account how the family nucleus is embedded in wider kinship relations. Yet if one combines his family forms with the general findings of Hajnal, Laslett and Mitterauer one can conclude that within Europe the significance of kinship compared to the family nucleus declines from south to north and from east to west. State–church relations are structured along the same axes. State and church were the two major historical ‘actors’ in family policy. The state started to interfere into the family when the church, but also guilds and cities, had already occupied regulatory positions. The Catholic church in particular refused to accept state legislation in family matters for a long time. In this sense state–church relations ‘pre-structured’ the legitimacy of state action on the family. In Europe, the Great Schism between Rome and Constantinople led to a first major dividing line between east and west (Rokkan 1999; Flora 1996). In the Orthodox world the state dominated whereas in the realm of the Roman Catholic Church a long conflict between state and church prevailed.
102
Thomas Bahle
The second divide was the Reformation which led to a unification of state and church in the reformed countries of the north. Between them and the countries of Catholic counter-reformation in the south a region of religious heterogeneity emerged in central Europe. In this area relations between state(s) and church(es) were complex with varying balances of power. In the long run, conflicts were usually solved by an institutionalisation of the subsidiarity principle. Moreover, the Reformation led to a stronger position of the individual within the nuclear family. Though at the beginning Protestantism was very patriarchal, the significance of literacy and individual consciousness paved the way for an ‘individualisation’ of family relationships whereas Catholicism kept a group-centred image of the family as an institution (Bahle 1995). 1.2 Family policy forms in ‘old’ Europe These historical variations in family forms and state–church relations are now combined into a conceptual framework for explaining the various families of nations of family policy in Europe. The first general hypothesis is that nuclear family systems based on values of equality are more ‘open’ to state intervention than extended or communitarian systems in which strong kinship ties and solidarity norms prevail. State interventions in these systems may be highly illegitimate if they try to impose equality norms which contradict prevailing social values. The stem family may be open to state intervention if this is executed in a subsidiary mode. Since the stem family is grounded on rules of inequality concerning siblings, state intervention may only be legitimate if it is executed indirectly and supports the family as a group rather than changing the balance between family members. The second general hypothesis is that state intervention into the family is more legitimate if the state succeeded in competition with the church or if conflict was absent. By contrast, it is less legitimate if the church was able to keep a dominant position for long. If there was a balance of power between state and church or religious heterogeneity state intervention may only be legitimate if grounded in the principle of subsidiarity. Synopsis 1 shows the various families of nations that one can expect on the basis of these arguments arranged on a simplified geographical map of Europe. Figure 4.1 (Synopsis 1)6 shows two ‘independent’ dimensions: family forms and state–church relationships. The borderlines between family forms are not strict, but should be interpreted as zones of relative predominance. In bold letters are identified the ‘dependent’ (hypothetical) families of nations of family policy. In the region in which Protestant state churches and nuclear (and stem) family forms with strong values of equality and weak kinship ties predominated (the North) one may expect a family policy characterised by strong individualisation of internal relationships and extensive socialisation of family functions. This family policy form is called universality. It is typical for Scandinavia, a culturally unique, homogeneous region in which state and church have been united since the Reformation with the state as the sole power centre. As a consequence, the family has become ‘integrated’ into society rather than protected as an autonomous institution. In the north-west of Europe the absolute nuclear family form predominates, which is not permeated by values of equality. Here we may expect a family policy characterised by individualisation of internal relationships, but not by a socialisation of
(Southern Europe)
Familism
Catholic church predominates + stem/extended families:
(Central West Europe)
Note: Bold letters: Families of nations of family policies.
← (South-east Europe: CO, SR, BG, RO)
Orthodoxy/Catholicism + communitarian family
← (Central East Europe: PL, CZ, SK, HU + LT)
← (Baltic states: EE, LV)
Subsidiarity
(Scandinavia)
(England) Catholicism/Protestant minorities + stem/extended families
Universality
Autonomy
Protestantism/Orthodox minority + nuclear/stem families
Religious heterogeneity or state dominates Catholic church + nuclear/stem families:
State-protestantism + (egalitarian) nuclear family and stem family:
Protestant pluralism + (absolute) nuclear family:
Figure 4.1 Synopsis 1: Conceptual map of family policy forms in Europe
Catholicism (state–church competition)
Mixed area
Protestantism (no state–church competition)
Former socialist model: Working family
104
Thomas Bahle
family functions. This form is called autonomy, the heartland of which is Britain, stretching out towards the Netherlands, parts of Denmark and (for historical reasons) Ireland. A major difference between Britain and Scandinavia is the character of the Reformation. The English Reformation was a separation from Rome rather than a religious reform; the Anglican state church retained major characteristics of the Catholic church. Moreover, Anglicanism has never represented the whole population. There have always been sizeable minorities of Catholics and Protestant sects. Thus religious pluralism rather than universal integration has characterised the British religious landscape. This constellation has been a fertile soil for the liberal principles of public non-intervention into ‘private’ religious, economic or family matters. In the centre of Europe we find a mix of nuclear and stem family forms. In addition, large parts of this area are religiously mixed. This is a fertile soil for a family policy characterised by limited individualisation of internal relationships and a limited externalisation of family functions. This form is called subsidiarity. It has strong affinity to the stem family form and is typically found in areas in which state and church competed for long and in areas of cultural heterogeneity.7 In this case the family is supported as a group without interfering strongly into internal relationships or socialising family functions extensively. In the south we find stem and nuclear family forms embedded in strong kinship relations.8 The power centres of these areas were also strongholds of Catholicism where the state could never gain high legitimacy.9 In the post-World War II period this characteristic was aggravated by late transition to democracy except in Italy. One can thus expect a rudimentary family policy with weak individualisation and almost no socialisation of family functions. This form is called familism. The guiding image of familism may be similar to subsidiarity, but since state intervention into the family is less legitimate, family policy is expected to be much less developed. To summarise, our hypothetical families of nations of family policy in ‘old’ Europe consist of the following country groups: autonomy (United Kingdom, the Netherlands, Ireland); universality (Sweden, Denmark, Norway, partly Finland); subsidiarity (Belgium, France, Luxembourg, Germany, Austria, Switzerland); familism (Portugal, Spain, Italy, partly Greece). The comparative ‘position’ of the new member states is discussed in the next section. 1.3 The new member states How do the new EU member states fit into this historical map of European variations? Large parts of this area fall east of Hajnal’s dividing line, Poland is on the border and the Czech Republic clearly to the west of it. In addition, large parts of Eastern and South-eastern Europe were dominated by Orthodoxy and/or the long rule of the Ottomans.10 In this area the stem family and the communitarian family predominated whereas the nuclear forms were under represented. The family institution was characterised by strong relations between generations and siblings. Both characteristics are not favourable to state interventions into the family. Historically, one would therefore expect a weak family policy, but one oriented towards the subsidiarity model where the family group is more important than the individual. Yet communism had changed this pattern profoundly. This has had a lasting impact on family policies.11 The family policy model that all socialist countries more or less shared can be labelled the ‘Working family’ (see Synopsis 1, Figure 4.1).
Family policy patterns
105
The family was largely integrated into society and many functions were taken over by state and society. Family policies were firmly based on the working couple family (Lobodzinska 1995, Robila 2004). Though socialist family policy covered the whole working population in fact, it was not really universalistic in principle, in contrast to Scandinavia. And also in contrast to Scandinavia the nuclear family form did not completely succeed as many traces of communitarian and stem family forms are still visible (see also Saraceno, this volume, Ch. 2). This explains why family policies in Eastern Europe under communism were work-related, but not based on individual rights and universalistic principles as in Scandinavia. And this had consequences after the fall of communism when the nexus between social protection and the workplace disappeared. Competitive markets, high unemployment, falling birth rates and rising poverty among families dramatically changed the context for family policy in transition countries. Yet there is no straightforward way out of these problems. Possible pathways can be simplified into three alternative hypothetical scenarios: persistence of distinctiveness, ‘creation’ or new institutional design, and historically conditioned institutional convergence. Each of these hypotheses can be defended with some plausibility.12 Hypothesis 1 Persistence of distinctiveness. This probably most straightforward scenario assumes that the Eastern European countries will constitute their own ‘family of nations’. One argument in favour of this scenario is that core elements of the socialist modernisation period survived the fall of the system, in particular a strong welfare and work orientation among the population (Pascall and Manning 2000). One could therefore expect a family policy that would come close to Scandinavia with respect to family– employment relationships. Less clear in this scenario, however, is the fate of universalism and individual rights, the two other major components of the Scandinavian family policy profile. Both universality and individual rights were never firmly rooted in the socialist world; this would speak for continuing ‘eastern exceptionalism’. Another argument in favour of ‘distinctiveness’ stresses the specific problems inherited from socialism and during the transition period. These problems could occupy these countries for long and may lead to a persistent focus on poverty policy. Hypothesis 2 ‘Creation’ or new institutional design. This scenario assumes that the breakdown of the socialist regime had opened a window of opportunity for adopting newly designed policy models imported from the west. An expectation may be that the former socialist countries will abandon the old ideas of welfare and work-relatedness in family policies and will withdraw from public service provision. The state would play a residual role focusing on unemployed and poor families. The route in this scenario would lead towards Britain or even the USA. Hypothesis 3 Historically conditioned institutional convergence. The third scenario goes back to earlier differences among the socialist countries that had characterised them already
106
Thomas Bahle
before and during communism. It postulates long-term effects of earlier variations on current developments.13 After the fall of communism some older institutional arrangements would be revived that may continue earlier paths of development.14 This may contribute to ‘renewed’ differences within new member states. In addition, historical cultural regions transcending the artificial west–east divide could be revived. Thereby the former socialist countries may once again join different ‘cultural’ country groups.15 The third scenario, in contrast to the first two, postulates a differentiated process of convergence (see Figure 4.1: synopsis 1). The Protestant Baltic states may move towards the Scandinavian form of universality, Catholic dominated Central-eastern Europe towards subsidiarity and South-eastern Europe towards Familism. Thus the new member states will not collectively transit ‘from Socialism to the periphery’16 or follow the Anglo-Saxon model of autonomy.
2. Variations in family policies This part of the chapter analyses family policy variations between EU member states in 2004.17 Do the cross-sectional data confirm the families of nations distinguished in the preceding part of the chapter and which scenario is confirmed for the new member states? The analysis first considers total social expenditure on families and children and goes on with specific policies: family allowances, maternity and parenting benefits and childcare services. 2.1 Social expenditures Table 4.1 shows social expenditures on family benefits in 24 European countries18 as a percentage of GDP and as a percentage of total social expenditures in 2004. This first indicator shows which share of a nation’s income is spent on family benefits. The share varies from 0.5 per cent in Spain to 3.9 per cent in Denmark, the EU-25 average is 2.1 per cent. The difference between the EU-15 and EU-25 averages is insignificant which shows that the new member states did not change the overall size of family benefits in Europe. The big family policy spenders are the Scandinavian countries (universality group) and the Western-central European countries (subsidiarity). The weak spenders are typically located in the south (familism group). Yet at the bottom of the distribution we also find two members of the autonomy group, the United Kingdom and the Netherlands. The new member states fall a bit below the European average, but spend more of their national incomes on the family than the Southern European countries. The share of social expenditure devoted to the family depends on the structure of the welfare state. The picture does not change significantly if we use this indicator. Countries with higher expenditures measured in terms of the GDP also tend to spend more of their overall social expenditure on the family. The share devoted to the family varies from about 2.5 per cent in Spain to 16.3 per cent in Luxembourg. The European average is at 7.7 per cent. Once more EU-15 and EU-25 averages are close to each other. The Scandinavian and Western-central European countries again appear in the upper half of the rank order, and the southern countries, the United Kingdom and the Netherlands at the bottom. Both the universality and the subsidiarity forms spend more of their social protection benefits on the family than the familism and the autonomy forms. The new member states perform even better with respect to
Family policy patterns
107
Table 4.1 Social expenditure on family benefits, Europe 2004 Country Denmark Luxemburg Germany Finland Austria Sweden France Hungary Ireland Belgium EU-15 Slovenia EU-25 Estonia Greece United Kingdom Poland Czech Republic Slovakia Latvia Lithuania The Netherlands Malta Portugal Italy Spain
As % of GDP 3.9 3.7 3.1 3.0 3.0 3.0 2.7 2.6 2.4 2.2 2.2 2.1 2.1 2.0 1.8 1.8 1.7 1.6 1.5 1.4 1.2 1.2 1.1 1.1 1.0 0.5
As % of total social expenditure 13.0 16.3 10.3 11.3 10.2 9.3 8.8 12.3 15.2 8.0 7.7 8.3 7.7 14.4 6.8 6.5 7.6 8.0 7.8 9.8 7.9 4.2 6.2 4.3 3.8 2.5
Incometested (%) 4.4 0.0 26.4 2.3 5.7 0.0 32.3 12.1 38.7 0.8 24.3 74.3 24.1 33.9 27.9 2.0 56.0 54.5 5.1 20.6 1.5 87.1 43.2 29.8 47.1
Source: Own calculations based on Eurostat Note: Countries are ranked by expenditures as % of GDP.
this indicator. Among the 15 countries lying above the European average we find 7 out of 10 new members (Ee, Hu, Lv, Sl, Cz, Lt and Sk). Thus the really pensionerbiased and ‘clientelistic’ welfare states are found in the south (Ferrera 1996). Yet income-tested benefits play an important role in the new member states. In Malta, Slovenia, the Czech Republic and Slovakia income-tested family benefits hold a share of more than 50 per cent of the total. But some older member states also rely significantly on income-testing, for example Portugal, Spain, Ireland and Greece. Moreover, in France and Germany the share of income-tested benefits is above the European average. In general, however, the familism and the autonomy groups have the highest proportions of income-tested benefits. In those countries family policy is focusing on low-income families. The new member states do not form a coherent group in this respect. As already mentioned, four of them are among the countries with high shares of income-testing, but in Poland, Latvia and Hungary figures are very low. Moreover, in the new member states income-testing is probably more a matter of scarce resources than of principles. One may expect that the significance of incometesting may decline if tax bases improve. By contrast, in many southern countries, and in the United Kingdom in particular, income-testing is an ‘institutional’ element of family policy.
108
Thomas Bahle
Thus the former socialist countries are closer to the European average than the Southern EU member states. In most of them, family policy is not a residual category, nor is income-testing exceptionally strong. Perhaps, however, the difference lies in the structure of policies. In the following sections three specific policies are analysed: family allowances, maternity and parenting benefits and childcare services. These policies together represent more than 80 per cent of expenditure on the family on average (see Figure 4.2). fi dk sl ee se pt lt lv sk gr es
Country
cz hu it at pl de eu25 eu15 lu fr uk be mt ie nl 0%
10%
20%
30%
maternity/parenting
40% 50% 60% % distribution childcare services
70%
80%
family allowances
90%
100%
other
Figure 4.2 Structure of social expenditure on family benefits, Europe 2004 Source: Own calculation based on Eurostat Note: Countries are rank-ordered by the sum of expenditures on maternity benefits and childcare services.
Family policy patterns
109
Yet the policy mix widely varies within Europe. In the figure, countries are ranked (in descending order) by the proportion of the total they spend on the sum of maternity and parenting benefits and childcare services. These two policies are taken together, because they show how strongly family policies are focusing on families with young children below compulsory school age. Thus they have an impact on the critical stage of family formation in the family life cycle. Family allowances, on the other hand, usually cover all families with children until the age of 18 or even above and represent more extensive income policies. Family allowances represent more than 60 per cent of total expenditure in Austria, France, Belgium, the United Kingdom, the Netherlands and Malta. On the other hand, these countries show low figures for childcare services. Yet these data should be interpreted with caution, because there is a major statistical problem. For example, in Belgium, France, the Netherlands and the United Kingdom most childcare is located in the preschool sector which belongs to education. And education expenditures are by definition excluded form social protection statistics. Still, what clearly stands out is the huge share that the Scandinavian countries as well as Slovakia and Estonia spend on maternity/parenting policies and childcare. Among the new member states, only Poland and Slovakia spend more than 50 per cent of the total on family allowances. Most Eastern European countries are characterised by a relatively high share of expenditure on young families. This policy profile clearly contrasts with the income approach prevailing in Britain and the Netherlands and some of the Southern European countries. Again we find no indication that the new member states are moving towards the ‘Autonomy’ model. 2.2 Family allowances The most important element of family policy in most countries is family allowances. Here we can test our three hypotheses with respect to two characteristics: eligibility conditions and benefit-rate variations by family size and age of children (see Table 4.2). In most European countries eligibility is based on residence. Only four countries (Es, It, Be, Gr) have a categorical system based on employment and thus exclude parts of the population. Most countries provide benefits without income conditions, but eight apply an income-test: five new and three old member states (see also Förster and Toth 2001). But if we locate these countries geographically, only three (Cz, Pl, Sl) belong to the former socialist states whereas five are in Southern Europe. Moreover, it seems that in the former socialist countries income testing is embedded in a universalistic approach. It thus reflects scarce resources rather than policy principles whereas in Spain and Italy there is no sign of universalistic ambitions. France deserves a special note, because it is the only country in Europe which does not provide allowances for the first child. Benefit rates tell us something about social or demographic aims related to family allowances. In most European countries benefit rates vary with family size, but not with children’s age. Pro-natalist elements are absent in a varied group of countries ranging from Portugal in the south-west to Estonia in the north-east of Europe. The United Kingdom and Germany also belong here. There seems to be no general pattern with respect to ‘families of nations’, but the countries that have benefits varying by both age and family size form an interesting group. They are all among the European family policy pioneers and have long histories of family allowances. Except for the Netherlands, they are predominantly Catholic.
110
Thomas Bahle
Table 4.2 Eligibility for family allowances and variations in benefit rates, Europe 2004 Eligibility
Income-tested
Not income-tested
Universal coverage (residence)
Czech Republic Malta Poland Portugal Latvia Slovenia Cyprus
Categorical system (employment)
Spain Italy
Germany Ireland Latvia Luxembourg Hungary Austria Slovakia United Kingdom Denmark Estonia (France)1 Lithuania Netherlands Finland Sweden Belgium Greece
Benefit rates
Varying by age
Not varying by age
Varying by number of children
Belgium France Luxembourg Netherlands Austria
Not varying by number of children
Czech Republic Denmark Portugal
Greece Ireland Italy Cyprus Latvia Hungary Malta Poland Slovenia Finland Sweden Germany Estonia Spain Lithuania Slovakia United Kingdom
Source: MISSOC 2004 Notes: Plain text: general age limit for eligibility; underlined text: extended period of payment for children in education or unemployed children. For disabled children most countries provide extended periods or unlimited benefits. 1 France: starting with second child only.
The amount of family allowances widely varies across Europe (see Figure 4.3). For a three-child family, benefit rates range from more than 20 per cent of GDP per capita in Germany to less than 5 per cent in Greece. Benefit rates for large families are highest in Western continental European countries (Germany, Belgium, Luxembourg and Austria; Hungary may also be included). High rates are also paid in two new Southern member states (Malta and Cyprus) and two Northern countries (Finland and Sweden). Among the countries with the lowest benefit rates we find six new member states and four ‘older’ southern members; Spain and Greece are by far the least generous countries. Apart from Malta, Cyprus, Hungary and Estonia, all new
Country
Family policy patterns
111
de mt be lu at fi hu se cy ie nl fr it1 ee dk uk lv cz it2 sk pl lt pt sl es gr 0
2
4 6 8 10 12 14 16 18 20 Benefit rates for families with one, two and three children as % of GDP per capita 1 child
2 children
22
3 children
Figure 4.3 Variations in family allowances, Europe 2004 Source: Own calculation based on Eurostat and MISSOC
member states provide allowances of less than 10 per cent of GDP per capita for a three-child family, but the situation in Southern Europe is even worse. In general, the one-child family gets much less. Only in Malta, the United Kingdom and Lithuania does the first child receive a relatively higher amount than the following siblings (if we look at the benefit share for each individual child). In Slovakia, Estonia, Spain and Germany benefit rates show a linear progression by family size. Progression is strong in Belgium, France, Luxembourg, Hungary and Italy. 2.3 Maternity and parenting benefits Maternity benefits belong to the oldest family policies and are nowadays found in all European countries, also due to EU directives (see Figure 4.4). By contrast, parenting benefits are relatively new. Here we still find huge variations within Europe. In 2004 all European countries have paid maternity leave programmes ranging from 14 weeks
Thomas Bahle
Country
112
dk cz fi sk fr lt ee hu pl at lu lv de se sl it uk be nl cy es ie pt gr mt 0,0
10,0 20,0 30,0 40,0 50,0 Maximum payment period (cumulated in months) Mb wage
Mb flat
Pb wage
60,0
Pb flat
Figure 4.4 Maternity and parenting benefits, Europe 2004 Source: MISSOC
(in Figure 4.4 calculated in months, i.e. 3.3 months) to more than one year in Sweden.19 In all countries except Greece and Malta maternity benefits are wage related. The important difference between countries is the parental benefit. In 2004 the majority of EU countries has a parenting benefit, usually paid parental leave. Only in some Southern countries (Spain, Cyprus, Portugal, Greece and Malta) and in the UK, Ireland, the Netherlands and Belgium is there no paid parental leave. All former socialist countries have a parenting benefit, most of them on a universal basis. There is thus no sign for a rudimentary family policy in the new member states, except in Malta and Cyprus. This finding is confirmed if we look at the total time period for which maternity and parenting benefits are paid in each country. In fact, the Eastern European and the Scandinavian countries are the most generous (see Figure 4.4). Among the eight former socialist new member states only Slovenia provides parental benefits for a shorter period than Germany. Quite clearly the Southern and the North-west European countries are the true representatives of rudimentary family policies in Europe, not at all the East Europeans.
Family policy patterns
113
With respect to the total amount of payments for maternity/parenting that can be accumulated over the whole period of eligibility, one has to distinguish between countries with wage-related and those with flat-rate benefits. Only four countries in Europe rely exclusively on wage-related benefits: Denmark, Lithuania, Slovenia and Italy. In Sweden, wage-related benefits form the largest but not exclusive part of parental insurance. Denmark is the most generous country. Here parents receive about 33 times the full monthly wage for the whole period of eligibility, in Lithuania 29, in Slovenia and Sweden about 12 and in Italy 6. Among countries with flat-rate benefits, Luxemburg, Austria, Finland and Germany are the most generous while Slovakia, Hungary, Poland, Estonia and Latvia pay much less. Thus the Scandinavian countries (the universality group) have the strongest employment-related profile.20 By contrast, the Western European continental countries (the subsidiarity group) provide comparatively high flat-rate benefits. Slovakia, Hungary and Poland may be added to this group. Second, most new member states (except Cyprus and Malta) introduced some kind of benefit whereas most southern countries and the autonomy group (United Kingdom, Ireland, the Netherlands) did not. 2.4 Childcare services The third and last family policy area studied in this chapter is childcare services. Here also we find wide variation across Europe (Table 4.3). In all countries coverage rate for children aged 3–5 is higher than 50 per cent and only in Lithuania and Poland are figures lower than 60 per cent (no data for Ireland). Five countries have coverage rates of more than 90 per cent: Fr, Be, Nl, It and Dk. All of them except Denmark have a long tradition of preschools going back to the nineteenth century. In fact, this is the most important factor explaining differences between countries in this age group. Preschools developed primarily in countries in which there was a strong competition between state and church. Countries in the subsidiarity group were the first which developed preschools and today they still provide most places. The picture looks different for children aged 0–2. Here, three groups of countries stand out. First of all, Denmark and Sweden are on top of the league. Finland provides less, but has the most highly developed system of paid home care. High coverage rates are also found in France and Belgium. These two countries have moved a bit towards the Scandinavian model. Today also the UK and Ireland provide a high number of places for this age group, but in the UK provision is often part-time and mostly paid by private sources. The same is probably true for Ireland. The bottom of the league is also interesting. Here we find a mixture of countries belonging to all family policy forms. Yet Eastern European countries in general are not overrepresented among the laggards. There is great variation among them: Slovakia has a very high rate, followed by Latvia and Lithuania (no figures for Estonia), and with some distance by Hungary, Poland and the Czech Republic. The Southern European countries do not have higher rates than the new member states. Most childcare is publicly funded or receives state subsidies, except in the UK and Spain where sizeable private markets exist. On the other hand, in most countries parents have to pay user fees, in particular for children aged 0–2. According to OECD (2005) figures the fee for a two-year-old child in centre-based day care varies from 6 per cent to 34 per cent of average production workers’ wages. Fees are particularly high in
114
Thomas Bahle
Table 4.3 Childcare coverage rates in Europe, around 2000 Country
Children 3–51
France Belgium Netherlands Italy Denmark Czech Republic Hungary Germany Spain Sweden Austria Latvia Portugal Slovakia Greece Finland Slovenia United Kingdom Lithuania Poland Ireland
100 99 98 95 91 88 88 87 81 80 79 78 75 71 70 66 64 60 55 51 —
Children 0–22
User Fees3
29 30 17 6 68 2 6 10 5 65 13 13 22 46 3 25 5 26 10 2 124
34 16 29 — 8 10 6 12 6 6 13 — 19 6 7 8 — 25 — — 30
Sources: Bahle and Pfenning (2003); Neyer (2003); for OECD countries: OECD (2001; 2003; 2005: 13, 15); EURYDIKE (2005); for new member states: UNICEF transMONEE database (1999) Notes: 1 Centre-based care in preschools, kindergartens and other day-care institutions. 2 Figures include full-time and part-time care in licensed childcare centres (public and private) and places provided by registered childminders; most private centres are publicly funded, except in the UK. 3 For a 2-year-old child as % of average production worker’s wages. 4 In the case of Ireland, figures shift between 12 and 38% depending on the source. In OECD Employment Outlook (figures in OECD 2003) it is 38%, in OECD (2005) it is 12%. Eichhorst and Thode (2002) refer to OECD (2003) figures. As in the case of the UK, most provision above the 12% may be privately funded.
France, Ireland, the Netherlands, the United Kingdom, Portugal and Belgium – that is in the Autonomy countries and in two subsidiarity countries with high childcare coverage rates for this age group (Fr, Be). Obviously, in these countries the state has not taken full financial responsibility for securing compatibility between family and work. Yet in France as in the UK and Germany, one has to take into account that parents can deduct part of childcare costs from taxable income. User fees are low instead in the Scandinavian countries and in those Eastern European countries for which data are available. Figure 4.5 combines the two policy elements focusing on children below compulsory school age: maternity/parenting benefits and childcare services. The figure shows the ‘policy package’ for children below the age of 3–4.21 The horizontal axis shows the childcare coverage rate for children below 3, the vertical axis the maximum duration of paid (!) maternity and parenting leave in months. If we divide both country distributions by their un-weighted arithmetic means, we get four groups of countries with typical policy combinations. In the right-hand upper cell are situated countries
Maximum payment of maternity and parenting benefits (in months)
Family policy patterns
115
60 dk
50
cz
fi
40 hu
lt
30 pl
de
fr
sk
lv at
20 se sl it
10
uk es
ie
nl
pt
be
gr
0
0
10
20 30 40 50 60 Childcare coverage rate for children aged 0–2
70
80
Figure 4.5 Childcare services and parenting policies, Europe around 2000 Source: Parenting: MISSOC; child care: see Table 4.3
with generous policies in both dimensions: Denmark, Finland, France and Slovakia. In the right-hand lower cell are located countries with relatively high childcare coverage, but limited parenting benefits: Portugal, the United Kingdom and Belgium. Sweden falls somewhat in between these two groups. In the left-hand bottom cell we have countries with low policy profiles in both dimensions: typically Southern European countries plus Slovenia, Ireland and the Netherlands. Except Slovenia and Slovakia all Eastern European countries are located in the left-hand upper cell: the group characterised by extended parenting policies and less developed childcare services. This is the typical subsidiarity group to which Austria and Germany belong. There are no figures for Luxembourg, but it would perhaps also be here. The empirical analysis thus shows two major results. First, the new member states do not constitute a group of their own characterised by underdeveloped family policy. In fact, with respect to most indicators, the Southern European and a few other countries are situated well below them. The new member states in most respects take an intermediate position within the European spectrum. Second, if we look at family policy profiles, the new member states do not constitute a homogeneous group either. Rather, like the old member states, they belong to different ‘families of nations’. In which way, then, have the new member states transformed the European landscape of family policies?
3. European family policy patterns In order to summarise the results of the analysis one can combine two key family policy dimensions: state policies with respect to the family–work relationship (including
116
Thomas Bahle State interference into family–work relationship Work compatibility
Mixed approach
Denmark Sweden
Finland Estonia Latvia
State-provided family income Universality and equality
1
Subsidiarity approach
Germany 3
No policy
Netherlands United Kingdom Ireland 4
Slovakia Preference for large families
Preference for lowincome families
France Belgium
2
Luxembourg Austria Hungary Czech Republic Poland
Slovenia
Italy Malta Cyprus
No policy
Lithuania
5
Portugal Spain Greece
Figure 4.6 Synopsis 2: Variations in family policies – a ‘cultural map’ of Europe Notes: Degree of institutionalization decreases from left-hand upper corner to the right and to the bottom of the table (e.g. strongest institutionalization in Denmark, weakest in Greece). Underline text: new EU member states. Explanation of categories: State interference into family–work relationship focuses on size and mix of parental benefits and childcare services. Work compatibility: childcare services more developed than parenting policies. Mixed approach: childcare services and parenting policies balanced. Subsidiarity approach: childcare services less developed than parenting policies. No policies: low provisions in both areas. State provided family income focuses on scope and size of family allowances. Universality and equality: cash benefits focus on all children without much differentiation. Preference for large families: benefits are strongly graded by family size. Preference for low-income families: benefits are either income-tested or vary strongly with income. No policy: family benefits play a minor role in general.
maternity/parenting benefits and childcare services); and state-provided income benefits for families (in particular family allowances) (see Figure 4.6: synopsis 2). For each of the two policy dimensions I distinguish four categories in order to ‘place’ individual countries. With respect to the family–work relationship (horizontal axis) the category ‘work-compatibility’ is characterised by a high level of childcare services and strong employment-relatedness of parenting benefits. The ‘subsidiarity approach’ is characterised by highly developed parenting policies, but few childcare services. The ‘mixed approach’ falls in between these two. ‘No policy’ means low benefits and few services. With respect to family income policy (vertical axis) ‘universality and
Family policy patterns
117
equality’ means universal eligibility to family allowances and uniform benefits for each child. ‘Preference for large families’ means that family allowances are strongly graded by family size. ‘Preference for low-income families’ implies a strong component of income-testing and ‘no policy’ stands for low and very limited benefits. Figure 4.6 (Synopsis 2) shows that in general countries with a high profile of workrelated policies (‘work compatibility’) tend to have universal uniform family allowances (1). By contrast, countries with a weak profile of work-related policies tend to have very limited family allowances or concentrate benefits on low-income families (5). They have poorly developed family policies in both dimensions. In between we find the largest group of countries characterised by policies which support parenting in the home and offer size-graded family allowances, sometimes also dependent on income (3). This group is very heterogeneous. Thus, in general, countries are ‘placed’ close to the diagonal line running from the left-hand upper corner to the right-hand bottom corner which means that family policies are either highly developed in all dimensions or poor in all – there is no substitutive relationship between policy fields. Two groups, however, deviate from this pattern. Countries in group 2 combine a high degree of work-related policies with a strong demographic component in family allowances. Countries in group 4 combine weak work-related policies with universal family allowances. A closer look at individual countries belonging to these five groups shows interesting overlaps, but also some deviations from our ‘historical’ families of nations postulated in synopsis 1 (Figure 4.1). Group 1 is consistent with our reflections at the beginning. It comprises the Scandinavian countries and two of the new Baltic member states with close cultural affinity to Scandinavia (Estonia and Latvia). These countries represent the universality model. Group 5 is also consistent with our model. It comprises only Southern European countries, among them the two new member states Cyprus and Malta. These countries are characterised by familism. Note that not a single former Socialist state joined this group. The autonomy model is present in group 4 consisting of the United Kingdom and the Netherlands, in part also by Ireland, which, however, in other aspects is closer to the southern country cluster. Note also here that no new member state joined this group.22 But what happened to the subsidiarity group, the one comprising most countries? I have already pointed out at the beginning that this is a very heterogeneous group, but the countries have one thing in common: their emphasis on supporting the family as a social institution. First of all, we can see that this group consists solely of Western and Eastern Central European countries; there is no Southern and no Northern country included. In this respect, our reflections at the beginning are confirmed. Also confirmed is the assumption that the new Central European member states (Hungary, the Czech Republic, Poland, also Lithuania and Slovenia) joined the subsidiarity group, though in slightly different ways with respect to income policies. Yet the Western part of the continent, Belgium and France, seems to have parted from this group and moved closer to the universality model of Scandinavia, at least with respect to work-related policies. Also Slovakia seems to have moved in this direction. In fact, Slovakia seems to be the only Central-eastern European country that clearly did not join the continental group. Do the country groups identified above (the ‘families of nations’ based on policy characteristics) also show typical problem constellations (in terms of social indicators)? Figure 4.7 gives an example for total fertility and child poverty rates, certainly two
118
Thomas Bahle tr
2,20
2,00
ie fr
Total fertility rate
4 dk
1,80
se
fi nl
1
2 be
uk
lu
1,60 cy at
1,40
de
5
ee
3
it hu
sl
1,20
1,00
5
cz
gr
lv
pt
mt lt
pl
es sk
10 15 20 25 30 Child poverty rate (at 60% of median equivalized household income)
35
Figure 4.7 Family policy problem constellation, around 2000 Source: Own calculations based on Eurostat
core indicators.23 All EU countries have fertility levels that fall into a small interval ranging from 1.2 to 2.0; Turkey is at 2.2; Child poverty rates range from about 6 per cent to more than 30 per cent. The Scandinavian countries form a distinct group characterised by very low child poverty rates and high but not exceptional fertility (group 1 in Figure 4.7). Close to them is France together with the three Benelux countries (2), although their poverty rates are higher. Below these two groups, we find the Central European cluster including Austria, Germany, the Czech Republic, Hungary and Slovenia (3) characterised by low fertility and low to moderate poverty levels. Group 4 with the United Kingdom and Ireland shows high fertility and high poverty, a really exceptional combination in the EU context. Group 5 clusters around the three Southern European countries Italy, Portugal and Spain and is characterised by low fertility and high poverty; Poland, Malta and Slovakia are close. Greece and Cyprus seem to be exceptions in the Southern European context: they show low poverty levels and are closer to the continental group 3. In between group No. 3 and group No. 5 we have the three Baltic countries showing medium to high poverty levels and low fertility. Though there is a loose correspondence between the ‘families of nations’ (Synopsis 2) and the country clusters on ‘problem’ indicators (Figure 4.7), it is obvious that other factors ‘intervene’ into this relationship. One important factor certainly is female employment. It is, for example, interesting that in most countries with fertility levels of above 1.6 (the upper part of Figure 4.7), female employment is high. By contrast, in countries with fertility levels of less than 1.4 it is low. Also policy efforts as such count. Countries with child poverty rates higher than 20 per cent are generally characterised by poorly developed family policies.
Family policy patterns
119
4. Conclusion The results thus reject hypotheses 1 and 2 on the new member states, but only partially confirm hypothesis 3. How can this be explained? The analysis in my view confirms the power of history. After the fall of communism, the new member states in Central East Europe revived parts of their older cultural and institutional inheritance that was common to the whole central region of the European continent prior to World War II. This included a developed set of intermediary institutions supported by the state, in particular the family, playing an important role in society. The Southern EU countries, to which Malta and Cyprus also belong, have a quite different history. Over long periods of time the church and not the state was the prime institution for social integration and control. This is not only true for the Catholic part of the Mediterranean region, but even more for Orthodox Greece and Cyprus which is part of the Greek culture. During the long reign of the Ottomans there was no Greek state and the Orthodox church was the only national power.24 Moreover, for the whole South late democratisation played an enormous role. Except for Italy, democracy has gained ground only since the mid-1970. The state until then was in the hands of conservative elites, often in coalition with the Catholic clergy, and was not an important agent for social integration. But what happened to France and Belgium? In the case of France, the state in the long run became much more powerful than the church. Although in the late nineteenth and early twentieth century, family policy in France was primarily rooted in the social-catholic movement (Schultheis 1988), the state transformed it according to its own views – often in conflict with the church. The ‘secularised’ character of French family policy is visible in two elements: a unique pro-natalism and strong efforts to integrate young mothers into the labour market. In France, the family is still an important social institution, but there is also a dense network of social services for families. As in Scandinavia one might say that the child is not primarily a member of a family, but a member of society. The three core characteristics of present French family policy – pro-natalism, child-orientation and work-integration – may show the way ahead to other continental European countries which all face similar problems: low birth rates, increasing poverty and low employment levels.25 Thus France may be the forerunner among the continental countries, which brings it closer to the relatively successful Scandinavian model of family policy that shows low rates of poverty and high rates of fertility and female employment. Therefore, the subsidiarity model is definitely in crisis. Moreover, Germany is starting to leave this model behind and moving towards the Scandinavian form. Childcare services for children under the age of 3 are being expanded, and the new coalition government introduced wage-related parental benefits according to the Swedish model of parental insurance. In contrast to these developments in the centre of Europe, family policies in the countries belonging to the familism model seem to stagnate, even though it is in severe crisis. In fact, the Southern European countries have the highest child poverty and lowest fertility rates in Europe. They also have the lowest level of employment. Despite this, there is no sign that family policy is changing fundamentally in these countries. By contrast, British family policy has been revived and partly re-oriented by New Labour governments. In particular, childcare services have been expanded strongly. Government is moving towards a policy by which employment is
120
Thomas Bahle
actively supported. Here also we can see a move towards the universality form of family policy. Similar developments can be seen in the Netherlands where childcare has also become an important issue since the part-time economy has grown to limits. To summarise, in family policy Eastern European countries found their place amidst older member states. Most of them joined the Central European subsidiarity model. At the same time, however, this model is changing. All over Europe except in the South family policy seems to be moving closer to the Scandinavian model. This would certainly not be the worst scenario. This development is pushed by strong political mechanisms. Family policies have gained prominence in the political process, foremost with respect to the electorate, but also within pressure groups.26 The south, however, seems to fall back seriously. This danger increased with the entrance of Romania and Bulgaria into the EU in 2007 and the envisaged further enlargements to other Balkan states or even Turkey. In these countries, intermediary institutions play a minor role and the state is not a major welfare provider Furthermore, these countries are characterised by strong extended family groups that usually have prevented public family policies. In the west, state family policy could only develop, because the dominant living form was the conjugal nuclear family. This can be regarded as part of the worldwide European exceptionalism. Extended family systems with their strong solidarity norms and forms of social control set up high barriers against any form of state interference. In this respect, family systems in candidate countries are much stronger than in Western Europe and even surpass the southern member states. Therefore the entry of these countries would even strengthen the profile of the Southern pattern. In this case the danger is that Europe will become more divided with respect to family policy – not between west and east, but between north/centre and south. To conclude, the chances for a common European family policy model are faint. Very likely family systems, religious and state traditions will continue to shape state– family relationships and family policies. And probably their impact will be heavier and will last longer than the fading heritage of communism that split the European continent into east and west for some forty years.
Notes 1 I’d like to thank Jens Alber, Franz-Xaver Kaufmann, Mathias Maucher and Chiara Saraceno for very helpful discussion and constructive critique on an earlier version of this chapter. 2 The EU is mainly active in policy areas related to employment. One is co-ordination of family benefits among member states with respect to EU citizenship and internal work migration. Another is gender equality that was already postulated in the founding treaties of Rome in 1957. 3 The term ‘families of nations’ is used in the sense of Castles (1993), i.e. identifying groups of nations which share central policy characteristics that at the same time distinguish them from other groups. 4 This is done by Keck and Blome, this volume Ch. 3. 5 The Catholic Church and the system of seigneurial domination had a major impact on the historical formation of the Western European family system in which the nuclear family form with weak kinship ties predominates. The Catholic church weakened kinship through strict exogamy rules and the principle of consensual marriage, not to mention that the Christian community per se is constituted by individual confession rather than descent. Seigneurial domination established an economy based on social contract rather than kinship. Moreover, servitude and patronage were limited to the family household.
Family policy patterns
121
6 The idea for this map was inspired by Rokkan (1999) and Flora (1996); further details are given in Bahle (2004). The reader who is familiar with the family policy literature will find differences to and similarities with other models. Somehow entered into my concept are models by Pfau-Effinger (2005), Gornick et al. (1997) and Vogel (2003). Further arguments can be found in the introduction to Pfenning and Bahle (2000). 7 The French case is very interesting in this respect. Historically, France was a very heterogeneous country that became unified by a powerful political centre located in the north. The nuclear family prevailed, but elements of the stem family played an important role. Napoleon introduced equal inheritance for all children. This became the official norm, but could not suppress others completely. Therefore one may say that the guiding image of French family policy is a mixture between nuclear and stem family characteristics. It was shaped by both Catholicism and the laic state that had strongholds in different parts of the country. This may also explain the typical mixture of tradition and modernity in French family policy. 8 There are significant variations within the south, however. In the southern parts of Italy and Spain, for example, we find a nuclear family form with strong norms of equality regarding siblings (Todd 1990). This is probably due to the agricultural system in which large estates dominated and the majority of the population were agricultural labourers with strong equality norms and no land to transmit. These areas also experienced a strong decline in religiosity in the eighteenth and nineteenth centuries still visible much later. In the modern period these areas are politically dominated by clientelism or left-wing radicalism. By contrast, the central and most northern parts of Italy and Spain were strongholds of Catholicism and dominated by the stem family system. Since these regions were the centres of nation-building in the nineteenth century, their cultural heritage became dominant in the institutional architecture of national society. 9 Not even the fascist period altered the balance of power significantly on the side of the state. The Franco regime, for example, was characterised by open collaboration of church and state. 10 Goody (1994) argues that only the Catholic Church had loosened ties between the nuclear family and wider kinship. The church tried to impose consensual religious marriage and thereby fought against the power of kinship groups. Most important, however, was the strategy to regulate inheritance in favour of the church itself. 11 In all former Communist countries the Socialist model of state–family relationships dominated and pushed them away from common European developments. For studies that ‘place’ these countries in a European context see Szücs (1994, 19832), Berend (1986, 1996), Spohn (2000) and Flora (2000). 12 For a more thorough discussion of the transition see Alber and Standing (2000), Deacon (2000), Fajth (1999), Götting (1998) and Kovacs (2002). On cultural variations see Gerhards (2005). 13 The literature on this subject is rich. I have mainly relied on Berend (1999), Heidenreich (2003), Manning (2004), Wagener (2002), Zielonka and Mair (2002) and Campbell and Pedersen (1996). 14 A good example is social security. In the interwar period Central-east European countries had followed the Bismarck approach. Today some ‘old’ institutional elements reappear. 15 A good example for a common cultural space is the Baltic Sea region including Scandinavia and the post-soviet Baltic states. 16 In order to rephrase the subtitle of Berend’s book (1996): ‘Detour from the periphery to the periphery.’ 17 The analysis is based on cross-sectional data provided by international sources. It should be noted that these data do not always ‘truly’ reflect institutional specificities of individual cases. Still, they offer a solid basis for studying broad variations within Europe. For quantitative data I have used EUROSTAT social expenditure data, because in contrast to OECD data it includes all EU member states. In addition, UNICEF transMONEE data are used for new member states. Website links are given at the end of the chapter. OECD data is used for childcare (OECD 2001; 2003; 2005). In a few cases I have taken additional data from Abramovici (2003), UNICEF (1999), Neyer (2003) and Bahle and Pfenning (2002). With respect to institutional information, the main source is MISSOC. In addition to the comparative tables, background reports and data on specific policy fields and the new member states are available on the web. All these sources are indicated as MISSOC. Information on new member states is also provided by UNICEF. EURYDIKE is used for preschool systems. Dörfler (2002) and
122
Thomas Bahle
Rostgaard (2004) have also collected information on family benefits in new member states. A general source for comparative information on family policies is the Clearinghouse on International Developments in Child, Youth and Family Policies at the Columbia University (New York). I have not used any national data. 18 Eurostat social expenditure data do not include figures for Cyprus. 19 In Sweden this is an integral part of parental insurance which combines maternity and parenting benefits to which both parents are entitled. 20 Finland and Sweden need a short note. Both countries have an exceptional combination of flat- and wage-related benefits. The Finnish home-care allowance comes close to the subsidiarity group whereas Sweden represents the typical profile of Scandinavian employmentoriented policies. 21 Denmark and Sweden have programmes in which parents can divide the total length of the leave several times over an extended period, also when children are already in school. For comparability reasons it is assumed in the figure that the whole period is taken until the child’s fourth birthday. In Denmark the total length of maternity and parental leave exceeds 48 months. 22 Fux (2002) and Künzler (2002) come to similar country groupings. Fux distinguishes between three family policy models on the basis of dominant ‘actors’: etatistic, familialistic and individualistic (2002: 372ff.). By factor analysis he shows that Sweden and Denmark closely fit the etatistic type, France, Italy and Germany the familialistic one and the Netherlands and Switzerland the individualistic form. The UK – though in principle part of the ‘individualistic’ world – differs by a strong focus on anti-poverty policy. Künzler refers to a typology originally developed by Hans-Joachim Schulze. He distinguishes between economic and ecological intervention into the family. The Southern European countries plus the Netherlands score low on both dimensions (2002: 277ff.) whereas the Scandinavian countries plus Belgium and France score high on both. Luxemburg, Austria and (West) Germany score high on the economic and low on the ecological dimension. This fits very well with my own synopsis. 23 This is not the place to discuss the nature of the relationship between policies and ‘problem’ or ‘outcome’ indicators. The problem indicators used here can be interpreted as ‘motives’ for policies, but in the longer run also as ‘outcomes’. In this chapter I simply want to find out whether the policy-based ‘families of nations’ are also reflected in specific combinations of ‘problem’ indicators. 24 In a similar way this is also the case for Ireland which may partly explain why this country shares some characteristics with Southern Europe. 25 On low fertility see Neyer (2003), on poverty Vleminckx and Smeeding (2001) and on employment and family Fine-Davis et al. (2004). 26 Further analysis of family policies needs to take into account the political process. In the German case, for example, it is crucial that in the last elections the Christian Democrats obviously had lost many votes among young, urban, well-educated women who want to combine family and employment. Women in the past have tended to vote more conservatively than men, but this has obviously changed in younger cohorts. In addition, there is increasing pressure from potent interest groups, not only the women’s movement. Employers’ organisations are in favour of more employment-related family policies which help them secure permanency among their young, well-educated female staff.
References Abramovici, G. (2003) ‘Social protection: cash family benefits in Europe’, Eurostat Statistics in Focus, theme 3, no. 19, 2003, Luxemburg: Eurostat. Alber, J. and Standing, G. (2000) ‘Social dumping, catch-up, or convergence? Europe in a comparative global context’, Journal of European Social Policy, 10, 2: 99–119. Bahle, T. (1995) Familienpolitik in Westeuropa. Ursprünge und Wandel im internationalen Vergleich, Frankfurt. a.M.: Campus. Bahle, T. (2004) ‘Staat, Kirche und Familienpolitik in westeuropäischen Ländern. Ein historisch-soziologischer Vergleich’, Politische Vierteljahressschrift, Sonderheft 33, Politik und Religion: 391–411.
Family policy patterns
123
Bahle, T. and Pfenning, A. (2003) Angebotsformen und Trägerstrukturen sozialer Dienste im europäischen Vergleich, Mannheimer Zentrum für Europäische Sozialforschung, Working Paper no. 42, Mannheim: MZES. Berend, I.T. (1986) ‘The historical evolution of Eastern Europe as a region’, International Organization, 40, 2: 329–346. Berend, I.T. (1996) Central and Eastern Europe, 1944–1993: Detour From the Periphery to the Periphery, Cambridge: Cambridge University Press. Berend, I.T. (1999) ‘The further enlargement of the European Union in a historical perspective’, European Review, 7, 2: 175–181. Campbell, J.L. and Pedersen, O.K. (eds) (1996) Legacies of Change: Transformations of Post-communist Societies, New York: de Gruyter. Castles, F.G. (1993) Families of Nations: Patterns of Public Policy in Western Democracies, Dartmouth: Aldershot. Deacon, B. (2000) ‘Eastern European welfare states: the impact of the politics of globalization’, Journal of European Social Policy, 10, 2: 146–161. Dörfler, S. (2002) Familienpolitische Leistungen in ausgewählten europäischen Staaten auflerhalb der Europäischen Union, Österreichisches Institut für Familienforschung, Arbeitspapier no. 30, Vienna: ÖIF. Eichhorst, W. and Thode, E. (2002) Vereinbarkeit von Familie und Beruf. Benchmarking Deutschland Aktuell, Bertelsmann- Stiftung, Gütersloh: Verlag Bertelsmann Stiftung. EURYDIKE (2005) Key Data on Education in Europe, Luxemburg: Eurostat. Fajth, G. (1999) ‘Social security in a rapidly changing environment: the case of the post-communist transformation’, Social Policy and Administration, 3, 4: 416–436. Ferrera, M. (1996) ‘The Southern welfare state in social Europe’, Journal of European Social Policy, 6, 1: 17–37. Fine-Davis, M., Fagnani, J. and Giovannini, D. (2004) Fathers and Mothers: Dilemmas of the Work-Life Balance. A Comparative Study in Four European Countries (France, Italy, Denmark, Ireland), Dordrecht: Springer. Flora, P. (1996) ‘Introduction’, pp. xi–xxxvi, in: P. Flora (ed.), Growth to Limits. The Western European Welfare States Since World War II, Berlin: de Gruyter. Flora, P. (2000) ‘Externe Grenzbildung und interne Strukturierung – Europa und seine Nationen. Eine Rokkan’sche Forschungsperspektive’, Berliner Journal für Soziologie, 2: 151–165. Förster, M. and Toth, I.G. (2001) ‘Child poverty and family transfers in the Czech Republic, Hungary and Poland’, Journal of European Social Policy, 11, 4: 324–341. Fux, B. (2002) ‘Which models of the family are encouraged or discouraged by different family policies?’, pp. 363–418, in F.-X. Kaufmann, A. Kuijsten, H.-J. Schulze and K. P. Strohmeier (eds), Family Life and Family Policies in Europe, Vol. 2: Problems and Issues in Comparative Perspective, Oxford: Oxford University Press. Gauthier, A.H. (1996) The State and the Family: A Comparative Analysis of Family Policies in Industrialized Countries, Oxford: Oxford University Press. Gerhards, J. (2005) Kulturelle Unterschiede in der Europäischen Union. Ein Vergleich zwischen Mitgliedsländern, Beitrittskandidaten und der Türkei, Wiesbaden: VS Verlag. Goody, J. (1994) The Development of the Family and Marriage in Europe, Cambridge: Cambridge University Press. Gornick, J.C., Meyers, M.K. and Ross, K.E. (1997) ‘Supporting the employment of mothers: policy variation across fourteen welfare states’, Journal of European Social Policy, 7, 1: 45–70. Götting, U. (1998) Transformation der Wohlfahrtsstaaten in Mittel- und Osteuropa. Eine Zwischenbilanz, Opladen: Leske und Budrich. Hajnal, J. (1965) ‘European marriage patterns in perspective’, pp. 103–143, in D.V. Glass, and D.E.C. Eversley (eds), Population in History: Essays in Historical Demography. London: Edward Arnold.
124
Thomas Bahle
Hantrais, L. (2004) Family Policy Matters. Responding to Family Change in Europe, Bristol: Policy Press. Heidenreich, M. (2003) ‘Territoriale Ungleichheiten in der erweiterten EU’, Kölner Zeitschrift für Soziologie und Sozialpsychologie, 55, 1: 1–28. Immervoll, H. and Barber, D. (2005) Can Parents Afford to Work? Childcare Costs, Tax-benefit Policies and Work Incentives, OECD Social, Employment and Migration Working Papers, no. 31, 2005, Paris: OECD. Kaufmann, F.-X. (1994) ‘Familienpolitik in Europa’, pp. 141–167, in Bundesministerium für Familie und Senioren (ed.), Vierzig Jahre Familienpolitik in der Bundesrepublik Deutschland, Neuwied: Luchterhand. Kaufmann, F.-X. (2002) ‘Politics and policies towards the family in Europe: a framework and an inquiry into their differences and convergences’, pp. 427–500, in F.-X. Kaufmann, A. Kuijsten, H.-J. Schulze and K. P. Strohmeier (eds), Family Life and Family Policies in Europe, Vol. 2: Problems and Issues in Comparative Perspective, Oxford: Oxford University Press. Keck, W. and Blome, A. (2007) ‘Is there a generational cleavage in Europe?’, this volume: Ch. 3. Kovacs, J.M. (2002) ‘Approaching the EU and reaching the US? Rival narratives on transforming welfare regimes in East-Central Europe’, West European Politics, 20, 3: 175–204. Künzler, J. (2002) ‘Paths towards a modernization of gender relations, policies, and family building’, pp. 252–298, in Kaufmann et al. (eds). Laslett, P., Wall, R. and Robin, J. (eds) (1983) Family Forms in Historic Europe, Cambridge: Cambridge University Press. Lobodzinska, B. (ed.) (1995) Family, Women and Employment in Central-Eastern Europe, London: Greenwood Press. Manning, N. (2004) ‘Diversity and change in pre-accession Central and Eastern Europe since 1989’, Journal of European Social Policy, 14, 3: 211–232. MISSOC (2004) Social Protection in the member states of the European Union, of the European Economic Area and of Switzerland: Situation on 1 May 2003, European Commission, Directorate General for Employment and social Affairs, Luxembourg, Office for Official Publications of the Europeon Communities. Mitterauer, M. (2003) Warum Europa? Mittelalterliche Grundlagen eines Sonderwegs, Munich: C.H. Beck. Neyer, G. (2003) Family Policies and Low Fertility in Europe, Max-Planck-Institut für Demographische Forschung, Arbeitspapier no. 21, Rostock: MPI. OECD (2001) Starting Strong. Early Childhood Education and Care, Paris: OECD. OECD (2005) Society at a Glance: OECD Social Indications, 2005 Edition, Paris: OECD. Pascall, G. and Manning, N. (2000) ‘Gender and social policy: comparing welfare states in Central and Eastern Europe and the former Soviet Union’, Journal of European Social Policy, 10, 3: 240–266. Pfau-Effinger, B. (2005) ‘Welfare state policies and the development of care arrangements’, European Societies, 7, 2: 321–347. Pfenning, A. and Bahle, T. (eds) (2000) Families and Family Policies in Europe. Comparative Perspectives, Frankfurt am Main: Lang. Robila, M. (ed.) (2004) Families in Eastern Europe. Contemporary Perspectives on Family Research, Vol. 5, Oxford: Oxford University Press. Rokkan, S. (1999) State Formation, Nation-building and Mass Politics in Europe: The Theory of Stein Rokkan Based on His Collected Works, ed. P. Flora, S. Kuhnle and D. Urwin, Oxford: Oxford University Press. Rostgaard, T. (2004) Family Support Policy in Central and Eastern Europe: A Decade and a Half of Transition, UNESCO Early Childhood and Family Policy Series no. 8, Paris: UNESCO. Saraceno, C. (2007) ‘Patterns of family living in an Enlarged Europe’, this same volume, Ch.2. Schultheis, F. (1988) Sozialgeschichte der französischen Familienpolitik, Frankfurt am Main: Campus.
Family policy patterns
125
Sleebos, J.E. (2003) Low Fertility Rates in OECD Countries: Facts and Policy Responses, OECD Social, Employment and Migration Working Papers, no. 15, Paris: OECD. Spohn, W. (2000) ‘Die Osterweiterung der Europäischen Union und die Bedeutung kollektiver Identitäten. Ein Vergleich west- und osteuropäischer Staaten’, Berliner Journal für Soziologie, 2: 219–240. Szücs, J. (1983; 2nd edn 1994) Die drei historischen Regionen Europas, Frankfurt a.M.: Verlag Neue Kritik. Todd, E. (1983) La Troisième Planète: Structures familiales et systèmes idéologiques, Paris: Editions du Seuil. Todd, E. (1990) L’Invention de l’Europe, Paris: Seuil. UNICEF (ed.) (1999) Women in Transition, Florence: UNICEF. Vleminckx, K. and Smeeding, T.M. (eds) (2001) Child Well-being, Child Poverty and Child Policy in Modern Nations: What Do We Know? Bristol: Policy Press. Vogel, J. (ed.) (2003) European Welfare Production: Institutional Configuration and Distributional Outcome, Dordrecht: Kluwer. Wagener, H.-J. (2002) ‘The welfare state in transition economies and accession to the EU’, West European Politics, 20, 2: 152–174. Zielonka, J. and Mair, P. (2002) ‘Diversity and adaptation in the enlarged European Union’, West European Politics, 20, 2: 1–18.
Data links EU Commission: MISSOC information system http://europa.eu.int/comm/employment_social/social_protection/index_de.htm EUROSTAT database http://epp.eurostat.cec.eu.int/portal/page?_pageid=1090,30070682,1090_33076576&_dad= portal&_schema=PORTAL Eurydike http://www.eurydice.org/portal/page/portal/Eurydice/DB_Eurybase_Home UNICEF transMONEE database on new member states http://www.unicef-icdc.org/resources/ Clearinghouse On International Developments in Child, Youth and Family Policies at Columbia University http://www.childpolicyintl.org/
Part II
Employment and working conditions
5
Employment patterns in the enlarged EU Jens Alber
Introduction To many observers inside and outside the European Union, the European social model lacks attractiveness, because even though subscribing to the goal of full employment in theory, it has for a long time produced high and persistent levels of unemployment in actual practice. As some have argued, Europe simply ‘is not working’, because it does not provide enough jobs (Brown 2005). Full employment has important repercussions for quality of life on several grounds. On the macro-level it is the indispensable basis of the tax state and of the sustainability of pay-as-you-go pension schemes. On the micro-level, work is the major determinant of individuals’ well-being in a dual sense. As the main source of income under modern conditions, it is the major safeguard against poverty and deprivation. As a symbol of belonging, having a job is one of the key underpinnings of people’s sense of inclusion in society, of self-respect and of subjective well-being. As additional determinants of well-being, certain aspects of the quality of jobs – including the rate of pay, the hours of work, and the degree of autonomy – are also important. Since these issues are dealt with in the chapter (6) by Wallace and Pichler on the quality of work, the focus in this chapter is on employment at the macro-level and on the comparison of labour market developments in old and new member states. Under the motto of a ‘Partnership for Growth and Jobs’, the European Commission (2005) re-launched the Lisbon agenda with an emphasis on full employment. A central question then is to what extent enlargement has moved Europe closer to or further away from the goal of work for all. In recent debates concerns have also been voiced that the European social model, which many see as an essential underpinning for quality of life for millions of Europeans, may partly be responsible for the malfunctioning of EU labour markets, because the social protections that form the core of the European social model may hamper either labour supply or labour demand, or perhaps both, and may thus do as much harm as good to quality of life. These issues are highly contentious, and the eastern enlargements have added further complexity and urgency to the debate, as labour market developments in the post-communist new member states have been more turbulent than anything witnessed in Western Europe since the 1930s. The consequences are not confined to the new member states (NMS), as worries about a potential spill-over to the rest of the EU have contributed to an unease about enlargement in several of the old states which seek to regulate access to their labour markets by the various restrictions allowed under the 2003 Accession Treaty, until the period of transitional
130
Jens Alber
arrangements will irrevocably end with the full integration of EU labour markets in April 2011. The chapter takes up some of these issues by showing how employment changed in the EU in recent times and how it varies between countries, groups and economic sectors. Based on official statistics, its main purpose is to identify similarities and differences between old and new member states. Whilst the descriptive part gives an overview of levels and recent changes in employment, as well as a more differentiated account of group- and sector-specific patterns,1 the analytical part takes up three questions. First, it examines to what extent economic growth was coupled with employment and if there were countries which experienced jobless growth. Second, it analyzes to what extent general employment developments trickled down to vulnerable groups such as women, younger workers, and the low-skilled. Finally, the question is raised as to what extent old and new member states share similar welfare state policies and to what extent these impeded or promoted employment developments in the most recent period.
1. A short description of European employment patterns 1.1 Levels and trends of employment After the collapse of the Soviet empire the former socialist countries of Central and Eastern Europe went through an economic slump which was much deeper than the decline Western nations experienced in the Great Depression. On average, economic output declined by 27 per cent, as compared to an average downturn of merely 9 per cent in twelve Western countries during the 1930s (see Figure 5.1).2 The massive economic decline translated also into a sizeable shrinking of the labour market which impinged particularly hard on the industrial sector, where between 20 and 35 per cent of the jobs were lost (Hafemann and van Suntum 2004: 18). In most post-communist new member states, employment rates shrank by 20 percentage points or more after the transition, and the downward trend was usually only reversed after the turn of the century (Figure 5.2).3 This trend reversal in countries which joined the Union has no match among the post-communist nations which remained outside.4 To some extent, then, there is a visible beneficial effect of the accession process which brought the labour markets of post-communist new member states (NMS-8) a comparatively strong upswing. A comparison of single new member states shows that there are also noteworthy country-specific differences, however. The timing of labour market recovery varied just as much as the recent levels of employment. At the end of the old era in 1989, employment rates of the population aged 15–59 had ranged from lows of 75 per cent in Poland and Slovenia to a high of 88 per cent in Estonia (Table 5.1). During the shrinkage following the transition the range of variation grew even somewhat larger, while the rank order of countries was subject to considerable change. The difference between developments in Poland and Slovenia is particularly striking. Departing from practically identical employment levels in 1989, the two countries followed very discrepant trajectories. While Poland remained at the bottom, Slovenia moved to the top of the rank order of the formerly socialist new member states. Slovenia also was first to reverse the downward trend in employment as early as 1993. In contrast, Poland belonged to the countries where labour market shrinkage continued right up to the
Employment patterns
131
115 110 105 100 95
Depression countries
90
Transformation countries
85 80 75 70
1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938
Figure 5.1 Development of economic output during the Great Depression and postcommunist transformation Source: 1928–1938, own calculations, based on Maddison (1964: 201–202). 1989–2003, own calculations, based on Unicef (2004: 91) Notes: Real GDP-development, Base year 1920 or 1989=100; unweighted country averages. Depression: BE, DK, FR, DE, IT, NL, NO, SE, CH, UK, CA, US; post-communist transformation: CZ, HU, PL, SK, SL, EE, LV, LT.
85.0
8 NMS (CZ, HU, PL, SK, SI, EE, LV, LT) 2 CC (BG, RO) 5 EEC (AL, BY, MD, RU, UA) 8 EEC (AL, BY, MD, RU, UA, HR, MK, SM)
80.0
75.0
70.0
65.0
60.0
55.0 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Figure 5.2 Employment rates of the population aged 15–59 in post-communist new member states and in Eastern European countries outside the EU, 1989–2004 Source: Based on Unicef (2006: 121)
132
Jens Alber
Table 5.1 Indicators of employment development in the post-communist new member states Country
Czech Republic Hungary Poland Slovakia Slovenia Estonia Latvia Lithuania Bulgaria Romania
1989 level (% 15–59) 86.9 83.0 74.7 79.6 74.5 87.9 — 83.9 81.5 77.4
Year with lowest employment rate 2004 1997 2003 2000 1992 2000 (1996)1 2001 2000 2004
Low compared to 1989 = 100 81.0 68.1 72.8 75.1 84.0 78.2 — 75.7 67.1 84.5
2004 level 2004 level indexed to (% 15–59) 1989 = 100 81.0 73.7 73.4 75.9 97.9 80.2 (109)2 79.9 72.9 84.5
70.4 61.2 54.8 60.4 72.9 70.5 70.1 67.0 59.4 65.4
Source: Own calculations, based on UNICEF (2006: Table 7.2, p. 121) Notes: 1 First year with data. 2 Compared to 1996 = 100.
eve of accession (2003). The two recent newcomers to the EU, Bulgaria and Romania, represent similarly different pathways out of the socialist past. While Romania still has a higher employment level than Bulgaria, the size of its employed population has been drastically shrinking from 1995 to 2004, whereas, recovering from an even deeper fall, Bulgaria was able to reverse the downward trend in 2000 and has since seen remarkable growth. These vastly discrepant country-specific trajectories suggest that there remains remarkable leeway for specific national policy responses in the presence of pressures from Europeanisation or globalisation. In contrast to UNICEF statistics, international comparisons of labour market policies as published by the European Commission or by the OECD usually express employment rates as a percentage of the population at working age (15–64). However, the Employment Reports issued by the European Commission present such data for the NMS only as from 1998. Even though not arriving at fully identical rank orders of the labour market performance of member countries, the two sources produce similar results with respect to the timing of the turnover and to recent levels of employment.5 There was a relatively early turnaround in Slovenia and Hungary as well as in Latvia, whereas particularly prolonged decline up to more recent years persisted in Poland and Romania, as well as in the Czech Republic where the shrinkage was, however, more moderate and occurring on higher levels of employment. The other countries are in between these extremes. On average, the NMS-10 have an employment rate 4 percentage points lower than the level in the old EU-15 (60.4 as compared to 64.7 per cent in 2004).6 The EU-25 aggregate was 63.3 per cent in 2004. Since the Lisbon European Council of 2000 the European Union has pursued the strategic goal of raising the EU employment rate to 70 per cent by 2010. The Stockholm European Council of 2001 then defined intermediate targets of an employment rate of 67 per cent overall and 57 per cent for women to be reached by 2005, and it also set a further target of 50 per cent for
Employment patterns
133
people aged 55–64 by 2010. It is obvious that enlargement has moved the European Union farther away from reaching these rather ambitious goals. The accession of Bulgaria and Romania only exacerbates the problem, as both countries have employment rates below 60 per cent. Table 5.2 shows, however, that similarities and differences in labour market performance cut across the divide between old and new member states, as there is much heterogeneity in both groups. In the old EU-15, four countries surpassed the target of 70 per cent by 2004, whereas two nations remained below even the 60 per cent level. In the group of new member states, Cyprus was the only one to come close to the target in 2004, whereas five more countries scored at least above 60 per cent. Including Bulgaria and Romania, six new member states still have very low employment rates under 60 per cent. Judging from the most current levels of employment, national differences in the EU thus do not fully coincide with the distinction between new and old member states. In terms of recent employment levels, we find, rather, the following four groups of countries: • •
•
•
four outstanding performers which are all in North Western Europe: Denmark, Sweden, the Netherlands and the UK; a group of eight above-average performers with employment levels over 64 per cent; this group includes three new member states – Cyprus, Slovenia, and the Czech Republic – and five old ones (Austria, Portugal, Finland, Ireland and Germany); Estonia closed up to this group in 2005; a group of seven below- or around-average performers which consists of France, Luxembourg, Belgium and Spain among the old member states, as well as of the three Baltic nations, among which Estonia was recently most successful; a group of six poor performers with employment levels below 60 per cent which includes four new member states – Slovakia, Hungary, Poland and Malta – as well as two old ones (Greece and Italy). In 2005 Greece moved just above the 60 per cent threshold. Having joined the Union in 2007, Bulgaria and Romania now swell the ranks of poor labour market performers in the EU-27.
Of course, the employment levels obtained should also be seen in the context of recent trends.7 Cross-tabulating the levels of employment with changes in the recent period since 2000, it becomes evident that strong employment growth occurred predominantly among hitherto poor labour market performers (Table 5.2). Sizeable increases of more than 2 percentage points were recorded in the three Baltic nations, in Slovenia, Cyprus and in Bulgaria, but only in three of the old member states (Spain, Italy and Greece). Both, the group with above-average increases, as well as the group with employment shrinkage, embrace old and new member states alike. Once again we thus find labour market developments to cut across the divide between old and new Europe. Since recent growth tended to be higher in countries with lower employment levels in 2000, there was some convergence of employment levels in recent years.8 Major exceptions to the general rule were Poland, Malta and Romania, where low levels of employment at the turn of the century led to even further shrinkage in subsequent years. Hungary and Slovakia barely fared better, as low employment levels combined with a stagnant job market. These five countries will thus require sizeable future support from EU structural funds, if the goal of convergence which dominates European cohesion policies is to be reached.
Netherlands (73.1 / 0.2) United Kingdom (71.6 / 0.4) Finland (67.6 / 0.4) Ireland (66.3 / 1.1)
France (63.1 / 1.0) Slovakia (57.0 / 0.2) Hungary (56.8 / 0.5)
Denmark (75.7 / −0.6) Sweden (72.1 / −0.9)
Portugal (67.8 / −0.6) Austria (67.8 / −0.7) Germany (65.0 / −0.6) Czech Republic (64.2 / −0.8)
Luxembourg (62.5 / −0.2) Belgium (60.3 / −0.2)
Romania (57.6 / −5.4) Malta (54.0 / −0.2) Poland (51.7 / −3.3)
High (> 70)
Above average (64 < 69)
Below average (60 < 64)
Low (< 60)
Source: Based on European Commission (2006: 257–288)
Low growth (< 2)
Shrinkage
Level 2004
Greece (59.4 / 2.9)
Estonia (63.0 / 2.6) Lithuania (61.2 / 2.1)
Slovenia (65.3 / 2.5)
Moderate growth (2–3)
Table 5.2 Recent employment records in EU-25, employment levels, 2004, and changes, 2000–4
Italy (57.6 / 3.9) Bulgaria (55.8 / 5.4)
Latvia (62.3 / 4.8) Spain (61.1 / 4.8)
Cyprus (68.9 / 3.2)
Strong growth (3+)
Employment patterns
135
1.2 Group-specific patterns of inclusion and exclusion National employment rates can only give a rough summary of labour market developments, since the crude aggregates conceal important differences between social groups. Compared to insiders such as men at prime working age, four groups have traditionally been vulnerable to becoming labour market outsiders who are hit hardest when the economy shrinks and who benefit less in periods of an upswing: women, young people in search of a first job, older workers, and the low-skilled. In the group of male workers at prime working age, the differences between old and new member states are not very marked. In the EU-15 on aggregate 86 per cent of all men aged 25–54 have a job. The margin of variation is rather small ranging from 84 per cent in Finland to 92 per cent in Luxembourg. In the NMS-10 the average is only slightly lower at 84 per cent. Poland and Bulgaria are the only countries where prime-age men have employment rates distinctly below 80 per cent. Much more sizeable national differences show up with respect to more marginal groups in the labour market (Table 5.39). Women hold jobs less frequently than men in old and new member states alike, but the size of the gender gap tends to be bigger in the old member states of the Union. In the EU-15 73 per cent of men at working age have jobs as compared to 57 per cent of women. The respective percentages for the NMS-10 are 68 and 53 per cent. The country-specific variation of male and female employment in Europe is stunning. In the case of women the range is from 33 per cent in Malta to 72 per cent in Sweden, in the case of men differences range from 57 per cent in Poland to 80 per cent in the Netherlands. Only five NMS and nine of the old member states have already reached the EU-employment target of 57 per cent for women. Even though the average gender gap is smaller in the NMS, there are very big nation-specific differences within both groups of countries. National similarities with respect to the size of the gender gap cut across the east–west divide. Relatively small gender gaps below 10 percentage points prevail in the three Nordic countries and the three Baltic states as well as in Slovenia. Particularly big gender gaps are found in Luxembourg and in the Mediterranean countries (Italy, Greece, Spain, Cyprus and Malta). The intermediate group with gender gaps between 10 and 19 percentage points is mixed, consisting of eight old and five new member states. Bulgaria and Romania both have below-average gender gaps in employment. Cross-national differences in the employment rates of more marginal groups are not only a function of the degree to which these groups meet particular employment impediments, but also reflect the general labour market situation in each country. In order to focus on the more specific first aspect, the emphasis in Table 5.3 is on groupspecific employment relative to the national means, rather than on the comparison of absolute levels which largely reflect the general employment situation. Among younger age cohorts (15–24) less than 40 per cent hold a job in Europe. This corresponds to less than two-thirds (60 per cent) of the aggregate EU-25 employment rate. Employment rates for this age group are shaped by labour market conditions as well as by different profiles of the educational systems. In the old EU-15 there are three distinct groups of countries. In the Netherlands the young have about the same employment chances as the economically active population at large. In a majority of nine countries their chance of holding a job is between one-half to around threequarters of the average. In the remaining five countries the young have much more limited prospects, as their employment rate is below one half of the national average
75.7 67.6 72.1 66.3 71.6 73.1 67.8 60.3 63.1 65.0 62.5 57.6 59.4 67.8 61.1 68.9 54.0 65.3 64.2 56.8 51.7 57.0 63.0 62.3 61.2 64.7 (60.4) 63.3 54.2 57.7
71.6 65.6 70.5 56.5 65.6 65.8 60.7 52.6 57.4 59.2 51.9 45.2 45.2 61.7 48.3 58.7 32.7 60.5 56.0 50.7 46.2 50.9 60.0 58.5 57.8 56.8 (53.2) 55.7 50.6 52.1
−8.1 −4.1 −3.1 −19.4 −12.2 −14.4 −14.2 −15.3 −11.6 −11.6 −20.9 −24.9 −28.5 −12.5 −25.5 −21.1 −42.4 −9.5 −16.3 −12.4 −11.0 −12.3 −6.4 −7.9 −6.9 −15.9 (−14.6) −15.2 −7.3 −11.3 0.8 0.6 0.5 0.7 0.8 0.9 0.8 0.5 0.5 0.6 0.4 0.5 0.5 0.5 0.6 0.5 0.9 0.5 0.4 0.4 0.4 0.5 0.4 0.5 0.3 0.6 (0.5) 0.6 0.4 0.5
0.8 0.8 1.0 0.7 0.8 0.6 0.4 0.5 0.6 0.6 0.5 0.5 0.7 0.7 0.7 0.7 0.6 0.4 0.7 0.5 0.5 0.5 0.8 0.8 0.8 0.7 (0.6) 0.6 0.6 0.6
(0.8) (0.8) (0.8) (0.8) (0.7) (0.8) (0.7) (0.7) (0.8) (0.7) (0.8) (0.8) (0.9) (1.0) (0.9) (0.8) (0.9) (0.8) (0.6) (0.6) (0.6) (0.4) (0.7) (0.7) (0.7) (0.8) (0.7) (0.8) (0.7) (0.8)
Low skill / total1 5.5 8.8 6.3 4.5 4.7 4.6 4.8 8.4 9.6 9.5 5.1 8.0 10.5 6.7 10.7 4.7 7.3 6.3 8.3 6.1 19.0 18.2 9.7 10.4 11.4 8.1 (10.1) 9.1 12.0 7.6
1.2 2.1 1.2 1.6 1.0 1.6 1.3 4.1 3.9 5.4 1.1 4.0 5.6 3.0 3.4 1.2 3.4 3.2 4.2 2.7 10.3 11.8 5.0 4.6 5.8 3.4 (5.2) 4.1 7.2 4.5
0.9 0.2 −0.4 −0.8 −0.8 0.5 0.9 2.0 1.8 1.8 3.3 4.1 9.6 1.7 6.5 2.4 2.2 1.0 2.8 0.0 1.7 1.8 −1.5 −0.4 0.8 2.1 (1.1) 2.2 −1.0 −2.1
Long- Gender term gap 8.2 20.7 16.3 8.9 12.1 8.0 9.6 21.2 21.9 15.1 16.5 23.6 26.9 15.4 23.4 10.5 16.2 16.1 21.0 15.5 39.6 33.1 21.7 18.1 22.7 16.7 (21.5) 18.9 25.8 23.2
1.5 2.4 2.6 2.0 2.6 1.7 2.0 2.5 2.3 1.6 3.2 3.0 2.6 2.3 2.2 2.2 2.2 2.6 2.5 2.5 2.1 1.8 2.2 1.7 2.0 2.1 (2.2) 2.1 2.2 3.1
0.7 0.5 0.6 0.3 0.4 0.4 0.2 0.1 0.2 0.6 0.1 0.2 0.2 0.4 0.3 0.5 0.1 0.1 0.3 0.1 0.2 0.3 0.3 0.4 0.5 0.4 (0.3) 0.3 0.3 0.1
(1.6) (1.6) (1.2) (1.7) (1.8) (1.4) (1.9) (1.7) (1.4) (1.9) (1.3) (1.2) (1.0) (1.1) (1.2) (1.5) (1.3) (1.7) (3.2) (2.2) (1.7) (2.9) (1.8) (1.6) (1.3) (1.4) (1.9) (1.4) (1.8) (1.2)
Youth Young / Elderly / Low skill / unempl. total total total1
Note: 1 Skill-specific data are only available for the population aged 25–64. Hence the reported data refer to low skill employment and unemployment as a proportion of total employment or unemployment in the age group 25–64.
Sources: Columns 1–5, 7–12: based on European Commission (2006: 257–288); columns 6+13: own calculations, based on EDS - online (30.11.2006 and 27.02.2007)
Denmark Finland Sweden Ireland United Kingdom Netherlands Austria Belgium France Germany Luxembourg Italy Greece Portugal Spain Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania EU-15 (NMS-10) EU-25 Bulgaria Romania
Elderly / total
Total
Young / total
Total
Female Gender gap
Measures of labour-market exclusion (unemployment)
Measures of labour-market inclusion (employment)
Table 5.3 Measures of labour-market inclusion and exclusion, 2004
Employment patterns
137
(Belgium, France, Greece, Italy and Luxembourg). Most new member states belong to this latter category of countries which offer very limited labour market opportunities for the young. Only Malta stands out as giving young people about average employment chances, whereas Cyprus, Latvia, Slovakia and Slovenia are the only other new member states where the employment rate of the young is at least half of the national aggregate. The difficult labour market position of younger people in the NMS is an important incentive for migration and if the prospects to find work do not improve, several NMS may become subject to a considerable youth and brain drain (European Commission 2006; Krieger 2004). Somewhat similar patterns of segmentation appear with respect to older segments of the labour force (ages 55–64). The employment rate of older European workers varies between 26 per cent in Poland and 69 per cent in Sweden. Only six countries surpassed the EU target of 50 per cent in 2004, among them the three Nordic countries, the United Kingdom and Portugal, as well as Estonia from the new member states. The group of countries with particularly low employment rates of around or even below 30 per cent is similarly mixed, including four of the old (Austria, Belgium, Italy and Luxembourg), and five of the new member states (Malta, Slovenia, Hungary, Poland and Slovakia). Compared to the aggregate employment rate, the employment rate of older workers is only about two-thirds as high. In the NMS, older workers do not differ very much from the national average in the Baltic nations, in the Czech Republic and in Cyprus. In the other countries they have less than 60 per cent of the average chance to hold a job, and Slovenia and Slovakia stand out for their particularly limited opportunities in absolute and in relative terms. In the old EU-15, the Scandinavian and Anglo-Saxon countries as well as Greece and Portugal stand out for their relatively high employment rates of older workers compared to the country mean. At the opposite pole, older workers score only very under-proportionate rates of economic activity in Austria, Belgium and Luxembourg. Data for the employment of people with low skills are only available for the population aged 25–64 so that they cannot immediately be compared to the information for the other marginal groups which relates to the total population at working age (15–64). With respect to low-skill employment there is a notable difference between old and new member states. In most Western European countries, the labour market chances of the low-skilled correspond to at least 80 per cent of the national average (for the population aged 25–64).10 Their relative chances tend to be lowest in Austria, Belgium, Germany and the United Kingdom, but particularly high in Southern Europe, where the agricultural sector still provides a rather high share of jobs. In the new member states, Cyprus and Malta resemble the other Mediterranean countries in that they offer the low-skilled almost average job prospects. Slovenia and Romania are the only other countries where the low-skilled do not deviate very far from the national average. In all other new member states people with low skills encounter very serious impediments in the labour market. Varying between 27 per cent in Slovakia and 52 per cent in Latvia, their employment rates are not only far below the Western European average (56 per cent), but also distinctly below the average employment rate in their home countries. Hence in most new member states special training programmes to promote the labour market chances of the less-skilled must be considered an urgent policy priority. Data on unemployment reveal to what extent the vulnerable groups are subject to over-proportionate risks of labour market exclusion. It is important to note that the
138
Jens Alber
exclusion indicators are not simply the mirror image of the data on labour market inclusion. Thus, Hungary has a similarly low employment rate as Slovakia, but it also has the second lowest unemployment rate among the NMS-10 (scarcely 6 per cent), whereas Slovakia has the second highest (18 per cent). This is due to the fact that the Hungarian labour force participation ratio – which the reports on Employment in Europe call the ‘activity rate’ – is roughly 10 percentage points lower than the Slovakian one. Malta and Romania are further examples of countries that combine a low employment rate with a comparatively low level of registered unemployment. This illustrates that low unemployment rates do not necessarily indicate successful labour market policies, but may also result from widespread resignation and labour market withdrawal. Whereas in the EU-15 Spain and Greece are the only countries with double-digit unemployment, five new member states (including Bulgaria) exceed this limit. In Poland and Slovakia almost every fifth member of the labour force is in search of a job. Among the new member states, Cyprus is the only one to join the group of five EU-15 countries which succeeded in keeping unemployment below 5 per cent. Unemployment soared particularly in the formerly socialist countries during the transition to market economies. By 1993, double-digit unemployment rates prevailed in all Central and Eastern European countries with the exception of the Czech Republic and the three Baltic nations.11 In subsequent years unemployment kept growing everywhere including even the four originally more successful nations. Slovenia was first to successfully break this trend in 1998. By 2004 the Czech Republic was the only formerly socialist country which had continuously managed to remain below the 10 per cent level of registered unemployment in the transition period. Survey data show that the proportion of people who have experienced prolonged or repeated unemployment during a period of five years is even about 2–3 times higher than the officially registered annual unemployment rate.12 In the new member states, unemployment is not only more frequent, but also more severe than in the EU-15, because higher proportions of jobless people are in longterm unemployment. On average, half of the unemployed in the post-communist new member states, as compared to 42 per cent in the EU-15 have been jobless for more than a year. Long-term unemployment is particularly widespread in Poland and Slovakia, where it affects more than 10 per cent of the labour force. On average, more than 5 per cent of the workforce in the NMS-10 are in long-term unemployment, as compared to roughly 3 per cent in the EU-15. The level of 3 per cent is surpassed in all new member states with the exception of Cyprus and Hungary, as well as in 7 of the 15 old member states. Throughout the European Union younger people at the beginning of the work career face vastly over-proportionate risks of unemployment. The size of the risk factor is roughly similar in old and new member states, as young people tend to have about twice the risk of the labour force at large across most countries. The remarkable feature, however, is the stunning margin of variation in the absolute size of youth unemployment which ranges from below 10 per cent in 4 old member states (Austria, Ireland, Denmark and the Netherlands) to highs above 20 per cent in 11 countries. Among the countries with extremely high levels of youth unemployment, we find 6 old and 5 new member states, as well as Bulgaria and Romania. Peak levels of youth unemployment in 2004 were reached in Slovakia with 33 per cent and Poland with 40 per cent.
Employment patterns
139
How do societies cope with such extraordinary levels of unemployment, especially among younger-age cohorts? Basically, there are four adaptation mechanisms. For young unemployed people, the family usually steps in as the prime agent of social integration, allowing younger people to reside in the parental household even after graduating from school. This extended household solution has been resorted to in the new member states as well as in Southern European countries (for data see Saraceno et al. 2005, Guerrero 1995). A second adaptation mechanism consisted in the expansion of the shadow economy which became particularly prominent in formerly socialist new member states. After the collapse of communism, national governments in Central and Eastern Europe tended to tolerate the growth of an informal sector, because it helped to relieve the tensions of the transition (Rose 1996). Around the turn of the millennium, estimates of the size of the shadow economy ranged from around 20 to 40 per cent, with the Czech Republic and Slovakia standing out for a relatively limited size (around 18 per cent), and Estonia and Latvia having the largest informal economies with shares of almost 40 per cent (Vaughan-Whitehead 2003: 178). A third adaptation mechanism to mass unemployment consisted in an extended reliance on household production and subsistence agriculture. On average 8 per cent of households in the NMS, and 14 per cent of households in rural areas helped to meet their needs for food by growing vegetables or fruits or keeping poultry or livestock in 2003. In the EU-15 the respective percentage nowhere exceeded 2 per cent (Fahey 2004). Migration across national borders to other EU countries may be considered a fourth adaptation mechanism. It has been of limited importance so far, but may come to play a more important role in future years as European labour markets become more integrated and labour exchanges are beginning to look across national borders (European Commission 2006, see also Krieger, this volume, Ch.15). Compared to the striking incidence of youth unemployment, the other vulnerable groups have been less prone to suffer vastly over-proportionate rates of unemployment. In 2004, women had higher unemployment rates than men in 12 of the 15 old member states. The three exceptions include Sweden, Ireland and the UK. Among the new member states, only Estonia as well as Bulgaria and Romania deviate from the pattern of higher unemployment among women. Older people do not stand out with over-proportionate unemployment rates, but this is largely due to the fact that they have frequently withdrawn from the labour market altogether. Low-skilled groups come closest to younger people in facing elevated unemployment risks. Compared to the national average,13 their risk factor is particularly high in new member states, where the Czech Republic, Slovakia and Hungary stand out for their vastly over-proportionate unemployment risk of low-skill people. 1.3 Sector-specific employment patterns The capacity of various member states to meet the employment targets of the Lisbon agenda is also a function of their national structures of employment. Having labour markets which used to be heavily dominated by industrial employment in the socialist era, the new member states had to undergo similar transitions of the employment structure as the old member states, seeking to compensate employment losses in the de-industrialisation process by gains in the service sector. In their case, however, the change was much more abrupt. They also have much bigger agricultural sectors. Even though shrinking, agricultural employment still plays a prominent role, thus
140
Jens Alber
confronting EU cohesion policies and structural funds with a major challenge. In the EU-15 the agricultural sector provides jobs for a little over 2 per cent of the population at working age on aggregate, and in 2003 agricultural employment exceeded the level of 6 per cent only in Greece and Portugal (Table 5.4).14 In the NMS-10, agricultural employment averaged 7 per cent in 2003, and the Czech Republic was the only country below 3 per cent. The agricultural sectors in Poland, Lithuania and Latvia still employ close to 10 per cent of the population at working age. In Romania even 20 per cent of the population at working age held agricultural jobs in 2003, whereas agricultural employment in Bulgaria was apparently much more limited at 5 per cent.15 Average employment rates in industry are by now very similar in new and old member states. Less than one in five members of the European population at working age now hold jobs in the industrial sector. In the old EU-15, the margin of variation is from 13 per cent in Greece and Luxembourg to 23 per cent in Portugal. Austria and Germany are the only other countries with an industrial employment rate above 20 per cent. In the NMS-10 the range is from 15 per cent in Poland to above 20 per cent in the Czech Republic, Slovakia and Slovenia. In both parts of Europe the recent phase of development was characterised by rather massive de-industrialisation. Among the old member states, Spain was the only country to experience noteworthy growth of industrial employment after 1998. Almost all other countries had more or less sizeable shrinkage. In the NMS-10, only the three Baltic countries had recent industrial employment growth, whereas industrial downsizing was particularly marked in Poland, Slovakia and the Czech Republic.16 The major difference between old and new member states is in service sector employment. Some 43 per cent of EU-15 residents at working age, but only 32 per cent of their NMS-10 peers found a job in the service sector in 2003. In the old member states, four countries (Denmark, Sweden, the Netherlands and the UK) had servicesector employment rates above 50 per cent, whereas the four southern countries were the only ones to remain below the level of 40 per cent. In the NMS-10, Cyprus is the only country where the service sector employs more than 40 per cent of the workingage population, whilst Poland stands alone with an employment rate below 30 per cent in services. Among the two newcomers to the Union, only Romania has an even lower rate of service sector employment than Poland. Which factors shape sector-specific employment patterns? According to an influential argument by Fritz Scharpf (2000) enhanced international competition in the globalised economy constantly induces firms to drive up productivity. Hence in sectors which are exposed to international competition, employment opportunities may decrease even in firms or countries that are doing well in global markets; employment gains can realistically be expected only in the sectors which are sheltered from international competition, as a basically local supply meets with a local demand. According to Scharpf, this sheltered sector includes consumption-related services such as wholesale and retail trade, restaurants and hotels (ISIC 6 in the International Standard Industrial Classification of Economic Activities), as well as community, social and personal services (ISIC 9). The exposed sector which is prone to employment losses includes all branches in which imports and exports play a prominent role. It thus embraces not only manufacturing industries, but also primary production, energy and construction industries, and production-related segments of the service sector such as transport, communication, financial and business services
(2.4) — (1.2) (2.3) (1.4) (0.8) (2.4) (8.6) — (3.5) (2.5)
2.2 3.7 1.0 2.7 1.5 1.4 2.6 8.7 6.0 3.3 2.4 3.0
14.8 20.7 14.8 15.7 20.5 12.9 17.9 12.9 23.3 18.4 18.1 17.3
17.4 18.3 16.7 18.2 16.9
3.1 8.8 3.5 3.8 8.8 11.4 — 5.8
(2.9) (9.9) (2.2) (3.7) (7.8) (9.7) (7.1) (5.7)
2.8 — — — 4.2 (6.9) 2.9 (2.6)
(12.9) — (12.5) (13.1) (17.2) (13.3) (16.6) (13.6) — (18.3) (15.5)
(16.1) (17.4) (16.3) (18.3) —
51.3 44.7 43.5 44.2 43.0 48.1 35.6 36.3 38.0 37.9 43.4 45.1
55.2 46.6 55.2 42.9 53.8
(57.7) — (46.6) (47.7) (46.3) (48.4) (38.6) (37.2) — (39.3) (46.7)
(57.2) (46.6) (54.1) (43.9) —
Services
19.1 14.9 22.1 19.8 16.8 17.4 — 19.9
(18.7) (13.9) (19.6) (21.9) (16.5) (17.2) (17.1) (19.5)
34.9 27.6 32.3 38.7 36.2 34.0 — 34.3
(35.2) (27.9) (35.2) (37.5) (37.9) (34.3) (317) (35.0)
16.2 — 50.2 — — — — — 23.4 (23.3) 34.4 (35.1) 26.0 (24.7) 35.9 (37.0)
Level 2003 (2004)
(2.4) (3.5) (1.7) (4.1) —
2.3 3.4 1.6 3.7 0.8
Level 2003 (2004)
Agriculture Industry
Employment rates in main sectors
-1.2 -4.9 -2.0 -2.0 -3.4 -2.2 -3.4 -2.3
— — -1.4 -1.1
1998–2004
-0.1 — -0.3 -0.3 -0.2 -0.2 -0.4 -1.5 — -0.2 -0.3 -0.5
-0.4 -0.6 -0.3 -1.3 —
1998–2004
0.2 -3.5 -2.7 0.5 1.2 -0.6 -2.4 -1.1
— — -0.8 -2.9
-1.6 — -0.8 -0.4 -2.3 -2.0 0.9 0.0 — 3.3 -0.7 -0.4
-1.6 -0.5 -0.9 1.0 —
4.1 1.1 1.2 -0.1 4.5 1.8 1.7 2.3
— — 4.6 0.9
4.5 — 4.0 3.6 3.5 4.2 5.2 4.9 — 6.7 4.3 4.3
2.6 4.1 2.8 6.1 —
Agriculture Industry Services
Change in employment rates in main sectors
4.7 0.7 3.9 2.3 4.2
4.5 7.1 12.0 7.1 6.3
-7.6 -11.3 -8.5 -7.8 -8.2
-0.3 — -3.2 -4.1 -1.5 -1.2 — -2.8
— — -4.3 -3.0 2.6 — 0.3 -0.5 1.5 1.0 — 0.4
— — 1.7 0.5
1.5 — 0.2 1.8 2.0 0.6 — 0.7
— — 1.4 0.1
2.8
4.3
-7.8
Change 1998–2003
7.8 5.1 2.9
12.5 7.6 3.9
4.9 3.4 0.5 9.7 5.4
-2.2 -7.9 -5.9
-5.8 -12.5 -11.9 -6.6 -10.5
4.9 1.9 4.9 5.8 6.5
Consumer Business services services (ISIC 6+9) (ISIC 7+8)
Change 1980–2000
Agriculture and industry (ISIC 1-5)
22.2 23.7 25.6 23.6 25.6 28.8 — 25.8
19.0 — 27.6 28.9
Level 2003
21.2
20.9 22.3 32.2 19.3
24.0
17.1 24.3 15.5
20.2 22.0 18.1 22.4 17.4
Level ~2000
Agriculture and industry (ISIC 1-5)
25.5 20.7 24.1 26.4 26.5 26.4 — 25.0
35.1 — 24.8 25.9
31.7
26.2 23.6 31.1 24.4
29.8
38.2 31.6 30.5
39.7 30.6 39.0 30.6 37.3
Consumer services (ISIC 6+9)
9.3 6.9 8.0 12.3 9.5 7.2 — 9.1
12.0 — 9.6 10.2
11.6
9.2 6.9 7.8 6.8
10.4
15.9 12.3 10.6
14.5 11.5 15.7 12.5 16.0
Business services (ISIC 7+8)
22.5 — 28.8 27.7 27.1 30.0 — 28.6
— — 31.9 31.9
Level 1998
29.4
28.5 33.6 40.7 27.1
31.8
19.3 32.2 21.4
26.0 34.5 30.0 29.0 27.9
Level ~1980
— — 8.2 10.1 7.8 — 7.8 10.5 7.5 6.6 — 8.4
22.9 — 23.8 26.9 25.0 25.4 — 24.6
7.4
25.4
— — 23.1 25.4
4.5 6.2 3.9 4.5
7.6
8.1 7.2 7.7
9.6 9.6 10.8 6.7 9.5
Business services (ISIC 7+8)
21.7 16.5 19.1 17.3
25.5
25.7 24.0 26.6
34.8 27.2 38.5 20.9 31.9
Agriculture Consumer and services industry (ISIC 6+9) (ISIC 1-5)
Employment rates in exposed and sheltered sectors
Sources: EU-15: columns 1, 3, 5: European Commission (2004: 145–157); columns 2, 4, 6–9: own calculations, based on European Commission (2006: 257–288). NMS: columns 1, 3, 5: European Commission (2004: 145–157); columns 2, 4, 6: own calculations, based on European Commission (2006: 257–288); columns 7–18: Own calculations, based on European Commission (2004: 145–157)
Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania NMS-8 agg. NMS-8 Ø
Denmark Finland Sweden Ireland United Kingdom Netherlands Austria Belgium France Germany Luxembourg Italy Greece Portugal Spain EU-15 agg. EU-15 Ø
Country
Table 5.4 Sector-specific employment rates in the enlarged EU
142
Jens Alber
(i.e. ISIC sectors 1–5 and 7–8). The crucial divide would thus be between the two relatively sheltered service sectors (ISIC 6 + 9) with considerable growth potential on the one hand and the more exposed other sectors with more limited prospects on the other. Data on sector-specific developments over a longer period which would show to what extent employment trends in Europe actually conformed to this pattern are only available for OECD countries. The exposed non-service sectors of traditional industrial society (ISIC 1–5) have indeed been shrinking everywhere in Europe. Within the EU-15, employment losses averaged 8 percentage points since 1980, ranging from 2 points in the Netherlands to 13 points in Finland. The losses tended to be particularly heavy in those countries where the level of traditional employment had been high in 1980. In other words, countries which started the transition to the service economy late were subject to particularly radical re-structuring. In 2003, employment rates in the traditional non-service sectors averaged 22 per cent in the EU-15, ranging from 16 per cent in Belgium to 32 per cent in Portugal. Among the NMS-10 the average was 25 per cent, with a variation from 19 per cent in Cyprus to 29 per cent in the Czech Republic and Lithuania. Assuming that the post-1980 developments in Western Europe herald the future for the new member states, the latter will experience similar changes in the years to come. This would mean that the industrial jobs lost in the transition from socialism will not be regained and that present levels of employment in traditional sectors may not yet have reached the bottom. It would also mean that especially those countries will have to go through further re-structuring which presently have relatively high levels of employment in the traditional sectors that are exposed to international competition. In the old EU-15 most countries could compensate for the jobs lost in the de-industrialisation process with increases in service-sector employment. Employment in the more sheltered consumption-related services (ISIC 6+9) has been growing sizeably across all nations. On average, the employment rate in this sector has increased by roughly 6 percentage points since 1980. While Sweden, which already had an extended service sector very early, stands out with the smallest change, the other countries registered increases in the range of 3 to 13 percentage points. In general, growth tended to be more sluggish in those countries which had reached a comparatively high level of service employment already in 1980 (i.e. the Scandinavian and Anglo-Saxon countries). In 2003, an average 32 per cent of West Europeans of working age held jobs in consumption-related services. The margin of variation in the EU-15 was from a low of 24 per cent in Italy and Greece to a high of 40 per cent in Denmark. The 2003-level of employment in the NMS-10 largely resembled the Southern European countries. On average 23 per cent of the population at working age worked in this branch of the service sector. With ranges from 21 per cent in Poland to 26 per cent in the Baltic countries and in the Czech Republic, there was only little country-specific variation. Only Cyprus stood out as an outlier with 35 per cent. In the old EU-15, employment in production-related or business services (ISIC 7-8) has also been growing everywhere since 1980. On the basis of the assumption that this sector is exposed to heavy international competition, the sizeable growth nevertheless is remarkable, showing that economic openness need not be an impediment to employment growth. On average, the increase was less marked than in consumptionrelated services, however, amounting to 4 percentage points, and ranging from merely
Employment patterns
143
Employment in consumption-related services (ISIC 6+9)
1 point in Greece to over 7 points in the Netherlands. In 2003 employment rates in business services averaged 12 per cent in the EU-15 and ranged from 7 per cent in Greece and Spain to 16 per cent in the UK. In the NMS-10, employment in business services is much lower, amounting to 8 per cent on average. Only the Czech Republic, Estonia and Cyprus surpassed the level of 10 per cent in 2003, whilst Poland and Lithuania occupied the bottom ranks with 7 per cent. These comparatively low levels suggest considerable growth potential in these sectors in the future. Looking at the two sub-sectors of service employment together, it becomes evident that both cluster to form a syndrome of service-sector employment. Countries with high employment in one service field thus also have high employment in the other. Both branches of service employment are specifically underdeveloped in the NMS and in Southern European countries which cluster together as countries with relatively few jobs in both sub-sectors of services (Figure 5.3). Future job growth will probably occur in these branches, where currently only Cyprus and Estonia stand out among the NMS-10 with higher levels of service employment. In the EU-15 the Nordic countries as well as the United Kingdom and the Netherlands stand apart from others for their relatively high levels of employment in both branches of the service sector. The fact that employment grew in both sub-sectors of services and that employment levels in these branches tend to cluster together suggests that social science scenarios linking employment problems in European societies to the growth of the service economy may have been overly pessimistic. Towards the end of the twentieth century rather dismal visions of the future had replaced the erstwhile optimistic views of work opportunities in post-industrial societies. Originally, scholars like
DK
40.0
SE UK
NL CY
35.0
AT DE FR
30.0
FI
BE IE LU
PT GR
LT
HU
25.0 SK
20.0 6.0
LV
ES CZ
IT
EE
SI
y = 1.4896x + 12.728 r 2 = 0.6892 r = 0.83
PL
8.0 10.0 12.0 14.0 Employment in production-related services (ISIC 7+8)
16.0
Figure 5.3 Employment rates in consumption-related and production-related services, 2003 Source: Based on European Commission (2001: 145–157)
144
Jens Alber
Jean Fourastié (1950) had assumed that with growing real incomes the structure of demand would shift from basic to superior goods thus leading to an insatiable demand for individual and collective services. Later, however, scholars like Baumol (1967), Gershuny (1978), Scharpf (1986, 2000) or Iversen and Wren (1998) developed a much more pessimistic view based on the assumption of a ‘cost disease’ in the service sector. The basic idea was that the service sector is much less prone to productivity gains than the industrial sector, but subject to similar wage demands. As a consequence, relative prices for services would increase, the price-elastic demand for services would shrink, and consumers would turn away from the market to look for cheaper alternatives in either the shadow or the do-it-yourself economy. Hence, officially registered employment would go down, a trend which could only be counteracted if services were subsidised by the government or would be delivered at lower prices. In contrast to such scenarios, Europe has seen a heavy downsizing of industrial work over the past decades and a growth of service employment even in those sub-sectors which are exposed to international competition. In international comparisons of OECD countries or EU member states we therefore find a positive statistical association between the relative size of the service sector and the level of employment (r =0.55 for 2004). This suggests that post-industrial societies tend to generate more rather than less work (Esping-Andersen 1999). The considerable growth in various branches of the service economy, including business services on the one hand and consumption-related services on the other, also suggests that there are several paths rather than just one road leading to postindustrial full employment. Thus, in addition to ‘high-tech’ jobs, there are work opportunities in ‘high-touch’ sectors of the service sector which require training in social rather than technical skills. And, even more important, there is not only a ‘high road’ to more professional work but also a ‘low road’ to growing service employment which is particularly important for those who are threatened to be left behind because of little formal schooling. This low road consists of personal services especially in the sector of private households, in restaurants and hotels, and in wholesale and retail trade, where several EU-15 countries have seen relevant growth during the past decades, and where the new member states presently have comparatively low levels of employment. In sum, similarities combine with remarkable differences between European countries, and it is the latter which have grown bigger after the recent enlargements. There is a general tendency for the decline in agricultural and industrial employment to be matched or even overcome by increases in service sector employment. Within the service sector the sheltered consumption-related services have been growing more dynamically than the more exposed business-related services. Beyond these general trends, there are marked country-specific differences, however. Regardless of whether we look at levels or changes, we usually find the leading European country separated from the laggard by factor 2 or more. A sector-specific analysis of employment structures thus reveals a high and even growing degree of variation in Europe rather than one common European employment regime. The two Eastern enlargements have increased this diversity even further. Once again, however, national differences do not overlap with the distinction between old and new member states, as the new member states tend to cluster together with the Southern European old members. The following section will further probe into similarities and differences in the enlarged EU from a more analytical perspective.
Employment patterns
145
2. Analyses of employment patterns The analysis of employment patterns will proceed in three steps. The first part explores the relationship between economic growth and employment growth and analyses to what extent there was jobless growth. In a second step, we investigate whether specific vulnerable groups have been de-coupled from employment developments. Finally, looking at differences in welfare state policies, we examine what impact these had on labour market developments, and to what extent new member states differ from the old ones to a degree that would justify speaking of social dumping policies. 2.1 Jobless growth: To what extent is economic growth translated into employment? From 1989 to 1993 economic output declined dramatically in the eight new member states which went through the transition to market economies. Negative annual growth rates averaged −7.5 per cent in this period, and in 1993 the average GDP-level amounted to merely 73 per cent of the value in 1989. Economic recovery began in 1994, and between 1994 and 2004 real annual growth rates in the eight post-communist NMS were almost twice as high as in the EU-15 (4.3 per cent vs. 2.4 per cent).17 Dynamic growth persisted also in the most recent period (2000–2004). Compared to a mean EU-15 growth rate of 2.1 per cent, growth in the eight transition countries averaged 4.9 per cent (4.4 per cent in NMS-10 – see Appendix, Table A5.1). Displaying growth rates around 7 per cent, the three Baltic countries even became nicknamed the ‘Baltic tigers’. For a long time this dynamic economic growth did not translate into a similar growth in employment, however. As illustrated in Figure 5.4, the average employment
110.0 105.0
Growth of GDP (in PPP $)(1989 = 100%) Growth of employment (pop. aged 15 - 59)(1989 = 100%)
100.0 95.0 90.0 85.0 80.0 75.0 70.0
1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Figure 5.4 Development of real GDP and employment rate in post-communist NMS-8 (1989 = 100) Source: Based on Unicef (2006: 91), Unicef (2006: 121)
146
Jens Alber 6.0
Change in employment (%)
5.0
ES
4.0
LV
IT CY
3.0
GR
EE
SI LT
2.0 FR
1.0 NL 0.0 0.0 −1.0
MT
BE 1.0 PT 2.0 DE DK AT
IE UK
FI 3.0
SE
HU
SK 4.0
LU
5.0
6.0
7.0
8.0
CZ y = 0.4854x − 0.7126 2 r = 0.2212 r = 0.47
−2.0 −3.0 PL −4.0
Average real GDP growth rate
Figure 5.5 Economic growth rates and change in employment rate in EU-25, 2000–2004 Source: Based on European Commission (2006: 249–288)
rate in the NMS-8 continued to decline up to 2000 even in the presence of rapidly accelerating economic growth. Even though employment began to increase again in the new millennium, there continued to be a remarkable bifurcation between the dynamic growth of the economy and the stagnant development of the labour market. Statistically there was no correlation between EU employment levels in 2004 and the amount of economic growth in the preceding period (1998–2004; r =−0.15). In fact, 8 of the 11 countries which had above-average economic growth also had below-average levels of employment. The recent period has been marked by a tighter coupling between the development of the economy and changes in the job market. Countries which had higher growth rates in the period 2000–4 now also tended to experience higher increases in employment. As Figure 5.5 shows, the statistical association now became clearly positive, even if not very strong (r =0.47). Except for Hungary and Slovakia, the new member states with above-average growth rates now also stood out for their above-average increases in jobs. The three Baltic countries finally abandoned their path of jobless growth and successfully turned the economic upswing into job creation. Even though employment levels in the post-communist countries are still far below the levels of the early 1990s, it now looks as if their hope that EU membership would bring them not only freedom, but also growth and prosperity, is about to materialise. 2.2 General labour market developments and group-specific employment chances: Have specific groups been de-coupled from employment growth? Europe’s poor labour market performance compared to the United States and the long persistence of mass unemployment gave rise to concerns that something in the continent’s institutional structures impedes full employment. Based on the assumption of a ‘big trade-off’ between equality and efficiency (Okun 1975), the most
Employment patterns
147
prominent and general idea in this respect is that overly generous welfare state schemes are at the root of the problem. Apart from this general charge – which will be dealt with in the next section – there were also more country-specific attempts to explain why the era of full employment should be considered to be behind us. Especially in continental European countries, where the industrial sector had long been the dominant branch of employment so that the de-industrialisation process hit their labour markets particularly hard, concerns were voiced that post-industrial knowledge societies might run out of work. Two ideas gained particular prominence, especially in Germany (for comparisons in historical perspective see Kaelble 1989; Kocka and Offe 2000). The first one relates persistent mass unemployment to an oversupply of labour which is supposedly linked to the mobilisation of women on the one hand, and to the recent change of European countries from emigrant to immigrant societies on the other (Kommission für Zukunftsfragen der Freistaaten Bayern und Sachsen 1996). A second interpretation points to the shrinking demand for unskilled labour which is linked to rising skill requirements in high-tech societies and to the growing competition from industrialising countries which have an abundant supply of cheap but literate labour and are thus able to take over the production of low-tech goods for mass consumption (Wood 1994). Especially if high social benefits contribute to making unskilled work in European welfare states even relatively more expensive, people without higher education credentials will be driven out of the labour market. A common denominator of these ideas was the assumption that specific groups, among them the low-skilled, young labour market entrants, and women were likely to become redundant and that only a reduction of labour supply through early retirement schemes or part-time work arrangements would be able to cope with the problem. Looking at the employment chances of men and women, one might indeed expect that the goal of full employment would become more difficult to achieve, as societies move away from the male breadwinner model to the ideal of dual career couples or career chances for everyone. In the EU-15, the sizeable increase in female employment over the past decades coincided with a marked decline of male employment rates (Figure 5.6). For the NMS longitudinal data are in scarce supply, but many accounts report that women have massively withdrawn from Central and Eastern European labour markets during the transition from socialism (Cazes and Nesprova 2003: 12–13). Hence there is some evidence supporting the notion that women and men may be crowding each other out in European labour markets. In contrast to the oversupply thesis, however, there is no evidence that unemployment increases as female activity rates rise.18 Moreover, there is also a mild positive correlation between 2004 male and female employment rates in the EU-25. In general, countries with a high number of jobs for men also tend to have more jobs for women. The example of the United States, where male employment kept growing despite a steep increase in female employment, provides further proof against the idea of labour market over-crowding. The empirical evidence suggests that instead of a fixed stock of employment which can merely be shuffled between various groups, there is a dynamic job market which fluctuates with variations in supply and demand. The employment patterns of people in different age-groups also fail to confirm oversupply concerns. In contrast to the idea that an extension of early retirement will make room for the young, the labour market does not function as a zero-sum game, where one age-cohort wins at the expense of another. There is actually a mild
148
Jens Alber
85.0 US males
80.0 75.0
EU-14 males 70.0 65.0 60.0 US females EU-14 females
55.0 50.0 45.0 40.0
1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Figure 5.6 Development of male and female employment rates in Europe and the United States, 1979–2003 (% of population aged 15–64) Source: Based on OECD (2000, 2004b) Note: EU-14 without Luxembourg because of unreliable data exceeding 100%.
Employment rate of popopulation aged 15–24
tendency for countries with high employment rates for people at prime working age (25–54) to have also more jobs for younger as well as for older workers.19 Moreover, there are no indications that early retirement translates into more jobs for the young. Instead, we find a weak, but positive correlation between employment rates of older and younger workers (Figure 5.7). In the eight post-communist new member states, 70.0 NL
65.0
DK
60.0 UK
55.0 AT
50.0 IE
MT
45.0
DE
40.0 CY
35.0 30.0 25.0 20.0 25.0
SE
ES
SI
LV
FR BE SK
FI PT
IT
PL LU
30.0
GR
CZ
HU
y = 0.4616x + 16.966 2 r = 0.1796 r = 0.42
EE LT
35.0
40.0
45.0
50.0
55.0
60.0
65.0
70.0
75.0
Employment rate of population aged 55–64
Figure 5.7 Employment rates of older (55–64) and younger (15–24) age cohorts in EU-25, 2004 Source: Based on European Commission (2006: 257–288)
Employment patterns
149
Employment rate of people with low educational attainment (ISCED 0–2)
widely discrepant levels of employment for older workers go hand in hand with almost identical levels of youth employment. Combining low levels of employment for both age-cohorts, Belgium, France, Luxembourg, Italy and Greece illustrate for the old member states that early retirement policies do not necessarily create more opportunities for the young. Particularly high proportions of young people are employed in countries with above-average employment rates of older workers, especially Denmark, the Netherlands, the UK and Ireland. This underscores that the Lisbon strategy of raising the employment rate of older workers to above 50 per cent in order to safeguard the sustainability of pension schemes does not necessarily diminish the labour market chances of younger workers. Total employment and low-skill employment also tend to co-vary in Europe, so that countries usually have more or less success in both dimensions jointly. However, Europe is characterised by very discrepant country-specific patterns in this respect. The Scandinavian and also the Anglo-Saxon countries combine high levels of general employment with a rather high number of jobs for people with low educational attainment. In contrast, all new member states except Cyprus and Slovenia combine their low levels of general employment with particularly low levels of low-skill employment. This distinguishes them also markedly from the Southern European old member states which combine their low levels of general employment with fairly high numbers of jobs in the low-skill sector. A distinct bifurcation of employment levels for high-skilled and low-skilled groups is found in the Czech Republic, Lithuania and Malta, where relative success in the employment of high-skilled people is coupled with a low employability of people with little schooling (Figure 5.8). Hence special training programmes for the low-skilled are called for if a social polarisation between labour market winners and losers is to be avoided in these countries. In contrast, Slovakia, Poland and Hungary provide insufficient labour market chances for
PT
70.0 SE CY DK
60.0
FR
ES
LU
GR
50.0
EE
NL
IE
SI
FI IT
AT
BE
LV
LT
UK MT
DE CZ
40.0 PL
HU
30.0 SK
20.0 80.0
y = 1.7199x − 92.828 2 r = 0.1715 r = 0.41
81.0 82.0 83.0 84.0 85.0 86.0 87.0 88.0 89.0 90.0 Employment rate of people with high educational attainment (ISCED 5–6)
Figure 5.8 Employment rates of people with high and low educational attainment in EU-25 (population aged 25–64), 2004 Source: Based on EDS – online (30.11.2006)
150
Jens Alber
low-skilled and high-skilled groups alike, suggesting that in their case more general and less-targeted activation policies should be in place. 2.3 Policy impacts: Does the welfare state impede or promote employment – and what kinds of welfare state policies do the new member states pursue?
Employment rate of population aged 15–64
In the debates on the European social model, a prominent concern is that the welfare state is the root cause of Europe’s employment problems. Arguing that overly generous welfare state schemes have the dual effect of inviting idleness and of raising the cost of labour to an extent that makes work in the low skill and low productivity sectors unprofitable for employers, critics claim that people of working age shy away from work, while producers are forced to either move to low-cost countries or to automate production. Growing unemployment, especially among unskilled workers, is seen as the typical result (Scharpf 2000; Zukunftskommission der Friedrich-Ebert-Stiftung 1998). Crude data on social outlays have the advantage of being available for old and new member states alike, but they only allow a first probing into the labour market effects of different welfare states. They illustrate, however, to what extent the new member states pursue restrictive social policies which distinguish them from the rest of the Union, and they also indicate to what extent there is a statistical association between spending levels and employment rates. Figure 5.9 shows that all new member states except Slovenia have below-average levels of social spending. Yet only the three ‘Baltic tigers’ stand out for their conspicuously low levels of social outlays. The other newcomers to the Union have expenditure levels which are roughly comparable to Ireland or Spain. In 2004, there was a weak positive correlation between spending levels and employment rates in the EU-25, as countries with higher social expenditure
80.0 DK
75.0 NL
SE
UK
70.0
CY PT IE
FI
SI
65.0
DE
CZ LV
AT
FR
EE LT
LU
ES
60.0
GR SK
55.0
HU
BE
IT y = 0.4545x + 53.298 r 2 = 0.2062 r = 0.45
MT PL
50.0 10.0
15.0
20.0 25.0 Total social expenditure (% of GDP)
30.0
35.0
Figure 5.9 Social expenditure ratio and employment levels in EU-25, 2004 Source: Based on EDS – online (24. 01. 2007); European Commision (2006: 257–288)
Employment patterns
151
ratios also tended to have higher employment rates (r =0.45; for raw data see the Appendix (Table A5.1) on possible determinants of employment development). The aggregate macro-statistics thus fail to support the hypothesis that generous welfare states come at the expense of employment problems in Europe. However, mixing many diverse elements such as social transfers and social services, crude data on social outlays are a complex compound of several components which do not lend themselves easily to a substantive interpretation. Even if generous social transfers had negative incentive effects, these might be counteracted by countervailing positive incentive effects of public services. As an ample supply of public services in the fields of child care and elderly care provides job opportunities while facilitating the task to combine work and family life, it has positive effects especially on female employment. Hence countries like Denmark or Sweden which have a high supply of public services are typically characterised by high female employment rates. A closer analysis of the incentive effects of social transfer programmes must distinguish between the effects of benefit levels on the one side and of financing modes on the other, and it should also disentangle supply-side and demand-side effects. On the supply side, generous social benefits may impede low-skill employment, because they raise the reservation wage below which jobless people will refuse work. On the demand side their financing drives a wedge between gross and net wages, so that employers find it increasingly difficult to pay wages which lead to net earnings above the level of welfare state benefits.20 The debate on these issues owes much to the work of Gøsta Esping-Andersen (1996, 1999, 2000) and Fritz Scharpf (1986, 2000) who spelled out some of the mechanisms linking social programmes to specific labour market consequences in more detail by distinguishing between various types of welfare arrangements and by highlighting above all the incentive effects of different financing structures. According to their analysis, a peculiar feature of continental European welfare states is that they rely heavily on social security contributions which are levied on income from work. Driving a wedge between the net earnings that workers receive and the gross costs which employers have to pay, they render the hiring of low-skill labour unprofitable for employers, thus lowering the highly price-elastic demand for unskilled work. On the supply side, payroll taxes reduce the attractiveness of work, because after exempting taxes and social contributions the remaining net pay is hardly above the level of state-provided public assistance which defines the reservation wage of workers. Generous social benefits in continental welfare states thus especially impede low-skill employment because of two effects: They raise the reservation wage below which jobless people will refuse to work, and they drive a wedge between gross and net wages which makes it increasingly difficult for employers to pay wages which would amount to net earnings above the public assistance level. On the demand side of consumers finally, high social security contributions contribute to diminishing the demand for services, because they augment the cost of an hour of service work, while reducing take home pay to an extent that the net earnings from work are no longer sufficient to pay the gross price of services. Hence consumers tend to resort to the self-service economy. OECD data on financing structures and benefit levels allow testing these ideas at least for a limited number of countries. If the mechanisms on the supply side are strong enough to make potential workers shy away from paid work, there should be a negative association between the employment rate and the level of social benefits
152
Jens Alber
Employment rate of people with low educational attainment (% of population aged 25–64; ISCED 0–2)
which replace earnings for people at working age such as social assistance or unemployment benefits. This negative correlation should be particularly strong in the lowskill sectors of the economy where the marginal productivity of workers is low. Thus, if the generosity of social transfers is a disincentive which has a major impact on the reservation wage and the willingness to work, we should expect countries with generous social assistance benefits or generous unemployment benefits to have conspicuously low levels of general employment and even lower levels among the unskilled. OECD data are available for 19 European countries including 4 of the new member states. In 2004 the net rate of social assistance for single unemployed people varied between 21 per cent in Slovakia to 61 per cent in the Netherlands (OECD 2006). The 4 new member states find themselves at the bottom of the rank order together with Portugal and Spain. A similar pattern emerges with respect to unemployment insurance benefits. Only Slovakia has an above-average replacement level, whereas the Czech Republic, Hungary, and Poland belong to the 6 countries with the lowest replacement rates in the EU. In sum, the new member states tend to offer rather meagre social benefits to the poor and the unemployed, but they do not deviate from Western European countries at similar levels of economic development to an extent that would suggest that they are actively pursuing social dumping policies. In a bivariate statistical analysis, the level of social assistance benefits is correlated positively with the general employment rate (r =0.64), as countries with above average benefit generosity also tend to have above average employment levels. There is also a positive correlation between the level of social assistance benefits and the level of low-skill employment (r =0.45; see Figure 5.10). Similarly there is no convincing evidence that generous unemployment benefits reduce labour-supply to a significant degree. In 2004, the association between the level of unemployment benefits and the 75.0 PT 70.0 SE 65.0 DK 60.0
LU
FR
ES
UK
55.0
FI
IE AT BE
50.0
NL
DE
45.0 CZ 40.0 HU 35.0
PL
y = 0.3757x + 36.603 2 r = 0.2005 r = 0.45
30.0 SK 25.0
20
25 30 35 40 45 50 55 60 Level of social assistance benefits (net replacement rate, unemployed single, 100% APW)
65
Figure 5.10 Level of social assistance benefits and low-skill employment in EU countries, 2004 Source: Based on OECD benefits and wages – online (2006); EDS – online (30.11.2006)
Employment rate of population aged 15–64
Employment patterns
153
80.0 DK
75.0
NL
SE
UK
70.0 AT
PT
FI
IE
65.0
DE
CZ
FR
60.0
BE
GR IT
HU
LU
ES
SK y = 0.0922x + 58.722 2 r = 0.0398 r = 0.20
55.0 PL
50.0
25
35 45 55 65 75 85 Unemployment insurance net replacement rate (single, no children)
95
Figure 5.11 Level of unemployment insurance benefits and employment rate, 2004 Source: Based on OECD benefits and wages – online (2006) European Commission (2006: 257–288)
general employment rate was mildly positive (r=0.20; see Figure 5.11). At the same time there was virtually no correlation between benefit levels and the unemployment rate (r =0.01). These findings should not be interpreted as a mere artefact of crude bivariate statistical analysis. More sophisticated econometric studies similarly failed to produce straightforward evidence for detrimental effects of generous benefits. Frequently cited early analyses concluded that the level and the duration of unemployment benefits significantly affect the level of unemployment and its persistence (Scarpetta 1996; Nickell 1997; Nickell and Layard 1999). Yet even the author of one of the pioneer studies was prudent in his conclusions, arguing that high benefits also make participation in the labour market more attractive, that the higher unemployment effect and the higher labour market participation effect tend to cancel out and that the impact of generous benefits might be offset by suitable active measures to push the unemployed back to work (Nickell 1997: 67–68). The most consistent finding found in several studies is that exit from unemployment increases at the moment of benefit exhaustion (for summary accounts see Esping-Andersen 2000; Vodopivec et al. 2003). Drawing attention to contradictory results of previous analyses, a generation of more recent studies headed in the direction of even greater complexity. Blanchard (2006) drew attention to the fact that unemployment kept increasing over time despite the retrenchment of benefits, and reported that in several studies none of the labour market institutions had significant effects on unemployment rates (Blanchard 2006: 31). In a time-series cross-section analysis of OECD countries, a study by the International Institute of Labour Studies found the replacement rate of benefits to have ‘almost always’ negative and statistically insignificant effects on unemployment levels (Baccaro and Rei 2005: 41). Another study highlighted the lack of robustness in empirical results and concluded that a higher replacement rate is associated with
154
Jens Alber
lower unemployment unless benefit duration is extremely high (Baker et al. 2003: 39). Several studies also found more generous benefits to result in a longer search leading to a better matching of job seekers and job qualifications (Acemoglu and Shimer 1999; Gangl 2002). The fact that a clear-cut empirical association between generous unemployment benefits and levels of employment or unemployment is hard to establish is related to a theoretical ambivalence concerning the effects of welfare state benefits. On the downside, generous benefits raise the reservation wage which job seekers are willing to accept as minimum pay, leading to an upward pressure on wages and hence to shrinking demand for labour; and they also allow unemployed workers to be more choosy thus promoting long-term unemployment. On the upside, generous unemployment benefits entail positive labour market effects, because they provide an incentive to find employment, as a job is linked not only to earnings but also to social insurance entitlements, and because the longer search made possible by the benefits promotes efficiency through a better matching of jobs and skills. Several sociological studies have furthermore shown that the unemployed rarely choose to remain on benefit rather than work. German studies found that the vast majority of social assistance recipients are actively searching jobs, and that the successful transition from benefits to work is primarily a function of the marketable skills of the applicants (Leisering and Leibfried 1999; Gangl 1998). British evidence similarly suggests that unemployed benefit claimants are and remain active in the pursuit of work (Walker 2001). Since the difference between take-home pay and the summed level of various welfare state entitlements is difficult to calculate, unemployed people usually lack the information that would be required to make a rational comparison between the level of benefits and earnings from work. Hence the empirical reservation wage – the lowest pay the unemployed would be prepared to accept – is actually not a function of the level of benefits, but of the amount the household needs to make ends meet. Thus in a survey of British job seekers, 87 per cent reported that they determined their minimum wage in light of the amount that they needed to live on, and only 20 per cent said that they took account of benefit levels when fixing their reservation wage (Walker 2001: 446). From various surveys we also know that the unemployed are far from happily enjoying income without work, but are actually much more frequently depressed or unhappy than other groups of the population (Winkelmann and Winkelmann 1998, Layard 2005). Hence, the activating strategies of an enabling state that help people find work need not necessarily be interpreted as a mere intensification of social controls, but are in line with the wishes of the majority of the unemployed. Even if welfare states do not impinge heavily on the supply side, they may have a detrimental effect on the demand side of the labour market. If especially contributionfinanced welfare states drive a wedge between take-home pay and the gross cost of labour which makes unproductive unskilled labour too expensive for employers, we would expect countries with a heavy dependence on contribution-financed social insurance schemes to stand out with low employment rates especially in the low-skill sector, and we would further expect that welfare states with similar institutional arrangements cluster closely together. With the exception of Slovakia the new member states for which the relevant OECD data are available belong to the countries with an above-average size of the tax wedge. With respect to the burden of social security contributions (relative to labour costs) they even occupy ranks 1 (Poland),
Employment patterns
155
Employment rate of people with low educational attainment (% of population aged 25–64; ISCED 0-2)
3 (Hungary), 5 (Slovakia) and 7 (Czech Republic) in the rank-order of 19 EU member states for which data are available. This relatively high burden of social security financing at least in the 4 Visegrad countries is counter to the idea that new member states are pursuing social dumping policies. Just like the newcomers to the Union after the southern enlargement, they are not only under pressures to boost their economic competitiveness, but also under democratic pressures of political legitimation, and this latter aspect calls for generous and embracing welfare programmes (Guillén and Matsaganis 2000; Alber and Standing 2000; Guillén and Palier 2004). The statistical association between the size of the tax wedge and the general employment rate in Europe was weak, but negative in 2004 (r =−0.27). The same is true for the correlation with the level of low-skill employment (r =−0.30). In contrast, the burden of social security contributions impinges more negatively on the general employment rate (r =−0.56). This is in line with the argument that high social security contributions are more detrimental to job creation than general taxation, because they differ in four important respects from income taxes: (1) they impinge only on income from work rather than on all forms of income; (2) they are proportional rather than progressive (3) in contrast to income taxes they usually do not exempt low income earners; (4) and they usually have contribution ceilings which exempt higher incomes from contributions. High social security contributions also have a negative impact on low-skill employment, where the four new member states stand out for their peculiar negative combination of very high contribution rates and employment levels far below average (r =−0.46; see Figure 5.12). In sum, there is no convincing evidence that the new member states seek to boost their competitiveness and employment rates by pursuing social dumping policies. Levying rather high payroll taxes for social security they look very similar to continental
75.0 PT
70.0 SE
65.0 DK
60.0
LU FI
IE
55.0
NL ES
FR
GR
UK AT
50.0
IT
BE DE
45.0 CZ
40.0 35.0 30.0
y= −0.6136x + 71.177 r2 = 0.2144 r = −0.46
PL
HU
SK
25.0 10.0
15.0
20.0
25.0
30.0
35.0
40.0
Social security contributions (employee + employer contributions as % of labour costs)
Figure 5.12 Burden of social security contributions and low-skill employment in EU countries, 2004 Source : Based on OECD (2005b: 18); EDS – online (30. 11. 2006)
2.4 3.7 3.0 7.7 3.0 2.7 2.3 2.3 2.2 1.6 4.4 1.6 3.6 2.6 3.5 4.3 3.0 4.0 2.5 3.8 4.6 4.3 5.4 5.3 4.1 2.4 (4.1)1 2.4 1.9 2.7
1.5 2.9 2.6 6.1 2.8 1.5 1.8 2.0 2.1 1.2 4.2 1.4 4.6 1.3 3.5 3.4 0.8 3.4 3.2 4.4 3.2 3.8 7.2 7.4 6.9 2.1 (4.4)1 2.1 4.9 5.3
1994–2004 2000–2004
Economic growth rates
30.7 26.7 32.9 17.0 26.3 28.5 29.1 29.3 31.2 29.5 22.6 26.1 26.0 24.9 20.0 17.8 18.8 24.3 19.6 20.7 20.0 17.2 13.4 12.6 13.3 27.6 (17.8)1 27.3 — —
2004
Social expenditure ratio
59 49 52 51 45 61 51 52 40 60 51 0 0 25 25 — — — 30 25 30 21 — — — (41)1 (27)1,2 — — —
2004
Social assistance net replacement rate
61 60 77 30 45 71 55 63 73 61 85 54 48 78 69 — — — 50 43 52 64 — — — (62)1 (52)1, 2 — — —
2004
Unemployment benefit net replacement rate
41.5 43.8 48.0 23.8 31.2 43.6 44.9 54.2 47.4 50.7 31.9 45.7 34.9 32.6 38.0 — — — 43.6 45.8 43.1 42.0 — — — (40.8)1 (43.6)1, 2 — — —
2004
Tax wedge
11.0 24.3 29.9 14.2 16.8 — 36.5 33.7 38.0 34.6 24.0 31.8 34.4 28.1 28.3 — 36.2 — 35.2 36.8 38.1 36.2 — — — (27.5)1 (36.6)1, 2 — — —
2004
Social security burden
Notes: 1 Unweighted average. 2 Four countries only.
Sources: Columns 1–2: own calculations, based on European Commission (2006: 257–288); column 3: EDS – online (21.02.2007); columns 4+5: OECD Benefits and Wages – online (2006); columns 6+7: OECD (2005b: 18)
Denmark Finland Sweden Ireland United Kingdom Netherlands Austria Belgium France Germany Luxembourg Italy Greece Portugal Spain Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania EU-15 (NMS-10) EU-25 Bulgaria Romania
Country
Table A5.1 Possible determinants of employment development in EU-25
Employment patterns
157
European welfare states such as Austria, Belgium, France or Germany, but their employment levels lag even behind poorly performing Western European countries. In designing their future social and labour market policies, policy makers in the new member states should be aware of two facts: (1) because a high level of social security contributions has detrimental effects on employment the example of the continental European welfare states is hardly worth pursuing; (2) there are two rather successful social models in the old European Union rather than just one – the model of the Anglo-Saxon countries which economists and international agencies such as the World Bank usually advocate, and the Scandinavian model which has received comparatively little attention in current policy debates (Alber 2006). Utilising welfare state programmes to boost the supply of social services, the Nordic countries have gone a long way in producing a policy mix which seeks to promote efficiency as well as equality. Policy makers in the new member states might benefit from considering this a policy model worth emulating.
Notes 1 The most recent report on employment in Europe was published by the European Commission in October 2006. This report presents employment data up to 2005. Since the 2005 data may be subject to revisions and are more than two years away from the timing of the European Quality of Life Survey which provides the basis of most other chapters in this volume, we decided not to go beyond 2004 as the most recent date for the statistical analysis. Reference to 2005 data is only made in cases where it is relevant for substantive conclusions. 2 In the Great Depression real output declined steepest in Canada (−30 per cent compared to 1929) and the United States (−28 per cent), but more moderately in Europe, where France (−18 per cent) and Germany (−16 per cent) were hit hardest (calculated on the basis of Maddison 1967: 220). 3 This refers to employed persons as a percentage of the population at age 15–59 for whom UNICEF (2006) publishes long time-series data covering also the early years of the transition. 4 Data since 1989 are available in the UNICEF collection for five countries (Albania, Belarus. Moldova, Russia, Ukraine). Post-1995 data also include Croatia, Macedonia and Serbia and Montenegro. Hence the figure presents a long time-series for the former five countries and a short one for the embracive group of eight countries which did not join the Union. 5 Given the fairly high degree of overlap in the two sources, the following discussion will be based on EU statistics which also give more detailed breakdowns of the employment data. 6 The NMS-10 number is an unweighted average, the EU-15 is an aggregate measure which refers to the whole of the old European Union rather than to an unweighted average of national data. 7 The data are based on the three most recent EU Employment Reports (European Commission 2004, 2005, 2006). It is noteworthy that there are some inconsistencies with OECD data. Thus the OECD Labour Force Statistics report a sizeable increase in the employment rate of Austria from 65.5 per cent in 1990 to 68.7 per cent in 2003, whereas according to the EU Employment Reports there was a slight decrease from 69.7 to 68.9 per cent. International variations in the changes of employment rates must always be interpreted with caution, as they occasionally reflect mere changes in statistical concepts. 8 Both the range and the standard deviation of employment levels shrank from 2000 to 2005 in the EU-25 as well as in the EU-27. 9 This table follows Vogel’s (2003) suggestion to distinguish between indicators of inclusion and exclusion in labour markets. 10 There is an unclear discrepancy between various accounts of low-skill employment in Europe and the United States. According to OECD-statistics, the EU-15 has a lower employment rate among people with less than upper secondary education (55.1 as compared to 57 per cent in the US), whereas European employment rates top American ones among people with higher education (OECD Employment Outlook 2004a: 308). On the
158
11
12 13 14
15 16
17 18 19 20
Jens Alber
other hand, the European Commission states in its 2004 Employment Report that ‘both labour market participation ratios and employment rates of the low-skilled are found to be around 5 percentage points higher in the EU 15 compared to the US’ (European Commission 2004: 106). So far, no explanation for this discrepancy could be found. CEE here refers to ten new member states with a socialist past, including Bulgaria and Romania. The long-term account is based on the data in UNICEF (2004: 94). Unemployment rates published in the Employment Reports by the European Commission are in some cases higher, in others lower. On average 29 per cent in the NMS-10, and 19 per cent in EU-15, with prolonged unemployment being here defined as more than three months, or several spells of shorter duration (Kapitany, Kovacs and Krieger 2005: Figure 4 and Table 5). Once again the comparison here is with the unemployment rate in the population aged 25−64. Sector-specific data for 2003 are here taken from European Commission 2004, Annex 6.5. Whereas more recent Reports on Employment in Europe only give employment shares of the various sectors, this report also shows sector-specific employment rates for the years 1998, 2000 and 2003. However, there are a number of printing errors in the source which make the data not fully reliable. For example it is stated that agricultural employment in the NMS-10 grew from 2.6 per cent in 1998 to 6.7 per cent in 2003. Whereas the rates for the three sectors correctly add up to the total employment rate in 2003 (56.4 per cent), the three sector-specific rates for 1998 – 2.6, 11.0, and 16.0 per cent – only add up to 29.6 per cent rather than to the total employment rate which is given as 60 per cent. It is here assumed that the correct values are 12.6 per cent for the employment rate in agriculture and 36 per cent for the rate in services. Rough estimates of sector-specific employment rates can also be calculated from the information in the most recent Employment Report which gives sector-specific employment shares together with the total employment rate. Again this is based on European Commission (2004). The two more recent Employment Reports published by the Commission do not give any information on sector-specific employment for Bulgaria and end the series for Romania with the year 2002. Despite an increase of more than a percentage point between 2000 and 2004, Lithuania has not yet regained the level of 1998. Hungary had shrinkage since 2000. The same is true for Ireland. This information is based on recalculations of the data on sector shares in European Commission (2006). Including Cyprus and Malta, the average growth rate in the ten NMS was 4.1 per cent. While being close to zero, the statistical association between the female activity rate and total unemployment in the EU-25 has a negative sign in 2004 (r = −0.12). Cross-sectional correlation coefficients are r = 0.32 in the case of younger people (15–24), and r = 0.51 for older people (55−64). A somewhat related hypothesis points to the detrimental effects of rigid employment protection legislation and claims that countries with excessive labour market regulations protect insiders who already hold a job at the expense of outsiders who fail to get in, because there is too little fluctuation (OECD 1994; Europäische Kommission 1993). Critical and empirically informed discussions of the labour market rigidity hypothesis may be found in Esping-Andersen (2000) as well as in the most recent report on Employment in Europe (European Commission 2006).
References Acemoglu, D. and Shimer, R. (1999) Productivity Gains from Unemployment Insurance, Working Paper series, no. 7352, Cambridge, MA: National Bureau of Economic Research. Alber, J. (2006) ‘The European social model and the United States’, European Union Politics, 7, 3: 393–419. Alber, J. and Standing, G. (2000) ‘Social dumping, catch-up, or convergence? Europe in a comparative global context’, pp. 99–119, in J. Alber and G. Standing (eds), Europe in a Comparative Global Context, special issue of Journal of European Policy, 10, 2.
Employment patterns
159
Baccaro, L. and Rei, D. (2005) Institutional Determinants of Unemployment in OECD Countries: A Time Series Cross-section Analysis (1960-08), Geneva: International Institute for Labour Studies. Baker, D., Glyn, A., Howell, D. and Schmitt, J. (2003) Labor Market Institutions and Unemployment, Center for European Studies Working Paper series no. 98, Cambridge, MA: Harvard University. Baumol, W.J. (1967) ‘Macroeconomics of unbalanced growth: the anatomy of urban crisis’, American Economic Review, 57, 2: 415–426. Blanchard, O. (2006) ‘European unemployment: the evolution of facts and ideas’, Economic Policy, January 2006: 7–59. Brown, G. (2005) Global Europe: Full-employment Europe, London: HM Treasury. Cazes, S. and Nesporova, A. (2003) Labour Markets in Transition: Balancing Flexibility and Security in Central and Eastern Europe, Geneva: International Labour Organization. Commission of the European Communities (2001) The Impact of Eastern Enlargement on Employment and Labour Markets in the EU and the Member States, European Commission’s Director for the Employment and Social Affairs, Brussels. Commission of the European Communities (2005) Working Together for Growth and Jobs: A New Start for the Lisbon Strategy, Communication to the Spring European Council Com (2005) 24, Brussels. Esping-Andersen, G. (1996) ‘Welfare states without work: the impasse of labour shedding and familialism in continental European social policy’, pp. 66–87, in G. Esping-Andersen (ed.), Welfare States in Transition, London: Sage Publications. Esping-Andersen, G. (1999) Social Foundations of Postindustrial Economies, Oxford: Oxford University Press. Esping-Andersen, G. (2000) ‘Who is harmed by labour market regulations? Quantitative evidence’, pp. 66–98, in G. Esping-Andersen and M. Regini (eds), Why Deregulate Labour Markets, New York: Oxford University Press. Europäische Kommission (1993) Weiflbuch zu Wachstum, Wettbewerbsfähigkeit und Beschäftigung in der Gemeinschaft, Brussels, Luxemburg. European Commission (2004) Employment in Europe 2004: Recent Trends and Prospects, Directorate-General for Employment and Social Affairs, Luxembourg: Office for Official Publications of the European Communities. European Commission (2005) Employment in Europe 2005: Recent Trends and Prospects, Directorate-General for Employment and Social Affairs, Luxembourg: Office for Official Publications of the European Communities. European Commission (2006) Employment in Europe 2006: Recent Trends and Prospects, Directorate-General for Employment and Social Affairs, Luxembourg: Office for Official Publications of the European Communities. Fahey, T. (2004) ‘Employment, education and skills’, pp. 23–32, in Foundation (eds), Quality of Life in Europe: First Results of a New Pan-European Survey, European Foundation for the Improvement of Living and Working Conditions. Ferrera, M. (1996) ‘The “southern model” of welfare in social Europe’, Journal of European Social Policy, 6, 1: 17–37. Fourastié, J. (1950) Le grand espoir du XXe siècle, Paris: Presses Universitaires de France. Gangl, M. (1998) ‘Sozialhilfebezug und Arbeitsmarktverhalten – Eine Längsschnittanalyse der Übergänge aus der Sozialhilfe in den Arbeitsmarkt’, Zeitschrift für Soziologie, 27, 3: 212–232. Gangl, M. (2002) Unemployment Benefits as a Search Subsidy: New Evidence on Duration and Wage Effects of Unemployment Insurance, Discussion Paper FS I 02–208, Berlin: Wissenschaftszentrum Berlin für Sozialforschung. Gershuny, J. (1978) After Industrial Society: The Emerging Self-service Economy, London: Macmillan.
160
Jens Alber
Guerrero, T.J. (1995) ‘Legitimation durch Sozialpolitik. Die spanische Beschäftigungskrise und die Theorie des Wohlfahrtsstaates’, Kölner Zeitschrift für Soziologie und Sozialpsychologie, 47, 4: 727–752. Guillén, A. and Matsaganis, M. (2000) ‘Testing the “social dumping” hypothesis in Southern Europe: welfare policies in Greece and Spain during the last 20 years’; pp. 120–145, in J. Alber, and G. Standing (eds), Europe in a Comparative Global Context; special issue of Journal of European Policy, 10, 2. Guillén, A. and Palier, B. (eds) (2004) EU Enlargement, Europeanization and Social Policy, special issue of Journal of European Policy, 14, 3. Hafemann, K. and van Suntum, U. (2004) Beschäftigungspolitik in Osteuropa, Gütersloh: Bertelsmann Stiftung. Iversen, T. and Wren, A. (1998) ‘Equality, employment, and budgetary restraint: the trilemma of the service economy’, World Politics, 50, 4: 507–546. Kaelble, H. (1989) ‘Was Prometheus most unbound in Europe? The labour force in Europe during the late 19th and 20th centuries’, Journal of European Economic History, 18, spring: 65–104. Kapitány, B., Kovacs, K., and Krieger, H. (2005) Working and Living in an Enlarged Europe, Dublin: European Foundation for the Improvement of Living and Working Conditions. Kocka, J. and Offe, C. (eds) (2000) Geschichte und Zukunft der Arbeit, Frankfurt: Campus. Kommission für Zukunftsfragen der Freistaaten Bayern und Sachsen (ed.) (1996) Erwerbstätigkeit und Arbeitslosigkeit in Deutschland – Entwicklung, Ursachen und Maβnahmen, Teil 1: Entwicklung von Erwerbstätigkeit und Arbeitslosigkeit in Deutschland und anderen frühindustrialisierten Ländern, Bonn. Krieger, H. (2004) Migration Trends in an Enlarged Europe, Dublin: European Foundation for the Improvement of Living and Working Conditions. Layard, R. (2005) Happiness, London: Allen Lane. Leisering, L. and Leibfried, S. (1999) Time and Poverty in Western Welfare States: United Germany in Perspective, Cambridge: Cambridge University Press. Maddison, A. (1967) Economic Growth in the West: Comparative Experience in Europe and North America, New York: Norton Library. Nickell, S. (1997) ‘Unemployment and labor market rigidities: Europe versus North America’, Journal of Economic Perspectives, 11, 3: 5–74. Nickell, S. and Layard, R. (1999) ‘Labour market institutions and economic performance’, pp. 3029–3084, in O. Ashenfelter and D. Card (eds), Handbook of Labor Economics, Vol. 3C, Amsterdam: Elsevier. OECD (1994) The OECD Jobs Study (3 vols), Paris. OECD (2000) Labour Force Statistics 1979–1999, Paris: OECD. OECD (2004a) Employment Outlook 2004, Paris. OECD (2004b) Labour Force Statistics 1983–2003, Paris: OECD. OECD (2005a) Labour Force Statistics 1984–2004, Paris: OECD. OECD (2005b) Taxing Wages 2003–2004, Paris: OECD. OECD (2006) Benefits and wages: gross/net replacement rates, country specific files and tax/benefit models (latest update: March 2006). Online. Available http: (accessed February 2007). Okun, A. (1975) Equality and Efficiency: The Big Tradeoff, Washington, DC: Brookings. Rose, R. (1996) What Is Europe? A Dynamic Perspective, New York and London: Addison Wesley Longman. Saraceno, C., Olagnero, M. and Torrioni, P. (2005) First European Quality of Life Survey: Families, Work and Social Networks, Dublin: European Foundation for the Improvement of Living and Working Conditions.
Employment patterns
161
Scarpetta, S. (1996) Assessing the Role of Labour Market Policies and Institutional Settings on Unemployment: A Cross-country Study, OECD Economic Studies no. 26, Paris. Scharpf, F. (1986) ‘Strukturen der post-industriellen Gesellschaft oder: Verschwindet die Massenarbeitslosigkeit in der Dienstleistungs- und Informations-Ökonomie?’, Soziale Welt, 37, 1: 1–24. Scharpf, F. (2000) ‘The viability of advanced welfare states in the international economy: vulnerabilities and options’, Journal of European Public Policy, 7, 2: 190–228. UNICEF (2004) Innocenti Social Monitor 2004: Economic Growth and Child Poverty in the CEE/CIS and Baltic States, Florence: UNICEF Innocenti Research Centre. UNICEF (2006) Innocenti Social Monitor 2006: Understanding Child Poverty in the Southeastern Europe and the Commonwealth of Independent States, Florence: UNICEF Innocenti Research Centre. Vaughan-Whitehead, D.C. (2003) EU Enlargement versus Social Europe? The Uncertain Future of the European Social Model, Cheltenham, Northampton: Edward Elgar. Vodopivec, M., Wörgötter, A. and Raju, D. (2003) Unemployment Benefit Systems in Central and Eastern Europe: A Review of the 1990s, Washington, DC: World Bank. Vogel, J. (2003) ‘The Labour Market’, Social Indicators Research, 64, 3: 349–372. Walker, A. (2001) ‘Can Work Work? A preliminary assessment of the “welfare to work” strategy’, pp. 145–171, in J. Alber and J. Kohl (eds), Arbeitsmarkt und Sozialstaat, Sonderheft der Zeitschrift für Sozialreform, Wiesbaden: Chmielorz. Winkelmann, L. and Winkelmann, R. (1998) ‘Why are the unemployed so unhappy? Evidence from panel data’, Economica, 65, 257: 1–15. Wood, A. (1994) North-South Trade, Employment and Inequality: Changing Fortunes in a Skill-driven World, Oxford: Clarendon Press. Zukunftskommission der Friedrich-Ebert-Stiftung (1998) Wirtschaftliche Leistungsfähigkeit, sozialer Zusammenhalt, ökologische Nachhaltigkeit. Drei Ziele – ein Weg, Bonn.
6
Working conditions and quality of work A comparison of Eastern and Western Europe1 Claire Wallace and Florian Pichler
Background Until the end of communism there was a distinctive tradition of work in the countries of Eastern and Central Europe. Full employment was mandatory, with both women and men working full time (except for periods of childcare leave) until retirement. The domination of heavy industrial and extractive concerns and state employment meant a predominance of work at lower professional levels and few service jobs (Vecernik 1996). The universally low wages were compensated for by a range of work-related benefits such as housing and subsidised holidays or even heating. Rewards were not based upon market demand or even professional qualifications but upon access to the political caste (mainly by joining the Communist Party) or by working in one of the better rewarded occupations such as mining. Hence, bus drivers could earn more than doctors. The implicit ‘citizenship’ contract between the employee and the state was that benefits such as health, education, holidays and pensions were earned by contribution to the national productive effort. Those who did not work were morally condemned as well as being financially penalised (there was no unemployment benefit) and excluded from the welfare system. Hence, there was a heavy moral commitment to work rather than a work ethic based upon financial incentives or career prospects. On the other hand, work was not very demanding. It was summed up in the adage ‘We pretend to work and they pretend to pay us.’ For women to work full time, childcare facilities were provided, often by the employer and for long periods, or even on a round-the-clock basis (Wallace 2003a, Wallace 2003b). Women comprised many of the professions such as doctors and teachers (which were under-rewarded) as well as many of the employees of the state services (Pollert 2003; Pollert 2005; Haas et al. 2006). Even though they were underrepresented at the higher levels of employment, they were relatively well situated in the labour market compared to many of their Western neighbours. This situation changed after the political collapse of communism. The rapid liberalisation of the economy that followed from the advice of those propounding the ‘Washington consensus’ of the time led to steep price rises and in some countries hyper-inflation. Wages however, stayed generally low and fell ever further behind prices. Industries were privatised or closed down, leading to the shedding of many jobs – the first time in 50 years that there was widespread unemployment in these countries. Rationalisation policies led to the social benefits associated with employment being scaled down or dispensed with – heating, health, holidays and housing
Working conditions in Eastern and Western Europe
163
were all severed from the employment contract. Workplace nurseries were closed. At the same time the central state, already in financial trouble, contracted even further, leading to job loss from state employment as well (Andor and Summers 1998; Pickles 1998; Vecernik 1996). The flat-rate taxes introduced in many Eastern European new member states since 2000 to boost economic competitiveness, have further contributed to the fiscal crisis of the state in the short term. At the same time new enterprises opened and old ones privatised – the most successful and obvious in the first round being small private businesses, creating a new layer of entrepreneurs (Earle et al. 1994). Private-sector jobs offered sometimes better money, but usually worse working conditions with longer hours and less security, so that the safer, older jobs seemed to be better even on lower pay. Many businesses were precarious and people who dipped their toes into the new waters of free enterprise often went under. Freedom came at a high price for workers and businesses alike. As a result there was a rapid expansion of the service industries, even though these are still underdeveloped compared to Western Europe and strong regional differentiation in employment developed.2 The old industrial areas were threatened with becoming ‘rust belts’ of high unemployment and economic decline. At the same time other areas such as capital cities and the Western border regions developed rapidly to become centres of growth and new technology. Insufficient housing, however, meant that the disadvantaged populations were immobilised and at least some of those taking advantage of the growth regions were immigrants, mostly from further East. In the process of transition, the various countries of Eastern and Central Europe started to diverge from one another (Agh 1998; EBRD 2003). The countries with the fast economic track reforms included Hungary, Czech Republic, Estonia and to a lesser extent Poland. Slovenia should be mentioned as a special case here, because although that country resisted many of the more liberalising reforms and expanded rather than contracting parts of the welfare state, the general level of development in the country ensured it a place at a higher level of prosperity than the other East Central European countries and in fact higher than some of the poorer EU-15 countries, as we will show below. Other countries are far further behind in development and here we should point to Romania and Bulgaria, who emerged much later from the post-transition traumas than the other countries we are considering, and Latvia, which was formerly part of the Soviet bloc. The remaining countries fall somewhere in the middle. Patterns of chain migration developed starting with cross-border commuting and work migration of short duration in the regions nearest the old EU (such as Germany and Austria), but migration soon started to widen to other regions, including the UK, Ireland and the Southern European countries such as Portugal, Spain, Italy and Greece (Wallace and Stola 2001). People exercising their new freedom to travel started to compare their living and working conditions (and especially their salaries) not with other communist countries (which were usually worse) but with western EU countries such as Germany, Austria and the UK. The accession process has encouraged a reorientation in work regulations and models of citizenship away from the neo-liberal paradigm and more in the direction of the various European Social Models, implying re-regulation and state intervention. Now the new member states have to conform to the same regulations, including the European Employment Strategy, Health and Safety Legislation, Social Dialogue, the Lisbon Targets and so on, as Western Europe, which has probably helped to improve working conditions.
164
Claire Wallace and Florian Pichler
However, there is still a lot of continuity in working conditions. The informal economy, a transformed continuation of the old ‘second economy’, still plays an important part, especially in some countries where avoiding regulations is an inherited skill. Small-scale and subsistence agriculture provides a ‘social safety net’ for some of the rural poor in countries like Poland and Romania and helps to absorb unemployment (Wallace and Haerpfer 2002). Wages are still relatively low, but unemployment rates are very high (see Table 6.1). In this chapter we consider attitudes to working conditions in 2003. This was already some fourteen years after the transition, so that people have had an opportunity to acclimatise to changes. What was considered transitional at first now appears normal. Furthermore, the new member states, after a period of crisis until the mid-1990s have all experienced growth by 2003, so that the benefits of transition have become more apparent. Here we compare ten post-communist countries including eight new member states (Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia and Slovenia) and two (former) Candidate Countries (Bulgaria and Romania) with the old EU. Using both official statistics and survey data drawn from the 2003 EQLS, we are able to bring together both subjective and objective elements of working conditions.
1. Working conditions 1.1 Long hours and low pay Working conditions are uniformly worse in Eastern Europe than in Western Europe. This is already apparent when we look at hours of work and rewards. Table 6.1 shows the mean of usual weekly working hours in people’s main job in ten Eastern European countries taken from Labour Force Survey. Average working hours in Eastern Europe are longer at 40.8 than in the old EU at 37.4. The difference amounts to 3.4 hours added to the average working week, or around half a working day. Apart from Lithuania, working time reaches or exceeds 40 hours a week in Eastern Europe, whereas in other member states of the European Union, only the Greeks, Spaniards, Cypriots and Maltese work comparably long hours. However, if we take into account all the jobs that people do, the hours in Eastern Europe are stretched even further and differences widen. The EQLS survey measures working hours in both the main employment and secondary jobs as well, so that using this source we find that hours are much higher in Eastern Europe, because doing additional work to make ends meet is relatively common. Using this measure, working hours range from 43 hours per week in Estonia to more than 51 in Romania. In comparison to Western Europe, Eastern Europeans thus almost work one day more per week (6.4 hours) than residents in Western Europe. Women work fewer hours than men, but the part-time option is neither common nor desired by women in the Eastern EU (Wallace 2003a). Wages are too low to make it a desirable option and what part-time work there is tends to be for preretirement or post-retirement purposes rather than as a way of balancing work and family. Eastern European workers of both genders therefore work longer hours for less money than do their western counterparts. Some people may undertake more than
40.8 37.4
Eastern Europe EU-15
46.0 39.6
43.1 43.6 43.0 44.3 44.8 44.8 45.2 51.3 43.8 44.7
SD
(14.43) (12.93)
(8.93) (9.07) (12.26) (12.56) (13.94) (13.66) (16.17) (15.82) (10.01) (10.56)
EQLS
Mean
781 2321
362 941 755 839 558 659 978 472 794 1450
Eurostat
Mean
11.1 8.0
13.7 7.8 10.0 5.9 10.5 12.4 19.6 6.8 17.6 6.7
Eurostat
Mean
Wages (E, PPS) Unemployment
3.6 4.3
2.7 3.5 3.7 4.1 3.3 3.2 3.6 3.8 3.5 4.1
Mean
(1.22) (1.06)
(1.29) (1.11) (1.21) (1.11) (1.29) (1.29) (1.14) (1.22) (1.12) (1.08)
SD
Job security SD
2.6 3.1
2.5 2.8 2.5 2.4 2.6 2.5 2.5 2.6 2.5 2.6
(1.08) (1.12)
(1.01) (1.07) (1.02) (1.05) (1.07) (1.04) (1.09) (1.09) (1.08) (1.08)
EQLS
Mean
Payment
3.7 4.0
3.8 3.7 3.7 3.8 3.7 3.7 3.5 3.9 3.5 4.1
Mean
(1.08) (1.04)
(0.97) (1.09) (1.03) (1.07) (1.05) (0.97) (1.13) (1.00) (1.15) (0.94)
SD
Intrinsic rewards
1 t-test for difference between mean in Eastern Europe and the EU-15: p < 0.01.
Working hours: ‘Average number of usual weekly hours of work in main job’ (EUROSTAT). Total working hours in main and secondary jobs (EQLS). Wages: ‘Mean Monthly Gross Earnings in Purchasing Power Parties (EURO)’ (EUROSTAT). Unemployment: Share of unemployed among the working population (EUROSTAT). Job security (Q11): ‘Using this card, how likely do you think it is that you might lose your job in the next 6 months? 1) very likely, 2) quite likely, 3) neither likely, nor unlikely, 4) quite unlikely, 5) very unlikely.’ Payment, intrinsic rewards (Q12), high means indicate positive evaluations: ‘How much do you agree or disagree with the following statements describing positive and negative aspects of your job? Scale from 1) strongly agree to 5) strongly disagree: I am well paid (recoded); My work is dull and boring’. Job Satisfaction (Q41b): ‘Could you please tell me on a scale of 1 to 10 how satisfied you are with each of the following items, where 1 means you are very dissatisfied and 10 means you are very satisfied? b) your present job.’
Notes:
Sources: EUROSTAT New Cronos: LFS Series (Working hours, wages); the EQLS 2003
40.8 42.1 39.9 40.6 42.0 37.9 41.3 41.4 40.5 41.1
Eurostat
BG CZ EE HU LV LT PL RO SK SI
Country
Mean
Working hours
Table 6.1 Working conditions in Eastern Europe I (working hours, unemployment, wages, job security, pay, intrinsic rewards and job satisfaction, means and standard deviations)
166
Claire Wallace and Florian Pichler
3.5 3.3 3.1 2.9 2.7 2.5 2.3 2.1 1.9 1.7 1.5
HU
BG
LT EE SK RO LV PL Eastern European countries
Professional Non skilled worker
Self-employed EU15 average
SI
CZ
Skilled worker
Perceived extrinsic rewards (Mean, scale 1-5)
Perceived extrinsic rewards (Mean, scale 1-5)
one job (as well as part-time farming), either in the formal or informal economies in order to make ends meet. This results in even longer hours. There are also large differences in wages between Eastern and Western Europe, with E781 gross earnings per month being the average in Eastern Europe compared with E2321 in the EU-15. However, there are very large differences within the NMS and candidate countries ranging from E362 a month in Bulgaria to E1450 in Slovenia. Slovenia is thus comparable with some of the EU-15 countries, since it has higher wages than in Portugal (E1164) and only a bit less than in Greece (E1514). These objective measures of wage levels are reflected in perceptions of pay (see under ‘Payment’ in Table 6.1). In no East European country were people more likely to agree with the statement ‘I am well paid’ than in any country in Western Europe – not even Slovenia, where in fact there is some overlap with the EU-15. However, there is surprisingly little variation in perceptions of reward across Eastern Europe, given the vastly differing wage levels, with Slovenians earning four times as much as Bulgarians on average. Nor are the relatively high salaries in Hungary (by Eastern European standards) reflected in perceptions of being well rewarded. Across Eastern Europe, most people feel poorly paid. However, there are important differences between occupational groups in this respect (Figure 6.1). The self-employed are much less likely to feel badly off; in Bulgaria, Romania, Slovakia and the Czech Republic their perceptions of their pay are respectably above the EU average. Those in professional jobs feel poorly rewarded, although they are still better off than those in manual jobs, with unskilled workers feeling the worst off. The divergence between professionals and manual workers is highest in the Czech Republic, Slovenia and Poland, so that it seems that at least in those countries the doctors might earn more than the bus drivers. However, women feel even worse rewarded then men. Everywhere apart from Slovenia, they feel worse off and gender differences are particularly stark in the Czech Republic and Romania.
3.5 3.3 3.1 2.9 2.7 2.5 2.3 2.1 1.9 1.7 1.5
HU PL BG SI EE LT SK LV RO CZ EU15 Eastern European Countries Male
Female
Figure 6.1 Perceptions of pay: gender and occupational differences across Eastern Europe. Source: EQLS 2003 Note: Means (scale of 1 to 5) by gender and occupational groups.
Working conditions in Eastern and Western Europe
167
Job insecurity In comparing the levels of unemployment, we find huge differences across Eastern Europe. According to data from Eurostat, unemployment rates in Hungary, Slovenia, Romania and the Czech Republic are lower than the EU-15 average. In these countries, approximately 7 per cent have no job. On the other hand, unemployment is extremely high in Slovakia (17.6 per cent) and Poland (19.6 per cent). The country average is 11.1 per cent in Eastern Europe. In the Eastern European new member states loss of job is a threat because high unemployment in some regions makes re-employment difficult and unemployment benefits are low and time restricted. As Table 6.1 shows, people in Eastern Europe are much more likely to feel that their jobs are insecure than in the EU-15. Eastern Europeans score 3.6 on average in terms of likelihood of losing their jobs in the next 6 months, whereas Western Europeans score 4.3. However, there is considerable variation within Eastern Europe, from a low of 2.7 in Bulgaria to 4.1 in Hungary and Slovenia (which is similar to the score of 4.0 reported in Greece, Portugal and France). Indeed this issue throws up the biggest variations across Eastern Europe where the variation of 1.4 points on a 5-point scale is a far more heterogeneous pattern than we find in the EU-15. Even so, all countries in Eastern Europe report feeling more insecure than in Western Europe. Hence, whilst in the past wages were low but jobs were secure and came with a variety of benefits, it is now the case that wages are still low, but jobs are insecure and the benefits have been lost. People in the new member states have to construct their working careers in an environment that is increasingly risky, uncertain and subject to rapid social change. Turning to differences between occupational groups (Figure 6.2), we find that selfemployed people report the highest levels of jobs security, in fact similar to the EU-15 average in many countries. Non-skilled workers report the lowest level of job security in a number of countries and professional employees feel relatively more secure. In two countries, we find hardly occupational differences. These are Poland and Slovenia. In countries such as Slovakia and Romania we find rather large occupational differences concerning perceived job security. By contrast, there are very few gender differences in this respect.
Dull and boring jobs Multivariate analyses have shown that intrinsic rewards are among the most important determinants of job satisfaction (Wallace et al. 2007). In general, people in the new Eastern member states find their jobs less intrinsically rewarding than do Western Europeans. However, with a mean score of 3.7 for people in new member states and 4.0 in the EU-15, there is more convergence on this item than on others. In fact, in Slovenia people are more likely to see their jobs as interesting than in the EU-15 average. Indeed, Eastern Europeans are more likely to see their jobs as interesting than are workers in Greece, Spain, Portugal and the United Kingdom. The stark contrasts between the EU-15 average and the Eastern European average is mainly accounted for by countries such as Belgium, Germany, the Netherlands and Sweden, where many people report having interesting work. Turning to within-country differences (Figure 6.3) we can see that professional workers and the self-employed are not far off the EU-15 average on this indicator,
Claire Wallace and Florian Pichler
4.5
Perceived job security (Mean, scale 1–5)
Perceived job security (Mean, Scale 1–5)
168 4.3 4.1 3.9 3.7 3.5 3.3 3.1 2.9 2.7 2.5
BG
LT
LV SK CZ PL RO EE Eastern European Countries
Professional Non skilled worker
Self-employed EU15 average
SI
4.5 4.25 4 3.75 3.5 3.25 3 2.75 2.5 2.25 2
HU
BG LV
LT SK CZ PL EE RO SI Eastern European countries Male
Skilled worker
HU EU15
Female
Figure 6.2 Perceived job security: gender and occupational differences across Eastern Europe Source: EQLS 2003 Note: Means of perceived job security (scale of 1 to 5) by gender and occupational groups.
5 4.5 4 3.5 3 2.5 2
LV
SK
PL CZ LT BG RO HU Eastern European countries
EE
SI
Perceived intrinsic rewards (Mean, scale 1–5)
Perceived intrinsic rewards (Mean, scale 1–5)
whilst those in manual jobs are more likely to find them dull and boring, with occupational differences being very small in Bulgaria and Romania. In Slovenia, however, the professionals and self-employed were strikingly more likely to find their jobs interesting – substantially exceeding the EU average. Gender differences are rather small as they are in the EU-15 generally and there is no consistent pattern. Only in Latvia, do we find some evidence that women are more intrinsically rewarded.
4.2
4
3.8
3.6
3.4
3.2
3 PL
Professional Non skilled worker
Self-employed EU15 average
Skilled worker
LV
SK LT CZ EE BG RO HU Eastern European countries Male
SI EU15
Female
Figure 6.3 Perceived intrinsic rewards: gender and occupational differences across Eastern Europe Source: EQLS 2003 Note: Means of perceived intrinsic rewards (scale of 1 to 5) by gender and occupational groups.
Working conditions in Eastern and Western Europe
169
Poor physical conditions, heavy workloads and time pressure Table 6.2 shows the differences between Eastern and Western Europe on other dimensions of working conditions. Starting with work load, it seems that it is people in Western Europe who feel that their work is demanding and stressful. Only in Estonia, Czech Republic and Hungary had work become as stressful as in Western Europe, probably reflecting the fact that these countries were among the most developed in terms of labour market reforms. In these countries, flexibilisation of work has been most extensively introduced (Cazes and Nesporova 2003). Another way of measuring work stress is the perception of tight deadlines. Eastern and Western Europeans do not differ from each other in this respect. However, the experience of tight deadlines varies widely across Eastern Europe. These findings suggest that work intensity may have increased in post-communist workplaces in some countries as a result of the introduction of market-oriented productivity measures. Eastern European workers are much more likely to say that they work in dangerous and unhealthy conditions than those in Western Europe. Workers in the most reformed countries – Estonia, Czech Republic and Hungary – are least likely to state this, indicating again that there is perhaps greater convergence between East and West in these labour markets than with others. Lack of work autonomy Given the traditional nature of many of the jobs in Eastern Europe, we would not expect there to be much work autonomy. The lack of worker autonomy is consistent with the tradition of the hierarchical mass workplaces inherited from former times. Indeed it is the case that work autonomy is generally lower in Eastern Europe. However, an exception is Slovenia, with higher levels of perceived work autonomy than the EU average. Poor promotion prospects The last working condition we present here concerns the perception of promotion prospects and the fulfilment of career opportunities. This working condition is the one found least often in Eastern Europe. An average score of 2.5 is not only significantly lower than in the EU-15, it is also the lowest score among all observed working conditions in Eastern Europe. Furthermore, the variation across Eastern Europe is not very large. This is also consistent with the mass industrial workplace that was a characteristic of the past. Little job satisfaction Another very central criterion of working conditions is job satisfaction. This indicator can be considered as an overall measure of the quality of work. Previous analysis has shown that the most important working conditions to influence job satisfaction are security of the job and perceived levels of payment on the extrinsic dimensions and not having a dull job as well as having promotion prospects, on the intrinsic dimension across all European countries (Wallace et al. 2007).
2.7 2.8
Eastern Europe EU-15
(1.14) (1.25)
(1.17) (1.04) (1.14) (1.19) (1.18) (1.05) (1.11) (1.22) (1.11) (1.15)
SD
2.9 2.9
3.4 2.7 3.0 3.0 2.9 2.6 2.8 3.1 3.1 2.4
Mean
(1.28) (1.27)
(1.16) (1.16) (1.18) (1.35) (1.22) (1.12) (1.28) (1.33) (1.31) (1.23)
SD
Tight deadlines
3.4 4.0
3.4 3.6 3.6 3.6 3.3 3.2 3.3 3.4 3.4 3.4
Mean
(1.32) (1.16)
(1.35) (1.22) (1.25) (1.32) (1.27) (1.29) (1.30) (1.39) (1.35) (1.51)
SD
Dangerous and unhealthy work environment
3.2 3.7
3.4 3.0 3.4 3.0 3.4 3.4 3.2 3.1 3.6 3.8
Mean
(1.28) (1.19)
(1.24) (1.29) (1.21) (1.31) (1.21) (1.10) (1.24) (1.38) (1.19) (1.13)
SD
Autonomy and influence
2.5 2.9
2.6 2.5 2.4 2.3 2.5 2.6 2.7 2.4 2.9 2.7
Mean
(1.19) (1.21)
(1.20) (1.14) (1.12) (1.16) (1.21) (1.14) (1.19) (1.23) (1.25) (1.24)
SD
Prospects
6.9 7.4
6.5 7.1 6.8 7.0 6.8 7.0 6.8 7.4 6.5 7.0
Mean
(2.31) (1.88)
(2.71) (2.19) (2.10) (2.26) (2.05) (2.26) (2.37) (2.07) (2.60) (2.14)
SD
Job satisfaction
1 t-test for difference between mean in Eastern Europe and the EU-15: p <0.01.
Work load, tight deadlines, dangerous and unhealthy work environment, autonomy and influence, prospects: How much do you agree or disagree with the following statements describing positive and negative aspects of your job? Scale from 1) strongly agree to 5) strongly disagree: My work is too demanding and stressful; I have a great deal of influence in deciding how to do my work (recoded); My job offers good prospects for career advancement (recoded); I constantly work to tight deadlines; I work in dangerous and unhealthy conditions. Job satisfaction: Could you please tell me on a scale of 1 to 10 how satisfied you are with each of the following items, where 1 means you are very dissatisfied and 10 means you are very satisfied? Your present job.
Notes:
Source: EQLS 2003
2.2 2.8 3.1 2.9 2.7 2.4 2.7 2.7 2.5 2.6
Country BG CZ EE HU LV LT PL RO SK SI
Mean
Work load
Table 6.2 Working conditions in Eastern Europe II (workload, tight deadlines, dangerous and unhealthy work environment, autonomy and influence, future prospects of promotion and career and job satisfaction, means and standard deviations)
Working conditions in Eastern and Western Europe
171
Job satisfaction is measured on a ten-point scale. Respondents were asked whether they are satisfied with their present job. Overall, job satisfaction is lower in Eastern than in Western Europe. Among East European countries, the Romanians report the highest level of satisfaction. Furthermore, the Romanians experience the same level of job satisfaction as the average worker in the EU-15. The Czechs, Hungarians and Slovenians also report high levels of job satisfaction. The long working hours and low pay suffered by Romanians mean that it is strange to find that they have high levels of job satisfaction, although they reported higher job security than many of the other countries as well as high intrinsic rewards. The relatively high quality of work we find in Slovenia, is also reflected in high levels of job satisfaction. Generally better working conditions are also reflected in higher job satisfaction in the Czech Republic and Hungary. On the other hand, high job insecurity and poor rewards make Bulgarians the most discontented of all the Eastern new Member State citizens and they are followed by Slovaks and Poles. Whereas Table 6.2 has shown that the level of job satisfaction varies considerably from country to country in Eastern Europe, we also find substantial variation in job satisfaction within each country. Job satisfaction among non-skilled workers varies greatly across these ten countries, but job satisfaction among professional groups is quite uniform and reasonably high across Eastern Europe. Altogether, the most satisfied occupational group is the self-employed. In many Eastern countries, their levels of job satisfaction are higher than the EU-15 average. Especially in Slovakia, Bulgaria and the Czech Republic, different occupations report very different levels of job satisfaction. For instance, in Slovakia and the Czech Republic, non-skilled workers report very low levels of job satisfaction, whereas professionals report levels of job satisfaction almost as high as the EU-15 average. On the other hand, in Latvia, Estonia, Slovenia, Poland and Lithuania, the gaps between occupational groups are much smaller. In these countries, non-skilled workers reach similar levels of job satisfaction to the skilled. Figure 6.4 also includes a gender perspective. The right-hand chart shows that there are gender differences in job satisfaction, but they are not consistent across all countries. 7.6 Job satisfaction (Mean, scale 1–10)
Job satisfaction (Mean, scale 1–10)
8.5 8 7.5 7 6.5 6 5.5 5 4.5 4
SK
BG
LV EE SI CZ HU PL Eastern European countries
Professional Non skilled worker
Self-employed EU15 average
LT
RO
Skilled worker
7.4 7.2 7 6.8 6.6 6.4 6.2 6
BG SK
PL LV EE LT SI HU CZ RO EU15 Eastern European countries Male
Female
Figure 6.4 Job satisfaction: gender and occupational differences across Eastern Europe Source: EQLS 2003 Note: Means of job satisfaction (scale of 1 to 10) by gender and occupational groups.
172
Claire Wallace and Florian Pichler
In some countries women are more satisfied than men – especially in Poland and Bulgaria – whilst in other countries men are more satisfied, especially in the Czech and Slovak Republics. Comparing these results to Western Europe, we find that gender differences in job satisfaction are larger in Eastern Europe, whilst in Western Europe no equivalent gaps can be found. Only in Finland and Germany, there are noticeable differences in job satisfaction, with women reporting higher levels of job satisfaction.
2. Conclusions Generally, Eastern and Western Europeans experience the same problems with working conditions with the important distinction that these problems are much more severe in Eastern Europe. The largest difference concerns wages and job security. Monthly salaries are extremely low in Eastern Europe, although Eastern Europeans work longer hours than Westerners. In Eastern Europe, people also consider their jobs less secure than in Western Europe. A further large difference is with respect to career prospects: Eastern Europeans seldom perceive any chances for future career advancement. The only aspect of working conditions where there were no significant differences between the East and West of the continent is the perception of tight deadlines: here the old and new member states were similar. There are, however, important variations within Eastern European countries. On many indicators, Slovenia had better working conditions than the poorer members of the EU such as Greece and Portugal, and on some indicators even above the EU average. We can also see that working conditions were generally better in the most reformed labour markets of the Czech Republic, Hungary and Estonia. Very poor working conditions were found especially in Bulgaria and Latvia on many indicators. These patterns indicate that working conditions are very diverse in Eastern Europe, yet, they share a common element: working conditions are on average worse than in the old member states of the European Union on all indicators. There were also important variations according to occupational status. On almost all indicators, the worst working conditions and lowest satisfaction was found among manual workers. The self-employed generally had the highest levels of satisfaction and best working conditions, followed by professional groups. Therefore there seem to be very fundamental divisions within the workforce; the deterioration in the situation of workers and their consequent discontent may reflect the introduction of more market-oriented societies where professional skills and entrepreneurship lead to greater rewards and those without such assets are left even further behind. Hence discontented doctors should no longer envy bus drivers – they should compare themselves instead to entrepreneurs. We see some sources of ‘path-dependent’ continuity in the nature of work in the eastern countries of Europe (Stark 1992). Namely, the low level of rewards, the low levels of promotion and autonomy and poor physical conditions were all characteristic of workplaces under the former regimes. Low intrinsic rewards are also found in jobs with lower professional levels. On the other hand, new aspects are the role of job insecurity, something that workers in the new member states feel acutely and high levels of work stress, which could also be seen as an outcome of market reforms aiming at higher worker productivity. The implicit employment contract based upon work-related benefits is gone but the new market conditions do not necessarily offer better prospects for most people.
Working conditions in Eastern and Western Europe
173
Whilst some countries (such as Poland) are associated with militant trade unionism, social dialogue is generally poorly developed and trade unions are not very good at defending or enhancing working conditions. Whether this improves with the integration of these countries into the European Union remains to be seen. Various ‘capacity-building’ exercises have aimed to do this. Many things have improved since the collapse of communism and the integration of Eastern European countries into the European Union has helped to improve living standards and enhance aspects of the quality of life. This report shows that in the field of satisfaction with working conditions, a lot still remains to be done to bring the new member states up to the level of Western Europe. The existing differences here diverged even further with the addition of Romania and Bulgaria in 2007.
Notes 1 We would like to thank the European Foundation for the Improvement of Living and Working Conditions in Dublin for the use of their survey, the EQLS 2003. The opinions expressed here are those of the authors and not the Foundation. 2 Based upon analysis carried out for the report The Social Situation in the European Union by Claire Wallace and Liliana Mateeva between 2002 and 2005, DG Employment and Social Affairs, European Commission.
References Agh, A. (1998) The Politics of Central Europe, London, Thousand Oaks, CA, New Delhi: Sage. Andor, L. and Summers, M. (1998) Market Failure: Eastern Europe’s Economic Miracle, London, Chicago, IL: Pluto Press. Cazes, S. and Nesporova, A. (2003) Labour Markets in Transition: Balancing Flexibility and Security in Central and Eastern Europe, Geneva: International Labour Office. Earle, J.S., Frydman, R., Rapaczynski, A. and Turkowitz, J. (1994) Small Privatisation: The Transformation of Retail Trade and Consumer Services in the Czech Republic, Hungary and Poland, Budapest: Central European University Press. EBRD (2003) Transition Report, London: European Bank for Reconstruction and Development. Haas, B., Steiber, N., Wallace, C. and Hartl, M. (2006) ‘Household employment patterns in an enlarged European Union’, Work, Employment and Society, 20, 4: 751–772. Pickles, J. (1998) ‘Restructuring state enterprises: industrial geography and Eastern European trasitions’, pp. 172–196, in A. Smith and J. Pickles (eds), Theorising Transition: The Political Economy of Post-Communist Transformation, London: Routledge. Pollert, A. (2003) ‘Women, work and equal opportunities in post-Communist transition’, Work, Employment and Society, 17, 2: 331–357. Pollert, A. (2005) Working Conditions and Gender in an Enlarged Europe, Luxembourg: European Foundation for the Improvement of Living and Working Conditions. Stark, D. (1992) ‘Path dependence and privatisation strategies in east central Europe’, East European Politics and Societies, 6, 1: 17–54. Vecernik, J. (1996) Markets and People: The Czech Reform Experience in a Comparative Perspective, Aldershot: Avebury. Wallace, C. (ed.) (2003a) Comparative Contextual Report, Vienna: Institute for Advanced Studies. Wallace, C. (ed.) (2003b) Country Contextual Reports, HWF Project, Vienna: Institute for Advanced Studies.
174
Claire Wallace and Florian Pichler
Wallace, C. and Stola, D. (2001) Patterns of Migration in Central Europe, Basingstoke: Palgrave. Wallace, C. and Haerpfer, C. (2002) ‘Patterns of participation in the informal economy in EastCentral Europe, 1991–1998’, pp. 28–48, in R. Neef and M. S. Stanulescu (eds), The Social Impact of Informal Economies in Eastern Europe, Aldershot: Gower. Wallace, C., Pichler, F. and Hayes, B. (2007) Monitoring Quality of Life in Europe, Paper 1: Quality of work and subjective life satisfaction in Europe, Dublin: European Foundation for the Improvement of Living and Working Conditions.
7
Extension through dilution? European integration, enlargement and labour institutions Jelle Visser
Introduction1 In May 2004 the European Union accomplished the largest enlargement of its history. Eight of the new members had made the transition to a democratic market economy following the collapse of communism. This transition and preparation for EU membership brought political, social and institutional changes, many of which affected the regulation of production and labour markets. Privatisation, lifting of price controls, opening markets to foreign investors, restructuring of production and soaring unemployment came with attempts to introduce or reform labour market institutions like trade unions, collective bargaining and works councils. This chapter is about that institutional extension of the EU to the new member states (NMS). In order to place this process in a proper perspective, it is necessary to include developments in labour market institutions in the EU-15 in this survey.2 It would be misleading to see these institutions as static or to simply project CEE as converging on the pattern set in the West. Labour market institutions in the EU-15 vary, reflecting different models of coordination and state–market interactions (Crouch 1993; Hall and Soskice 2001). Moreover, European labour markets and institutions have changed in past decades, not least due to product market integration and enlargement. Transition and enlargement extended European markets and created new opportunities for low-cost competition nearby. The adoption of the Single European Act helped to tear down barriers to the free flow of goods. This prepared the conditions for a Single European currency. The Treaty of Maastricht (1992), which set the conditions and timetable for EMU, expanded the powers of the European Parliament and added a separate Agreement on Social Policy. The Amsterdam Treaty (1998) expanded EU policy in matters of economic coordination and employment. The proposed Constitutional Treaty was botched in popular referendums in France and the Netherlands in May 2005, not least because of anxieties over liberalisation of services and low-cost competition from the NMS. The conflict over the draft Services’ Directive of January 2004 illustrates how European integration, enlargement and labour regulation interrelate. The original draft proposed free movement of services by applying the country-of-origin principle,
176
Jelle Visser
already used in EJC case law. Its opponents fear that the possibility of protecting social standards by extending the reach of national regulations to foreign service providers and workers posted by foreign companies, allowed under the Posted Workers Directive (96/71/European Commission), is weakened by the removal of administrative controls, brandished as a pretext for protectionism by the directive’s framers. Lack of regulation of the European market for temporary employment agencies3 and ambiguity over the legal definition of a ‘dependent employee’ opens the door to evading labour standards by using quasi self-employed contractors operating out of state. In response to pressure from unions and member states like France and Germany, the EP voted in February 2006 in favour of a diluted version of the directive, removing the country-of-origin principle and exempting social and health services as well as temporary agencies. Despite bitter criticisms from the NMS, the Council and the Commission have resigned their ambitions to the political realism of the parliament. Surely a disappointment for those who regret that the diluted version, unlike the original draft, creates ‘unfortunately no legal possibility of east European or any other service provider of floating, say, France’s 35-hour working week’ (Financial Times 2006). A similar conflict smoulders over the free movement of workers, another basic freedom under the Treaty of the European Union. Notwithstanding their solemn and unanimous declaration, on the occasion of signing the Accession Treaty in 2003, that the acquis would be honoured on this point, 12 of the EU-15 decided to impose restrictions. Claiming that labour migration has been unproblematic and helped to alleviate skill bottlenecks in the three member states without restrictions (Ireland, the UK and Sweden), the Commission advocates the end of these transitional measures. Arguing that restrictions drive migrants into the ‘underground economy’, the European Trade Union Confederation favours free labour migration in exchange for better controls on national labour standards. Free movement of services and workers raises fundamental questions regarding social cohesion, convergence, labour standards and institutions. How much convergence and cohesion is there? How much difference in standards can Europe sustain? Does Europe still support a dynamic of upward harmonisation? How realistic is it in view of the massive differences caused by this and further enlargements? The chapter proceeds in four steps. Highlighting some key changes in labour markets and employment regulation sets the scene. Institutional changes in collective bargaining and wage setting are analysed next. Then follows a look at the macroworld of social pacts, tripartism and European social dialogue. Adding the perspective from the micro-world of firms and employee representation concludes the chapter.
1. Changing labour markets and employment contracts At the time of accession, the NMS added one-fifth to the population, one-sixth to employment, one-third to unemployment and just over one-twentieth of total production of the EU-15.4 Differences in per capita GDP are massive, also within the NMS, and productivity per hour worked is still much lower, reflecting various factors, including the structure of capital, investment, skills, management, working practices and infrastructure. Average annual earnings are two to four times lower compared with the EU-15 (European Commission 2005: 167). However, recent
Extension through dilution
177
growth trends in hourly productivity, GDP per hour worked and real wages are much stronger in the NMS than in the EU-15. These positive trends must be set against problematic developments. Employment rates in the NMS tend to be lower and declining in recent times, while unemployment rates have risen (Table 7.1). Restructuring from industry and agriculture to services is occurring in old and new member states, but with a smaller and less growing service sector this is associated with falling employment levels and rising unemployment in the NMS. As a consequence of various emergency measures with which politicians have addressed rising unemployment, different contractual arrangements at variance with standard employment contracts have proliferated in the EU, fuelling fears of a dual labour market (Supiot 2001). The incidence of non-standard employment contracts has risen to almost 12 per cent and a growing number of people are employed on ‘freelance’ contracts or work as ‘quasi-independent’ contractors. Although this form of employment hardly existed in CEE, fixed-term work has witnessed spectacular growth, especially in Poland, the Czech Republic, Latvia and Lithuania. The shift to a free market economy has brought radical changes in employment contracting and remuneration. While previously employment contracts that were not permitted by law were deemed prohibited, new legislation tends to espouse the opposite philosophy that anything not prohibited is permitted. The most common form of flexibility is self-employment. Still strongly related to agriculture, one-quarter of the self-employed in the EU-15 and one-half in the NMS work outside farming or fishing, and their numbers are rising. Many are professionals or skilled workers who work very long hours. In Poland and Table 7.1 Key changes in employment, unemployment and contracts, 1998–2004 (%) EU-15
NMS
1998
2004
+/−
1998
2004
+/−
160.8 61.4 71.2 51.6
200.5 64.7 72.7 56.8
24.7 3.3 1.5 5.2
29.4 60.8 68.2 53.0
28.4 56.0 62.0 50.2
−3.5 −3.2 −5.8 −2.8
Total unemployment1 15.9 Unemployment rate 9.3 Male 7.8 Female 11.2
14.7 8.1 7.1 9.3
−7.7 −1.2 −0.7 −1.9
3.3 9.3 8.6 10.9
4.7 14.0 13.4 15.0
44.8 4.7 4.8 4.1
Share in % Agriculture Industry Services
4.5 26.5 68.9
3.9 24.3 71.9
−7.2 −1.9 11.7
12.9 34.3 52.7
12.4 30.7 56.4
−7.4 −13.5 3.3
Self-employed Part-time Non-standard
15.5 17.3 13.0
14.9 19.4 13.6
2.9 20.0 12.0
22.6 8.2 5.2
22.0 7.4 14.3
−6.2 −13.2 264.2
Total employment1 Employment ratio2 Male Female
Sources: Own calculations from Eurostat, European Labour Force Survey; see also: European Commission, Employment in Europe 2005, European Commission, DG Employment and Social Affairs, annex Notes: 1 in millions. 2 employment–population ratio 15–64 years.
178
Jelle Visser
Hungary skilled workers with a standard employment contract often hold a second contract as ‘jobber’ under civil law with the same employer. This is a functional equivalent to flexibility based on variable overtime hours, but outside the framework of labour and social security regulation. All CEE countries have longer average working weeks than in the EU-15 with the exception of Greece, Portugal and the UK. Reflecting lower hourly productivity and wages, long hours help firms and families to make ends meet. Thanks to stronger increases in productivity and earnings, average working hours in CEE tend to be falling, whereas in Western Europe the long trend towards shorter working hours was recently halted or reversed, as currently in Germany (EIRO 2004). Part-time work is not very diffused and tends to be involuntary in the NMS, reflecting the lack of alternative full-time employment. According to data from the ILO (European Commission 2004: ch. 4), in Hungary nearly 20 per cent and in Poland almost 40 per cent of part-time workers are on some sort of disablement or retirement pension. Lack of development of part-time work is also traced back to low wages and the absence of regulations granting part-time workers similar pro rata social security and employment rights as full-time workers. The low incidence of part-time employment in Southern Europe has a similar background, though another factor is that the informal economy may offer better opportunities nearer home.
2. Change or decline in collective bargaining 2.1 The Western European mainstream In post-1945 Western Europe collective bargaining over the terms and conditions of employment and based on voluntary negotiations between unions and employers became the common rule. After the upsurge in worker militancy in the late 1960s and the strengthening of worker rights, employers in France and Italy resigned to the common European pattern, thus guaranteeing conditions and rules above the level guaranteed by law. With the end of authoritarian regimes in Spain, Portugal and Greece, unions regained their autonomy and labour–management relations converged towards the mainstream. Sectors or branches of economic activity became the main organising units for collective bargaining, since they were most homogenous in technology, competition, legislation and occupational structure. Eager to rule out competition on wages between them, especially when labour markets were tightening, employers tended to favour centralisation. Unions favoured this as well, since sector bargaining incorporates a solidaristic element which is absent in company bargaining. By setting standards for the entire industry and orienting these towards the performance of average or aboveaverage performing firms, collective agreements create incentives for laggards to catch up or leave the industry. This can work as a modernising or upward harmonisation process with benefits for remaining firms and workers, and is socially and economically efficient if capital is freed to move elsewhere and workers are retrained for other activities and redeployed without long delays (Streeck 1992). 2.2 Collective bargaining in Central and Eastern Europe Unable to build strong and autonomous employers organisations, with weak and divided unions, and an economy fragmented between struggling state enterprises,
Extension through dilution
179
foreign-owned firms and a myriad world of SMEs, the Western European model of collective bargaining never stood a chance in CEE. Slovenia, with its obligatory employers’ associations based on Chambers, is the exception confirming the rule that employer organisation is the key issue.5 Fifteen years into transition towards a market economy, it is clear that wages in the private sector are decided at company level, with a large role for minimum wages determined by governments. Sectoral bargaining is mostly restricted to the state sector. In most small and many large foreign-owned companies there is no collective bargaining or union representation. With the exception of Slovenia, bargaining practices are unstable, with large parts of the economy beyond the reach of collective bargaining and limited effectiveness of the agreements that exist (Table 7.2). In Poland, the Baltics and Hungary, collective bargaining activity has in fact decreased. Most agreements in Poland simply recite what is already in the law and there are examples, for instance in Estonia and Lithuania, that agreements set conditions below those provided for by law, demonstrating the negotiators’ lack of knowledge or inability to update agreements on time. The Labour Inspectors’ report for 2004 finds that a quarter of all collective agreements in Poland contain provisions in contravention with the law. Under-reporting of wages or providing workers with cash payments is widespread as a way to escape taxes and earnings-related social contributions (Renooy et al. 2004). A survey of Hungarian employers in 2003 reveals that 46 per cent see this as the most commonly used method to save on labour costs. In Poland this practice is strongly related to employment in micro-firms and in 2003 ‘cash-in-hand’ wages affected nearly 20 per cent of employees in firms with fewer than five employees (European Commission 2004). We have to bear in mind that in CEE a far larger share of industrial and service employment is located in these very small firms compared to Western Europe, though the situation in Greece and Italy offers parallels. 2.3 Flexibility and the customisation of agreements Changes in the Western European model of collective bargaining aree documented in various comparative studies (Calmfors et al. 2001; European Commission 2004; Katz and Darbishire 2000; Marginson and Sisson 2004; Traxler, Blaschke and Kittel 2001). This is, however, hardly reflected in the extent of bargaining coverage, which has remained high in spite of union decline (Visser 2006). On average, the outcome of collective bargaining affects nearly 70 per cent of EU-15 employees compared to an average union density rate of 26 per cent. In the NMS and disregarding the special case of Slovenia, collective bargaining reaches only a minority of the employees, slightly above the average level of union density (20–25 per cent). Figure 7.1 shows that across countries bargaining coverage increases with union density, though the association (R2 = .480) is mainly driven by, in one corner, countries with high union membership and bargaining coverage (Scandinavia) and, in the other, those with low membership and coverage (CEE, UK).6 Figure 7.1 shows that in many countries (France, Spain, the Netherlands, Austria) coverage exceeds density by a large measure. Association of employers for bargaining purposes and extension of agreements to non-organised firms play a key role (Table 7.2). Mandatory extension is possible in many countries, including CEE, but it is not applicable if unions and employers fail to reach a sectoral agreement for even
180
Jelle Visser
Table 7.2 Collective bargaining (private sector) and wage setting, recent years Coverage National Finland Sweden Denmark
83 92 82
UK Ireland
35 —
Germany Austria Belgium Netherlands
63 98 93 82
France Italy Spain Portugal
95 80 81 87
Slovenia 100 Hungary 30 Slovakia 48 Czech Republic 37 Poland 34 Estonia 29 Latvia 19 Lithuania 13
1 3
Sector
Company
Pacts
2 1 1
3 3 2
yes tacit tacit
yes no no
agreement agreement agreement
1 2
no yes
no n.a.
law law
1 1 2 1
3 3 2 3
failed tacit partly yes
yes (yes) yes yes
agreement agreement law law
2 1 2 2
2 2 2 3
no partly partly yes
yes (yes) yes yes
law courts law law
2 3 2 3 3 3 3 3
3 1 2 1 1 1 1 1
yes partly partly no no no partly partly
(yes) n.a. n.a. n.a n.a. n.a. n.a. n.a.
law law law law law law law law
1
1 3 3 3 2 1 3 3
Extension Min. wage
Sources: Adapted (and updated) from European Industrial Relations 2004; updated EIRO publications Notes: Bargaining coverage is defined as the share of employees working in firm in which wages and terms of employment are determined by collective agreements negotiated by trade unions or works councils. Bargaining level or unit at which bargaining is conducted and agreements are signed: 1 indicates that this is the principal or dominant bargaining level; 2 important but not dominant level; 3 existing level of bargaining. Pact: data refer to recent years (1999–); ‘partly’ indicates that there is an agreement with some unions for some years and issues, but no general pact; ‘tacit’ indicates that there is informal agreement about wage setting arrangements (conflict resolution) and/or a wage norm. Extension: legal (public law) provision to bind non-organized firms to conditions set by voluntary agreement. The presence of a functional equivalent for legal extension is indicated between brackets: in the case of Austria and Slovenia through obligatory membership of employers in Chambers to which the right to sign agreement is delegated. n.a. = not applicable, i.e. the legal possibility to extend collective agreements exists, but the absence of sector agreements makes the law non-applicable. Minimum wage setting: can be based on mandatory public law, court rules, or voluntary agreement.
a minority of firms. When the law does not provide for extension by public law (Scandinavia, UK), inclusion of non-organised firms depends on union pressure inviting employers to ‘follow the contract’ or face industrial action. In Scandinavia this pressure is still formidable. Bargaining coverage in the UK halved in the course of two decades, from 70 to 35 per cent. The proximate cause was the withdrawal of employers from sectoral bargaining. Given the ‘low degree of control and an unrealistically narrow scope’ (Brown 1993: 190) of sector agreements with undisciplined trade unions, this was a rather
Extension through dilution
181
90 80
SE DK
FI
70 60 BE
NO
50 Rsq.=.480
40 SK
30
LV
UK
EL
CZ
20
HU EE
LT
AT
IT
LU
PT
DE
NL
ES
PO
10
FR
0 0
10
20
30
40
50
60
70
80
90
100
Figure 7.1 Bargaining coverage and union density Source: As in Tables 7.1 and 7.2: European Industrial Relations 2004
costless option for them. The German story is different. Here, sector bargaining is highly institutionalised, addressing wage and non-wage issues, based on shared rules and supplemented by congenial institutions like the works council in the company (Streeck 1992). However, when bargaining units are very large7 and include large and small firms with sharply different abilities to pay, the pressure to break up agreements rises. If decentralisation or differentiation is blocked and it is impossible to make bargaining units smaller for a larger range of issues, employers with the lowest capacity to sustain the costs of sectoral agreements will withdraw and coverage decreases. This is what happened in the East German Länder after unification when labour market institutions of the West were extended wholesale. Smaller firms in western Germany followed their example. Bargaining coverage in the private sector was stable until the late 1980s but has decreased by about 10 to 15 percentage points since (Hassel 2006; Ellguth and Kohaut 2005). Customisation of agreements, offering more choice to individual companies and workers, may be a solution (Streeck and Rehder 2003). With the diffusion of ‘opening’ and ‘exemption’ clauses firms gain the formal right of what they were already doing, namely to suspend particular aspects of a legally binding sector agreement negotiated by their representatives. Plant-level agreements specifying additional concessions in order to maintain employment and investment exist in one-third of the firms under the collective agreement, while another 15 per cent violate the sectoral agreement (Hassel 2006). Concession bargaining has also spread to other European countries (Visser 2005). For instance, in Italy ‘area agreements’ for particular regions with high unemployment swap lower wages for additional public investment. ‘Inability-to-pay-clauses’ exist in Spanish and Irish agreements. In France, sectoral agreements have almost fully lost their importance for wage setting. Two-thirds of the collective agreements sampled by the Labour Ministry in July 2004 contained wage rates below the statutory minimum norms (Schmid and Schulten 2006: 118).
182
Jelle Visser
Recent legal reforms in France introduced the possibility of derogating from the legal statute by collective agreement, but the legal minimum wage was explicitly excluded. Consequently, in many collective agreements wage clauses have no legal basis. In Northern Europe, sectoral agreements have kept their importance for job classification and for assigning minimum pay rates to skill and seniority levels, but actual pay is increasingly decided in the firm. The ‘most favourable’ rule, a cornerstone of labour law and guaranteeing that deviations from law and higher-level agreements can only happen in a direction favourable to employees, is clearly under attack (Sciarra 2004). The development towards allowing firms to exit from common standards if they cannot sustain them, is an attempt to combine the protection and predictability of common minimum standards with flexible adjustment and tighter management control over actual costs. This re-interpretation of the postwar model is a response to increased international competition and organisational diversity. However, even if retaining an element of solidarity across firms, this development is problematic for trade unions, especially when representation in the workplace is limited or beyond their control, as is surely but not only the case in CEE. 2.4 Rising pay inequality, in West and East Fixing the standard pay rate for jobs across firms in the industry was the ‘common rule’ of union wage policy of the past (Flanders 1970; Slichter et al. 1960). Objective pay criteria, based on job descriptions and seniority, diminish the power of supervisors and risks of discrimination. Trade unions advocate standardised employment contracts in order to protect workers against uncertainty, de-couple the economic situation of workers from that of their employing organisation, and suspend competition between workers as far as possible, in order to enable them to act in solidarity (Streeck 2005). Technological changes and intensive global competition change the organisation and remuneration of work. High-cost producers tend to reorganise work around decentralised management of human resources, customised products and flexible working schedules, and reorganise tasks to make it easier to subcontract some tasks, employ part-time workers and hire temporary staff, while core work is multi-tasked and carried out in teams. Employment security and remuneration are defined less in terms of the seniority and job status of employees than in terms of the competences they bring to the job or acquire while working. Breaking away from centralised agreements gives firms more scope for merit- and performance-based pay (Lindbeck and Snower 2001). This is usually associated with an increase in earnings dispersion and more discretion for management to set individual pay by open-ended appraisal procedures (Brown et al. 1998). As was mentioned before, sector bargaining and, a fortiori, economy-wide bargaining incorporates a solidaristic element. In exchange for wage discipline, Swedish unions conducted between the 1950s and 1980s a wage solidarity policy based on national agreements favouring workers with low bargaining power. Hibbs and Locking (1996) and Iversen (1999) show that pay differentials in Sweden started to rise following the move from nationwide to industry bargaining after 1983. There is empirical evidence and theoretical proof that associates centralisation of wage bargaining with pay compression (Agell 1999; Blau and Kahn 1999; Iversen 1999; OECD 2004; Wallerstein 1999). Wage dispersion has increased most where unions
Extension through dilution 5.00 LV
EE PO
LT
4.50 4.00 D1/D9 ratio
183
Rsq.=-.734
3.50
HU
LU SK
UK
FR
PT ES
DE
AT
3.00 CZ
NL
EL
BE
2.50
80 DK NO
2.00
SE
FI
1.50 0
10
20
30
40 50 60 % employees covered
70
80
90
100
Figure 7.2 Earnings inequality and bargaining coverage Source: As in Tables 7.1 and 7.2; for earnings inequality, data from Eurostat (earnings survey) as provided in Employment in Europe 2004
have declined most and collective bargaining is most reduced or decentralised (e.g. the US, the UK, Canada, New Zealand). Figure 7.2 shows the positive association between bargaining coverage and equality, here measured as the ratio of first to ninth deciles in the earnings distribution (before taxation, but including social charges).8 We find the Nordic countries with relatively low (but rising) earnings inequality at one end and the NMS (and the UK) with high (and also rising) inequalities at the other. If earnings differentials are regressed on an index of centralisation of pay bargaining, as proposed by Iversen (1999) and Visser (1990), we find a similar albeit weaker association (R2 = −0.410). A comparison through time suggests that the association between centralisation and earnings equality has decreased (OECD 2004). This might be explained by the rise of ‘atypical’ employment contracts beyond the reach of collective bargaining and the diffusion of ‘exemption’ clauses. Incentives for upward harmonisation of wages will be weakened if inefficient and unprofitable employers can choose not to apply agreements or side with works councils and workers who bid for lower standards in an attempt to save their jobs. Massive unemployment and the increased capacity of firms to move investment and production abroad undermine the upward harmonisation dynamic based on the virtuous circle of labour seeking a bigger share in productivity increases and capital promoting stronger productivity growth. 2.5 Minimum wage regulation Minimum wages can be powerful instruments to diminish inequalities at the bottom of the wage ladder (European Commission 2005; Dolado et al. 1996). Since the adoption of a statutory minimum wage in the UK and Ireland, 18 of 25 EU members
184
Jelle Visser
and all of the CEE countries have a legally defined minimum wage (European Commission 2004). In Sweden, Finland, Denmark, Germany, Austria and Italy minimum wages may vary across sectors and implementation depends on the coverage of collective bargaining with various mechanism providing for the inclusion of non-members: judges’ law in Italy; compulsory membership in chambers in Austria; pressure on unorganised employers in Sweden and Denmark, legal extension to nonorganised firms in Finland. In Germany, minimum wages are determined in sector negotiations that cover only 50 per cent or less in low-paid industries (personal services, retail, textile and clothing), below the threshold needed for extension (Funk and Lesch 2005). Unlike in the US, where a national minimum wage has existed since 1937, there is no provision for European-wide wage regulation. Implementation of EU Directive (96/71/European Commission) on the posting of workers, which allows member states to impose national standards on firms and workers operating in their territory if they apply the same standards to domestic firms and workers, has spawned some minimum wage regulation in member states, among others in the German construction industry. In Sweden, a recent conflict over building workers posted by a low-cost competitor operating from Latvia, has indicated the potential weakness of a model that cannot guarantee full application of negotiated standards. Although the Swedish labour court decided in an interim ruling that a union boycott against foreign firms offering wages below the sectoral standard is lawful, it is uncertain whether the European Court of Justice will take the same view. If not, Sweden will have to revise its labour legislation away from voluntary principles. Needless to say, the issue divides not only unions and employers, but also Latvian and Swedish unions and the two governments (Woolfson and Sommers 2006). The level of statutory minimum wages, measured in purchasing power parities, varied in 2003 from E1,225 in the Netherlands, via E983 in the UK and E617 in Spain, to E351 in Poland and E239 in Latvia (Table 7.3) (European Commission 2003: 80).9 In line with increases in average real wages, statutory minimum wages have in recent years increased strongly in the CEE than in EU-15 countries, though the decline in the value of the minimum wage in Poland stands in radical contrast to convergence. The relative importance of minimum wages, measured by the so-called Kaitz index, shows the minimum as a percentage of average wages. With the exception of Slovenia and Slovakia, minimum wages in CEE are relatively low, similar to Spain, the only Southern member state in this sample. In recent years, the relative value of the minimum wage declined in many member states, with minimum wage adjustments lagging average wage growth (Table 7.3).10 This may have its reasons in high unemployment rates among unskilled workers, but tends to increase the incidence of the working poor. In Latvia, for instance, unions claim that the minimum wage has been below subsistence level since the 1990s. The incidence of the statutory minimum, the proportion of workers actually earning the minimum wage, varies across countries and is higher when collective bargaining is poorly developed.
3. The macro-world of social pacts, dialogue and tripartism The start of EMU in 1999 was a key change in European economic policy, provoking a realignment of wage policies towards the non-inflationary Deutschmark regime. The institutional means varied: pattern bargaining in Austria and Germany with
Extension through dilution
185
Table 7.3 Statutory minimum wages in Europe Country
Level in PPP (Euro)
Increase
Share in % of average wage
% adult workers affected
2003
2001–4
2001
2004
910 1.162 1.150 1.225
117.3 106.1 114.1 107.2
51 49 47 46
51 46 47 45
— −3 — −1
4.5 — 13.0 2.1
Medium (40–45%) Slovenia Slovakia UK
668 265 983
134.0 146.5 118.3
41 40 37
44 41 40
3 1 3
2.7 2 to 4 5.0
Low (<40%) Latvia Lithuania Czech Rep. Hungary Poland Estonia Spain
239 254 389 384 351 264 617
133.3 116.0 134.0 132.5 89.3 155.9 113.4
38 44 34 39 37 29 35
38 38 37 36 36 34 33
— −6 3 −3 −1 5 −2
15.7 18.4 2 to 3 — 4.2 7.0. 1 to 3
High (≥45%) Ireland Belgium France Netherlands
2003
Sources: European Commission (2004); Funk and Lesch (2005); Schulten et al. (2006)
export industries taking the lead; national framework agreements to guide increasingly decentralised bargaining in the Netherlands; government imposition of an international wage norm in Belgium; state-imposed minimum wage norms and freedom for company wage bargaining in France; modernisation of the rules with safety valves at company level in Spain, national reform pacts and wage agreements in Finland, Ireland, Italy and Portugal. 3.1 Social pacts Many EU-15 countries retained or re-introduced a modicum of national coordination within a more diverse and decentralised economy, by tacit agreement or by means of a negotiated pact with unions and employers. In CEE the history of social pacts (Poland, 1993; Lithuania, 1995, again in 2005; Latvia, 1996; Hungary, 1994, again in 2005) is patchy and it has been extremely difficult to repeat and institutionalise the early pacts. Pacts that were reached were often shallow, limited to a narrow set of issues and poorly connected to lower levels (see Avdagic et al. 2005). Slovenia is the only NMS which can claim an institutionalised pattern of broad social pacts since 1994, somewhat similar to Ireland or Finland (the history of pacts in Italy or Portugal is more uneven). In Table 7.2 we saw that the setting of wages and employment terms involves bargaining activities at different levels. Nowadays, sectoral bargaining is nearly always supplemented with bargaining or adjustments at enterprise level; in nearly half of all member states also with some form of national bargaining or coordination. After the eclipse of sectoral bargaining, the UK is currently the only EU-15 country where all
186
Jelle Visser
private sector wage bargaining takes place at company level, presenting an industrial relations environment resembling the US. Sectoral bargaining is also non-existent or fragile in Ireland, France and the CEE, except Slovenia and Slovakia. Ireland is here the only country in which cross-industry agreements put a cap on wage increases at company level. In France and most CEE countries, the state tries to influence wage setting by means of minimum and public sector wages. 3.2 Tripartism Social pacts may result in national councils for consultation and concertation between unions, employers and the government. However, the existence of these tripartite bodies need not result in meaningful pacts or agreements. In many countries tripartite bodies are talking shops and places to exchange information, possibly not even the most important ones. Union representatives in CEE often complain that they received more information from Brussels and European sister organisations than from their own governments (European Commission 2004; Lado and Vaughan-Whitehead 2003). All CEE countries have founded tripartite councils, as this seemed a good thing to do. But many councils wield little influence and have been ignored by unfriendly governments. Tripartism is also common in the EU-15, albeit only in narrow policy domains in the UK (minimum wage setting, vocational training), Germany (public employment office) and Sweden (labour courts). Even where tripartite councils do have a broad remit, their real function and influence varies massively. An estimated 80 to 85 per cent of all recent legislation in social and economic matters in the Netherlands is prepared in the Social and Economic Council. In France, where it has a similar position on paper, the council is typically consulted after the cabinet takes its decisions and in the 2006 conflict over the lifting of dismissal protection in firstjob employment contracts for youth under 26 year of age the council was completely invisible. Tripartism was expected to ease the path to reform and preserve social peace in transition economies. Endorsed by international organisations like the EU and ILO, it was presented as an important tool to help them ‘join Europe’ (Pollert 1999). Preparing for accession, the EU devoted considerable resources to capacity building through its PHARE programme. However, the weakness of union and employer organisation, the poor state of their mutual relationships and dependence upon politics and governments, are a rather common predicament of industrial relations and tripartism in CEE. Some scholars argue that tripartism has nonetheless facilitated capacity building among unions and employers and generated support for reforms (Iankova 2002, Iankova and Turner 2004). Others maintain that these exercises in corporatism should be categorised as ‘illusionary’ rather than ‘transformative’ (Ost 2000). There is some consensus that weak governments have often used tripartite talks and agreements instrumentally as a way to appease the unions, usually after they have taken to the streets (Avdagic 2005; Kohl and Platzner 2003). 3.3 European social dialogue and cross-national coordination Social pacts and tripartism are attempts at national coordination and policy-making. Within the EU there is also an elaborate structure for international coordination
Extension through dilution
187
and policy-making. Since 1985 the Commission has promoted social dialogue with unions and employers. Its anniversary, at a recent special summit, praised the social dialogue as ‘an essential tool for the future’ (EIRO 2005). Since 1997, the European Council meets with officials of the European ‘social partners’ on the eve of its halfyearly summit meetings. Since 2003, under the responsibility of the Council Presidency there are special social summits dealing with issues of macro-economics, employment, social protection, and education and training. Since 1999 a so-called Macro-Economic Dialogue involves meetings of European Union and employers with representatives of the European Central Bank, the Commission and the Presidency. In a number of policy domains, like social security of migrant workers, the European Social Fund, vocational training, equal opportunity policies, and health and safety, there are tripartite advisory committees. This complex structure has no parallel in other countries or regions of the world. In fact, it is more elaborate than what exists in most member states, of which the larger ones, i.e. the UK, Germany, France, Spain and Poland, hardly engage in structured or formalised tripartite consultation. It is possible and even likely that the excess of formalism and complexity at the European level hides very thin structures of social relationships and trust between unions and employers, and the lack of steady informal practices. In CEE but also in France and the UK, bipartite relationships between employers and unions are absent, unsteady or restricted to isolated occasions (European Commission 2004; Lado and Vaughan-Whitehead 2003). Despite recent attempts to promote joint activities autonomously conducted by social partners, bipartism at the European level has remained frail (European Commission 2002). Employers have no incentive to do serious business with the unions unless threatened by political initiatives from Council, Commission or Parliament. The Commission promotes a shift from intersectoral to sectoral negotiations, hoping that greater homogeneity of problems and interests propels the social partners into a more active role. But in many countries, in particular CEE but also Southern Europe, sectoral organisation of interests and bargaining has hardly developed and unions and employers may depend more upon political supports than upon their members or each other. As a consequence, the sectoral level may be ‘the weakest link’ compared to what happens in politics or firms (Ghellab and VaughanWhitehead 2003). In the sectoral social dialogue committees the possibility of joint action of unions and employers is mostly determined by the increased likelihood of gaining financial support from the Commission or exemption from regulations after privatisation (Benedictus et al. 2002). It is certainly premature to see these committees as constitutive for European collective bargaining, as is suggested by Marginson (2005). Pay, the right of association, to strike or impose lockouts are explicitly and completely excluded from the competence of European-level institutions. In key sectors of the European economy there are no transnational European employer associations, and in major EU states (CEE, the UK, Ireland), there are no sectoral employers’ associations or none with a mandate to negotiate with trade unions. 3.4 EMU and wage coordination When product markets become more international through completion of the internal market, enlargement or globalisation, the effectiveness of national regulations decreases and unions must seek to extend the reach of domestic labour standards
188
Jelle Visser
beyond existing borders. Because standard setting and upward harmonisation is sometimes also in the interest of leading producers, countries or the public interest, the regulatory response of unions has frequently provided a common agenda with domestic employers, national politicians and ‘progressive’ international forces like the US after 1945, the ILO and the EU. However, with vastly greater competitive pressures and with investment and ownership less tied to the country-of-origin, these coalitions are harder to form and easier thrown in a minority. Through its ‘one-size-fits-all’ monetary policy, the disallowance of competitive currency adjustments and restraint on fiscal policies, EMU has added to these competitive pressures. European trade unions have tried to respond by stepping up attempts at transnational coordination. German unions and those in countries and regions bordering Germany have been particularly active, anxious to prevent ‘beggar thy neighbour’ competition for investment and jobs. Prompted by IG Metall, at the time the largest and strongest private sector union in Europe, the European Metalworkers’ Federation (EMF) began to create instruments for a transnational wage policy based on the adoption of a common wage and working time norm, a system of information exchange and mutual surveillance of wage bargaining across borders (Schulten 2002). The results have been modest. In 2001, the EMF concluded that its coordination rule had little effect on national bargaining (Marginson 2005, Schroeder and Weinert 2004). Other unions tried to follow the EMF example, but their capacities are even more limited. Expectations of a jointly governed wage norm in the old Deutschmark zone, uniting trade unions in Belgium, Luxembourg, Germany, the Netherlands and later France, have been toned down in recent years. It is very difficult to implement international coordination on a voluntary basis. Employers keep their distance and EU law offers no support. It must be doubted whether interests are sufficiently aligned to make a voluntary coordination rule stick. Unions face a clear prisoner’s dilemma. If they knew how to generate a higher growth path and redistribute jobs to places or countries most in need of them, and if it were possible to punish those unions breaching the cartel, a common wage norm might work and be Pareto-optimal. The history of national incomes policies shows that such solutions required a high degree of centralisation and sanctioning power by law or union statute, a fair degree of cooperation from employers and political instruments like popular referendums or membership ballots in order to outwit militant insider groups (Baccaro 2003; Lange 1984). These conditions are nowhere near met in Europe. 3.5 The turn to voluntary policy and implementation failure Since Maastricht (1992) the Treaty gives the European social partners a special role as potential co-legislators in the social policy domain (Articles 135–139). Through this channel, they negotiated agreements on parental leave (1995), part-time work (1997) and fixed-term employment (1999), preparing the basis for EU directives. More recently, they negotiated agreements on telework (2002) and work-related stress (2004), which will be implemented ‘in accordance with the procedures and practices specific to management and labour and the member states’, using Article 139.2 of the Treaty. Given hugely different bargaining coverage rates and the weakness of legal monitoring in many member states, especially in CEE and Southern Europe, this implies unequal implementation. For this reason, legal experts have
Extension through dilution
189
expressed doubts concerning the applicability of article 139.2. They argued that agreements between the social partners, being a substitute for legislation, ought to have an erga omnes effect, which requires a state guarantee (Barnard 2000: 92; Bercusson 1992: 181; Blanpain 2002: 102). Member states may want to amend legislation to help implement voluntary agreements, as is routinely done in, for instance, Belgium. But there is nothing in European law that requires them to do so. This is different in the case of directives, for the implementation of which member states are held responsible. The view that European social policy is about establishing uniformity in labour standards or adjustment to a common model of regulation is false. There is however, potentially, a dynamic rather than static harmonisation view, which projects social policy as a dynamic process in which enforceable international labour standards interact with economic modernisation to produce an upward movement in labour conditions and social provision. This does not imply uniformity or equal speed, but what it does require is that member states, and firms within them, are prevented from regressing to lower levels of provision resulting from competitive underbidding (Deakin 1997). In order to gain support in the Council for its social policy agenda, with proposals stalled since the early 1980s, the Commission diluted its original proposals and used new legal techniques and methods to gain support. Streeck (1995) has described the turn in post-Maastricht social policy as ‘neo-voluntarist’ and Kilpatrick (2003: 137) characterised the change in EU legal technique as ‘using types of legal instruction other than legally binding commands’. Apart from overcoming the obstacles of diversity and stalwart member states, the new approach was also justified in terms of legal efficiency, based on the reasoning that only voluntarily adopted labour standards will bear no costs while those forced upon employers will be resisted and have an employment cost (Collins 2002). This recalls the old union joke that if they had always heeded the advice of stalwart employers, there would never have been a wage rise in history. The unequal application and partial unenforceability of minimum labour regulation across the territory of the Union is problematic from the perspective of traditional labour law and if one is concerned with equality. It is, however, fully consistent with the voluntarist turn in collective bargaining and European social policy, allowing firms and states to exit from common standards if they cannot sustain them. The idea that coercive standards and inalienable rights, which prohibit long working hours or sub-subsistence wages, force employers to make beneficial investment choices which they would not make of their own accord, is no longer popular among policy makers and increasingly out of reach in EU policy. This is certainly not a problem limited to CEE or created by enlargement, but it has also not become easier to solve.
4. Firms, partnership and employee representation There are several ways to gauge the position of trade unions in labour markets and society: the proportion of employees who are members (union density); the proportion of employees who work in enterprises where unions or union-like institutions are present (workplace representation); the proportion of employees whose terms of employment are affected by collective agreements (bargaining coverage); and their
190
Jelle Visser
recognition and presence in bi- and tripartite institutions (union representation). I have been concerned with coverage and macro-level representation, its variation across the EU and how it is affected by enlargement. In this final section I add the micro-dimension of firms and workplace representation. 4.1 Benefits of workplace representation The implementation of many employee rights, especially where they involve choice, depends on awareness and, if need be, support. Unions and other intermediaries at the place of work can play an important role. Based on evidence for the UK, Brown et al. (1998: 627) report ‘an independent and significant association’ between union density and the provision of written details of rights. Not only do trade unions raise the awareness of choice and knowledge of laws and contracts, employer compliance is also improved. ‘Collective procedures are the custodians of individual rights’, they conclude. Their research design, based on a sophisticated comparison of union and non-union firms in the UK, is not easily replicated on the European mainland. The division between firms with and without unions is less stark when there are many firms with only a handful of union members and the effects of union activity accrue to non-members as well. From the British evidence, however, we may conclude that policies that promote workers’ rights (equal opportunity, choice of flexible or part-time working hours, leave of absence to take care of parental duties, etc.) on an individual basis, but at the same curtail workers’ rights of collective representation, are contradictory. While formally workers are given more rights of choice, in reality they are given less support and confidence to use such rights. This is why legislators promoting health and safety standards or equal opportunity policies usually favour public guarantees for collective, equally accessible, employee representation in firms. The directive on information and consultation of employees (2002/14/European Commission), which was accepted against the opposition of employers and member states like the UK and Germany, considers that workplace representation can create legitimacy for company restructuring and improve access to training in the flexible enterprise. It is a long held view that implementation of health and safety regulation cannot be based on state inspection alone but also depends on self-policing. 4.2 Extent of workplace representation My estimate is that between 50 and 60 per cent of all employees in firms with 50 or more employees in the EU-15 work in enterprises with union representation. A recent Europe-wide enterprise-survey of the Foundation for the Improvement of Living and Working Conditions found that in 2004–5 one-third of companies in establishments with 10 or more employees in the EU, varying from a high 73 per cent in Denmark, via 32 per cent in Germany to a low 21 per cent in the United Kingdom, 19 per cent in Poland and only 11 per cent in Portugal, had some form of workplace representation (Riedmann 2006). In most member states, national laws or agreements guarantee employee information, consultation, and in some cases (i.e. Germany, Sweden, Denmark, Austria, the Netherlands, Belgium and France) codetermination, either by assigning rights to unions or to elected works councils (Rogers and Streeck 1995). Usually, small firms (under 50 employees) are exempted from mandatory statutes.
Extension through dilution
191
Before recent EU legislation, workplace representation was voluntary in the UK and Ireland and only available in a minority of firms, typically excluding foreign-owned multinationals, especially when owned by US parent companies. This is also the common situation in CEE. During transition, there were several attempts at legislation, but outside Slovenia and Hungary the legal powers of works councils are extremely limited. In many CEE countries, foremost in Poland, trade unions perceive works councils as potential competitors and unions have rallied against legislation, sometimes together with employers (Tóth and Ghellab 2003). The political fragmentation of the unions, each claiming ‘single union’ representation in a particular company, explains why they reject works councils as Trojan horses bringing in rival unions, possibly manipulated by employers. (For similar reasons UK and US unions have opposed works councils.) In EU-15 countries with politically or religiously divided unions, practice and legal doctrine supports joint representation and single union bargaining and representation is rare (Rogers and Streeck 1995). Works council elections have in fact become a test for measuring relative support for different unions (Ebbinghaus and Visser 2000). Figure 7.3 compares the extent of workplace representation with union density and bargaining coverage, although for several countries data are missing. We observe that in most countries the proportion of employees who report union presence in their workplace is much larger than the proportion reporting membership. For instance, in France only 8 per cent of the employees join a union but 39 per cent report union presence in their workplace (Amossé 2004); in Hungary these figures are 17 and 33 per cent respectively, in Poland 17 and 25 per cent (based on national Labour Force Survey data). The arithmetic is simple. In many workplaces, union members are in a
Finland Sweden Denmark Norway Austria France Belgium Netherlands Germany Luxembourg Spain Italy Portugal Greece Ireland United Kingdom Slovenia Slovakia Czech Republic Poland Hungary Estonia Latvia Lithuania
bargaining coverage union workplace union density
0
10
20
30
40 50 60 % employees (with jobs)
Figure 7.3 Union presence, 2003–04 Source: European Industrial Relations 2004
70
80
90
100
192
Jelle Visser
minority and yet they may wield considerable influence through their elected positions in institutions that derive their power from a legal mandate or agreement with employers with effects that extend to all employees. This creates free-rider incentives but creates a radically different world compared to one in which costs and benefits of unionism accrue to members only. In that world, common in the US, but also found in the UK and CEE, there tends to be a sharp divide between a minority of fully unionised workplaces and a majority of workplaces without any union member. The small difference between union density and workplace representation in Poland, compared to France, indicates closeness to the US model in the Polish case. 4.3 EU legislation Disregarding the early directives on the transfer of undertakings and health and safety, two legal initiatives stand out. In 1994, the Council accepted a directive establishing representation in multinational firms with a minimum of 1,500 employees and 150 employees in at least two member states. This European Works Council (EWC) directive (94/45/European Commission) guarantees employees located in different countries access to common structure of information and consultation on international company policy. The EWC does not provide for sanctions in case management refuses to consult or reconsider its policies (as would be possible under German or Dutch law). The directive used an innovative legal technique, offering firms which negotiated an early deal with unions or staff representatives exemption from even the minimum requirements of the law. Applying to an estimated 1,800 firms with a total of 17 million employees, by 2003 the EWC has been implemented by 45–50 per cent, covering some 70 per cent of the employees concerned (European Commission 2004). Of 547 companies having a subsidiary in the NMS and falling within the scope of the directive, a majority has provisions extending representation but this process has not yet been completed. In 2002 Council and Parliament adopted a directive (2002/14/European Commission) establishing a framework for informing and consulting employees in domestic firms. This directive applies to firms employing at least 50 employees (with transitional arrangements for CEE, the UK and Ireland). Like the EWC directive, this piece of EU law was heavily contested by European and American employers and by some member states, the UK in particular. The compromise is that member states have latitude to find their own ways of implementation, without upsetting existing structures of representation. This creates considerable headaches in the UK, Poland or the Czech Republic, given the animosity between unions and works councils. In Denmark, on the other hand, where labour legislation has generally come from union-employers agreements rather than from parliament, the social partners have extended their joint regulations to non-members in order to satisfy the requirements of EU law. Implementation is more problematic when state capacities to enforce the law are poor and social partners have no joint platform or track record to ease gradual enforcement. This is already an issue in the NMS with regard to the acquis communautaire of EU regulations and EJC jurisprudence (European Commission 2004). According to a recent study, only in Hungary and Slovenia are councils comparable with the fully fledged types in Germany or Austria and there tends to be no relevant workers’ representation in the growing number of companies without trade unions (European Foundation 2003).
Extension through dilution
193
Some scholars expect that EWCs will be harbingers of transnational collective bargaining (Lecher et al. 2002; Marginson and Sisson 1998). There is certainly considerable EWC activity varying from meetings to negotiations over the functioning of the institution itself and its extension to the NMS and candidate countries. There is a dearth of research on outcomes, however. In her study of 18 Dutch-based EWCs, Lamers (1999) found that they helped to mobilise support for company objectives and ease the resolution of conflict over international restructuring proposals. In their joint declaration, published in 2005 and evaluating the decade since the directive was adopted, the European social partners also sound positive about improved information flows between workers and management. In their study of Swedish-based EWCs, Huzzart and Docherty (2005) show that both unions and management want to keep the EWC informal and that neither sees the EWC as a vehicle for transnational bargaining. Like their German sister organisations, but unlike those in member states with weaker legislation and less established trade unions, Swedish unions consider that national institutions give them more control over international development than European ones. This national sentiment is likely to have been reinforced rather than weakened by enlargement. Concluding, this review finds that the impact of European legislation on workers’ and employment rights depends substantially on how such legislation integrates with national laws and traditions. When acceding to the European Union, the new member states had to overhaul their labour and employment laws completing a process begun after the end of communism. However, within rather weakly developed national institutions of collective representation of employees at firm and national levels, European legislation – itself providing only a floor of minimum rights while giving national legislators considerable leeway to implement EU law according to national traditions and preferences – cannot be expected to have a very strong harmonising effect. With the neo-voluntarist turn in EU labour legislation, the extension of worker rights to the new member states was only possible by accepting a diluted version of rights.
Notes 1 The author thanks Peter Auer (ILO) and the editors for their helpfully critical comments. The usual disclaimer applies. 2 This chapter compares developments in the NMS, or rather a subset of eight ex-communist countries from Central and Eastern Europe (CEE) with the EU15, although the lack of reliable data for Portugal, Greece and Luxembourg does not always warrant their inclusion. Since reliable and comparable institutional data for the three candidate countries (CC) are not (yet) available, they are omitted from this chapter. 3 In 2001 European unions and employers failed to reach agreement on temporary agency work, in particular on the principle of non-discrimination compared with similar jobs under standard contracts. 4 Figures calculated from the European Labour Force Survey (ELFS). Additional information on the NMS comes from the 2001 survey of the European Foundation for the Improvement of Living and Working Conditions, Working Conditions in the Acceding and Candidate Countries, 2003; and from the country studies in Working and Employment Conditions in Future EU Member States, an ILO project directed by Daniel Vaughan-Whitehead and François Eyraud, published in chapter 6 of Industrial Relations in Europe 2004, edited by Jelle Visser for the European Commission (European Commission 2004). 5 Both the EU and ILO put pressure on Slovenia to change its labour laws and disallow bargaining based on involuntary organisation, which violates the European Social Charter,
194
6 7 8 9 10
Jelle Visser
article 6, and ILO conventions 98 (1949) guaranteeing the right of voluntary bargaining and 154 (1981) charging the state with the responsibility to uphold that guarantee. Unfortunately there are no comparable figures for Ireland. Slovenia, with compulsory membership of employers and therefore 100 per cent coverage, does not compare and has been omitted. Industry bargaining in Germany takes place at the regional level, but agreements are highly patterned. Contractual rates in the East German Länder are lower than the national average, but differences have narrowed, as has been a consistent union objective since 1989. Data on earnings equality are from the Structure of Earnings Survey (SES) of Eurostat, released in April 2005. Figures are as of 1 January 2003. The frequency of adjustment is annual in Belgium, the Czech Republic, Estonia, France, Hungary (frozen in 2003), Slovakia, Slovenia and the UK, twice a year in the Netherlands (frozen in 2003) and irregularly in Poland and Spain. Statutory minimum wages are adjusted regularly but at intervals longer than one year in Ireland, Latvia and Lithuania.
References Agell, J. (1999) ‘On the benefits from rigid labour markets: norms, market failures, and social insurance’, Economic Journal, 109: 143–164. Amossé, T. (2004) ‘Mythes et réalités de la syndicalisation en France’, DARES, Premières synthèses et informations, 44/2, Paris. Avdagic, S. (2005) ‘State-labour relations in East Central Europe: explaining variations in Union effectiveness’, Socio-Economic Review, 3, 1: 25–53. Avdagic, S., Rhodes, M. and Visser, J. (2005) ‘The emergence and evolution of social pacts: a provisional framework for comparative analysis’, European Government Papers, EUROGOV N-05-01. Online. Available http: (accessed 4 December 2006). Baccaro, L. (2003) ‘What is alive and what is dead in the theory of corporatism’, British Journal of Industrial Relations, 41, 4: 683–706. Barnard, C. (2000) EC Employment Law. Oxford: Oxford EC Law Library. Benedictus, H., de Boer, R., van der Meer, M., Salverda, W., Visser, J. and Zijl, M. (2002) The European Social Dialogue: Development, Sectoral Variations, and Prospects, Amsterdam, The Hague: Amsterdam Institute for Advanced Labour Studies, Ministry of Social Affairs and Employment. Bercusson, B. (1992) ‘Maastricht: a fundamental change in European labour law’, Industrial Relations Journal, 23: 177–190. Blanpain, R. (2002) European Labour Law, Deventer and Boston: Kluwer Law International. Blau, F. and Kahn, L. (1999) ‘Institutions and laws in the labour market’, pp. 1399–1461, in O. Ashenfelter and D. Card (eds), Handbook of Labour Economic, Amsterdam: North-Holland. Brown, W. (1993) ‘The contraction of collective bargaining in Britain’, British Journal of Industrial Relations, 31, 2: 189–200. Brown, W., Deakin, S., Hudson, M., Pratten, C. and Ryan, P. (1998) The Individualisation of Employment Contracts in Britain, research paper, London: Department of Trade and Industry. Calmfors, L., Booth, A., Burda, M., Checchi, D., Naylor, R. and Visser, J. (2001) ‘The role of collective bargaining in Europe’, pp. 1–156, in T. Boeri, A. Brugiavini and L. Calmfors (eds), The Role of the Unions in the Twenty-first Century, Oxford: Oxford University Press. Collins, H. (2002) ‘Is there a third way in labour law?’, pp. 449–470, in J. Conaghan, R. Fischli and K. Klare (eds), Labour Law in an Era of Globalization: Transformative Practices and Possibilities, Oxford: Oxford University Press. Crouch, C.J. (1993) Industrial Relations and European State Traditions, Oxford: Clarendon Press.
Extension through dilution
195
Deakin, S. (1997) ‘Integration through law? The law and economics of European social policy’, pp. 118–151, in J.T. Addison and W.S. Siebert (eds), Labour Markets in Europe: Issues of Harmonization and Regulation, London: Dryden Press. Dolado, J., Kramarz, F., Machin, S., Marfgolis, D. and Teulings, C. (1996) ‘The economic impact of minimum wages in Europe’, Economic Policy, 23: 319–372. Ebbinghaus, B. and Visser, J. (2000) Trade Unions in Western Europe since 1945, London: Palgrave-Macmillan. EIRO (2004) ‘Working time developments 2003’. Online. Available http: (accessed 13 March 2007). EIRO (2005) ‘Summit marks 20 years of EU social dialogue’. Online. Available http: (accessed 13 March 2007). Ellguth, P. and Kohaut, S. (2005) ‘Tarifbindung und betriebliche Interessenvertretung: Aktuelle Ergebnisse aus dem IAB Betriebspanel’, WSI-Mitteilungen, 58: 7. European Commission (2002) Report of the High-level Group on Industrial Relations and Industrial Change, European Commission, DG Employment and Social Affairs, Luxembourg: Office for the Official Publication of the European Communities. European Commission (2003) Employment in Europe 2003, DG Employment and Social Affairs, Luxembourg: Office of Official Publications of the EU. European Commission (2004) Industrial Relations in Europe, 2004, DG Employment and Social Affairs, Luxembourg: Office for the Official Publication of the European Communities. European Commission (2005) Employment in Europe 2005, DG Employment and Social Affairs, Luxembourg: Office of Official Publications of the EU. European Foundation for the Improvement of Living and Working Conditions (2003) Dialogue and Conflict Resolution in the Acceding Countries, Report no. 11, Luxembourg: Office for Official Publications of the European Communities. European industrial relations observatory online (2004) ‘Working time developments 2003’. Online. Available http: (accessed 6 December). Financial Times (2006) ‘Dilution is disservice to EU’s draft directive’, editorial comment, 10 February: 14. Flanders, A. (1970) Management and Unions: The Theory and Reform of Industrial Relations, London: Faber and Faber. Funk, L. and Lesch, H. (2005) ‘Minimum wages in Europe’. EIROnline. Available http: (accessed 4 December 2006). Ghellab, Y. and Vaughan-Whitehead, D. (2003) ‘Sectoral social dialogue: a link to be strengthened’, pp. 1–42, in Y. Ghelleb and D. Vaughan-Whitehead (eds), Sectoral Social Dialogue in Future EU Member States: The Weakest Link, Geneva: International Labour Office. Hall, P.A. and Soskice, D. (2001) ‘An introduction to varieties of capitalism’, pp. 1–68, in P.A. Hall and D. Soskice (eds), Varieties of Capitalism: The Institutional Foundations of Comparative Advantage, Oxford: Oxford University Press. Hassel, A. (2006) ‘What does business want? Labour market reforms in CMEs and its problems’, chapter forthcoming in M. Rhodes, B. Hancké and M. Thatcher (eds), Institutional Change in Contemporary European Capitalism: Conflict, Contradiction and Complementarities, Oxford: Oxford University Press. Hibbs, D.A. Jr. and Locking, H. (1996) ‘Wage dispersion and productive efficiency: evidence for Sweden’, Journal of Labour Economics, 18: 755–782. Huzzard, T. and Docherty, P. (2005) ‘Between global and local: eight European works councils in retrospect and prospect’, Economic & Industrial Democracy, 26, 4: 541–568.
196
Jelle Visser
Iankova, E. (2002) Eastern European Capitalism in the Making, Cambridge MA: Cambridge University Press. Iankova, E. and Turner, L. (2004) ‘Building the new Europe: western and eastern roads to social partnership’, Industrial Relations Journal, 35, 1: 76–92. Iversen, T. (1999) Contested Economic Institutions: The Politics of Macroeconomics and Wage Bargaining in Advanced Democraciesm, Cambridge, MA: Cambridge University Press. Katz, H. and Darbishire, O. (2000) Converging Divergencies, Ithaca, NY: Cornell University Press. Kilpatrick, C. (2003) ‘Has New Labour reconfigured employment legislation?’, Industrial Law Journal, 32, 3: 135–163. Kohl, H. and Platzer, H-W. (2003) ‘Labour relations in Central and Eastern Europe and the European social model’, Transfer, 9, 1: 11–30. Lado, M. and Vaughan-Whitehead, D. (2003) ‘Social dialogue in candidate countries: what for?’, Transfer, 9, 1: 64–87. Lamers, J. (1999) The Added Value of European Works Councils, Haarlem: AWVN (Algemene Werkgevers Vereniging Nederland). Lange, P. (1984) ‘Unions, workers and wage regulation: the rational basis of consent’, pp. 98–123, in J.H. Goldthorpe (ed.), Order and Conflict in Contemporary Capitalism: Studies in the Political Economy of Western European Nations, Oxford: Clarendon Press. Lecher, W., Platzer, H-W., Rueb, S. and Wiener, K-P. (eds) (2002) European Works Councils: Negotiated Europeanization between Statutory Framework and Social Dynamic, Aldershot: Ashgate. Lindbeck, A. and Snower, D.J. (2001) ‘Centralised bargaining and reorganised work: are they compatible?’, European Economic Review, 45, 10: 1851–1875. Marginson, P. (2005) ‘Industrial relations at European sector level: the weak link?’, Economic & Industrial Democracy, 26, 4: 511–540. Marginson, P. and Sisson, K. (1998) ‘European collective bargaining: a virtual prospect?’, Journal of Common Market Studies, 36, 4: 505–528. Marginson, P. and Sisson, K. (2004) European Integration and Industrial Relations: Multi-level Governance in the Making, London: Palgrave-Macmillan. OECD (2004) ‘Wage-setting Institutions and Outcomes’, pp. 127–182, in Employment Outlook 2004, Paris: Organisation for Economic Cooperation and Development. Ost, D. (2000) ‘Illusory corporatism in Eastern Europe: neoliberal tripartism and post-communism’, Politics and Society, 28: 503–530. Pollert, A. (1999) Transformation at Work: In the New Market Economies in Central Eastern Europe, London: Sage. Renooy, P., Ivarsson, S., van der Wusten-Gritsai, O. and Meijer, R. (2004) Undeclared Work in an Enlarged Union: An Analysis of Undeclared Work. An In-depth Study of Specific Items, study commissioned by European Commission, DG EMPL, Brussels. Riedmann, A. (in cooperation with H. Bielenski, T. Szczurowska and A. Wagner (2006) Working-time and Work-Life Balance in European Companies, European Foundation for the Improvement of Living and Working Conditions, Luxembourg: Office for Official Publications of the European Communities. Rogers, J. and Streeck, W. (eds) (1995) Works Councils: Consultation, Representation, Co-ordination, Chicago: University of Chicago Press. Schmid, B. and Schulten, T. (2006) ‘Der französische Mindestlohn SMIC’, pp. 102–126, in T. Schulten, R. Bispinck and C. Schäfer (eds), Mindestlöhne in Europa, Hamburg: VSA-Verlag. Schroeder, W. and Weinert, R. (2004) ‘Designing Institutions in European Industrial Relations’, European Journal of Industrial Relations, 10, 2: 199–217. Schulten, T. (2002) ‘Europeanisation of collective bargaining: trade union initiatives for the transnational co-ordination of collective bargaining’, pp. 112–136, in B. Keller and H.-W. Platzer (eds), Industrial Relations and European Integration, Aldershot: Ashgate.
Extension through dilution
197
Schulten, T., Bispinck, R. and Schaefer, C. (eds) (2006) Mindestlˆhne in Europa, Hamburg: VSA Verlag. Sciarra, S. (2004) The Evolution of Labour Law (1992–2003), Report for the European Commission, DG Emp, Florence: University of Florence. Slichter, J., Healy, J. and Livernash, R. (1960) The Impact of Collective Bargaining on Management, Washington, DC: Brookings Institution. Streeck, W. (1992) Social Institutions and Economic Performance: Studies of Industrial Relations in Advanced Capitalist Economies, London: Sage. Streeck, W. (1995) ‘From market making to state building? Reflections on the political economy of European social policy’, pp. 389–431, in S. Leibfried and P. Pierson (eds), European Social Policy: Between Fragmentation and Integration, Washington, DC: Brookings Institution. Streeck, W. (2005) ‘Labour markets and trade unions’, pp. 254–283, in N.J. Smelser and R. Swedberg (eds), The Handbook of Economic Sociology, Princeton, Oxford: Princeton University Press, with Russell Sage Foundation. Streeck, W. and Reholer, B. (2003) ‘Der Flächentarifvertrag: Krise, Stabilität und Wandel’, Industrielle Beziehungen, 10, 3: 341–362. Supiot, A. (2001) Beyond Employment: Changes in Work and the Future of Labour Law in Europe, Oxford: Oxford University Press. Traxler, F., Blaschke, S. and Kitter, B. (2001) National Labour Relations in Internationalized Markets: A Comparative Study of Institutions, Change, and Performance, Oxford: Oxford University Press. Tóth, A. and Ghellab, Y. (2003) The Challenge of Representation at the Workplace in EU Accession Countries: Does the Creation of Works Councils Offer a Solution Alongside Trade Unions? Report for Tripartite Conference, Geneva: International Labour Office. Visser, J. (1990) ‘In search of inclusive unionism’, Bulletin of Comparative Labour Relations, 18: 5–278. Visser, J. (2005) ‘Beneath the surface of stability: new and old methods of governance in European industrial relations’, European Journal of Industrial Relations, 11, 3: 278–306. Visser, J. (2006) ‘Union membership statistics in 24 countries’, Monthly Labour Review, January: 38–49. Wallerstein, M. (1999) ‘Wage-setting institutions and pay inequality in advanced industrial societies’, American Journal of Political Science, 43: 649–80. Woolfson, C. and Sommers, J. (2006) ‘Labour mobility in construction: European implications of the Laval un Partneri dispute with Swedish labour’, European Journal of Industrial Relations, 12, 1: 49–68.
Part III
Material living conditions
8
Poverty, deprivation and economic vulnerability in the enlarged EU Christopher T. Whelan and Bertrand Maître
Introduction1 The term ‘quality of life’ refers to the overall well-being of individuals. Its distinctive feature is the attempt to move beyond narrow or one-dimensional views of human personality in assessing people’s situation. In this chapter our focus is on one of the fundamental components of quality of life – objective living conditions and the manner in which individuals evaluate their material situation. The core notion within which these issues are addressed is that it is not simply outcomes that matter, but, rather, the capacity to affect outcomes (see Fahey et al. 2003 for a more detailed discussion of these issues). This chapter will focus on income and current lifestyle deprivation while recognising that, in capturing the resources and opportunities open to people, collective as well as individual resources need to be assessed. Social provision in areas such as health care, housing and social services are fundamental and are addressed in other parts of this book. However, previous research suggests that the aspects on which we concentrate in this part are crucial to individuals’ evaluations of their well-being (Whelan et al. 2001). Monitoring living conditions and quality of life cannot be a purely ‘scientific’ exercise but, rather, must tap into the central concerns and goals of society – what is counted as a good quality of life is as much a matter of values and social philosophy as of scientific fact. Here we take the European social policy agenda as a practical expression of underlying basic principles in this area and so use it as a reference point in shaping our approach. For many years the predominant policy focus of the EU was economic rather than social. However, in recent years that focus has shifted. The EU has an increasing interest in social issues and a growing competence in social policy, partly in the light of the perceived need to offset some of the potentially negative social effects of the single market. Such concerns take on even greater significance with the enlargement of the EU. Important landmarks in the development of social competence were the 1989 Social Charter, the Social Protocol of the Maastricht Treaty in 1989 and articles 136 and 137 of the Amsterdam Treaty requiring the Community to support the member states’ action to combat social exclusion. The concerns of European policy now encompass raising living standards and improving living conditions, strengthening social cohesion and combating exclusion. It also now increasingly sees the link between economic and social spheres as of central importance. The linkage between the policy domains is highlighted in the Lisbon European Council’s identification of
202
Christopher T. Whelan and Bertrand Maître
a fresh set of challenges that must be met so that Europe can become ‘the most competitive and dynamic knowledge based economy in the world capable of sustaining economic growth with more and better jobs and greater social cohesion’. It is within the context of the new challenges created by transformation in the economic sphere that social exclusion is being identified as a key focus for social policy. The shifting balance between economic and social concerns and the need to continually review the relationship between them, particularly in the context of EU enlargement, is reflected in the interpretation of key elements in the conceptual architecture of EU integration policy. Used in this context, the concept of ‘social cohesion’ refers to equality between countries and regions within the EU, particularly with regard to level of economic development. It is measured by the degree to which key economic indicators at the national or regional level, such as GDP per capita, converge towards an EU-wide mean. A socially cohesive EU, in this sense, is one where no country or region is much poorer or less economically developed than the norm for the EU as a whole. Social inclusion/exclusion can almost be thought of as the within-country counterpart of the cross-country concept of social cohesion. Its concern is with within-country inequalities between individuals or households, rather than cross-country inequalities between countries or regions. It does not take account of the full range of the social distribution within countries but focuses on a dichotomy between the bottom tail of the distribution – the socially excluded minority who are ‘cut off’ – and the rest, who are assumed to constitute the ‘mainstream’. A socially inclusive society, in this sense, is one where no individuals or households fall below the threshold of living conditions that is thought to provide the minimum necessary basis for participation in the normal life of a society. In this chapter we endeavour to make use of the information available in the European Quality of Life Survey 2003 (EQLS) to develop an approach that addresses both social inclusion and social integration issues.
1. Indicators of income and lifestyle deprivation in the EQLS As mentioned earlier, our focus in this chapter is both on objective living conditions and how people evaluate their situation. The two key measures of objective conditions we use are income and material deprivation, while people’s evaluation of their situation is captured in a measure of economic stress, which refers to people’s perceived difficulty in making ends meet. 1.1 Income The income measure used in the EQLS was relatively crude, as it consisted simply of a global estimate of net household income on the part of the respondent. The resulting income data cannot be expected to yield a precise estimate of household income. Nevertheless, a comparison at the national level between the EQLS income data and aggregate economic indicators shows that the EQLS data perform reasonably satisfactorily, in that, for example, national median household income as measured by the EQLS correlates almost perfectly with GNP per capita as measured by national accounts (Fahey et al. 2005: 14). The income measure we use here is household equivalent income which adjusts for differences in household size using the modified OECD equivalence scale (the first adult in a household is given a value of 1, any
Poverty, deprivation and vulnerability
203
additional adult a value of 0.5 and every child aged under 14 a value of 0.3). The income data also take account of price differences between countries by converting nominal values to Purchasing Power Standards (PPS is an artificial common currency developed by Eurostat to give a measure of purchasing power that is standardised across EU countries – see Stapel et al. 2004). 1.2 Lifestyle deprivation The ten-item deprivation index we employ is constructed from items relating to food, clothing, heating, consumer durables and ability to participate in social life. For each of these items the respondent was asked if the household possessed it and if not if this was because they could not afford it. An individual is considered as deprived of an item only where both conditions are fulfilled. The resulting scale, which is scored from 0 (no deprivation) to 10 (deprived on all items) therefore is designed to measure enforced deprivation. The measure, which we label current lifestyle deprivation (CLSD), is intended to capture exclusion from participation in a manner generally identified as appropriate in the relevant community. For ease of presentation we will refer to this indicator throughout as ‘deprivation’. 1.3 Economic stress Our measure of subjective economic stress treats those who were having ‘great difficulty’ or ‘difficulty’ in making ends meet as experiencing economic stress and all others as not.
2. Income and deprivation inequalities In Figure 8.1 we make use of Eurostat data to show net median monthly income broken down by country with further distinctions between the new member states (NMS), the EU-15 and the EU-25 (the income measure is expressed in PPS and equivalised for household size as mentioned earlier). Data for clusters of countries are calculated to take into account variations in population size. The pattern of distribution of household income between the 28 countries reveals a number of distinct groups. The first is composed of the EU-15 countries where 11 out of 15 countries have a median monthly equivalent income over 1,000 PPS, Greece, Portugal and Spain having the lowest values between 700 and 1,000 PPS. For the NMS countries the median is less than half that of the EU-15. In the NMS no country reports a monthly equivalent income above 1,000 PPS with the exception of Cyprus (no data were available for Malta). Within the CC-3 the income median range is very narrow and the overall median in each case represents a fifth of the EU-15 one and just above half of the NMS one. Drawing on the EQLS income data, we can fill in further detail on the picture of differences in household incomes between countries by looking at absolute income levels at different points in the income distribution in each country and in the EU-15, EU-25 and the NMS. From Figure 8.2 we can see that if we divide the population of each country into household income quartiles, the middle quartiles in all the countries of the EU-15, with the exception of Spain, Greece and Portugal, have a median income that is higher than the median EU-15 income. For most of the NMS, the
T R urk om ey Li an th ia u Bu ani lg a Es aria to n La ia Po tvia Po lan rtu d G ga r l H eec un e S g C lov ary ze a ch kia R e Sp p. Sl a ov in e C nia yp ru s U M K Be alt lg a i Ire um l Fi and nl an d Ita N Fr ly et a he nc rla e Au nds D st en ria m Lu Sw ark xe ed m en bo ur g om Bu ani lg a a Tu ria rk e Li Lat y th vi ua a Es nia to Po nia N lan M d Sl S 1 ov 0 a C Hun kia ze g ch ar R y Po ep rtu . G ga re l Sl ec ov e en ia It Sp aly a EU in C 25 yp Fi rus nl a Ire nd Sw lan ed d Fr en an EU ce Au 15 Be stri a G lgiu er m m D an en y m ar N et k Lu her UK xe lan m ds bo ur g
R
204 Christopher T. Whelan and Bertrand Maître
2000
1800
1600
1400
1200
1000
800
600
400
200
0
Figure 8.1 Median monthly household equivalised (modified OECD scale) incomes in purchasing power standards, by country Source: Eurostat (New Cronos database) 2001–2003
3500
3000
2500
2000
1500
1000
500
0
Low quartile
Medium quartile
High quartile
Median EU-15
Figure 8.2 Median monthly household income, by income quartile across countries (in PPS) Source: EQLS 2003
Poverty, deprivation and vulnerability
205
Luxembourg Denmark Austria Netherlands Sweden Germany Finland Belgium France Ireland Slovenia EU-15 Cyprus UK Spain EU-25 Italy Malta Czech Greece Portugal Hungary Slovakia NMS-10 Estonia Poland Latvia Lithuania Turkey CC-3 Romania Bulgaria 0
1
2
3
4
5
6
Figure 8.3 Mean deprivation (10 items) across countries Source: EQLS 2003
median household income of the top quartile is at a similar level to that of the middle quartiles of the EU-15. In countries such as Hungary, Portugal and Turkey, the top income quartile fail to reach the level of the EU-15 median but are clearly above the lowest EU-15 quartile. At the extreme end of the continuum are Romania and Bulgaria where the median of the top quartile is at the same level as that of the bottom quartile of the majority of EU-15 countries.2 Thus the scale of income differences between economic clusters requires that both absolute and relative perspectives are required to grasp the realities of income distribution. In Figure 8.3 we break down our ten-item index of deprivation by country. The pattern of deprivation is consistent with our a priori expectations. For the EU-15 as a whole, the level of deprivation is low at 1.1. The corresponding value for the NMS countries is three times higher at 3.2 and that for the CC-3 over four times higher at 4.8. The fact that the population is weighted ensures that the figures for the EU-25 and the 28 countries overall are closer to the EU-15 figure than the NMS and CC-3 levels (given the large population of the EU-15). The disparities between clusters of countries reported above are remarkably similar to those found earlier based on
206
Christopher T. Whelan and Bertrand Maître
median equivalised household income. Thus both income and deprivation approaches locate groups of countries at almost identical points on a continuum of disadvantage. Focusing on deprivation within clusters of countries reveals that Portugal and Greece are sharply differentiated from the remaining EU-15 countries. Portugal has a deprivation level almost three times the EU-15 average and Greece is over two times higher. As a consequence, these two countries have deprivation levels higher than those of four of the NMS, namely, in order of deprivation levels, Slovenia, Cyprus, Malta and the Czech Republic. The first two in fact have values that are not very different from the majority of the EU-15. Thus neither group constitutes an entirely homogeneous block, though the CC-3 do display distinctively high levels of deprivation. In Figure 8.4, we combine the income and deprivation indicators by showing average levels of deprivation by income quartile for each country and cluster of countries. The data show that in the poorer countries even the higher income brackets are quite deprived by the standards of the richer parts of Europe. For example, the mean deprivation level for the highest income quartile in the CC-3 and NMS is respectively higher than and equal to that for the lowest income quartile in the EU-15. If we look at the low-income bracket in the poorer countries, the disparity in level of deprivation compared to their counterparts in the richer countries is very wide indeed. For example, the mean deprivation level in the lowest income quartile in the CC-3 is three and a half times greater than in the lowest income quartile in the EU-15.
8 7 6 5 4 3 2 1
Tu
rk ey R CC om -3 Bu an lg ia a La ria Li tv th ia ua Es nia Po ton rtu ia Po ga l N lan M d Sl S-1 ov 0 H ak un ia C ze g ch Gr ary R eec ep e ub lic Ire UK l C and yp r EU us Fr 25 an c M e Sl al ov ta en Sp ia EU ain Be -1 lg 5 iu m I N Fi taly et nl he an rla d Au nds Sw stri e a Lu Ge den xe rm m an D bou en rg m ar k
0
Low quartile
Medium quartile
High quartile
Figure 8.4 Mean deprivation (10 items), by income quartile, by country Source: EQLS 2003
Poverty, deprivation and vulnerability
207
3. Income poverty Both the income and lifestyle deprivation indicators provide broadly consistent pictures of inequalities between countries and clusters of countries. However, as we shall see, official statistics on income poverty entirely fail to capture such differences. The EU social policy perspective continues to define ‘risk of poverty’ in purely relative terms as falling below a percentage of national median income. However, the widening in income inequalities associated with enlargement raises questions about the continued validity of this approach. If we focus on the Laeken indicator of riskof poverty – those who are below 60 per cent of national median income – we find similar risk-of-poverty rates in rich and poor countries (Figure 8.5), with the average country-level rate being 16 per cent. The countries with the highest poverty rates are Slovakia, Ireland and Greece (21 per cent). Intermediate poverty levels are observed in Portugal, Italy, Spain, the United Kingdom and Estonia (16–19 per cent). The lowest levels are found in the Czech Republic, Luxembourg, Hungary and Slovenia (8–10 per cent). The poverty rate in Latvia – a poor country – is slightly lower than in the UK, a quite rich country. Enlargement has no effect on the numbers estimated to be at risk of poverty. However, even after adjusting for differing purchasing power standards, there are substantial variations in the absolute levels at which such thresholds are set. Consequently, while some of the NMS have relatively low poverty risks, nine of the ten of such states have a threshold that is below the EU-25 average. Expressed in terms of the EU-25 average of E15,913, values range from 28 per cent in Latvia to 188 per cent in Luxembourg (Eurostat 2005a). Poverty, as generally understood in advanced societies, has two core elements: it is about inability to participate, due to inadequate resources. Most quantitative research on poverty in such societies then in fact employs a unidimensional approach to distinguishing the poor: it uses income. The most common approach to establishing a cut-off point has been to rely on relative income lines, with thresholds such as 50 per cent, 60 per cent or 70 per cent of median income being used.3 The broad rationale is that those falling more than a certain ‘distance’ below average income are
% 25 20 15 10 5
NO
BG RO HR TR
FR NL AT DE CY MT BE EU-25 LV EU-15 LT PL EE UK ES IT PT EL IE SK
CZ LU HU SI FI SE DK
0
Figure 8.5 Share of persons in poverty at 60% median equivalised household income, 2003 (modified OECD scale) Source: Eurostat, Statistics in Focus 13/2005
208
Christopher T. Whelan and Bertrand Maître
unlikely to be able to participate fully in the life of the community. However, it has been recognised for some time that low income may be an unreliable indicator of poverty in this sense (Ringen 1987). In teasing out why current income might have serious limitations in capturing poverty, both the concepts of resources and needs and their empirical measurement are relevant.4 A household’s standard of living will depend crucially on its longer-term command over resources. While disposable cash income is a key element in the resources available to a household, it is by no means the only one. In addition, a range of measurement problems mean that one cannot be confident that income has been measured comprehensively and accurately.5 Turning to needs, these also differ across households, in a manner that is difficult to capture adequately at the conceptual, much less empirical, level. The evidence for a range of countries suggests it is hazardous to draw strong conclusions about whether a household is poor from current income alone. This has been demonstrated in a variety of studies of different industrialised countries employing non-monetary indicators of deprivation. It has also been shown that the overlap between income and deprivation approaches is least in the more affluent societies.6 Thus the EU enlargement process does not, as such, exacerbate the problems associated with relative income lines. The major problems with such indices relate, predominantly, to ‘old’ EU members rather than the new member countries. The strength of the argument that ‘poverty is about more than income’ is not dependent on the consequences of enlargement. Of course, both the general limitations of such measures and the variability in the efficiency of such measures across countries creates considerable difficulties for the interpretation of cross-national comparisons. It is entirely reasonable, however, for advocates of the relative income line approach to argue that such measures were never intended to capture cross-country differences in living standards and that the purely national perspective is entirely consistent with the limiting character of the EC’s social policy remit. Thus, despite the EU’s increasing interest and competence in the field, social policy could be considered as primarily a member state competence with the EC’s role in the field being defined largely in terms of a coordination function.7 On the other hand, for EU regional policy the divergence in living standards between regions and member states is the main focus of interest. Regional policy is also firmly grounded in EU law and confers on the EU the power to distribute funds between the EU regions for the purpose of promoting the development of the disadvantaged regions. The policy goal in this context is to promote economic and social cohesion by bringing about convergence in economic development and living standards between the rich and poor member states and regions of the EU (European Commission 2004). While the social policy perspective takes member state ‘thresholds’ as its point of reference, the regional perspective uses EU-wide thresholds based principally on GDP per capita (expressed in PPS). This approach captures differences comparable to those presented earlier in relation to median income and lifestyle deprivation. The perceived limitations of state-level relative income lines and a desire to encompass regional differences has led authors such as Fahey (2007) to propose that EU-wide income poverty lines should be used alongside state-level approaches. The former produces rankings of countries very similar to those produced by GDP per capita, with very high poverty rates in the poorer countries and low rates in the richer countries. However, the EU-wide approach appears to require a particularly strong
Poverty, deprivation and vulnerability
209
case for treating the EU-25 as a single society and for focusing on EU-wide rather than national reference groups. In the absence of an explicit justification of this kind, we find it difficult to see how such measures can be connected to the rationale that those falling more than a certain ‘distance’ below median income are excluded from participating fully in the life of the community. If poverty is understood in this sense, then EU-wide income thresholds seem likely to significantly overestimate poverty rates in poorer countries and correspondingly underestimate levels in more affluent countries. A recent Eurostat (2005b: 5) analysis confirms that, at the macro-level, using an EU-wide poverty line leads us to underestimate deprivation problems in countries where the level of median income is higher than the EU threshold but where a non-negligible part of the population can still face deprivation. The juxtaposition of national and EU-wide relative income poverty lines does not appear to us to have a sufficiently coherent rationale. Overall, we are not persuaded that the use of such EU-wide lines provides us with much greater information than is available from indicators such as GDP. Both the development of a broader European Commission social policy remit to include a common set of indicators of social exclusion that are monitored through national action plans for social inclusion, and the concerns encapsulated in the regional perspective, point to the need for a consideration of a wider range of indicators of economic and social circumstances (see Atkinson et al. 2002 and Ferrera et al. 2002). In this context it seems unlikely that national at-risk-of-poverty rates that appear to be counter-intuitive are likely to be taken seriously as a basis for evaluating the comparative impact of policy interventions, unless an explicit rationale for such comparison is developed. On the other hand, an EU-wide income poverty measure, while capturing aspects of national diversity that the current at-risk-of-poverty figures miss, does not provide a satisfactory alternative. In responding to these issues, we follow Fahey (2007) in arguing that social policy and regional perspectives can be viewed as complementary rather than contradictory. However, we argue that a multidimensional approach is needed in order to develop a sufficiently comprehensive understanding of the policy relevance both within- and between-country inequalities.
4. A multidimensional approach to measuring economic exclusion Even prior to enlargement, a large body of literature had argued the need to shift focus from income poverty to multidimensional social exclusion and from a static to a dynamic perspective – see Ashworth et al. (2000), Bane and Ellwood (1986), Duncan et al. (1993), Halleröd (1996), Kangas and Ritakallio (1998), Nolan and Whelan (1996). Such demands have been given added impetus by the difficulties that EU enlargement has created for policy on disadvantage. As Förster (2005: 29) observes, agreement that poverty and social exclusion should be defined on a common basis across countries must now be implemented in the context of the accession of ten countries, the vast majority of which underwent a deep ‘transitional recession’ during the early 1990s and whose shared legacy from the former era included the denial of poverty as a social policy concern (for a discussion of some of the consequences of such changes for individual reference groups see Kelley and Zagorski 2005).
210
Christopher T. Whelan and Bertrand Maître
In pursuing a multi-dimensional agenda we will argue for the value of a vulnerability perspective. As De Haan (1998: 15) observes, notions of vulnerability are closely associated with the social exclusion perspective. Following Chambers (1989: 1), we can define vulnerability as not necessarily involving current deprivation either in income or other terms but rather insecurity and exposure to risk and shock. To develop measures of vulnerability we require point-in-time proxies for the kind of exposure to risk of persistent disadvantage that is captured in panel surveys. This objective is combined with a concern to develop a genuinely multidimensional perspective. The IMF (2003), the United Nations (2003) and the World Bank (2000) have developed a range of approaches to measuring vulnerability at the macro-level. Consistent with the approach developed here, the World Bank sees vulnerability as reflecting both the risk of experiencing an episode of poverty over time but also a heightened probability of being exposed to a range of risks (World Bank 2000). Most attempts to measure vulnerability have operated at the macro-level based on composite indices, as in the case of the UNDP’s Human Development Index (HDI) (World Bank 2000, United Nations 2003). Such approaches must confront the usual issues relating to the aggregation of indicators. How do we combine measures of factors such as life expectancy, income poverty rates, unemployment levels and educational standards? As a consequence of such difficulties, the EU Laeken indicators were deliberately presented individually with no attempt to produce an overall ‘score’ across dimensions – indeed Atkinson et al. (2002) argue that this should be avoided precisely because the whole thrust of the European social agenda is to emphasise the multidimensionality of social disadvantage. The approach we adopt analyses manifest indicators at the individual level in order to identify underlying or latent vulnerability. The successful implementation of such a measurement strategy requires multiple indicators to fulfil the multidimensionality condition. However, as we have stressed, it also involves a dynamic perspective. Our objective is not simply to document those who experience a specific deprivation at a particular point in time but rather to identify those who are vulnerable to such deprivation. From a policy perspective this allows us to think in terms of options that may prevent such vulnerability from being translated into actual negative outcomes. In the longer run, it allows us to study the processes involved in the differential routes that lead from vulnerable status to positive or negative outcomes. The focus is very clearly on process rather than simply point-in-time outcomes.8 We achieve this objective by the application of latent class analysis, which can be used as a tool to gain deeper understanding of the observed relationships between dichotomous (or polytomous) indicators. It can be thought of as a log-linear model where there are more dimensions in the estimated table than in the observed table. Such models generate tables of expected frequencies that can be compared to the observed frequencies using goodness-of-fit tests. The basic idea underlying such analysis is long established and simple.9 The associations between a set of categorical variables, regarded as indicators of an unobserved typology, are accounted for by membership of a small number of latent classes. As Moisio (2004) notes, implicit in the notion of multidimensional measurement of exclusion is the assumption that there is no one ‘true’ indicator of the underlying concept. Instead we have a sample of indicators that tap different aspects of a complex phenomenon. Latent class analysis assumes that each individual is a member of one and only one of N underlying classes and that, conditional on identification of latent class membership, the
Poverty, deprivation and vulnerability
211
observed variables are mutually independent of each other. Conditional independence is a version of the familiar idea that the correlation between two variables may be a result of their common dependence on a third variable. In estimating latent class models the logic is identical but the explanatory variable is unobserved and must be identified statistically. The analysis presented below will make use of the clustering of countries employed by Fahey et al. (2005), using GDP per capita as a classification variable, as adapted from a classification developed by the European Commission (European Commission 2004). The four clusters of countries are as follows: 1
2
3
4
Twelve high-income EU member states whose GDP per capita exceeds the mean of the EU-25 (EU-12 HI). These comprise Finland, Sweden, Denmark, Germany, Luxembourg, Austria, Belgium, the Netherlands, France, Ireland, the UK and Italy. Seven intermediate EU member states whose GDP per capita lies between 60 per cent and 100 per cent of the EU-25 mean (EU-7 INT). These comprise Spain, Greece, Portugal, Malta, Cyprus, Slovenia and the Czech Republic. Six low-income EU members states whose GDP per capita lies below 60 per cent of the EU-25 mean (EU-6 LO). These comprise Poland, Estonia. Hungary, Slovakia, Latvia and Lithuania. Three candidate countries (CC-3), Bulgaria, Romania and Turkey. GDP per capita in the CC-3 is below 35 per cent of the mean of the EU-25.
Our focus is on three key indicators – an individual’s quartile position within economic cluster for both income and deprivation and exposure to economic stress as captured by the item relating to ‘difficulty in making ends meet’. In the first two cases it is clear that our indicators refer entirely to relativities within clusters. For the final indicator the fact that 80 per cent of the variation on this item is within cluster indicates that it serves primarily as a relative measure but there are also significant between-cluster differences. Our objective is to identify groups who are vulnerable to economic exclusion in the sense of being distinctive in their risk of falling below a critical resource level, being exposed to life-style deprivation and experiencing subjective economic stress. For each cluster of countries we identify a vulnerable and a non-vulnerable class.10 The key results arising from such analysis are the size of the vulnerable class in each cluster and the contrasting multi-dimensional deprivation profiles of the vulnerable and non-vulnerable classes. The latter outcomes are provided by the conditional probabilities, given the latent class of which one is a member, of being located in the bottom income quartile, in the top deprivation quartile and experiencing great difficulty or difficulty in making ends meet. A striking finding to emerge from this analysis, as illustrated in Figure 8.6, is that, unlike the case for analysis based solely on relative income, a multidimensional approach focusing on within-cluster inequalities reveals substantial between-cluster variations in the level of economic vulnerability. The lowest level of 23 per cent is observed in the EU-12 HI. This rises to 31 per cent for the EU-7 INT cluster and to just over 40 per cent for the EU-6 LO and the CC-3 clusters. In Figure 8.7 we set out the patterns of differentiation between the economically vulnerable and all others. Our analysis reveals generally similar outcomes in terms of income differences between vulnerable and non-vulnerable classes across clusters.
212
Christopher T. Whelan and Bertrand Maître
50 40 30 20 10 0
EU-12 HI
EU-7 INT
EU-6 LO
CC-3
Figure 8.6 Vulnerable class size, by economic clusters Source: Whelan and Maître (2005b)
For those in the vulnerable class the probability of being located in the bottom income quartile displays extremely modest variation across economic cluster with the highest value being 54 per cent and the lowest 45 per cent. It is also true that the likelihood of being found in the top income quartile remains within the narrow range running from 2 per cent to 8 per cent. In the case of deprivation, however, there is a sharp contrast between the more and less affluent clusters. In the former, deprivation is a much stronger factor in distinguishing the economically vulnerable from the non-vulnerable and risk levels are significantly higher with the likelihood reaching 90 per cent in the EU-12 HI before falling to just below 80 per cent in the EU-7 INT. In contrast, the risk levels are below 60 per cent in the remaining clusters. Not only are relative deprivation levels for the economically vulnerable significantly higher in the better-off clusters but deprivation also serves to differentiate them much more sharply from the non-vulnerable than in the less affluent clusters. 1 0,8 0,6 0,4 0,2 0 EU-12 HI
EU-7 INT Bottom income quintile V
EU-6 LO
CC-3
Bottom income quintile NV
Deprivation V
Deprivation NV
Economic strain V
Economic strain NV
Figure 8.7 Conditional probabilities for latent class models for income deprivation and economic stress (V = Vulnerable, NV = Not Vulnerable) Source: Whelan and Maître (2005a)
Poverty, deprivation and vulnerability
213
The relatively large vulnerable classes in the EU-6 LO and CC-3 display extremely high-risk levels of economic stress with nine out of ten meeting this condition. Six out of ten fulfil the deprivation condition and one in two are in the bottom-income quartile. For these clusters economic stress is the most powerful factor differentiating between the vulnerable and non-vulnerable groups. For the EU-7 INT on the other hand, economic stress is a much less powerful distinguishing factor. As in the case of deprivation the sharpest contrast in relation to economic stress occurs for the EU-12 HI cluster.11 Unlike a relative income poverty approach, the latent class multidimensional perspective enables us to capture both between-economic-cluster variation in the scale of economic vulnerability and within-cluster variation in the intensity and patterning of such vulnerability. Thus while the vulnerable class in the EU-12 HI is half the size of that in the EU-6 LO and CC-3 and their absolute income levels and living conditions and ability to avoid economic stress are clearly superior, the relative disadvantages experienced by this group in relation to both lifestyle deprivation and economic stress are substantially greater than those found in any of the less affluent clusters. In that sense they are more excluded than their counterparts in the less affluent clusters. The multidimensional approach we have pursued shows the importance of simultaneously taking into account both within- and between-economic-cluster differences. The results we have reported in relation to variation in levels of economic vulnerability across economic clusters are entirely consistent with our earlier findings relating to cross-national variations in income and deprivation levels. In each case the variability that is minimised by the relative income poverty approaches emerges. On the other hand, while the economically vulnerable constitute a much smaller group in the more affluent countries the degree of polarisation in terms of deprivation and economic stress is much sharper. This rather different form of variability would be entirely obscured, for example, by using an overall EU income poverty threshold. The importance of being able to take both forms of inequality into account is shown by the fact that, as Whelan and Maître (2005a) demonstrate, taking both types into account is important in understanding the consequences of economic variability for social cohesion measured at the level of the individual; as reflected in interpersonal trust, alienation and perceived inter-group tensions.12 Levels of cohesion, understood in this sense, vary across economic clusters in a manner consistent with the magnitude of vulnerability.13 However, the association between economic vulnerability and a range of indicators of social cohesion is significantly stronger in the more affluent economic clusters. This conclusion is in line with that of Böhnke in a later chapter (13) in this volume that polarisation, in terms of a feeling of ‘belonging’ between economically worse and better-off groups, is far more pronounced in the group of EU-15 nations.
5. Conclusions In this chapter we have sought to show that the limitations of the unidimensional relative income approach, irrespective of whether a national or EU-wide threshold is chosen, mean that a multidimensional perspective is needed to understand differences in economic vulnerability in an enlarged EU. The limitations of relative income lines have little to do with the process of enlargement. Instead, the major problem with such measures is that of the already wellknown difficulties arising from the weak association between income and deprivation
214
Christopher T. Whelan and Bertrand Maître
in the more affluent countries. However, a consequence of such difficulties is that it becomes increasingly hard to provide entirely meaningful comparison of levels of disadvantage across major regions of the EU. A switch to a focus on absolute income, or EU-wide relative income lines, will not resolve such difficulties. Here we have proposed that the solution lies in developing a more rigorous approach to the multidimensional measurement of poverty and social exclusion. Unlike the situation with relative income poverty lines, when employing a multidimensional perspective we do observe substantial regional differences in vulnerability to economic exclusion of the kind that we would anticipate. Together with such differences, we observe a remarkable similarity in the distribution of economically vulnerable over income quartiles for each of the regional clusters. However, the vulnerable groups in the more affluent clusters are much more sharply differentiated in terms of life-style deprivation and, in the case of the EU-12 HI, economic stress. Thus, while the economically excluded constitute substantially larger groups in the poorer clusters, they are much more sharply differentiated from others in the richer clusters. These results, taken together with the fact that the consequences for social cohesion of economic vulnerability is greater in the more affluent countries, bring out the importance of being able to combine both regional and social policy perspectives within the same conceptual and measurement framework.
Notes 1 This chapter draws substantially on Fahey et al. (2005). In particular the distinction between the social policy and regional perspectives is set out in more detail in that paper. It also draws on Whelan and Maître (2005a, 2007). Readers are referred to these sources for more detailed treatment of measurement and methodological issues. We would like to thank the editors, Hans-Jürgen Andreß and the remaining participants at a seminar held at the WZB in December 2005 for comments on an earlier version of this paper. We have benefited substantially from discussions with Tony Fahey relating to a number of issues in this paper. The authors remain entirely responsible for the interpretations offered in this version of the paper. 2 While the EQLS data tend to underestimate income levels the pattern of cross-national variation is very close to that exhibited by the Eurostat data. 3 See for example Förster and Pearson (2002), Eurostat (2000). 4 See the discussions in for example Atkinson et al. (2002) and Mayer (1997). 5 For a more detailed discussion of these issues see Nolan and Whelan (2007). 6 See Whelan et al. (2001). 7 See Begg and Berghman (2001: 306). 8 This perspective views poverty as a structural position rather than simply a consequence of behavioural choice. For further discussion of this issue see Somers and Block (2005: 275–276). 9 See Lazarsfeld and Henry (1968) and more recently Magidson and Vermunt (2004) and McCutcheon and Mills (1998) for discussion of latent class models. Recent applications to the analysis of social exclusion include Moisio (2004) and Dewilde (2004), Whelan and Maître (2005a and 2005b). 10 This model misclassifies approximately 2 per cent of cases in all clusters other than the CC-3 where it rises to 6 per cent. The poorer performance in the latter is primarily a reflection of the fact that the model fails to take into account that for those outside the most deprived quartile the impact of income position is rather uneven. 11 These conclusions are based on the relevant odds ratios. 12 For a discussion of levels of social cohesion and appropriate indicators see Chiesi (2002) and Friedkin (2004). 13 However, differences between economic clusters cannot be simply accounted for corresponding differences in economic vulnerability. For a discussion of the complex issues involved in interpreting such differences see Inglehart and Klingeman (2000), Hagerty and Veenhoven (2003), Frey and Stuzer (2002) and Easterlin (2005).
Poverty, deprivation and vulnerability
215
References Ashworth, K., Hill, M.S. and Walker, R. (2000) ‘A new approach to poverty dynamics’, pp. 210–229, in D. Rose (ed.), Researching Social and Economic Change: The Uses of Household Panel Studies, London: Routledge. Atkinson, A., Cantillon, B., Marlier, E. and Nolan, B. (2002) Social Indicators: The European Union and Social Inclusion, Oxford: Oxford University Press. Bane, M.J. and Ellwood, D.T. (1986) ‘Slipping into and out of poverty: the dynamics of spells’, Journal of Human Resources, 21: 1–23. Begg, I. and Berghman, J. (2001) ‘The future role of the EU in dealing with social exclusion: policy perspectives’, pp. 306–327, in D.G. Mayes, J. Berghman and S. Salais (eds), Social Exclusion and European Social Policy, Cheltenham: Edward Elgar. Chambers, R. (1989) ‘Vulnerability: how the poor cope’, Editorial IDS Bulletin, 20, 2: 1–7. Chiesi, A.M. (2002) ‘Social cohesion and related concepts’, in N. Genov (ed.), Advances in Sociological Knowledge, Paris: International Social Science Council. De Haan (1998), ‘Social exclusion: an alternative concept for the study of deprivation?’, IDS Bulletin, 29, 1: 10–9. Dewilde, C. (2004) ‘The multidimensional measurement of poverty in Belgium and Britain: a categorical approach’, Social Indicators Research, 68: 331–369. Duncan, G.J., Gustafsson, B., Hauser, R., Schmauss, G., Messinger, H., Muffels, R., Nolan, B. and Ray, J.C. (1993) ‘Poverty dynamics in eight countries’, Journal of Population Economics, 6: 215–234. Easterlin, R.A. (2005) ‘Feeding the illusion of growth and happiness: a reply to Hagerty and Veenhoven’, Social Indicators Research, 74, 3: 429–443. European Commission (2004) A New Partnership for Cohesion: Convergence Competitiveness Cooperation. Third Report on Economic and Social Cohesion, Luxembourg: Office for Official Publications of the European Communities. Eurostat (2000) Income, Poverty and Social Exclusion in Member States of the European Union, Luxembourg: Office for Official Publications of the European Communities. Eurostat (2005a) ‘Monetary poverty in the new member states and candidate countries’, Statistics in Focus, 05/2005, by I. Dennis and A.-C. Guio, Luxembourg. Eurostat (2005b) ‘Material deprivation in the EU’, Statistics in Focus, 13/2005, by Dennis, I. Guio, A.-C., Luxembourg. Fahey, T. (2007) ‘The case for an EU-wide measure of poverty’, European Sociological Review, 23, 1: 35–47. Fahey, T., Nolan, B. and Whelan, C.T. (2003) Monitoring Quality of Life in Europe, Luxembourg: Office for Official Publications of the European Community. Fahey, T., Whelan, C.T. and Maître, B. (2005) First European Quality of Life Survey: Income Inequalities and Deprivation, Luxembourg: Office for Official Publications of the European Communities. Ferrera, M., Matsoganis, M. and Sacchi, S. (2002) ‘Open coordination against poverty: the new EU social inclusion process’, Journal of European Social Policy, 12: 227–239. Förster, M.F. (2005) ‘The European social space revisited: comparing poverty in the enlarged European Union’, Journal of Comparative Policy Analysis, 7, 1: 29–48. Förster, M.F. and Pearson (2002) Income Distribution and Poverty in the OECD Area: Trends and Driving Forces, OECD Economic Studies no. 34, OECD: Paris. Frey, B.S. and Stuzer, A. (2002) ‘What can economists learn from happiness research’, Journal of Economic Literature, 40: 402–435. Friedkin, N.E. (2004) ‘Social cohesion’, Annual Review of Sociology, 30: 409–425. Gallie, D., Paugam, S. and Jacobs, S. (2003) ‘Unemployment, poverty and social isolation: is there a vicious circle of social exclusion?, European Societies, 5, 1: 1–31.
216
Christopher T. Whelan and Bertrand Maître
Hagerty, M.R. and Veenhoven, R. (2003) ‘Wealth and happiness revisited: growing national income does go with greater happiness’, Social Indicators Research, 64, 1: 1–27. Halleröd, B. (1996) ‘The truly poor: direct and indirect measurement of consensual poverty in Sweden’, Journal of European Social Policy, 5, 2: 111-129. Inglehart, R. and Klingemann, H. (2000) ‘Genes, culture, democracy and happiness’, pp. 165–183, in E. Diener and E.M. Suh (eds), Subjective Well-being Across Areas, Cambridge, MA: MIT Press. International Monetary Fund (2003) ‘Vulnerability indicators: a factsheet’. Online. Available http: (accessed 20 February 2007). Kangas, O. and Ritakallio, V.-M. (1998) ‘Different methods – different results? Approaches to multidimensional poverty’, pp. 167–203, in H.J. Andreß (ed.), Empirical Poverty Research in Comparative Perspective, Aldershot: Ashgate. Kelley, J. and Zagorski, K. (2005) ‘Economic change and the legitimation of inequality: the transition from socialism to the free-market in Central-East Europe’, Research in Social Stratification and Mobility, 22: 321–336. Lazarsfeld, P.F., and Henry, N.W. (1968) Latent Structure Analysis, Boston: Houghton Mifflin. McCutcheon, A. and Mills, A. (1998) ‘Categorical data analysis: log-linear and latent class models’, pp. 71–94, in E. Scarborough and E. Tannenbaum (eds), Research Strategies in the Social Sciences, Oxford: Oxford University Press. Magidson, J. and Vermunt, J. (2004) ‘Latent class models’, pp. 175–98 in D. Kaplan (ed.), Handbook of Quantitative Methodology, London: Sage. Mayer, S.E. (1997) What Money Can’t Buy. Family Income and Children’s Life Chances, Cambridge, MA: Harvard University Press. Moisio, P. (2004) ‘A latent class application to the multidimensional measurement of poverty’, Quantity and Quality: International Journal of Methodology, 38, 6: 703–717. Nolan, B. and Whelan, C.T. (1996) Resources, Deprivation and Poverty, Oxford: Clarendon Press. Nolan, B. and Whelan, C.T. (2007) ‘On the multidimensionality of poverty and social exclusion’, in J. Micklewright and R. Jenkins (eds), Inequality and Poverty Re-examined, Oxford: Oxford University Press. Ringen, S. (1987) The Possibility of Politics, Oxford: Clarendon Press. Sen, A. (2000) Social Exclusion: Concept, Application and Scrutiny, Social Development Papers no. 1, Office of Environment and Social Development Asian Development Bank. Somers, M. and Block, F. (2005) ‘From poverty to perversity: ideas, markets, and institutions over 200 years of welfare debate’, American Sociological Review, 70, 2: 260–287. Stapel, S., Pasanen, J., and Reinecke, S. (2004) ‘Purchasing power parities and related economic indicators for EU, candidate countries and EFTA: final results for 2002 and preliminary results for 2003’, Eurostat Statistics in Focus, 53/2004. Online. Available http: (accessed June 2005). United Nations (2003) Report on the World Situation: Social Vulnerability: Sources and Challenges, New York: United Nations Department of Economic and Social Affairs. Whelan, C.T. and Maître, B. (2005a) ‘Economic vulnerability, multi-dimensional deprivation and social cohesion in an enlarged European Union’, International Journal of Comparative Sociology, 46, 3: 215–239. Whelan, C.T. and Maître, B. (2005b) ‘Vulnerability and multiple deprivation perspectives on economic exclusion in Europe: a latent class analysis’, European Societies, 7, 3: 423–450. Whelan, C.T. and Maître, B. (2007) ‘Income deprivation, deprivation and economic stress in an enlarged Europe’, Social Indicators Research.
Poverty, deprivation and vulnerability
217
Whelan, C.T., Layte, R., Maître, B. and Nolan, B. (2001) ‘Income, Deprivation and Economic Strain: An Analysis of the European Community Household Panel’, European Sociological Review, 17, 4: 357–372. World Bank (2000) World Development Report 2000–01: Attacking Poverty, New York: World Bank.
9
Minimum income policies in old and new member states Bea Cantillon, Natascha Van Mechelen and Bernd Schulte
Introduction Nearly all industrialised countries have, in the second half of the twentieth century, established elaborate and comprehensive systems of social protection and now spend a considerable proportion of their national incomes on these schemes. Resources are redistributed in multiple directions: between the young and the old, between families with children, single persons and childless couples, between the healthy and the sick, between the rich and the poor, and so forth. The redistributive impact of the welfare state is so great that some population groups actually receive the largest part of their income from the public purse. Yet, a sizeable proportion still lives in economic distress in all industrial countries. This persistence of poverty in advanced welfare states casts serious doubt on the virtues of the market economy as well as on the fundamental operating procedures of the tax and transfer systems and, more specifically, of social protection schemes, in both their social insurance and social assistance components. This issue is particularly challenging in Europe, since the European social model, notwithstanding its internal divergences, is perceived on the whole to provide greater protection against social risks than other social models within the democratic, industrialised world. The European welfare states emerged within the framework of comparatively small and homogenous nation states. Consequently, substantial differences can be observed across them in terms of both social expenditure and the architecture of social protection. In this sense, it is hard to find common denominators in Europe’s social architecture. However, in a large majority of EU countries – the ten new member states included – minimum income provisions have been introduced as general safety nets for the poor. Although these schemes differ substantially in terms of generosity, eligibility rules and organisation, they would appear to be good candidates to develop into a common denominator of the European social model. The existence in a large majority of EU countries of general safety nets for the poor is indeed a significant, yet often overlooked, distinguishing feature between the social models in the European Union and, say, in the USA. In this chapter we describe the basic characteristics of minimum income schemes in the ‘old’ and ‘new’ EU member states as crucial policy tools for alleviating poverty and social exclusion. We shall demonstrate that enlargement, on the one hand, confirms the hypothesis that there is a shared European social model in goals and values, if not in the detail of instruments. On the other hand, it has not added further to the diversity of schemes within the EU. In most new member states, a general assistance scheme is in place and diversity of the schemes is a common feature of both new and
Minimum income policies: old and new member states
219
old member states. We also discuss social assistance in the light of European social subsidiarity and the new strategy of the Open Method of Coordination (OMC).
1. The emergence of universal safety nets Traditionally, in almost all countries, we may distinguish between social insurance and social assistance, which together are referred to as social protection. As a specific approach to social protection, social assistance has always played a relatively minor role, even though historically speaking it preceded social insurance. However, the balance between these two basic forms of social protection within a given system may vary quite considerably. Social protection may, for example, be strongly insurancebased, with a general social assistance scheme acting as a last resort ‘safety net’ (e.g. in Germany). Conversely, the social assistance approach may prevail, for example through specific schemes covering well-defined risks or population groups (e.g. the United Kingdom). Social assistance has been defined by the International Labour Organisation (ILO) as ‘a service or scheme which provides benefits to persons of small means as of right in amounts sufficient to meet minimum standards of need and financed from taxation’ (ILO 1942). This ILO definition contains four crucial elements: (1) benefits should be provided only to persons in need; (2) benefit amounts should be set at levels adequate for meeting ‘minimum standards of need’, implying that benefit levels depend on the definition of ‘need’ in a given social and historical context; (3) benefits should be provided as of right, meaning that they should not be granted as a kind of charity or at discretion, but as a legal entitlement, and (4) schemes should be funded from general taxation (and not from social contributions). Moreover, social assistance schemes are residual in nature, invoked when all other means to acquire a decent living standard (work, social security benefits, categorical benefits) have been exhausted. Social assistance schemes may be distinguished first of all on the basis of their being either universal or categorical. Social assistance systems based on a universal approach are targeted to all those who reside legally in a given society and are below the level of income or standard of living defined as identifying a situation of need. Social assistance schemes based on the categorical approach either offer differentiated forms of support to persons and social groups located differently in the social structure, or provide support only to special categories of people. In this chapter, we are concerned with universal or general social assistance schemes. The range of their possible beneficiaries depends on the level of diversification and the extent of other forms of social protection (i.e. unemployment benefits, disability benefits, pensions, child benefits, housing benefits). The scope of these general schemes guaranteeing minimum resources is thus doubly circumscribed: on the one hand, by exclusions in the shape of the conditions for entitlement: particularly income threshold, qualifying age, residence conditions (which causes problems for the homeless), nationality and/or length of residence in a country (e.g. 5 years in the case of Luxembourg). On the other hand, it is defined by the overall structure of social protection, by the kind, level, duration of benefits it provides to different categories otherwise at risk: elderly without an occupational pension, unemployed, single parent households, households with many children, disabled persons and so forth. Despite the great variety in European welfare states, most countries, including the newly acceded members, have a general minimum income or social assistance scheme
220
Bea Cantillon, Natascha Van Mechelen and Bernd Schulte
in place. Today, in at least 19 of the 25 member states, minimum income protection is supported by a general social assistance scheme, a non-categorical safety net covering all needy persons. The United Kingdom, Ireland, Hungary, Malta, Greece and Italy are exceptions in this respect. The United Kingdom, Ireland and Malta have no universal safety net, but various complementary categorical schemes. In the United Kingdom, for instance, non-able-bodied non-workers (i.e. working fewer than 16 hours a week) with insufficient resources can claim Income Support, able-bodied non-workers are entitled to an income-based Job Seekers Allowance, and low-income households working at least 16 hours a week are eligible for the Working Tax Credit. Because these programmes are closely aligned, they are often considered together as a universal guaranteed minimum income (e.g. Walker and Wiseman 2003). Ireland has a very similar arrangement. Greece is a more important exception, because there are only very limited categorical welfare schemes (e.g. a means-tested benefit for households in ‘mountainous and less favourable areas’) that by no means constitute a universal safety net (Matsaganis et al. 2003; Matsaganis 2005). Also in Italy there is no universal minimum income provision. In the absence of a nation-wide regulation on social assistance, regions may (though, unlike in Spain1, are not required to) establish general assistance schemes. In practice, very few such schemes exist. They are found most commonly in the Centre-Northern regions and are usually decided and organised at the municipal level. In Milan, for example, there is a general safety net, but it applies a maximum entitlement period of 3 to 12 months, which is quite exceptional (Bonny and Bosco 2002). Moreover, in Italy, most existing minimum income provisions both at the national and at the municipal level are categorical (Eardley et al. 1996; Saraceno 2002; Sacchi and Bastagli 2005). From a historical perspective, these universal safety nets constitute the final addition to the welfare state, in the Western as well as in the Eastern European member states (Loedemel and Schulte 1992). In the ‘old’ member states, however, this finalisation process began about three decades earlier than in the new member states. In most of the former countries, social insurance regulations and categorical social assistance schemes for specific groups, such as the aged and the disabled, took shape well before World War II, while general social assistance schemes were usually introduced in the 1970s (Belgium, Denmark and Ireland) and 1980s (Sweden, Finland, Luxembourg, France, some Italian and Spanish regions). Only the German, Dutch and UK minimum income schemes date back to the 1960s. A latecomer is the Portuguese system, which was established in 1996. In Eastern Europe, basic social assistance safety nets to supplement labour-centred social security were developed only recently. One of the important legacies of the socialist era was an ideological rejection and denial of poverty as a structural concern for social policy. Near-universal employment at low pay was accompanied by workbased welfare systems, which, together with subsidised prices and services, largely prevented income poverty. In accordance, policymakers saw poverty as a social pathology – experienced by individuals who for some reason were unable to work. Highly selective and stigmatised services were developed to cater for them. The shock of transition created, among other things, unprecedented levels of post-war poverty, while the changeover to democratic politics and policymaking made responding to poverty an essential but contentious area of debate and action. Early policy design tended to focus on reforming existing contributory pensions and other benefits and on introducing unemployment insurance. Poverty emerged as a
Minimum income policies: old and new member states
221
major issue only in the mid-to-late 1990s (European Commission 2003; Schulte 2002a). Subsidies for utilities were in general withdrawn and cash assistance programmes grew. Most Eastern European income support measures for working-age people experiencing poverty were introduced in the 1990s. Today, in such member states as Slovakia, Slovenia and the Czech Republic, expenditures on combating social exclusion through means-tested cash benefits are, however, as high as or even higher than those in the old member states, at least as a proportion of GDP (see Figure 9.1). Their part in overall poverty reduction is probably substantial, as unemployment is quite high, while the old state-run services for the poor have disappeared and the coverage of work-based protection is very low, due to the growth of employment in the informal economy, that does not offer any social security protection.
2. Diversity in eligibility rules, organisation and administration The existing general social assistance schemes in Europe display substantial variation, both within and between ‘new’ as well as ‘old’ member states. In this chapter, we discuss the diversity that exists among member states in relation to decentralisation, entitlement conditions and adjustment mechanisms for benefit amounts in social
Mt At Pt Lt EU-25 Ee Be Se Ie Fi Nl Lu Fr De Cz Si Sk
0,00
0,10
0,20
0,30
0,40
0,50
0,60
0,70
Figure 9.1 Expenditure in % of GDP, social exclusion (NEC), means-tested cash benefits, 2003 Source: Eurostat, ESPROSS (Eurostat website, 24.11.2006)
0,80
222
Bea Cantillon, Natascha Van Mechelen and Bernd Schulte
assistance programmes. Other differences with regard to eligibility, that we do not consider here, concern the degree to which potential beneficiaries may appeal against what they consider an unjust denial, and the way in which the beneficiary’s household is accounted for. First, social assistance schemes are usually regulated at the national level, but administered at the local level. They are, in fact, one of the most decentralised items of the welfare state (Saraceno 2002). The involvement of local authorities facilitates the alignment of social assistance provisions to the regional situation and individual needs. Yet, the degree to which decentralisation affects the uniformity of rules and implementation differs greatly, in particular in the ‘old’ member states. Social assistance schemes are nationally standardised and highly codified – permitting little discretion with regard to eligibility and amounts to local administrators – in France, Denmark, Germany, the Netherlands, and the UK (see Paugam 2003; Trickey 2000). Highly decentralised systems are in place in Austria and Spain. National laws in these countries only lay down the main principles. Specific eligibility requirements and benefit levels vary between regions and municipalities. In other countries (e.g. Belgium, Finland, Portugal, Sweden), the eligibility criteria and benefit standards are determined by national law, while supplementary benefits (e.g. for rent, heating or health costs) are granted at the discretion of local administrators. Eastern European social assistance schemes are also regulated at the national level, although the decisionmaking powers of local governments are often more substantial. For example, in Poland national legislation defines minimum and maximum benefits, but it is up to local Social Assistance Centres to decide how much a particular family should receive (OECD 2004). Further decentralisation is on the political agenda, as it is in Slovakia (World Bank 2007). In Latvia, the nationally established Guaranteed Minimum Income is merely a basic rate and municipalities are free to increase benefit amounts in accordance with local needs and financial possibilities (Gassmann 2005). The case of Latvia illustrates quite clearly the risks associated with decentralising both financial responsibilities and power of decision on social assistance schemes: the deprived regions, where poverty is most widespread, are least able to provide an adequate safety net (World Bank 2007). Second, benefit levels and eligibility rules vary considerably from country to country. With regard to eligibility rules, differences in relation to age, residence, nationality, willingness to work and the ways in which resources are evaluated are significant. The enlargement of the EU hardly added to this diversity. For example, many ‘old’ as well as ‘new’ member states have not set a minimum age (this is the case, for example, in the Nordic countries, Germany, the Czech and the Slovak Republics, Estonia, Latvia, Lithuania) and, in those that have, it is usually 18 years (e.g. Belgium, Ireland, Hungary, Poland). However, in France, Luxembourg and Spain, the under-25s are excluded from social assistance, unless they have children of their own. Third, there is substantial variation not only in generosity (see below) but in the way benefits are kept in line with the evolution of general living standards. There are basically four mechanisms to adjust social assistance benefits: ad hoc revisions and regular adjustments based on a price index, a wage index or a consumption index. Eastern European member states do not differ substantially from their Western counterparts in terms of their use of these mechanisms. The most commonly used – both among the old ‘old’ and the ‘new’ member states – are price indices, usually the Consumer Price Index (see Table 9.1). In five out of ten ‘old’ member states and in
Minimum income policies: old and new member states
223
one out of 10 ‘new’ member states, gross wages constitute the basis for benefit adjustments. In the Slovak Republic, just as in Germany and Sweden, the adjustment mechanism for social assistance payments is based on surveys regarding consumption. Social assistance schemes without any automatic price or wage linking exist in Estonia, Latvia, Lithuania, Ireland and Portugal.
3. The generosity of European safety nets Figure 9.2, based on OECD data, compares the generosity of net social assistance benefits for two family types among 19 EU member states. The purchasing power (in purchasing power parities (PPP)) of households on social assistance varies considerably from country to country. For lone-person households of working age, the net social assistance amount is about EUR 54 a month in Latvia and EUR 1,002 in the Netherlands. For lone-parent households, net social assistance varies from EUR 117 in Latvia to EUR 1,613 in Denmark. If we compare net assistance incomes with the incomes of households on one average wage, we also notice considerable – though less pronounced – differences in benefit levels. For lone-person households of working age, the proportion is below 30 per cent in the Slovak Republic, Portugal, Spain and Hungary, and about 60 per cent in the Netherlands and Denmark. In the case of a lone parent, this proportion varies from roughly 35 per cent in Hungary and the Slovak Republic to about 70 per cent in Denmark, Germany and the Netherlands. In all ‘new’ member states displayed in Figure 9.2, purchasing power of households on social assistance is significantly lower than in the ‘old’ member states (with the exception of Portugal and Spain). Compared to average wages, however, the generosity of Polish social assistance for lone parents is comparable to generosity in France, Luxembourg and Ireland, while benefits in the Czech Republic are close to those in Sweden and Finland.
Table 9.1 Statutory adjustment mechanisms relating to social assistance benefit standards in EU countries, 2004 Old EU member states Price index Austria Belgium Denmark Finland France Germany Ireland Luxembourg Netherlands Portugal Spain Sweden UK
Wage index
New EU member states Consumer survey
x x x x x x x x x x
x
x
x x
Price index Cyprus Czech Rep. Estonia Malta Latvia Lithuania Poland Slovak Rep. Slovenia
x
Sources: Cantillon et al. 2004, European Commission 2004
Wage index
Consumer survey
x x x x x x
Lv Lt Sk Hu Pt Pl Cz Sp Fr Be At Se Fi Uk Ge Ir Lu Dk Nl
Monthly net social assistance (in Euro (PPP))
… 72
Single person
122 134 192 211 241 345 565 578 698 703 737 790 792 875 967 969 1002
0
250
500
750
1.000
1.250
1.500
1.750
Net social assistance in % of APW Sk Pt Sp Hu Ir Cz Pl Be Fr Uk Fi At Lu Ge Se Dk Nl
Single person
21 24 25 26 30 30 31 37 43 45 48 49 51 51 52 59 61
0
Lv Lt Sk Hu Pt Sp Cz Lu Be Fr At Pl Uk Se Ge Fi Nl Ir Dk
20
40
60
80
100
Adequacy of net social assistance (in % of poverty line)
30 37
Single person
38 40 48 55 64 72 76 77 85 92 96 98 102 111 120 121
126
20
40
60
80
100
120
140
Figure 9.2 Monthly net disposable income of social assistance recipients (working age), 2004 Source: Own calculation based on OECD benefits and wages series (OECD website 27.3.2006)
Minimum income policies: old and new member states Lv Lt Hu Sk Pl Cz Pt Sp Fr Se Be Fi At Ge Ir Nl Uk Lu Dk
Monthly net social assistance (in Euro (PPP))
117 217
Lone parent
234 249 434 516 534 572 939 1000 1084 1149 1185 1216 1222 1347 1384
1538 1613
0
250
500
750
1.000
1.250
1.500
1.750
Net social assistance in % of APW Hu Sk Sp Cz Se Fi Be Ir Pl Pt Lu Fr Uk At Nl Ge Dk
36 39 52 57 58 58 60 60 61 61 61 65 66 67 67 71
0
Lv Hu Sk Sp Lt Lu Fr Pt Cz Se Be At Ge Nl Uk Ir Fi Pl Dk
20
40
60
80
100
Adequacy of net social assistance (in % of poverty line)
40 44
Lone parent
49 57 70 72 80 83 86 87 89 90 98 101 105 105 108 118
131
20
Figure 9.2, cont’d.
Lone parent
35
40
60
80
100
120
140
225
226
Bea Cantillon, Natascha Van Mechelen and Bernd Schulte
The ranking of countries does not always correspond to what one would expect on the basis of the traditional welfare state classification, according to which the social democratic welfare states of Scandinavia are the most generous, followed by the so-called corporatist welfare states of the Netherlands, Belgium, Germany, France and Luxembourg, and with the liberal Anglo-Saxon and Southern European welfare states being the least generous. This should not come as a surprise, since the traditional welfare state classification is based on the design of social security systems rather than on the characteristics of social assistance programmes. Existing welfare state classifications do not, as a rule, take into consideration the new EU member states (an exception is Deacon et al. 1992). The purchasing power of social assistance recipients in these countries is usually very low, offering a comparable protection level to those seen in Portugal and Spain, or even lower. However, compared to average wages or relative poverty standards, minimum income protection in these countries does not always seem undersized. More specifically, the net income of Polish lone parents on social assistance amounts to 60 per cent of average wage, which is higher than in either Sweden or Finland. The level of generosity of social assistance benefits is also determined by the implicit or explicit equivalence scales used to asses both equivalent thresholds and equivalent benefits for households of different sizes (Bonny and Bosco 2002). Indeed, as the two top graphs in Figure 9.3 show, social assistance regimes (in conjunction with family allowances and housing benefits) differ significantly in the way they treat families with children relative to childless families (Bradshaw and Finch 2002). All new member states in these graphs – Hungary, Latvia, Lithuania, Poland, Slovakia and the Czech Republic – have higher net incomes for families with children on social assistance than for childless families (as a percentage of the respective poverty lines); in the Netherlands and Ireland, the opposite is the case. Indeed, in the Eastern European countries, child benefits are an important tool in the fight against poverty (see also World Bank, 2007). Figure 9.2 also compares benefit levels to the relative poverty threshold. Measured in terms of the 60 per cent poverty line, it appears that only Denmark, Finland, Ireland and the Netherlands offer socially adequate benefits to single persons as well as to lone-parent households. In four other countries (Poland, Germany, the UK and Sweden), benefits are close to adequate, falling short of this poverty line by about 10 per cent. In Austria, Belgium, the Czech Republic, France, Luxembourg and Spain benefits lie more than 10 per cent below the poverty threshold. In Hungary, Latvia, Lithuania, the Slovak Republic and Portugal, the protection level amounts to only 50 per cent of the poverty line or less. In Figure 9.4, average benefit levels for 4 family types (singles and couples, with 2 children and without children) are plotted against poverty risk, measured as 60 per cent of median equivalised disposable income, and against the poverty gap of the working-age population, as measured by the difference between the median equivalised disposable income of persons below this poverty threshold and this poverty threshold expressed as a percentage of the poverty threshold. Benefits below the poverty threshold presumably have no impact on the poverty risk of the working population, but they do affect the poverty gap. Although empirical research by Behrendt 2002), Hölsch and Kraus (2004), Nelson (2004) and Sainsbury and Morissens (2002) demonstrates that the contribution of social assistance schemes to poverty reduction is quite substantial in most countries, the pattern of the relationship
Minimum income policies: old and new member states
227
Dk
125
Pl Uk
Nl
Ge
100 Cz
Pt
75
Lt
50
Sk
Be At
Ir
Se
Fr
Sp
Pl
Fi
Couple + 2 children
Single + 2 children
125
Lu
Lv
Uk
Pt Cz Lu
75
Ir
Fi
Dk
100 At
Nl
Se Ge
Be Fr Sk
50
Sp
Hu
Hu
25
25 25
50
75 100 Single no children
125
125
25
50
75 100 Couple no children
125
125
Pl
100
Fi
Dk
Pt At
75
Lu
Cz Fr
Sk
50
Se
Uk Ge Nl
Be
Sp
Couple no children
Couple + 2 children
Ir Ir
Fi
100
75 Pt
50
Be At
Lu
Fr
Ge Uk
Sp
Sk
Hu
Cz
Nl Dk
Pl Se
Hu
25
25 25
50
75 100 Single + 2 children
125
25
50
75 100 Single no children
125
Figure 9.3 Cross-country correlations between net incomes of various household types on social assistance, 2004 (net income as a % of the poverty line) Source: Own calculations based on OECD benefits and wages series (OECD website 27.3.2006)
in Figure 9.4 shows that neither the poverty rate nor the poverty gap depend solely (or even partially) on the generosity of social assistance benefits. On the one hand, many countries with below-average poverty risks indeed provide rather generous benefits (e.g. Denmark, Finland and the Netherlands). However, generous social assistance benefits cannot always guarantee low poverty risks or a narrow poverty gap (see Poland and Ireland). The effectiveness of social assistance schemes on poverty reduction also depends on eligibility rules and on take-up of social assistance. If eligibility rules exclude large segments of the population, for instance through stringent means or work-testing, the impact of generous minimum incomes on poverty rates is attenuated. And even if social assistance schemes provide adequate benefits for all those in need, poverty may still occur if benefits are not claimed by those entitled to them. In the absence of comparative research on eligibility rules and take-up rates, it is impossible to tell whether these factors can explain the position in Figure 9.4 of countries such as Ireland and Poland, where generous social assistance benefits are combined with high poverty risks for the working-age population. Perhaps this counterintuitive observation reveals more about the quality
Sk
20
100%
At risk of poverty rate (age 16-64)
Gr
Pt Lv
It
Sp
Ir Uk
Lt
15
Pl
Average Fr
Ge
Be Hu
Lu
At
Nl Fi Dk
Se
10
Cz
5 0
20
40
60
80
100
120
Adequacy net social assistance (in % of poverty line) (average 4 families)
100%
45 Relative at risk of poverty gap (age 16-64)
Sk
Pt
30 Lv Hu
Sp
Se Be
Lt
Ge
Pl
Dk Average
At
Fr Lu
Wc
N1
Cz
Ir
Fi
15
40
60
80
100
120
Adequacy net social assistance (in % of poverty line) (average 4 families)
Figure 9.4 Net disposable income of social assistance recipients (working age, average of four family types), 2004 (as a % of poverty line and poverty risks) Sources: Own calculation based on OECD, Benefits and wages series (OECD website 27.3.2006) (Monthly net disposable income of social assistance recipients) and Eurostat, Laeken indicators (Eurostat website, 7.9.2006). (At-risk-of-poverty rate, Relative at-risk-of-poverty gap, Poverty line (= 60% of equivalised median income, using equivalence factors 1, 0.5 and 0.3) and Eurostat, Purchasing Power Parities (Eurostat website 27.3.2006)) Note: Except 2003 for at-risk-of-poverty rates and poverty gap for Czech Republic, Latvia, Lithuania, the Netherlands, Portugal and the UK.
Minimum income policies: old and new member states
229
of survey data on low-income households than about the exclusion of the poor from social assistance schemes. Conversely, low social assistance benefits do not necessarily result in high poverty risks. If other income arrangements, such as social insurance schemes that compensate for income losses due to disability, sickness, old age and unemployment, are very effective, then poverty risks might be modest or even low despite low social assistance benefits. We know, for example, that in Belgium and Germany other social transfers effectively lower the poverty rate, and social assistance schemes play only a minor role in poverty reduction (Nelson 2004, Sainsbury and Morissens 2002). Although social transfers play a significant role in reducing the Czech poverty rate, it is the country’s relatively equal distribution of primary income that seems to contribute most to the low level of income poverty (Mora 2006). Finally, it should be considered that differences in generosity are not restricted to amounts only. They include also the range of additional benefits to which recipients are entitled. In many old and new social insurance based member states, social assistance recipients enjoy free health insurance or free medical services, or receive supplementary allowances for substantial health costs.
4. General social assistance and ‘the active welfare state’ Minimum incomes are, in principle, intended merely as a temporary measure to alleviate exceptional situations the recipient may be expected to escape from, with or without further assistance, and particularly through paid work. Still, the temporary nature of social assistance benefits is rarely imposed by means of a statutory maximum duration, except in a number of Italian municipalities (see above) and in some central and eastern European countries such as Latvia, where social assistance entitlement is restricted to 9 months per year. But in all countries working age recipients are expected to be available for work or vocational training; they must be actively seeking work and, in principle, be willing to accept any suitable job. Exceptions are granted in the case of illness, disability, old age or caring for young children or disabled adults. With the economic and budgetary constraints all countries have been facing since the 1990s, the availability-for-work criterion applied in minimum income schemes has taken on renewed significance. Legal provisions now refer to being actively available for work, while the concept of ‘suitable’ or ‘appropriate’ jobs is interpreted more loosely. Since the 1990s, activation has become the catchword in the social policies of EU member states and indeed across Europe. In the Eastern European countries, and in those old member states where social assistance was introduced in the 1980s and 1990s (France, Luxembourg and Portugal), the activation concept has been ingrained in the system. In Eastern European countries, with each reform of social assistance programmes, the number of activation measures tends to increase further. In Latvia, for example, activation allowances were introduced with the 2002 reforms. Although policy discourses are apparently similar across Europe, policy practices are still very diverse. For example, while Danish social assistance beneficiaries are activated after 3 (under 30-year-olds) or 12 months (aged 30 and over), through a wide range of activation measures (individual job training, rehabilitation programmes, pre-rehabilitation programmes, flex jobs, light jobs and so forth), in other countries the work test merely requires them to register as unemployed and to
230
Bea Cantillon, Natascha Van Mechelen and Bernd Schulte
confirm that they are actively looking for work. Even policy goals may vary across Europe, depending not only on the situation in national labour markets, but also on the type of national welfare state and prevailing social citizenship culture (Torfing 1999, Taylor-Gooby 2004, Castles 2004, Barbier 2005). Barbier, for example, distinguishes between two ideal types. The first emphasises incentives, sanctions and inwork benefits, and is mainly found in Anglo-Saxon countries. The second ideal type is universalistic: it provides relatively generous and unconditional income support for the poor. Availability for paid work is requested, but not strictly enforced. It also provides a variety of services for the unemployed, including training and placement in protected, more or less temporary, jobs. Thus, in this case, too, paid work is perceived and proposed as the main road to social integration and fully active citizenship. However, this goal is in principle achieved by providing opportunities and skills rather than through financial incentives and sanctions. According to Barbier, the continental welfare states do not have a model of their own, but rather combine elements of these two ideal types. Yet, it is precisely in some of the continental welfare states, and particularly in France, that the goal of ‘activation’ was first programmatically included in a minimum income scheme – the RMI. Moreover, in the original RMI design, social integration was not restricted to insertion in the labour market. And activation – becoming active in the social assistance relationship – affected social services as much as it did social assistance beneficiaries.
5. EU anti-poverty and social inclusion policies In 1992, the European Council enacted a recommendation on common Criteria Concerning Sufficient Resources and Social Assistance in Social Protection Systems.2 It aimed at combating social exclusion by recognising the individual’s fundamental right to sufficient resources and assistance to live in a manner compatible with human dignity. This right should also provide access to health protection and improve access to rights, services and benefits needed for economic and social integration. The recommendation called upon the member states to cover all exclusion situations as broadly as possible. To this end, the member states are called upon to recognise the basic right of a person to sufficient resources and social assistance to live in a manner compatible with human dignity. Every person should have access to such right, subject to active availability for work or for vocational training. The level of minimum income should take account of living standards and price levels in the country concerned. This national minimum should not be seen as a strict subsistence line, but as a means to help ensure that even the poorest citizens are integrated into society, social integration being considered key in the European Union’s policy on social assistance. The minimum protection to which the recommendation refers is merely residual, as it applies only if all other rights have been exhausted. It is an ultimate safety net, but at the same time an effort should be made to reintegrate the poorest people into the system of general rights. This may be achieved through support measures such as integration into the labour market or vocational training. This recommendation has proved ineffective insofar as the European Community was not able to force the member states to introduce a system of universal social assistance. The current, more pragmatic, approach no longer focuses on policy instruments, but on common objectives. With regard to social assistance and social inclusion, this has meant a shift from the notion of a harmonised system of minimum
Minimum income policies: old and new member states
231
income protection to a common striving for lower financial poverty rates. To this end, a consultation method between member states has been introduced that is known as the Open Method of Coordination (OMC) (Schulte 2002b, 2005a, 2005b; Zeitlin et al. 2005). The procedure of the OMC consists in a fixed cycle of analysis and reporting. The basis of this cycle is formed by National Action Plans (NAPs). In these periodic reports, the member states are required, first, to outline the present situation and, second, to evaluate the policies pursued with a view to improving that situation and to indicate which policy changes they intend to implement in the short term. The European Commission subsequently summarises these national reports in a Joint Report in which it indicates what it considers to be good practices. In this manner, the Commission provides a starting point for a learning process between the member states as well as processes of peer-review and benchmarking. This method has been developed gradually. At the Lisbon summit (2000), the OMC was extended successively to the policy areas of social inclusion, pensions, health and education (Schulte 2005a; Zeitlin et al. 2005). With the open method of coordination, Europe today finds itself in a rather different situation than at the time when the notion of social security harmonisation was being debated at the EU level. The OMC social inclusion has, however, the potential to generate a dynamics that would make it possible to arrive at binding European agreements regarding common minimum standards. This presupposes that the link between policy outcomes (poverty as measured by 60 per cent of national median incomes) and income protection become stronger (Atkinson 2004; Cantillon 2004). Yet, the OMC on social inclusion has thus far not been very successful in stimulating an overall rethinking of policies and evidence-based evaluation processes. Due also to the lack of sanctioning power by the Commission in this field, most national reports on social inclusion (NAPincls) are merely descriptive, the link between policy and outcomes being rather weak. Neither the social partners nor the national and sub-national parliaments, or indeed the media, have devoted much attention to the process (Marlier et al. 2006). In a report on the future of social policy in the enlarged Europe, a group of experts argues in favour of a basic income for every child (Report of the High Level Group on the Future of Social Policy in an Enlarged Union 2003). This income would be determined relative to the economic circumstances of each member state and it would have to be paid for by the member states, in accordance with the available tools, either in the shape of child benefit, or as a tax credit or social benefit. This proposition obviously reverts to the 1992 recommendation.
6. Conclusion In this chapter, we have considered universal social assistance schemes in the EU after enlargement. Today, in at least 19 of the 25 member states, minimum income protection is supported by benefits provided as of right for all citizens. Although in Eastern Europe basic social safety nets were developed only recently, enlargement did not add to the degree of diversity within Europe. Certainly the systems display substantial variation in terms of generosity, eligibility rules, organisation and administration, but this is equally true for ‘old’ and ‘new’ member states. According to OECD data, the generosity of Polish social assistance relative to average wages is comparable to generosity in France, Luxembourg and Ireland, while benefit levels in the Czech Republic are close to those in Sweden and Finland.
232
Bea Cantillon, Natascha Van Mechelen and Bernd Schulte
Compared to the official EU poverty line (defined as 60 per cent of median equivalised income) only Denmark, Finland, Ireland and the Netherlands provide adequate minimum income support. In four other countries (Poland, Germany, Sweden and the UK) benefits are close to being adequate, falling short of the poverty line by only about 10 per cent. In all other countries, however, benefits drop more than 10 per cent (Belgium, the Czech Republic, France, Luxembourg and Spain) and sometimes over 50 per cent below the poverty line (Hungary, Latvia, Lithuania, the Slovak Republic and Portugal). The existence in a large majority of EU countries of general minimum income provisions for the poor is an important, yet often neglected, distinguishing feature between the social models of the European Union on the one hand and that of, for instance, the USA on the other. Therefore, these schemes would appear to be good candidates to develop into a common denominator of the so called ‘European Social Model’. However, to this end general safety nets should be implemented in all member states and benefit levels should be adequate. The open method of coordination social inclusion has the potential to lead to a dynamics that will encourage member states to move in that direction. In addition, the activation of the ’92 ‘Recommendation on common criteria concerning sufficient resources and social assistance’ could prove very helpful.
Notes 1 The Spanish case represents a different exception: a national law stipulates the autonomous regions’ responsibility to set in place such a scheme, but schemes vary from a high degree of regulation and institutionalisation to a high degree of discretionary powers, not so different from the Italian situation, although things are in a continuous process of change and there is a kind of virtuous competition between regions (Arriba and Moreno 2005). 2 O.J. NO. L 245/1992.
References Arriba, A. and Moreno, L. (2005) ‘Spain: poverty, social exclusion and “safety nets”’, pp. 141–203, in M. Ferrera (ed.), Welfare State Reform in Southern Europe, London: Routledge. Atkinson, A.B. (2004) ‘Could the open method of coordination lead to a basic income for Europe?’, in B. Cantillon and J. Vandamme (eds), The Open Method of Coordination and Minimum Income Protection in Europe: Liber Memorialis Herman Deleeck, Leuven: Acco. Barbier, J.-C. (2005) ‘Citizenship and the activation of social protection: a comparative approach’, pp. 113–134, in J.G. Andersen, A.-M. Guillemard, P.H. Jensen and B. Pfau-Effinger (eds), The Changing Face of Welfare, Bristol: Policy Press. Behrendt, C. (2002) At the Margins of the Welfare State, Aldershot: Ashgate. Bonny, I. and Bosco, N. (2002) ‘Income support measures for the poor in European cities’, pp. 81–125, in C. Saraceno (ed.), Social Assistance Dynamics in Europe, Bristol: Policy Press. Bradshaw, J. and Finch, N. (2002) A Comparison of Child Benefit Packages in 22 Countries, UK Department for Work and Pensions, Research Report no 174, Leeds: Corporate Document Services. Cantillon, B. (2004) ‘Minimale inkomensgarantie in Europa tussen subsidiariteit en federalisme’, in B. Cantillon and J. Vandamme (eds), The Open Method of Coordination and Minimum Income Protection in Europe: Liber Memorialis Herman Deleeck, Leuven: Acco.
Minimum income policies: old and new member states
233
Cantillon, B., Van Mechelen, N., Marx, I. and Van den Bosch, K. (2004) ‘De evolutie van de minimumbescherming in 15 Europese welvaartsstaten in de jaren negentig’, Belgisch Tijdschrift voor Sociale Zekerheid, 46, 3: 513–549. Castles, F.G. (2004) The Future of the Welfare State: Crisis Myths and Crisis Realities, Oxford: Oxford University Press. Deacon, B., Castle-Kanerova, M. and Manning, N. (1992) The new Eastern Europe London: Sage. Eardley, T., Bradshaw, J., Ditch, J., Gough, I. and Whiteford, P. (1996) Social Assistance in OECD Countries: Synthesis Report, London: HSMO. European Commission (2003) Social Protection in the 13 Candidate Countries: A Comparative Analysis, Brussels. European Commission (2004) Social Protection in the Member States of the European Union, of the European Economic Area and in Switzerland, Brussels: European Commission. Gassmann, F. (2005) ‘How to improve access to social protection for the poor? Lessons from the social assistance reform in Latvia’, paper presented at the conference on Social Protection for Chronic Poverty, IDPM, 23–24 February, Manchester. Hölsch, K. and Kraus, M. (2004) ‘Poverty alleviation and the degree of decentralization in European schemes of social assistance’, Journal of European Social Policy, 14, 2: 143–164. ILO (1942) ‘Approaches to social security: an international survey’, Montreal. Loedemel, I. and Schulte, B. (1992) ‘Social assistance: a part of social security or the poor law in new disguise?’, paper presented at the conference Social Security Fifty Years after Beveridge, September, University of York. Marlier, E., Atkinson, A.B., Cantillon, B. and Nolan, B. (2006) The EU and Social Inclusion: Facing the Challenges, Bristol: Policy Press. Matsaganis, M. (2005) ‘Greece – fighting with hands tied behind the back: anti-poverty policy without a minimum income’, pp. 33–83, in M. Ferrera (ed.), Welfare State Reform in Southern Europe, London: Routledge. Matsaganis, M., Ferrera, M., Capucha, L. and Moreno, L. (2003) ‘Mending nets in the south: anti-poverty policies in Greece, Italy, Portugal and Spain’, Social Policy & Administration, 37, 639–655. Mora, M. (2006) ‘Social cohesion in the Czech Republic: a blessing or a trap?’ ECFIN Country Focus, III: 1–6. Nelson, K. (2004) ‘Mechanisms of poverty alleviation: anti-poverty effects of non-means tested and means-tested benefits in five welfare states’, Journal of European Social Policy, 14: 371–390. OECD (2004) Benefits and Wages. Country-chapters Information: Poland, Paris, OECD. Paugam, S. (2003) ‘The Revenu Minimum d’Insertion (RMI) in France: the limits of a progressive social policy’, pp. 29–54, in G. Standing (ed.), Minimum Income Schemes in Europe, Geneva: International Labour Office. Sacchi, S. and Bastagli, F. (2005) ‘Italy – striving uphill but not stopping halfway: the troubled journey of the experimental minimum insertion income’, pp. 84–140, in M. Ferrera (ed.), Welfare State Reform in Southern Europe, London: Routledge. Sainsbury, D. and Morissens, A. (2002) ‘Poverty in Europe in the mid-1990s: the effectiveness of means-tested benefits’, Journal of European Social Policy, 12: 307–327. Saraceno, C. (2002) ‘Introduction: exploring social assistance dynamics’, pp. 1–34, in C. Saraceno (ed.), Social Assistance Dynamics in Europe, Bristol: Policy Press. Schulte, B. (2002a) ‘Die Transformation von Systemen sozialer Sicherheit in Mittel- und Osteuropa. Die Rolle des europäischen Gemeinschaftsrechts am Beispiel des europäischen koordinierenden Sozialrechts’, pp. 649–669, in W. Boecken, F. Ruland and H.-D. Steinmeyer (eds), Sozialrecht und Sozialpolitik in Deutschland und Europa. Festschrift für Bernd Baron von Maydell, Luchterhand: Neuwied.
234
Bea Cantillon, Natascha Van Mechelen and Bernd Schulte
Schulte, B. (2002b) ‘Die “Methode der offenen Koordinierung” – Eine neue politische Strategie in der europäischen Sozialpolitik auch für den Bereich des sozialen Schutzes’, Zeitschrift für Sozialreform, 48: 1–27. Schulte, B. (2005a) ‘The open method of co-ordination as a political strategy in the field of immigrant Integration policy’, pp. 114–121, in R. Süßmuth and W. Weidenfeld (eds), Managing integration: The European Union’s Responsibilities towards Immigrants, Gütersloh: Bertelsmann-Stiftung. Schulte, B. (2005b) ‘Die “offene Methode der Koordinierung” (OMK) als politische Strategie in der europäische Sozialpolitik’, Sozialer Fortschritt, 54: 105–113. Taylor-Gooby, P. (ed.) (2004) New Risks, New Welfare: The Transformation of the European Welfare State, Oxford: Oxford University Press. Torfing, J. (1999) ‘Workfare with welfare: recent reforms of the Danish welfare state’, Journal of European Social Policy, 9: 5–28. Trickey, H. (2000) ‘Comparing workfare programmes: features and implications’, pp. 249–294, in I. Loedemel and H. Trickey (eds), An offer you can’t refuse. Workfare in international perspective, Bristol: Policy Press. Walker, R. and Wiseman, M. (2003) The Welfare We Want: The British Challenge for American Reform, Bristol: Policy Press. World Bank (2007) Social Assistance in Central Europe and the Baltic States, Washington, DC: World Bank. Zeitlin, J., Pochet, P. and Magnusson, L. (2005) The Open Method of Coordination in Action: The European Employment and Social Inclusion Strategies, Brussels: PIE Peter Lang.
10 Housing conditions1 Henryk Doma´nski
Introduction Housing conditions are an important contributor to quality of life and an important axis of differentiation between social conditions in the ‘old’ EU and the new member states (NMS). As the following chapter on the institutional context of housing systems in the NMS makes clear (Norris, this volume, Ch. 11), the former communist countries of central and Eastern Europe under-invested in housing over many years and made poor use of the limited funds they did provide for the sector. It is hardly surprising to find, therefore, that most of the former communist NMS suffer from poor housing conditions. One of our concerns here is to document how far this is so but also to assess to what degree poor housing conditions are equally found in the EU-15, particularly in the southern EU states. A second concern is to document the peculiar differences in housing tenure that now characterise the east–west divide in Europe. As communism collapsed in Central and Eastern Europe in the late 1980s and early 1990s, it left a distinctive legacy in the housing system in the form of the transformation of housing tenure wrought by the transition. State-owned housing, which in most of the countries accounted for the bulk of the housing stock, was suddenly transferred to ownership of the sitting occupants, typically at very low or even zero purchase prices. Thus owner-occupation became the dominant housing tenure virtually overnight in some NMS countries, especially in Poland, Slovakia and the Baltic countries (Shinozaki 2005). However, this transformation did not occur to the same degree in all countries, nor did it uniformly displace other tenures, so it is useful also to examine tenure patterns in the NMS and compare them with those of the EU-15. Based on the 2003 European Quality of Life Survey, this chapter examines these aspects of housing in the new member states and candidate countries, viewed in comparison to the EU-15. While the EQLS deals with housing only as one among a number of dimensions of quality of life and therefore is limited in the range of housing indicators it provides, it nevertheless is a valuable source for present purposes because of the standardised coverage it provides for 28 European countries. The present chapter draws on five housing indicators available from this source: housing tenure, dwelling size (as measured by number of rooms per person), dwelling defects (in relation to leaking windows, rot in woodwork, indoor toilet and affordable heating), neighbourhood environment, and subjective satisfaction with housing.
1. Tenure In developed western countries, home ownership is usually taken to indicate well-being and high socio-economic status and to contribute to creating an ownership mentality.
236
Henryk Doma´nski
Some have argued that home ownership has given rise to a new social cleavage, that between home owners and non-home owners, which has replaced the ‘old’ class system (Saunders 1990). In the former communist NMS, one would not expect the same associations to hold, given the sudden manner in which mass home ownership came about as the old state-dominated housing systems were dismantled. Our data on ownership structures in Europe provide only a tentative basis for comparison, since different pathways to ownership and related financial obligations exist in different countries. This is especially so in connection with ownership acquired through mortgage purchase which in post-communist countries is underdeveloped. The EQLS allows us to map variations of house ownership in European societies in a general way (see Table 10.1). Contrary to what Western Europeans might expect home ownership is more widespread in the new member states (NMS) and in the candidate countries (CC) than in the old member states of the EU (see Figure 10.1). In the former EU-15 scarcely 60 per cent of the adult population are home owners, and there are several countries where less than half of the respondents own the home they live in. In the NMS the home ownership ratio exceeds 70 per cent on average, and the Czech Republic and Latvia are the only countries with ownership ratios below 50 per cent. This leads to the conclusion that in the wealthier western countries home ownership is less widespread than in the less affluent new member states. The home ownership ratio among even the poorer people in the NMS – those in the lowest income quartile – is quite high, even by comparison with richer people in the old member states – those in the highest income quartile of the old member states (67 as compared to 70 per cent). An analysis by type of ownership casts more light on this question. While ownership without a mortgage is in most countries more common than ownership with a mortgage, fully fledged proprietors are more frequently found in eastern European countries, where roughly two-thirds of householders own their homes without a
EU-15
NMS
EU-25
CC-3
0%
10%
20% own outright
30%
40%
own (mortgage)
50%
60%
rent (private)
70% rent (public)
80%
90%
other
Figure 10.1 Tenure status according to accommodation, by country grouping Source: Table 10.1
100%
Housing conditions
237
Table 10.1 Housing tenure (%) Ownership Total
Tenant in Without With private mortgage mortgage sector
Tenant in social/ voluntary/ municipal housing
Rent Total free/ other total
EU-15 Denmark Finland Sweden Ireland UK Austria Belgium France Germany Luxembourg Netherlands Greece Italy Portugal Spain
60 63 67 60 73 59 51 70 48 45 77 48 68 76 59 76
38 11 40 21 39 24 30 37 34 27 54 4 62 63 41 52
22 51 27 39 34 35 21 33 14 18 23 44 6 13 18 24
22 12 14 24 13 13 17 15 31 34 17 6 29 15 24 17
15 20 17 13 12 26 26 9 17 19 2 42 1 5 8 2
4 5 2 3 3 2 6 6 4 2 5 4 3 4 9 4
100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
NMS Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania
72 79 77 86 45 90 70 82 82 49 89
66 60 68 80 41 76 68 74 73 47 83
6 19 10 6 4 14 2 8 9 2 5
4 9 13 5 9 3 3 3 8 12 6
19 1 3 5 38 4 23 7 4 28 3
5 11 7 4 9 3 3 8 6 10 3
100 100 100 100 100 100 100 100 100 100 100
CC-3 Bulgaria Romania Turkey
67 86 84 57
66 86 83 56
1 0 1 1
19 4 4 27
2 2 1 2
13 9 11 14
100 100 100 100
Source: EQLS 2003: Q18: ‘Which of the following best describes your accommodation? 1 Own without mortgage (i.e. without any loans), 2. Own with mortgage, 3. Tenant, paying rent to private landlord, 4. Tenant, paying rent in social/voluntary/ municipal housing, 5. Accommodation is provided rent free, 6. Other, 7. (Don’t know)’
mortgage, as compared to only 38 per cent in the EU-15. In contrast, ownership with a mortgage is more widespread in the EU-15 countries, which not only have a more developed financial sector but also more affluent consumers whom private banks consider worthy of receiving loans. The main impediments to the development of the mortgage lending market in Eastern Europe are, among other things, the immature stage of banking legislation (which, for example, limits the proportion of the market value of dwellings that banks can lend) and the underdevelopment of mortgage banks with a resulting dominance of universal banks in mortgage lending (Merrill and Kozlowski 2001). Figure 10.2 provides an illustration of this tendency: the percentage
238
Henryk Doma´nski
60 Denmark
% ownership with mortgage
50
Netherlands
40
Sweden UK Ireland Belgum
30
Finland Luxembourg
Spain
20
Cyprus Portugal
France
Hungary
10
Malta Estonia Slovakia Slovenia Greece Lithuania Czech Republic Romania Latvia Poland Turkey Bulgaria
0 0
50
100
150
Austria Germany Italy
200 250 300 GDP per capita
350
400
450
500
Figure 10.2 Relationship between Gross Domestic Product per capita and percentage of ownership with mortgage Sources: Ownership: EQLS 2003; GDP per capita: Eurostat news release 14/5/2004
of owners with a mortgage is plotted against GDP per capita across 28 countries (GDP values are presented in terms of the Purchasing Power Standards (PPS) to eliminate differences in price levels between countries). It shows that with rising GDP the percentage of owners with a mortgage clearly increases.2 The higher prevalence of fully fledged proprietors in the NMS countries, of course, is not a reflection of their purchasing power but of the privatisation policies that accompanied the post-communist transformation in the early 1990s. New ownership laws were enacted that allowed people to buy the accommodation they lived in on very favourable terms, while at the same time unburdening public authorities of responsibility for the quality of housing. Subsidies for the construction of state owned housing, which had previously driven the bulk of housing output, were largely withdrawn, and in the beginning of the 2000s only 10 per cent of housing output in these countries was publicly funded (Shinozaki 2005). Rapid privatisation can also explain why home ownership does not depend on family income and occupational status in the NMS. Table 10.2 makes clear that home ownership varies with occupational category and income in the EU-15 but not in Eastern Europe. The lack of an income effect on tenure in post-communist countries reflects the greater importance of political as opposed to market influences on the distribution of home ownership in those countries. Across occupational categories, farmers are the most likely to own their homes in both the EU-15 and NMS, with professionals/managers and the self-employed coming in second and third place. At the same time, while working-class categories in general have less access to home ownership, the gap for non-skilled workers is small in the NMS, where 72 per cent are home owners, and is
Housing conditions
239
Table 10.2 Persons living in own homes, by age, income, area of residence and occupational status (%) CC-3
NMS
EU-15
EU-25
Age 18–34 35–64 65 and over
57 61 89
58 71 78
44 56 67
47 58 68
Income quartiles Highest 2nd 3rd Lowest
68 67 66 71
75 75 68 67
70 63 54 43
69 64 58 48
Area Rural Urban
81 60
83 61
70 53
72 54
Household occupational status Professional managerial Other non-manual Self-employed Farmers Skilled workers Non-skilled workers
68 70 67 83 76 64
81 70 79 95 66 72
69 56 71 82 57 47
70 58 72 85 59 51
Source: EQLS 2003
wide in the EU-15. This again indicates that in Western European societies home ownership tends to be more correlated with social position than in post-communist countries. The two other associations presented in Table 10.2 are worth noting. One might expect ownership to be related to age, because it is an important factor in the process of accumulation of wealth over a person’s lifetime. Table 10.2 shows that this association is present in all the major European regions. Detailed cross-country analysis (not presented here) reveals that deviations from this rule occur only in Italy, Slovenia, and Cyprus, where the association between age and ownership is weaker. Place of residence also matters: home ownership prevails in the rural areas and is less frequent in cities.
2. Living space The accommodation space available to householders is difficult to measure in general population surveys, especially across countries, since what counts as housing space varies between countries and householders often lack precise data on the size of their dwelling. As a fall-back, survey researchers often use the number of rooms as a proxy for dwelling size, and this is the measure of living space used in the EQLS. We must keep in mind, however, that this measure is highly approximate, as it takes no account of how room sizes may vary in different housing systems and of the consistency with which householders report spaces such as bathrooms, hallways and storage areas as rooms. Historical data suggest that there is a long-standing disparity in housing space, measured as the number of persons per room, in dwellings in Western and Eastern European countries. As shown in Table 10.3 sizeable differences date back at
240
Henryk Doma´nski
least to the 1950s or 1960s, when the average number of persons per room was much higher in what are now the NMS. In the post-war period only Finland and Greece ranked as low as the NMS countries. The historical data also show, however, that the number of persons per room declined throughout all European countries and that the east–west gap is narrower now than it was around the 1970s. Yet despite this trend, the size of dwellings in Eastern Europe continues to be smaller than in the west in 2003, with a particularly severe shortage of space in Poland. The persons per room data can be supplemented by statistics on floor space as measured in terms of square metres presented in the last column of Table 10.3. One can see that number of rooms translates into floor space. The pattern of international differentiation looks mostly the same with Denmark followed by Sweden, the UK, and Netherlands on the top of this ranking. Residents of Eastern Europe – with Poles having smallest floor area per person – fare worst. As the shortage of housing space in the NMS coincides with the relatively high incidence of ownership, it adds to the overall impression that home ownership differs in meaning across Europe. The smaller size of homes in the NMS has it roots in the rapid pace of industrialisation of predominantly agrarian societies and in the massive
Table 10.3 Average floor space Average density (persons/room) 1950–1962 Denmark Finland Sweden Ireland UK Austria Belgium France Germany Luxembourg Netherlands Greece Italy Portugal Spain Malta Czech Republic Hungary Slovak Republic Slovenia Poland Bulgaria
0.7 1.3 0.8 0.9 0.7 0.9 0.6 1.0 0.9 0.8 0.8 1.5 1.1 1.1 1.1 1.3 1.3 1.5
2003
(1960) (1960) (1960) (1961) (1961) (1961) (1960) (1962) (1960) (1960) (1956) (1961) (1961) (1960) (1950) (1957) (1961) (1963)
0.6 0.8 0.7
0.7 0.9 1.2
1.7 (1960) 1.8 (1956)
1.5 1.1
0.5 0.7 0.5 0.7 0.6 0.7 0.6 0.9 0.7 0.9
Floor area per person (in m2) 2001–2003 50.6 36.3 44.0 35.0 44.0 38.3 37.5 40.1 41.0
34.3 28.2 28.0 26.0 29.5 22.2
Sources: Ratios of persons/room for the 1950s and 1960s are from United Nations Statistical Yearbook 1966, 702–771. For 2003 they were calculated from EQLS 2003: Q17: How many rooms does the accommodation in which you live have, excluding the kitchen, bathrooms, hallways, storerooms and rooms used solely for business? For no.of persons: HH1: Including yourself, can you please tell me how many people live in this household? Data on floor space in m2 are from: Housing Statistics in the European Union, 2005. National Board of Housing and Planning. Boverstat’
Housing conditions
241
mobility from rural to urban areas which started in Eastern Europe in the 1950s. This created a strong demand for the quick construction of new accommodation. This goal was realised by building houses from standardised, prefabricated elements and by restricting their space as much as possible. The latter was linked to the ideological view of the communist welfare states that larger dwellings were not a priority, because many family functions were taken over by state institutions and placed outside private homes. In some of these countries the official regulations provided norms of only 7–10 sq m per person as an adequate living space. The permanent undersupply of investment funds, materials, energy, parts, machinery and tools – coupled with the socialist state acting as a main investor – resulted in a constant shortage of apartments. The introduction of capitalist markets in the 1990s has not reduced this shortage in a significant way up to now. Even though it lacks information on the exact size of dwellings, the EQLS does give data on the number of persons per room (Table 10.4). This measure reveals the persisting dichotomy in the size of accommodation between Eastern and Western Europe, though with some overlap at the margins. In EU-15 countries there is on average 0.7 persons per room, compared to 1.2–1.3 for the NMS and for the three candidate countries. Within the EU-15 countries the best conditions are in Belgium and UK (0.5) and the worst in Southern Europe, where Greece, Italy, Portugal and Spain have between 0.8 to 0.9 persons per room. In the NMS only Malta surpasses the low Southern European standards. All other countries have distinctly more persons per room. Poland is the only country which reaches 1.5 persons per room. The particularly low position of Poland may be related to the massive destruction and devastation of housing during World War II and to the necessity after the war to provide accommodation at the lowest possible cost. The size of accommodation varies strongly across age groups. Older people have relatively more and younger people less spacious accommodation. The younger generation is particularly disadvantaged in Poland where the standard of one person per room is reached only by those aged above 65 years. Strong space deficits are also experienced by the younger age categories in Hungary, Latvia and Lithuania. The number of persons per room is correlated with income, a tendency which is stronger in the NMS. In the EU-15 the number of rooms per person is 0.8 among those in the lowest quartile, and decreases to 0.6 in the highest bracket. Within the NMS differences by income range from 1.6 in the bottom to 1.1 in the top quartile. Urban–rural differences are not very marked in either group of countries.
3. Housing deficits The size of dwelling is not necessarily linked to standards in other dimensions of housing. The EQLS allows us to inspect the prevalence of five housing deficits, namely: (1) perceived shortage of space; (2) leaking windows, (3) rotting doors, window frames and floors, (4) lack of indoor flushing toilet, and (5) lack of funds to ‘keep [the] home adequately warm’. According to the data in Table 10.5, there are consistently fewer complaints on these issues in the EU-15 than in the NMS and CC-3. Poor housing conditions are especially acute in the Baltic States (Estonia, Lithuania, Latvia), with Turkey, Romania, Poland and Bulgaria almost as bad. Only the Czech Republic and Slovenia come close to EU-15 standards. With the exception of Portugal, where damp and rot
242
Henryk Doma´nski
Table 10.4 Number of persons per room, by ownership status, income, region and age Total Tenure
Income quartiles
Region
Age
Non- Owners Lowest 2nd 3rd Highest Rural Urban 18–24 25–49 50–64 65+ owners EU-15 Denmark Finland Sweden Ireland UK Austria Belgium France Germany Luxembourg Netherlands Greece Italy Portugal Spain
0.7 0.6 0.8 0.7 0.7 0.5 0.7 0.5 0.7 0.6 0.7 0.6 0.9 0.9 0.9 0.8
0.7 0.6 0.7 0.6 0.7 0.5 0.7 0.5 0.6 0.6 0.6 0.6 0.9 0.8 0.8 0.8
0.7 0.7 0.8 0.8 0.8 0.5 0.8 0.6 0.7 0.7 0.8 0.6 1.0 0.9 0.9 0.9
0.8 0.7 0.9 0.8 0.8 0.5 0.9 0.6 0.8 0.7 1.0 0.8 1.2 1.2 1.0 0.9
0.7 0.7 0.8 0.8 0.7 0.5 0.8 0.5 0.7 0.7 0.7 0.6 1.0 0.9 0.9 0.8
0.6 0.6 0.8 0.6 0.7 0.5 0.6 0.5 0.6 0.6 0.6 0.5 0.9 0.8 0.8 0.8
0.6 0.6 0.7 0.6 0.5 0.4 0.6 0.4 0.5 0.5 0.5 0.5 0.8 0.6 0.8 0.6
0.7 0.6 0.7 0.7 0.7 0.5 0.7 0.5 0.6 0.6 0.6 0.6 0.9 0.8 0.8 0.8
0.7 0.6 0.8 0.7 0.7 0.5 0.8 0.5 0.7 0.7 0.8 0.6 0.9 0.9 0.9 0.8
0.8 0.7 0.9 0.9 0.8 0.6 0.9 0.6 0.8 0.8 0.9 0.7 0.9 1.1 1.0 0.9
0.8 0.8 0.9 0.8 0.8 0.5 0.8 0.6 0.7 0.7 0.8 0.7 1.1 0.9 1.0 0.9
0.6 0.5 0.6 0.5 0.6 0.5 0.6 0.4 0.6 0.6 0.6 0.5 0.9 0.8 0.8 0.8
0.5 0.4 0.6 0.5 0.4 0.4 0.6 0.3 0.4 0.5 0.4 0.4 0.7 0.6 0.6 0.6
NMS Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania
1.3 0.8 0.7 1.1 0.9
1.2 0.8 0.7 1.0 0.8
1.4 0.8 0.8 1.3 1.0
1.6 0.8 0.9 1.2 1.1
1.2 0.8 0.7 1.1 0.8
1.2 0.7 0.7 1.1 1.0
1.1 0.5 0.6 0.9 0.8
1.3 0.9 0.7 1.1 0.9
1.2 0.8 0.7 1.1 1.0
1.4 0.9 0.8 1.1 1.1
1.4 0.9 0.8 1.2 1.0
1.1 0.7 0.7 1.0 0.8
0.9 0.6 0.6 0.8 0.7
1.2 1.5 1.0 0.9 1.2 1.2
1.1 1.4 1.0 0.9 1.1 1.1
1.7 1.7 1.4 1.1 1.3 1.5
1.6 1.8 1.5 1.0 1.3 1.6
1.3 1.5 0.7 0.8 1.4 1.0
1.0 1.4 1.0 1.0 1.1 1.1
1.0 1.2 0.9 0.9 1.1 1.1
1.2 1.5 1.0 0.8 1.1 1.1
1.1 1.4 1.0 1.0 1.3 1.2
1.5 1.6 1.2 1.2 1.5 1.4
1.4 1.6 1.2 1.0 1.4 1.4
1.0 1.3 0.8 0.8 1.0 0.9
0.9 1.0 0.6 0.7 0.9 0.7
CC-3 Bulgaria Romania Turkey
1.2 1.1 1.1 1.3
1.2 1.0 1.0 1.3
1.3 1.4 1.3 1.3
1.7 1.3 1.3 1.9
1.3 1.3 1.3 1.2
1.0 0.7 0.8 1.1
0.9 1.1 1.0 0.8
1.3 1.0 1.1 1.5
1.2 1.1 1.1 1.2
1.3 1.3 1.2 1.3
1.4 1.2 1.2 1.4
0.9 0.9 0.9 0.9
0.9 0.8 0.8 1.0
All countries 0.9
0.8
0.9
1.1
0.9 0.8
0.7
0.8
0.9
1.0
1.0
0.7
0.6
Sources: EQLS 2003: Q.17, HH1 (as in Table 10.3)
are reported as often as in Eastern European countries, standards in the EU-15 countries do not vary widely. There is a clear relationship between housing conditions on the one hand, and area of residence, household income quartile and occupational category on the other. Detailed analysis not shown here reveals that problems with rot and damp tend to be more common in rural areas, especially in eastern European countries and Turkey, and they are also more frequently reported by people with lower incomes. This tendency is clearly marked in almost all countries except Denmark, Germany and Slovakia. As regards occupational category, rot and damp problems, not surprisingly, are least frequent among professionals and self-employed. Conversely, they are much more common among workers, especially unskilled workers. Both tendencies hold true for the EU-15, the NMS and CC-3. As far as distribution according to age category is concerned, no consistent patterns emerge.
Housing conditions
243
Table 10.5 Households that reported deficits in accommodation (%) Shortage of Rot in Damp space window, doors and leaks or floors
Lack of indoor At least two flushing toilet problems
EU-15 Denmark Finland Sweden Ireland UK Austria Belgium France Germany Luxembourg Netherlands Greece Italy Portugal Spain
17 19 22 20 17 22 14 14 21 11 25 16 21 20 25 14
8 5 8 2 9 7 5 9 11 4 5 9 11 12 16 5
12 11 15 6 13 8 8 13 14 10 7 11 19 13 40 14
11 1 2 1 2 1 1 3 1 1 — 2 4 1 5 2
91 7 10 3 10 7 5 9 12 5 — 7 13 11 24 7
NMS Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania
24 17 13 15 15 18 30 13 30 29 26
25 15 21 14 6 24 28 41 40 32 35
19 20 31 13 13 15 21 13 31 29 19
10 4 1 5 5 8 11 7 17 20 25
22 17 19 11 9 17 25 20 36 31 30
CC-3 Bulgaria Romania Turkey
31 21 28 33
30 19 30 31
30 25 29 31
21 30 39 11
32 26 35 31
Source: EQLS 2003 Note: 1 Luxembourg is excluded from the EU-15 and EU-25 mean as the data on ‘indoor flushing toilet’ are inadequate
Table 10.6 shows the proportion of households which do not have an indoor flushing toilet, lack of which may be regarded as indicative not only of poor housing conditions but also a low standard of living. The differences between countries are substantial, with a basic division between the EU-15 as one group and the NMS and CC-3 as another. Within the EU-15 a greater number of households in Portugal, Greece and Belgium do not have a toilet but, even in these countries, the number does not rise above 5 per cent. The only NMS/CC-3 countries where lack of an indoor toilet is equally rare are Malta, Cyprus, the Czech Republic and Slovenia. The numbers in the remaining countries are much higher, reaching 20 per cent in Latvia and Lithuania and 30 per cent in Romania and Bulgaria. Cross-country differences concerning the lack of a toilet clearly correspond to other housing problems, for example, rot and damp.
244
Henryk Doma´nski
Table 10.6 Households that do not have an indoor toilet, by age of respondent and area of residence (%) Total
CC-3 NMS EU-15 EU-25
21 10 1 3
Age
Area
18–24
25–34
35–49
50–64
65+
15 6 1 2
18 9 1 2
20 9 1 2
26 12 2 3
35 14 2 4
Rural Urban 48 15 2 4
7 6 1 2
Source: EQLS 2003
The lack of an indoor flushing toilet is associated with area of residence in that it is generally seen more often in the countryside. The urban–rural divide appears to be stronger in Eastern European countries. In the rural areas of some countries (Bulgaria, Lithuania, Romania), one in two households lack an indoor toilet, while in Latvia, Estonia, and Turkey the proportion is around one-third. These data emphasise that it is in rural areas in the NMS/CC-3 that poor housing is a particular problem. Age is a much more important axis of differentiation in Eastern Europe than in the EU-15, with older respondents in the east much more likely to suffer from the lack of an indoor toilet especially in Latvia, Estonia, Lithuania, Bulgaria and Romania. In order to derive an overall picture from the various dimensions of housing conditions, we can construct an aggregate (country-level) measure of good housing based on the items examined so far. This measure is defined as the percentage of people who perceive none of the five deficits (lack of space, leaks, rot, no toilet, inadequate heating) and have at least one room per person. The results of this measure are presented in Table 10.7. Three main conclusions can be drawn from this comparison of housing conditions in old and new member states. First, it confirms the gulf in average housing standards between the EU-15 and the east: a two-thirds majority of the EU-15 citizens has good housing, compared to only one-third in the NMS and less than one-fifth in the CC-3. Second, home owners tend to have better housing conditions than non-owners in old and new member states alike. However, the difference in favour of owners is larger in the EU-15. Owners in the NMS and CC-3 live in even poorer housing conditions than non-owners in the EU-15. Third, fewer of those in the highest income quartile in the NMS (and of the candidate countries) have good housing conditions (42 per cent) than those in the lowest income quartile in EU-15 countries (54 per cent). The table thus reiterates the housing paradox in the new member states where a high level of home ownership is associated with a low quality of housing. The table also shows, however, that the contrast between new and older member states should not be exaggerated, because there are some borderline countries that blur the distinction. Thus, the Czech Republic and Cyprus have similar housing conditions to those of Italy and Spain, whereas Portugal and Greece have poorer housing standards than the leading countries in the NMS. However, not even Portugal, the country with the poorest quality in the old member states, comes close to the poor standards in the three Baltic nations and in Poland, where less than one-quarter of the population have satisfactory housing conditions by our yardstick.
Housing conditions
245
Table 10.7 Good housing (having at least one room per person and perceiving none of five housing deficits) (%) Country
All
Ownership
Income quartiles
Region
Age
No
Yes
Lowest 2nd 3rd Highest Rural Urban 18–24 25–49 50–64 65+
EU-15 Denmark Finland Sweden Ireland UK Austria Belgium France Germany Luxembourg Netherlands Greece Italy Portugal Spain
66 70 63 74 67 67 74 70 64 78 62 70 49 59 31 59
56 57 61 68 47 60 66 50 49 71 50 64 43 39 13 46
72 78 65 79 75 72 82 79 80 88 66 77 51 66 43 63
54 68 61 67 43 56 63 55 53 80 43 59 30 24 13 45
58 65 60 71 57 64 74 66 53 64 59 62 39 54 16 58
71 71 60 78 68 73 72 69 71 84 57 72 50 65 32 64
76 73 69 80 87 71 83 71 73 84 67 83 52 80 55 75
68 77 65 74 65 70 77 69 69 83 61 71 47 62 31 61
64 66 62 74 70 66 71 72 60 75 63 70 49 57 30 58
61 69 62 60 67 64 62 68 53 76 53 67 45 52 39 64
59 57 53 63 64 59 68 62 53 69 56 64 46 56 34 57
72 81 72 87 69 72 86 78 71 87 69 76 55 61 27 60
77 87 77 89 78 83 82 82 85 89 76 84 50 68 22 62
NMS Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania
32 57 45 50 58
24 33 34 35 50
35 63 48 53 68
18 42 31 41 44
24 43 53 41 60
35 68 44 51 55
42 82 53 69 69
29 52 43 50 61
35 58 46 51 56
29 60 51 50 47
27 56 45 46 52
38 56 44 61 69
41 58 42 49 73
41 22 37 24 25 14
29 10 17 22 22 11
42 28 41 25 30 14
28 8 24 10 14 5
24 15 26 25 18 11
47 26 41 24 25 12
55 32 49 30 40 28
35 18 35 18 21 12
46 27 40 27 28 15
32 22 39 22 27 14
31 18 36 22 23 12
53 25 43 25 27 17
52 32 32 28 27 13
CC-3 Bulgaria Romania Turkey
18 16 12 21
14 8 10 15
20 17 12 26
7 4 5 8
8 9 6 9
21 18 18 22
37 29 22 45
12 12 9 14
22 19 15 24
19 9 13 22
16 17 12 17
21 16 11 28
21 17 11 32
Source: EQLS 2003: Rooms per person: Q17, HH1 (as in Table 10.2), deficits: Q19.1. + Q19.2. + Q19.3. + Q19.4. ‘Do you have any of the following problems with your accommodation? 1. Shortage of space 2. Rot in windows, doors or floors 3. Damp/leaks 4. Lack of indoor flushing toilet. & Q20.1. There are some things that many people cannot afford, even if they would like them. For each of the following things on this card, can I just check whether your household can afford it if you want it? 1. Keeping your home adequately warm’
The breakdowns in Table 10.7 by various social categories help to identify who the most privileged groups in the housing sector are. Rural–urban differences are substantial in the three candidate countries (where 12 per cent of rural households have good housing compared to 22 per cent of urban households) and are present but less marked in the NMS (with 29 per cent of good housing in rural areas compared to 35 per cent in urban areas). In the EU-15, rural–urban differences are not only narrower but also are reversed: rural households are slightly more likely than urban households to have good housing (68 per cent versus 64 per cent respectively). Differences with respect to income, age and ownership status are also important and also vary by region. In the NMS and the CC-3 income differentials are more marked than in the EU-15. This is related to the fact that in the more affluent of the EU-15 countries, good housing as we define it has become very widespread, with only small
246
Henryk Doma´nski
% of those having at least one room per person and perceiving none of five deficits
minorities falling below our good housing threshold. Even in the lowest income quartile of EU-15 countries more than half of the people enjoy good housing conditions by our standard. In contrast, not even half of those in the highest income quartile in NMS (and CC-3) can boast of satisfactory standards. The poorest strata in Sweden, Denmark, Germany and Austria fare better than the top income category in Lithuania, Malta, Poland, Latvia, Estonia, Bulgaria, Romania or Turkey. In old and new member states alike, quality of housing substantially increases with age so that older categories are more privileged than younger people. Only in the new member states, however, do huge majorities of the young and old lack adequate housing standards. The differences in housing standards in old and new member states reflect differences in national wealth as well as different policy legacies. As Figure 10.3 shows, there is a clear positive association between GDP per capita and our measure of the quality of housing. The corresponding regression line indicates that every one unit of increase in GDP per capita (measured in purchasing power parities) increases the expected percentage of people living in high-quality housing by 0.41 of a percentage point. However, the figure also shows that several western countries have even higher standards of housing than their level of wealth would suggest, whereas Poland, the Baltic nations, and the three candidate countries have even lower standards than their level of GDP would make us predict. This indicates that policy legacies such as differences in the degree of market reliance also have an impact on housing conditions.
100 90 80
Germany Austria Sweden Belgium Netherlands Denmark United Kingdom France Ireland Spain Italy Finland Czech Republic Cyprus
70 60
Luxembourg
SIovenia
50
Malta Hungary
40
Slovakia
30
Latvia
Greece
Portugal Turkey
20
Estonia Poland
Buglaria Lithuania
10
Romania
0 0
50
100 150 GDP per capita (PPS, EU-25 = 100)
200
250
Figure 10.3 Relationship between GDP per capita and index of housing quality Sources: GDP per capita: Eurostat news release 14/5/2004; index of housing quality calculated from EQLS as in Table 10.7
Housing conditions
247
4. Local environment The quality of housing is related not only to properties of the dwelling but also to the quality of the local environment. Such characteristics as quality of water and air, level of noise, access to green areas, personal safety, and accessibility of local infrastructure (roads, shops, offices, schools, etc.) clearly impact on the quality of life. Satisfaction of these needs requires an appropriate ecological policy by the state, investments in infrastructure, and other macro-systemic conditions that highly depend on economic development, the political system and the cultural heritage of various countries. The inferior ecological situation in Eastern Europe dates back to pollution of the environment arising from rapid industrialisation in the communist era. Pressure to increase production, the use of outdated technologies, and a lack of consideration for environmental issues led to the growth of many zones of ecological risk. For a long time health risks linked to pollution were largely unrecognised, as state authorities were not interested and the ecological awareness of people was low. It may seem paradoxical nowadays that for many years fuming factory chimneys were considered to be the symbols of progress and growth. The collapse of the communist system initially made some aspects of the social environment worse by weakening established social controls and social ties, and by contributing to increased crime rates and feelings of insecurity, especially in urban areas. East–west differences in local environmental conditions are clearly reflected in our data. Respondents were first presented with a list of four aspects of their environment – noise, air pollution, lack of access to recreational or green areas, and water quality – and were asked to indicate if they had ‘very many reasons to complain’, ‘many reasons to complain’, ‘a few reasons’, or ‘no reason to complain’ about each of these. A fifth dimension, safety at night, was measured by the question: ‘How safe do you think it is to walk around in your area at night?’ (four categories from ‘very unsafe’ to ‘very safe’). All variables were here scaled on a 4-point scale from 3 to 0. Summing up the scores gives us a summary index of the quality of the environment which ranks from 0 to 15 – where lower values indicate less complaints and thus better quality. Table 10.8 presents the mean values of this summary measure. The data show that the big international contrast in housing conditions noted earlier between the NMS and CC-3 on the one hand and the EU-15 on the other is blurred in the case of environmental problems, as the west–east difference is now complemented by a stark contrast between the north and south within the EU-15. Southern Europeans have many more complaints about their environment than people living in the North. Among those with high levels of dissatisfaction concerning noise, air pollution, lack of access to green areas, etc., we now find countries such as Italy, Greece, France, and Portugal alongside not only Cyprus and Malta but also Lithuania, Latvia, Hungary, Romania, and Turkey. On the other hand, environmental conditions pose much less of a threat for Czechs, Slovaks, Estonians, or Slovenians, who appear to be on a par with inhabitants of Belgium, Luxembourg, and Spain. Thus, east–west differences are here cross-cut by a south–north axis. On average, however, citizens of the NMS perceive the local environment less favourably than EU-15 citizens, and once more even those in the highest income quartile of the NMS consider themselves worse off than those in the bottom quartile in the EU-15. These findings are consistent with other data. According to statistics collected by the European Environmental Agency emissions from manufacturing
248
Henryk Doma´nski
Table 10.8 Complaints about environment (summary scale from 0 (no complaints) to 15) Country
Total
Ownership
Income quartiles
Region
Age
No
Yes
Lowest 2nd 3rd Highest Rural Urban 18–24 25–49 50–64 65+
EU-15 Denmark Finland Sweden Ireland UK Austria Belgium France Germany Luxembourg Netherlands Greece Italy Portugal Spain
3.4 1.0 1.7 1.6 2.2 2.5 2.0 3.3 4.3 2.2 3.4 1.9 4.9 5.6 4.0 3.9
3.7 1.5 2.2 2.2 3.2 2.7 2.3 4.0 5.3 2.6 3.6 2.2 5.9 6.3 4.5 4.1
3.2 0.7 1.5 1.2 1.8 2.4 1.6 3.0 3.3 1.7 3.3 1.6 4.4 5.4 3.7 3.8
3.8 1.1 1.8 1.7 2.8 3.4 2.2 4.2 4.4 2.5 3.3 2.2 4.7 6.6 3.9 4.1
3.6 1.1 1.6 1.7 3.2 2.8 1.8 3.8 4.2 2.5 3.5 2.1 5.4 6.1 3.8 4.3
3.1 1.0 2.0 1.5 2.1 2.1 1.9 3.1 4.3 2.1 2.6 1.7 5.1 4.7 3.9 4.2
3.0 0.9 1.6 1.3 2.0 2.2 2.2 2.9 4.3 1.8 3.2 1.6 5.6 4.8 3.8 3.7
2.6 0.5 1.3 1.0 1.6 2.1 1.3 2.8 2.9 1.6 3.0 1.4 3.0 4.5 3.6 2.8
4.0 1.3 2.2 1.8 2.8 2.8 2.7 3.9 5.6 2.7 4.1 2.3 5.9 6.5 4.9 4.3
3.3 1.2 2.0 1.8 2.3 2.5 2.0 3.3 4.9 2.0 3.6 1.7 5.4 4.5 3.4 4.0
3.5 1.1 1.8 1.6 2.3 2.7 2.0 3.5 4.8 2.0 3.2 1.9 4.9 5.7 4.2 4.0
3.3 0.9 1.5 1.4 2.1 2.4 2.0 3.2 4.2 2.3 3.4 1.9 4.8 5.4 4.4 3.7
3.4 0.9 1.6 1.6 1.7 2.5 1.8 3.0 3.1 2.4 3.7 2.0 4.5 6.1 3.7 3.8
NMS Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania
4.0 4.3 6.0 3.1 3.6
5.0 4.6 6.9 3.7 4.4
3.6 4.2 5.7 3.0 2.6
3.8 3.7 7.5 3.0 3.9
3.9 4.2 5.4 3.2 3.6
4.2 4.5 6.1 2.9 4.0
4.1 4.1 5.2 3.3 3.1
2.6 3.7 4.7 2.4 2.6
5.2 4.4 6.5 4.3 4.4
3.8 4.9 4.5 3.1 3.6
4.1 4.2 5.6 3.0 3.7
4.0 4.0 6.8 3.0 3.5
4.0 4.1 6.8 3.3 3.1
4.0 4.0 3.9 3.8 4.9 5.4
5.0 5.3 5.3 3.7 4.9 4.9
3.9 3.4 3.6 3.8 4.8 5.4
4.0 3.4 4.4 3.7 5.1 4.9
3.7 3.8 4.2 3.9 4.7 5.2
4.3 4.2 3.6 3.9 4.8 5.6
4.2 4.1 4.1 3.7 4.9 5.7
2.9 2.2 3.6 2.8 3.6 4.2
5.1 5.6 4.3 4.5 5.6 6.0
3.6 3.7 4.0 4.1 4.2 5.1
4.0 4.1 3.8 3.8 4.9 5.4
4.2 3.9 3.9 3.9 4.8 5.5
4.2 4.0 4.0 3.3 5.3 5.2
CC-3 Bulgaria Romania Turkey
5.4 4.5 4.4 5.9
6.1 4.8 4.5 6.4
5.1 4.5 4.4 5.6
4.9 4.2 3.4 5.8
5.7 4.2 4.7 6.4
5.5 4.5 5.3 5.7
5.5 5.1 4.8 5.9
3.8 3.1 3.2 4.6
6.3 5.8 6.0 6.4
5.5 4.0 4.3 6.0
5.6 4.6 4.5 6.2
5.0 4.6 4.7 5.2
4.7 4.5 4.0 5.6
Source: EQLS 2003: Q56.1 + Q56.2 + Q56.3 + Q56.4: ‘Please think about the area where you live now – I mean the immediate neighbourhood of your home. Do you have very many reasons, many reasons, a few reasons, or no reason at all to complain about each of the following problems? 1. Noise 2. Air pollution 3. Lack of access to recreational or green areas 4. Water quality + Q57: How safe you think it is to walk around in your area at night? (Very safe, rather safe, rather unsafe or very unsafe)’
industries are larger in Eastern than in Western Europe (European Environment Agency 2003). This results from, among other things, the higher network of railroads in the former communist bloc (and lower share of road transport) and from more developed environmentalist policies in the EU-15. Ground-level ozone which is detrimental to health exceeds safety norms in Eastern and Southern European countries to much higher degrees. The same pattern is displayed with respect to the contamination of drinking water supplies, where NMS and CC-3 are, for example, more intoxicated by salt. Water shortages are also more common in Southern and Eastern Europe (European Environment Agency 2003). Within single countries, perceptions of the local environment tend to be similarly stratified as other dimensions of housing quality, with some noteworthy exceptions. Thus, the perception of the environment shows only little variation by age. In the old member states people in lower-income echelons see their local environment more
Housing conditions
249
critically than the better-off, but in the NMS those with higher incomes hold even more critical views of the environment. Here we cannot determine whether such differences reflect discrepant objective conditions or rather income-specific degrees of environmental consciousness. Home ownership is associated with fewer perceived environmental problems in old and new Europe alike. In both country groups, problem perceptions also tend to be more frequent in urban than in rural areas.
5. Satisfaction with housing and subjective well-being Subjective perceptions of housing conditions tend to reflect objective conditions, but in the light of the previous findings, it is interesting to know which objective conditions impact more on satisfaction with housing in new member states, as for example between high ownership rates and poor housing standards. As might be expected, housing satisfaction is higher in the EU-15 than in the NMS. On a scale from 0 to 10 the mean satisfaction with accommodation is 7.7 in the EU-15 compared to 6.7 in the NMS and the CC-3. This difference of one full point on the scale is higher than the difference which distinguishes the highest from the lowest income quartile in the EU-15. Again we find that those in the bottom income quartile of old member states are more satisfied (7.2) than those in the highest income quartile in the NMS (7.0). The CC-3 top income quartile (7.3) corresponds to the bottom quartile in the West. The lowest levels of satisfaction with housing prevail in Latvia (6.3), Estonia (6.4), Poland and Turkey (both 6.5), the highest in Denmark (8.4), Austria (8.3) and Luxembourg (8.2). These subjective evaluations more closely reflect the differences in objective housing conditions than differences in home ownership. In order to determine the relative weight of these factors, we regressed satisfaction with accommodation on various measures of objective conditions (see Table 10.9). Assuming that satisfaction results also from the quality of general infrastructure we added quality of public services to the model. Respondents were asked to assess quality of public services in their countries taking into account five items: health services, education system, public transport, state pension system, and social services. All variables were scaled from 1 to 10. Parameters in the first three columns of the table show how strongly satisfaction with accommodation is affected by size, quality of housing, quality of local environment, ownership, and quality of public services once other factors such as sex, age, occupational category, place of residence, and family income are controlled for.3 The parameters in each cell show the percentage decrease in the explained variance if the variable in the respective line is removed from the analysis. The clear result is that the strongest determinant of satisfaction with accommodation is the quality of housing as indicated by the absence of four serious deficits. In the NMS this quality measure explains 13.6 per cent of the variance, whereas the second most important factor – the number of rooms per person – explains only 3.6 per cent of the variance. Home ownership has comparatively little impact on housing satisfaction in Eastern Europe, explaining only 1.5 per cent of the variance, once other factors are controlled. In the old member states, the quality of housing is also the most important determinant of satisfaction explaining 8.1 per cent of the variance. In the west, however, ownership precedes the other factors as the second most important determinant of housing satisfaction, and its impact is four times greater than in the NMS. The quality of local environment and quality of public services are more important in old
250
Henryk Doma´nski
than in new member states, but in both contexts they are less important than characteristics of the dwelling. Interestingly, in both the NMS and EU-15, satisfaction with housing is more strongly affected by assessments of the broader services context defined in terms of quality of health services, education, transport, etc., than by quality of the physical environment. Other factors such as sex, age, occupation, place of residence or family incomes do not affect satisfaction with accommodation in a significant way once the quality of housing is controlled for. We may conclude that while home ownership is more widespread in the NMS than in the old member states, it has less impact on housing satisfaction in the east than in the west. This may be so because ownership is less a function of socio-economic status in the east and is much less associated with high quality housing than in the west. In the next step of the analysis, we ask what impact housing conditions have on general satisfaction with life. Parameters given in the last three columns of Table 10.9 compare the net effect of various aspects of housing with effects of quality of public services, incomes, occupational category, place of residence, sex and age. The result is that quality of housing plays an important role in determining general life satisfaction even if other factors such as income are controlled for. The new element is that
Table 10.9 Relationships between satisfaction with accommodation and general satisfaction with life and objective characteristics of housing Independent variables
Satisfaction with accommodation EU-15
Sex 0.7 Age 0.9 Occupation 0.7 Place of residence 0.0 Income 0.1 Ownership 6.1 Number of rooms per person 0.3 Quality of housing (rotten 8.1 windows, damp and leaks, lack of flushing toilet, problem with heating) Quality of environment (number 1.2 of complains of environment, insecure environment) Perceived quality of public 3.0 services in country (health services, education system, public transport, social services, state pension system) R2 * 100 26.4
General satisfaction with life
NMS
CC-3
EU-15
NMS
CC-3
0.0 0.3 1.1 0.0 0.1 1.5 3.6 13.6
0.1 0.4 0.4 0.1 0.3 2.6 0.8 9.8
0.0 0.9 1.0 0.0 0.9 2.4 0.2 2.3
0.0 1.3 1.7 0.1 1.7 0.3 0.4 4.8
0.0 0.4 0.8 0.1 1.3 0.8 0.1 6.8
0.3
1.2
0.4
0.5
0.4
2.8
7.3
7.7
9.7
10.9
27.6
27.4
18.5
22.6
24.4
Source: EQLS 2003: Q41 and Q31: ‘Could you please tell me on scale of 1 to 10 how satisfied you are with each of the following items, where 1 means “very dissatisfied” and 10 means “very satisfied”. Respondents were presented a list of items that included “your accommodation” and general “satisfaction with life”’ Notes: The table shows semi-partial coefficients in the OLS models, multiplied by 100.
Housing conditions
251
general satisfaction appears to be preponderantly affected by quality of public services. In case of the EU-15, it explains 7.7 per cent of variance in satisfaction and in NMS countries its net effect amounts to 9.7 per cent. However, the main difference between EU-15 and NMS lies in the net effect of home ownership. Once again we see that home ownership has a different meaning in Eastern and Western Europe. In the EU-15 it signifies higher status and higher housing quality whereas in the NMS it has a much less distinctive character. The higher share of owners is coupled with poorer housing quality and less space, and this also implies that home ownership has little impact on general life satisfaction in the NMS. One major contribution to crisis legitimacy theory is the emphasis on the creative side of social trust. Robert Putnam (1993) in his influential book Making Democracy Work argued that a civic culture of ‘generalised trust’ and social solidarity between citizens is an important societal prerequisite of working democracy. It has not yet been tested in Europe whether willingness to cooperate and a positive experience of mutual help are positively related to higher material status reflected in ownership of housing and housing conditions. Theories of legitimation crisis have originally referred to Western democracies. They may not apply to Eastern European societies bearing in mind that with the fall of the communist system, long-lasting decades of the ‘deficit’ of legitimation was ended. However, paradoxical as it may be, a deficit of legitimation has returned despite the majority support for the new democratic regimes. In sum, we expect ownership and better housing conditions produce higher trust. In order to check this hypothesis I regressed the EQLS measure of trust on ownership and housing resources, controlling for our standard set of variables (see Table 10.10). The results show a significant effect of housing conditions on interpersonal trust – the belief that people can be trusted is dependent on the number of complaints
Table 10.10 Relationships between trust and objective characteristics of housing Independent variables Sex Age Occupation Place of residence Income Ownership Number of rooms per person Quality of housing (rotten windows, damp and leaks, lack of flushing toilet, problem with heating) Quality of environment (number of complaints about environment, insecure environment) R2 * 100
EU-15
NMS
CC-3
0.1 0.1 0.4 0.3 0.2 0.4 0.1 0.6
0.0 0.0 0.9 0.1 0.0 0.0 0.0 0.8
0.1 1.2 0.4 0.0 0.1 1.4 0.4 1.2
2.1
0.9
1.8
6.0
3.7
7.9
Source: EQLS 2003: Q28. ‘Generally speaking, would you say that most people can be trusted, or that you can’t be too careful in dealing with people? Please tell me on a scale of 1 to 10, where 1 means you can’t be too careful and 10 means that most people can be trusted’ Notes: The table shows semi-partial coefficients in the OLS models, multiplied by 100.
252
Henryk Doma´nski
regarding rotten windows, damp and leaks, lack of flushing toilet, and problems with heating. In the EU-15 quality of housing explains 0.6 per cent and and in the NMS 0.8 per cent of variance in trust – these values are not big but are statistically significant. As might be expected from patterns noted earlier, home ownership matters for trust in Western Europe but not at all in the NMS.
6. Conclusion There are important differences in housing tenure and housing conditions across European countries. The basic divide runs between the old and new member states of the EU, with the latter having higher levels of home ownership but poorer housing conditions. High levels of home ownership in the NMS reflect the hurried privatisation of former social housing during the transition from communism and thus have a different social significance than the gradual build-up of private housing assets represented by home ownership in the EU-15. As far as housing conditions are concerned, cross-national variations within the former eastern bloc are also marked and parallel the differences between northern and southern countries within the EU-15 – Slovenia, which has relatively favourable conditions, is clearly different from the Baltic states and Poland, which together with the three candidate countries represent the opposite pole of adverse conditions. In general, international rankings of housing conditions tend to reflect the levels of economic development of countries so that GDP per capita explains a large portion of variation, but there are also clearly discernible traces of different policy legacies. The analysis also sheds some light on the subjective perception of living conditions and the role of housing in determining the general satisfaction with life. Reflecting differences in objective housing conditions rather than differences in home ownership, average levels of satisfaction with accommodation are much higher in Western European societies. With respect to general life satisfaction, the importance of housing for the perceived quality of life is clearly established. This underlines the potential of housing to become a focus of social conflict and the role of housing policy to contribute to the stabilisation and legitimation of political regimes. In terms of policy, the further spread of home ownership is not a crucial issue in new member states, since it is already extensive. The key challenge ahead is to stimulate investment aimed at improving the quality of housing. This will not only require economic incentives such as credits, grants or tax subsidies, but also reliable and long-term rules of housing market regulation which will encourage private investors by providing them with stable and reliable expectations. Apart from such more specific measures of housing policy, the primary condition that must be met in order to reduce the ‘housing gap’ between Eastern and Western Europe is economic development of the former. Reducing unemployment, promoting employment and increasing the wealth of the new member states is the key challenge which, when met, is likely to produce sizable spill-over effects into the housing sector.
Notes 1 I thank Dariusz Przybysz and Grazyna Drazyk for very helpful technical assistance. 2 However, as Norris (this volume, Ch. 11) discloses, although the mortgage debt to GDP ratio was much lower in the NMS than in the EU-15 in 2004, mortgage debt grew at a much faster rate in the former group of countries (by 44.4 per cent compared to 9.6 per cent).
Housing conditions
253
3 Values presented in Table 10.9 are semi-partial coefficients in regression of satisfaction with accommodation, and (separately) general satisfaction with life on 27 variables. The set of independent variables includes: 4 (dummy coded) categories of age, 6 (dummy coded) sociooccupational categories, 4 (dummy coded) categories of ownership, number of rooms per person, rotten windows, damps and leaks, lack of flushing toilet, problems with heating (all three coded 0–1), sex, family incomes, place of residence (urban–rural) and scale of complaints on environment, with the set of complaints including: A. noise; B. air pollution; C. lack of access to recreational or green areas; D. water quality. Last set of independent variables includes five items concerning assessment of public services, i.e. health services, education system, public transport, social services and state pension system. In order to facilitate the interpretation of the coefficients, all values are expressed in percentages by multiplying them by 100.
References Burns, L.S. and Grebler, L. (1986) The Future of Housing Markets: A New Appraisal, London, New York: Plenum Press. Butler, T. and Savage, M. (eds) (1995) Social Change and the Middle Classes, London: UCL Press. Castles, F.G. (1998) Comparative Public Policy. Patterns of Post-war Transformation, Cheltenham: Edward Elgar. Delhey, J. (2004) Life Satisfaction in an Enlarged Europe, European Foundation for the Improvement of Living and Working Conditions, Luxembourg: Office for Official Publications of the European Communities. European Environment Agency (2003) Europe’s Environment: The Third Assessment Summary 2003, Luxembourg: Office for Official Publications of the European Communities. Flippen, C. (2004) ‘Unequal returns to housing investments? A study of real housing appreciation among black, white and Hispanic households’, Social Forces, 82: 1523–1552. Kiel, K.A. and Mieszkowski, P. (1990) ‘An examination of systemic differences in the appreciation of individual housing units’, Journal of Real Estate Research, 5: 301–318. Merrill, S. and Kozlowski, E. (2001) ‘Developing housing finance in a transition economy: the case of Poland’, Journal of Housing Economics, 10, 3: 363–392. Myers, D. and Wolch, J. (1995) ‘Polarization of housing status’, pp. 269–334, in R. Farley (ed.), State of the Union, New York: Russel Sage Foundation. Putnam, R. (1993) Making Democracy Work: Civic Traditions in Modern Italy, Princeton, NJ: Princeton University Press. O’Rand, A.M. and Henretta, J.C. (1999) Age and Inequality: Diverse Pathways Through Later Life, Boulder, CO: Westview Press. Saraceno, C. (2005) ‘Family-work system in Europe’, pp. 57–83, in J. Alber and W. Merkel (eds), Europas Osterweiterung: Das Ende der Vertiefung, WZB Jahrbuch, Berlin: edition sigma. Saunders, R. (1990) A Nation of Home Owners, London: Unwin Hyman. Shinozaki, S. (2005) ‘A comparative assessment of housing finance markets in transition economies’, pp. 7–81, in OECD (ed.), Housing Finance Markets in Transition Economies: Trends and Challenges, Paris: OECD Publishing.
11 Institutional drivers of housing inequalities in the enlarged EU Michelle Norris
Introduction Domanski’s chapter (10) in this volume examines the data on the quality of housing and the local environment from the 2003 European Quality of Life Survey (EQLS). This analysis reveals that housing quality is lower in the 10 new member states (NMS) and three candidate countries (CC-3) than in the EU-15. However, he also highlights variations in housing quality within these groups of nations. For instance, Malta, Cyprus, the Czech Republic and Slovenia enjoy better housing conditions than the other new member states and the candidate countries, and housing standards in the southern states of the EU-15 (Greece, Italy, Portugal and Spain) are generally lower than in their northern counterparts. This chapter aims to map the institutional context within which these international patterns of housing inequality have arisen. Housing lies on the cusp of economic and social policy, and it is financed by public and private sector institutions as well as individual households. Thus the institutional factors which could potentially have impacted on these patterns are manifold. Previous research by the author on housing provision, policies and output in the EU member states and the candidate countries, which was based in part on national official statistics and data supplied by housing ministries, indicates that four institutional factors are particularly important in this regard (Norris and Shiels 2004). These are: housing tenure policy, finance and subsidy systems, construction systems and trends, and governance arrangements. The main body of this chapter is organised around discussion of these factors, while also trying to take into account the considerable degree of overlap and interaction that occurs between them. Given the primary interest of the present volume on the ‘new’ rather than the ‘old’ EU (that is, the NMS and candidate countries as opposed to the EU-15) and the limited coverage given to the ‘new’ EU in the comparative housing literature, this chapter devotes more space to discussion of this part of Europe than to the EU-15. In addition, because housing supply is slow to respond to changes in demand, considerable attention is given to the historical development of the institutional drivers of housing inequalities. There is also evidence that the differences between housing quality in different parts of Europe have widened in recent years. Tsenkova and Turner (2004: 133) report that in many of the new member states ‘housing production has remained historically low, the existing stock has deteriorated and homelessness has increased’ since the 1990s. Thus, the contemporary drivers of these patterns of housing inequality are also examined here. The concluding section considers possible implications of these inter-country housing inequalities for the future of the European Union.
Institutional drivers of housing inequalities
255
1. Housing tenure policy 1.1 Historical trends and their implications for housing inequality Variations in housing tenure have until recently been the primary focus of comparative housing policy analysis (for instance: Barlow and Duncan 1994; Harloe 1995; Oxley and Smith 1996). Review of the various typologies proffered in these studies indicates that, broadly from the end of World War II until the 1980s, three housing tenure systems prevailed in the 25 countries that are now members of the European Union – the dual and unitary systems, which Kemeney (1995) distinguishes in the EU-15, and the ‘East European model’ identified by Hegedüs and Tosics (1996). In the dual system, which operates in Ireland and the United Kingdom, home ownership is the dominant tenure, the private rented sector is both unregulated and unsupported by government, the social rented sector caters mainly for disadvantaged households and is therefore strongly residualised and often (but not always as the UK case illustrates) small in size. The social housing sectors in countries in this category are also characterised by state ownership and management, rent setting policies which are insensitive to demand, and allocation driven by bureaucratic procedures. In contrast, in the unitary system that operates in Germany and Denmark, tenure patterns are driven by household preferences rather than government interventions as the latter do not favour one tenure over another. Thus, Table 11.1 demonstrates that the owner occupied sector tends to be smaller than is the norm in dual tenure systems and the social rented sector is larger and not targeted exclusively towards low income groups. The private rented sector competes on equal terms with social renting. In this system, social housing is generally owned and managed by non-governmental, non-profit agencies, rents usually reflect the cost of providing the dwellings rather than market imperatives, but also tend to be sensitive to demand. The former communist new member states and the candidate countries of Bulgaria and Romania adhered with varying levels of definitude to a third tenure system, which Hegedüs and Tosics (1996) have termed the ‘East European housing model’. The key characteristics of this model are as follows: • •
• •
•
Housing is defined as a state guaranteed social right, rather than a commodity. The state controls the production, consumption and allocation of housing and the role of the market in this area is severely constrained. Consequently as Table 11.1 reveals, the level of homeownership was relatively low in 1980. In many instances, a large proportion of the housing stock was state owned, and indeed prior to 1990, practically all rented housing was in this category. Although state-owned housing tenure shared many of the characteristics of social housing in the EU-15 (non market allocation systems, rent regulation, subsidization, it was not necessarily allocated according to need. Housing was cheap for consumers but expensive for society because of the high level of state subsidy required, coupled with inefficient arrangements for housing production and management.
As a result of financial constraints, some communist countries began to diverge from the East European housing model from the 1960s onwards by introducing ‘marketisation’ reforms which involved the removal of barriers to the operation of the market in this sector and the re-privatisation of housing (Pinchler-Milanovich 2001).
43 38 16 40 43 30 41 61 69 25 29 24 36 Nav Nav 39 52 Nav 39 Nav Nav 21 42 42
52 59 61 40 45 63 47 39 31 75 71 76 59 Nav Nav 60 42 Nav 52 Nav Nav 73 42 58
0 0 0 13 1 0 0 0 0 0 0 0 0 Nav Nav 0 0 Nav 0 Nav Nav 0 16 0
5 3 23 7 1 7 12 0 0 0 0 0 5 Nav Nav 1 0 Nav 5 Nav Nav 6 0 0
41 33 13 40 40 25 39 58 74 24 26 18 26 79 Nav 30 55 Nav 28 28 28 15 44 35
Rented
Cooperative Other
Rented
Owner occupied
1990
1980
55 67 64 38 45 72 54 42 26 76 74 79 68 21 Nav 64 45 Nav 67 49 49 78 39 65
0 0 0 19 5 0 0 0 0 0 0 0 0 0 Nav 0 0 Nav 0 22 22 0 17 0
4 0 23 3 1 3 7 0 0 0 0 3 6 0 Nav 6 0 Nav 5 1 1 7 0 0
Owner Cooperative Other occupied Nav 32 14 29 39 32 39 Nav Nav 20 7 Nav Nav 30 7 26 47 26 21 21 9 10 39 31
Nav 68 68 47 52 64 55 Nav Nav 74 92 Nav Nav 70 91 70 53 55 75 75 74 84 46 69
Nav 0 18 7 2 4 7 Nav Nav 6 1 Nav Nav 0 0 4 0 0 4 4 2 6 0 0
Notes: Nav means not available. GDR means German Democratic Republic. Data for the following countries are not available: Bulgaria, Estonia, Malta, Romania.
Nav 0 0 17 7 0 0 Nav Nav 0 0 Nav Nav 0 0 0 0 19 0 0 15 0 15 0
Rented Owner Cooperative Other occupied
2000
Source: National Board of Housing Building and Planning, Sweden and Ministry for Regional Development of the Czech Republic (2004)
Austria Belgium Cyprus Czech Republic Denmark Finland France Germany Ex GDR Greece Hungary Ireland Italy Latvia Lithuania Luxembourg Netherlands Poland Portugal Slovak Republic Slovenia Spain Sweden United Kingdom
Country
Table 11.1 Occupied dwellings, by tenure, in European countries, 1980, 1990, 2000 (%)
Institutional drivers of housing inequalities
257
As a result, as Table 11.1 demonstrates, by the end of the 1980s owner occupied housing was the dominant form of tenure in Hungary. This was also the case in Bulgaria (92 per cent owner occupied) and Romania (76 per cent), while Poland developed a large co-operative rental sector. Furthermore, the East European housing model was not applied evenly in all regions of the countries in question. Home ownership was more dominant in rural areas, while the vast majority of dwellings in cities were state owned. In the EU-15, large social housing sectors are broadly associated with greater housing equality and higher housing standards, because in addition to addressing housing shortages, the key impetus behind the expansion of this sector was to eliminate poor housing conditions, especially in the urban private rented sector. Indeed in many cases social housing was built on the sites of demolished slums (Harloe 1995). As mentioned above, large social rented sectors are generally associated with unitary tenure systems, but this is not universally the case. In the United Kingdom, for instance, an ambitious programme of construction of local authority owned social housing in the 1950s and 1960s affected a marked improvement in housing standards (Harloe 1995). In contrast, the social housing sectors of the southern states of the EU-15 were generally too small to address poor housing conditions in any significant way and as a consequence of governance problems in this tenure and also in the private rented sector, dwellings were often not well maintained after construction (see section 5) (Allen et al. 2004). Sections 3, 4 and 5 of this chapter discuss the negative effect of a range of factors – shortage of finance for new building and renovation of existing dwellings, construction methods and arrangements for governance of the housing sector – on housing standards in those countries subject to the Eastern European housing model. However, Hegedüs and Tosics (1996: 37) argue that the tenure policy which underpinned the East European housing model was the fundamental cause of these problems because ‘housing was regarded as a social good but no society could really afford this’. Tosics (1998: 227) also makes the point that ‘Even in the period of relatively high budget expenditures on housing ... [the East European housing model] was quite an inefficient way of allocating this money.’ Thus he argues that this model was associated with over-centralised housing policies which did not reflect the needs of consumers in general or of different regions or localities, and being completely dependent on the state budget were often radically changed at the first sign of budget difficulties. 1.2 Contemporary trends and their implications for housing inequality Table 11.1 reveals that the period 1990 to 2000 saw marked changes in the tenure structure of many of the countries under examination. In many of the northern countries of the EU-15 where public spending on housing was traditionally high, state capital subsidies for social housing construction were radically reduced and were only partly replaced by personal subsidies in the form of housing allowances for low income households (Gibb 2002). The termination of these construction incentives effected a sharp decline in output of social rented dwellings and around the same time some countries, notably the United Kingdom and the Netherlands, also began to privatise the social rented stock by means of sales to tenants (Forrest and Murie 1988). The combination of these measures, coupled with rising household incomes
258
Michelle Norris
and easier access to credit due to liberalisation of mortgage lending, led to an absolute and relative diminution in size of the social rented sector, generally accompanied by a rise in the proportion of households living in owner-occupied accommodation (van der Heijden 2002). Changes in tenure patterns effected by the collapse of the East European housing model in the early 1990s were far more dramatic. Although this system had been in crisis for the preceding decade – indeed Hegedüs and Tosics (1996) suggest since its inception – five developments signalled its demise. These were: • • • • •
reform of property rights to provide for owner occupation and the introduction of market institutions in the housing sector; restitution of dwellings which had been seized by the state to their original owners; reform of the state-owned sector, which generally involved transfer of ownership to local authorities; privatisation of the formerly state-owned housing stock by means of sales to tenants; and withdrawal of central government subsidies for the construction of state-owned housing. This is now the responsibility of local authorities and funding shortages severely constrained new building (Pinchler-Milanovich 2001).
In many of the relevant countries, the design of these measures was heavily influenced by recommendations of the World Bank (1990) but this was not the case universally. Thus Roberts (2003) distinguishes between ‘fast privatisers’, countries where between 40 and 90 per cent of the state-owned housing was privatised between 1989 and 1994, often by means of transfer at no or nominal cost to tenants, and ‘slow privatisers’, where between 1 and 9 per cent of relevant dwellings were privatised. Hungary, Slovenia, Bulgaria, Romania and Lithuania are in the first of these categories; the Czech Republic, Poland, Slovakia, Estonia and Latvia are in the second. The tenure patterns that characterised these countries by 2000 are a function of the speed at which the former state-owned housing stock was privatised and the size of the owner occupied housing stock prior to the end of the East European housing model. Table 11.1 reveals that in 2000, Bulgaria, Hungary and Romania, where owner occupation levels have traditionally been high and privatisation was extensive and quick, have the highest levels of home ownership among the NMS and CC nations. These changes in tenure policy and patterns had very different implications for housing quality in the EU-15 compared to the new member states and candidate countries. Dunleavy (1981) links the withdrawal of capital subsidies for social housing construction in the longstanding EU members with the ascendance of neo-liberal ideology and also with the ‘delegitimisation’ of the tenure which, in the public mind, had become increasingly associated with social problems. However, the fact that poor housing standards have been largely eradicated in many of these countries and that, in the context of low population growth, the housing shortage was resolved was also an important impetus behind these reforms. In this part of Europe the potentially negative impact of these reforms on housing standards was also mitigated by the redirection of the remaining social housing capital subsidies towards the refurbishment and regeneration of problem estates in the sector and by the strong financial position of social landlords and their access to private sector finance (see section 3; Power 1997).
Institutional drivers of housing inequalities
259
In the countries that formerly employed the East European housing model recent tenure changes have had a more negative impact on housing quality. Tosics (2003) reports that, although the formerly state owned dwellings were not allocated strictly on the basis of need, a large number of former tenants who purchased these dwellings have low incomes and cannot afford to pay the fees necessary for their upkeep. This is confirmed by Andrews and Sendi’s (2001) study of Slovenia which describes widespread defaulting on both rent and management fees. Particularly in those countries where the vast majority of the state owned stock was privatised, the incomes of households who remained tenants of this sector are generally even lower than their counterparts who purchased. This makes it difficult for landlords to raise rents to levels sufficient to pay for maintenance and upgrading (Tsenkova and Turner 2004). In addition, in countries where owner occupation levels are very high, lack of private and social rented dwellings has severely impeded housing access for low income groups, who have little option but to purchase if they are to secure a dwelling. Consequently, some of the NMS and CC countries have halted sales of state-owned dwellings in recent years, while others have reintroduced central government subsidies for new building in this tenure (Norris and Shiels 2004).
2. Housing finance and subsidy systems 2.1 Historical trends and their implications for housing inequality Comparable data on long-term investment by government and the private sector in housing are not available for all of the countries under examination here. However, the available research on housing finance systems in individual countries and groups of countries points to some generalisations on the significance of historical housing investment patterns for the quality of dwellings. The preceding discussion highlighted marked variations in the extent of social housing provision in the EU-15. As would be expected in those countries with a large social housing stock, government investment in this sector has been sizeable over the long term. Maclennan et al.’s (1997) analysis of data from the early 1990s found that in the longstanding EU member states with the largest social housing sectors (the Netherlands, Sweden and the United Kingdom), public spending on housing accounted for the highest proportion of GDP (more than 3 per cent) in the early 1990s. Governments in the countries with the smallest social housing sectors (Spain, Portugal and Greece) committed the smallest proportion of GDP (less than 1 per cent) to this area. Significantly, from the perspective of the discussion at hand, in those EU-15 countries where public investment in housing was high, private investment was generally high also. This is because until recent years the mortgage lending market was relatively underdeveloped in the southern states of the EU-15 and house purchase was often funded by non-credit methods such as self-building, personal savings and gifts or inheritance from relatives (Allen et al. 2004). Thus, Fahey et al. (2004) report that in 1996, 69 per cent of home owners in Greece and 62 per cent in Italy and Spain owned their dwelling outright, as compared to only 27 per cent of UK and 7 per cent of Dutch home owners. Non-credit methods of funding such a bulky purchase as housing are associated with poorer housing standards because the initial investment is lower and strategies such as building a temporary dwelling, or a part of a dwelling which is completed as funds allow, are often employed. In the southern
260
Michelle Norris
EU-15 countries overcrowding related to multi-family living arrangements is also more common than among northern countries, as house purchase is deferred until the requisite savings have been accumulated (Allen et al., 2004). This may explain Domanski’s (this volume, Ch. 10) finding that households in southern Europe enjoy considerably less living space than residents in the north. In the new EU members that employed the East European housing model until 1990, a number of aspects of the housing finance system had negative implications for housing standards. Firstly, as a result of state control over housing production, consumption and allocation, the majority funding for housing construction and maintenance also emanated from the state. However Sillince (1990) reports that the prioritisation of investment in industry led to underinvestment in housing between 1950 and 1975, which in turn gave rise to persistent housing shortages and overcrowding in many countries in this region. Although the post-1960 marketisation reforms did help increase output of housing by non-governmental sources, the impact was lessened by the economic crisis of the late 1970s and early 1980s, which reduced the opportunity for both households and the state to invest in housing. Kozlowski’s (1988, cited in Hegedüs and Tosics 1996) study of Poland reveals that in this context targets for state owned housing output were achieved by decreasing per unit costs through reduced average apartment sizes and decreasing specifications particularly in terms of equipment and finishes. According to Tosics (1998) lack of funds also led to skimping on investment in the infrastructure necessary for housing development. The available investment was concentrated on cheaper amenities such as water supply, rather than on expensive items such as sewage systems. This investment strategy explains the limited availability of certain facilities and the smaller size of dwellings in this part of Europe, as revealed by Domanski’s chapter (10) in this volume. 2.2 Contemporary trends and their implications for housing inequality Recent changes in housing finance systems and subsidy policy have had varying implications for housing quality in the southern and northern countries of the EU-15. In the latter countries, the availability of private finance and the strong financial position of social landlords helped to reduce the potentially negative impact of cutbacks to state capital subsidies on the output and renovation of social housing, as described earlier. Table 11.2 reveals that in the mid-1990s, social landlords in Denmark, the Netherlands, France and Sweden derived the vast majority of their funding for the construction of new buildings from borrowing, and British, German and Belgian social landlords are also heavily reliant on this source of income. With the exception of Great Britain, social landlords in these countries also received various forms of government support for borrowing (subsidised loans, loans provided through intermediary borrowers, state guarantees). What Kemeny (1995) terms the maturation of the social rented stock in many of these countries (meaning that loans for the construction of older dwellings in the sector have been largely repaid) also enables social landlords to use their own funds for social housing construction and to set aside part of these funds for renovation and upgrading. However, Gibb (2002) argues that the reduction of capital subsidies will have negative implications for social rented estates in the future, because they have forced social landlords to increase rents, which means that only welfare dependants who qualify for housing allowances
Institutional drivers of housing inequalities
261
Table 11.2 Sources of funding for new construction in seven longstanding EU member states, 1994–1995 (%) Country Belgium Denmark France Germany Great Britain Netherlands Sweden
Own assets 20 2 3 33 0 9 3
Borrowing 59 91 80 49 58 83 95
Lump-sum subsidies from government 21 7 17 18 42 8 2
Source: Priemus and Boelhouwer (1999)
can afford to live in newer, more expensive, social rented dwellings. In addition, the drop in social housing output associated with the withdrawal of capital subsidies has, on occasions, led to housing shortages. Consequently, at various times since 1990, Belgium, France, Finland and Ireland increased the level of public investment in social housing (van der Heijden 2002; Norris and Shiels 2004). In contrast, in several of the southern states of the EU-15, notably Spain and Portugal, the wider availability of mortgage credit since the mid-1990s as a result of the liberalisation of lending rules has helped to improve housing standards by increasing investment in new house building and rehabilitation. Figure 11.1 reveals that in 2000, new house building rates in Spain and Portugal were among the highest in the European Union and candidate countries. As a result of these developments Ball (2005) argues that the gap between housing standards in the north and south of the EU-15 has narrowed in recent years. The collapse of the East European housing model prompted radical changes in housing finance systems in those new member states that formerly employed it. As mentioned above, subsidies for the construction of state-owned housing, which has previously driven the bulk of housing output, were largely withdrawn, and currently only 10 per cent of housing output in these countries is publicly funded (Shinozaki 2005). Although measures to support and regulate the development of private finance for housing were introduced initially on the direction of international institutions such as the United Nations Economic Commission for Europe (2005) and subsequently to meet the requirements for European Union membership, these have had mixed success. Figure 11.2 demonstrates that in 2004, the ratio of mortgage debt to GDP was much higher in the EU-15 (on average 46.4 per cent) than in the NMS (with an average of only 11.2). Housing output in the countries formerly subject to the East European housing model declined radically in the first half of the 1990s, and has only recovered marginally in the years since then. Figure 1 demonstrates that, excluding Cyprus and Malta, housing output in the NMS and CC averaged only 1.8 per 1,000 inhabitants in 2000, compared to 6.3 in the EU-15. Shinozaki (2005) attributes this trend primarily to under-development of the mortgage lending system rather than to the macro economy (which has recovered much more strongly than the housing market), lack of demand (the proportion of households of working age is similar to the EU-15) or
Au st
ria
No. dwellings per 1,000 inhabitants
Housing output per 1,000 inhabitants
G G C C Ire R Fr Po Sl Sw Li Lu Sp Po Bu La U H N er re yp R ze Den Fin om Slo th un et ov an la xe tv rtu ai la lg lg ed Kin nite la va m ec iu he ru ep ch m g n i e n n iu ar c a a m n a g a ar d ki ni en gd d e d e s ub ni ar d rla m ny ia al na bo a a y a k om nd lic ur s g
Be
16 14 12 10 8 6 4 2 0
Notes: All data are for 2000, with the exception of: Spain (2002 data), Luxembourg (2001 data), Germany (2002 data) and Greece (average annual output since 1998). Data for Estonia, Italy, Malta and Turkey are not available.
Source: Norris and Shiels (2004)
Figure 11.1 Number of dwellings and housing output per 1,000 inhabitants in European countries, 2000
0
100
200
300
400
500
600
ria
st
Au
um
lg i
Be
Mortgage debt to GDP ratio
G C Ita Fi Ire Es H D Fr z er Gre un e nl an ly la to m an ec ru Rep ech nm g n n c an ar d ia e s u e ar d y bl y k ic
yp
C ia
tv
La th iu an a
Lu xe m bo ur g
al ta
M N et he
rla
Growth in mortgage debt
Li nd
s
d
la n
Po
Po
rtu
ga
l
Sl
Sl ia
ak
ov
Note: Data for Bulgaria, Romania and Turkey are not available.
Source: European Mortgage Federation (2004)
Figure 11.2 Ratio of outstanding mortgage dept to GDP in European countries and annual growth in mortgage debt, 2004
0
20
40
60
80
100
120
en
ov
ia
Sp
ai
n
Sw
U ed Ki nite n en g d do m
264
Michelle Norris
the capacity of the construction industry (it has concentrated on the commercial sector which accounts for 80 per cent of construction in these economies). Specifically, she argues that the mortgage lending system in these countries is inefficient because of its ologopolistic domination by small numbers of commercial banks. Lenders of this type often require high deposits and short repayment terms, which, together with the relatively high interest rates in this part of Europe, render mortgages unaffordable for most households. The specialist mortgage lending market in the NMS has also been slow to develop, and most households rely on contract savings to fund house purchase. This arrangement, under which the amount lent is linked to savings, limits the amount of money home buyers can borrow, and also constrains them from entering the housing market until they have accumulated the requisite savings. These problems in credit access mean that demand for dwellings cannot be translated into purchasing power, which in turn discourages commercial builders from catering for this market. These developments in housing finance have had a number of negative implications for housing quality. Shinozaki (2005) argues that they have contributed to a housing shortage in relevant countries. This contention is supported by the data presented in Figure 11.1. The number of dwellings per 1,000 inhabitants stood at 438.4 on average in the EU-15, compared to 383.8 in the NMS, and the latter figure is significantly inflated by the inclusion of Malta. Although, population decline in many NMS means that assessing housing shortages is not a straightforward task and some commentators (cf. Lux 2003) argue that few countries in this region suffer from problems in this regard at the national level, there is a widespread consensus that housing availability problems exist among certain groups and in certain localities in this region. This is the case in economically successful cities for instance, and among recently formed households whose housing access has been constrained not only by low output of dwellings, but also by the type of dwellings constructed – which often target the high end of the market (Norris and Shiels 2004; Shinozaki 2005). Pinchler-Milanovich (2001) acknowledges that in most countries in this region, housing reconstruction and rehabilitation activities have increased in recent years as housing output has fallen. However these activities are not evenly distributed around the housing stock. They generally target the high end of the property market and potentially lucrative locations such as city centres, because demand for developments of this type is relatively strong. In contrast, owners of dwellings in high-rise blocks who cannot afford maintenance or upgrading have little prospect of selling.
3. Housing construction systems, styles and trends 3.1 Age of the housing stock Table 11.3 sets out the most up-to-date available data on the age of the housing stock in the countries under examination. As would be expected considering the recent inter-country variations in housing output described above, it reveals that the 13.3 per cent of the housing stock in the EU-15 has been constructed since 1990, whereas 8.25 per cent of dwellings in the NMS date from this period. Moreover, the latter figure is significantly inflated by the inclusion of Malta and Cyprus – in the eight Central and Eastern European new EU members only 6.5 per cent of the housing stock has been constructed since 1990. The proportion of stock that was constructed
Institutional drivers of housing inequalities
265
Table 11.3 Age distribution of the housing stock in European countries, various years Country
Year to which Pre 1945 1945–1970 1970–1990 1990–2004 data refer
Austria Belgium Bulgaria Cyprus1 Czech Republic Denmark Estonia4 Finland France1 Germany1 Hungary1 Ireland1 Italy Latvia Lithuania Luxembourg Malta Netherlands Poland Portugal Slovakia Slovenia Spain Sweden United Kingdom1 Mean (R)
2002 1996 2000 2000 1991 2000 2002 1996 2002 2002 1996 2002 1995 2000 2001 2001 1995 2002 2002 1991 2001 2002 2001 2002 1991 n/a
26.8 31.8 24.3 231 41.9 38.9 18.9 12 29.41 27.91 29.5 20.5 29.5 25 27 27.3 25.5 20 23.2 24.4 11.5 23.4 22 21 41 25.8
28 29.8 48.2 10.7 24.6 29.9 22.8 32.9 15.1
28.7 34.1 24.5 38.1 33.5 25.9 31.5 46.2 32.2
16.4 4.2 1.9 28.2 02 5.3 4.1 7.9 5.7 11.1 4 25.7
61 27.2 17.6 40.7 28 34 22.2 27 26.9 31.2 35.2 39.8 29.8 43 22 28.9
38.9 36.2 29.8 43 32 72.7 43.1
4 7 9.1 53
37 44.2 46.6 28.9 34.3 28 38 33.6
11.6 0 6.7 7.8 13.8 6 03 11.0
Source: Norris and Shiels (2004) Notes: Data for Greece, Romania and Turkey are not available. 1 In these cases data were reclassified for the purposes of inclusion in this table. 2 These data only cover the period to 1990. 3 These data only cover the period to 1996. 4 22.8 per cent of the housing stock in Estonia is categorised as unknown in terms of age, for this reason the mean figures for this table do not total 100%. The Czech Republic, Italy, Portugal and the United Kingdom were excluded when calculating the mean number of dwellings build since 1990, because data for these countries dates from 1991. Nav means not available, n/a means not applicable, (R) means rounded.
prior to 1970 is also higher in the EU-15 compared to the NMS-10 – 55.6 per cent compared to 53 per cent respectively. This indicates that a higher proportion of dwellings in the 10 new member states were constructed between 1970 and 1990 than is the case in the longstanding EU members. The higher proportion of the housing stock in the EU-15 that is of recent construction has a number of positive implications for the quality of dwellings. For instance, newer dwellings are more likely to include amenities such as indoor flushing toilets than their older counterparts, because the market requires this and also because of more rigorous government regulations governing to the quality of new and refurbished buildings (Meijer 2002; Sheridan 2003). Although all European Union
266
Michelle Norris
members currently have nationwide building regulations which specify the standards of new dwellings, these measures have impacted on a larger proportion of the housing stock in the EU-15 because more dwellings in these countries are of recent construction (Norris and Shiels 2004). To date the EU has been only marginally involved in the regulation of construction standards (although building products are heavily regulated) and as a result, building regulations vary widely between member states. Meijer (2002) and Sheridan (2003) report that these regulations were introduced earlier and are more rigorous in northern countries of the EU-15, which may have contributed to the higher quality of dwellings in these countries compared to the southern member states. 3.2 Dwelling design The higher rate of housing output between 1970 and 1990 in the Central and Eastern European new member states also has implications for housing quality because of the construction methods and building designs adopted during this period. By the 1960s, a widespread housing shortage, together with the influence of modernist architects and land-use planners and advances in building technology, inspired a fashion for high-rise dwelling designs and semi-prefabricated or ‘system’ building techniques. The construction of city-edge estates also spread steadily across Europe (Taylor 1998; Wassenberg et al. 2004). However the influence and durability of this trend varied considerably between countries. Most of city-edge the northern EU-15 countries abandoned high-rise residential construction and mass housing estate development by the mid-1970s, as a result of their unpopularity among residents, growing evidence regarding social and management difficulties (see influential critiques by Jacobs 1961; Newman 1972) and construction problems which were well publicised by the collapse of the Ronan Point tower block in London in 1968. Consequently, as Figure 11.3 demonstrates, a relatively small proportion of dwellings in these countries are high-rise. In contrast, in several of the southern states of the EU-15, most notably Spain and Italy, mass high-rise construction continued for long after this, which has resulted in a relatively high proportion of dwellings in high-rise estates. The requisite technology to build high-rise dwellings became available in many of the NMS-10 only during the 1970s. However it was embraced with particular enthusiasm in this part of Europe, as a solution to shortages of dwellings, finance and employment for unskilled workers. Output levels were high and this type of construction continued until the 1990s (Wassenberg et al. 2004). As a consequence, high-rise dwellings now account for 34.1 per cent of the housing stock in the NMS, compared to 14.7 per cent in the EU-15 (Figure 11.3). As is evidenced by the large proportion of high-rise estates in the EU-15 that are currently subject to regeneration projects, this design type is commonly (but not universally in this part of Europe) associated with poor-quality accommodation and management and social problems (Power 1997). However, Andrews and Sendi (2001) argue that the quality of high-rise dwellings is even more problematic in those countries which formerly employed the East European housing model. The state owned housing stock in these countries was built almost exclusively in this style, but due to the aforementioned funding problems, together with the shortcomings in construction systems discussed in the following section, their initial build quality was often poor.
Institutional drivers of housing inequalities
267
45 40 35 30 25 20 15 10 5 United Kingdom
Spain
Slovenia
Slovakia
Portugal
Poland
Netherlands
Luxembourg
Italy
Hungary
Germany
France
Denmark
Czech Republic
Belgium
0
Figure 11.3 % of dwellings in high-rise buildings in European countries, 2004 Sources: National Board of Housing, Building and Planning, Sweden, and Ministry for Regional Development of the Czech Republic (2004) Note: Data for Austria, Bulgaria, Cyprus, Estonia, Finland, Greece, Ireland, Latvia, Lithuania, Malta and Romania are not available.
3.3 Construction systems During the ascendancy of the East European housing model, the building of the large high-rise estates described above necessitated the establishment of large companies and factories to manufacture the concrete panels from which they were constructed. Sillince (1990: 50) highlights a number of problems with these arrangements: Such organisations often exerted … a strong monopolistic control over (and hence simplification of) design and price (amounting to protectionism of increasingly out-of-date technology and methods). The lack of participation of the eventual customers – the residents – or at least of any organisation (such as a local authority) to represent their interests, was one factor in … poor quality work. The result was a higher demolition rate and even higher costs to repair poorly constructed dwellings. Hegedüs and Tosics (1996) argue that building costs were higher than necessary due to the inefficiency of these construction companies. In support of this contention, they cite research conducted in Poland in the early 1980s which concluded that investment in system building could have yielded 70 per cent more dwellings if channelled into traditional construction methods. In addition, they attribute the persistence of systematic high-rise building construction methods in Eastern Europe in the face of widespread public dissatisfaction and after it had been largely abandoned in the EU-15 to the political and economic power of the construction companies.
268
Michelle Norris
Many of the traditional problems with the construction systems in these countries persisted after their economic and political reorganisation in the early 1990s (Pinchler-Milanovich 2001). The efficiency of the industry failed to improve because the former state construction enterprises often retained their privileged role in the sector. This is due in part to their status and established business contacts, but also to barriers to entry of new firms into the market, the most significant of which is the difficulty of obtaining credit, mentioned above, but problems in purchasing building material and land are also contributory factors.
4. Governance 4.1 Trends in the EU-15 and their implications for housing quality There is some evidence that arrangements for governance of the housing sector has contributed to the lower dwelling standards in the southern states of the EU-15 compared to their northern counterparts (Domanski and Ostrowska 2004). Governance of the social and private rented sectors is of particular consequence in this regard. The introduction of government regulation of private residential rents in most of the EU-15 countries, following World War II, has been criticised for constraining investment in the purchase and maintenance of dwellings in this sector (Harloe, 1995). However, these measures had a more negative impact on housing in the southern states of this region than in the north. This is because rent controls remained in place for longer in the southern countries, while in the north they were often accompanied by measures to improve housing quality. Prior to the 1980s, the latter measures included such things as the clearance of private rented slums and their replacement by social housing prior to the 1980s, and since then, incentives to encourage private landlords to improve housing in poor repair and regeneration of inner cities where most private rented housing is located (Maclennan et al. 1997). Among the northern countries of the EU-15, the private rented sector in Germany is also distinctive because it has remained large and high quality due to generous fiscal support. As a result of these differences in the governance of the private rented sector, Maclennan et al. (1997) report that generally no more than 5 per cent of private renting households in the north of the EU-15 live in substandard dwellings, whereas in the south, the proportion is typically closer to 15 per cent. According to Allen et al. (2004) in those southern EU-15 countries where some social housing was supplied, such as Spain and southern Italy, its quality deteriorated rapidly after construction due to poor management standards and the inability to evict tenants. As mentioned earlier, the period since the 1980s has been marked by growing concern about social housing in the northern members of the EU-15, and these problems have often been linked to governance issues. However, the evidence in this regard is far from definite. Thus many commentators (most notably Power 1997) argue that provision and management of social housing directly by local authorities, which has until recently been the norm in Ireland, the Netherlands and the United Kingdom, is associated with more widespread problems and greater difficulties in regeneration than is the case in Denmark and the Scandinavian countries where social housing is provided by non-profit bodies or by agencies controlled by but separate from local government. Although many researchers (cf. Priemus and
Institutional drivers of housing inequalities
269
Boelhouwer 1999) take the view that the case has yet to be proven, this evidence, together with concerns about the extent of public subsidies for social housing, have encouraged governments to promote the development of other non-profit social landlords as an alternative to municipal providers. 4.2 Trends in the NMS and their implications for housing quality In those new and candidate EU members that were subject to the East European housing model prior to the 1990s, inadequate arrangements for the maintenance and upgrading of dwellings also helped precipitate poor housing standards and often caused the fabric of dwellings to deteriorate significantly after their construction. Several commentators (for instance: Sillince 1990) emphasise the role which rent determination methods in the state owned housing sector played in these problems. Prior to the 1980s (by which time the rent setting regime in many of the countries in question had been liberalised), rents in this sector were fixed by central government and were generally set at between 1 and 3 per cent of average family earnings, rising to between 5 and 8 per cent when utility costs were included (Hegedüs and Tosics 1998). This was far too low to generate sufficient income for the maintenance of dwellings, and moreover, did not reflect variations in the costs of maintaining different types of dwellings or dwellings in different locations. Although in theory the shortfall was made up by central government, Pinchler-Milanovich (1994) reports that, in practice, housing subsidy programmes rarely provided for maintenance and refurbishment and so very little was carried out. Blunt and Muziol-Weclawowicz’s (1998) case study of Poland highlights the contribution of rigid and inefficient management practices on the part of the state housing agencies to poor maintenance standards. Their research found that the productivity of Polish housing enterprise workers was less than one-third that of their UK counterparts. However, in the years since the economic and political reorganisation of eastern Europe, reform of governance of the housing system has failed to keep pace with efforts to promote the role of the market in this sector and this also has had negative implications for housing standards. In this regard, four aspects of governance arrangements are particularly problematic. First, rent controls have been applied to private rented dwellings in many of these countries. As in the EU-15, the application of rent control in the NMS has constrained investment in the maintenance of existing private rented dwellings and the purchase of dwellings for rent (Blunt and Muziol-Weclawowicz 1998). Consequently, the governments of Estonia, Slovakia and Poland are currently examining the reform of these provisions (Norris and Shiels 2004). Second, in the case of formerly state owned dwellings that were not privatised, housing management and maintenance practices have seen only marginal reform since the early 1990s. Although ownership of these dwellings has been transferred to the local authorities, in most cases responsibility for their management and maintenance still rests with the state enterprises which carried out this task prior to the 1990s. Competition in this sector has been slow to take off, because few organisations have the requisite skills to challenge the state housing enterprises for housing management contracts. Furthermore, due to the strength of workers’ representative associations in these enterprises, local authorities have generally been hesitant to push through reform of work practices (Blunt and Muziol-Weclawowicz 1998).
270
Michelle Norris
Rent determination systems in the state-owned housing sector have also not been reformed in most of the relevant countries. Indeed Hegedüs and Tosics (1998) report that, in this region of Europe, the proportion of income which tenants of this sector devote to rent fell from 5.1 to 1.9 per cent between 1990 and 1994. Their research reveals that rent rises were highest in those countries where the state-owned housing stock is largest. Thus they speculate that the opportunities to raise rents are limited in those countries where the social rented sector is small and dominated by lowincome families who can afford neither to avail themselves of the opportunity to purchase their dwelling nor to pay higher rents. Arrangements for the governance of privately owned dwellings in high-rise blocks are also a significant contributor to inadequate maintenance in many NMS. For instance, Andrews and Sendi’s (2001) case study of Slovenia reveals that condominium law in this country allows home owners in each individual block within an estate to choose separate housing managers. Where this management structure is in place, the organisation and execution of refurbishment and improvement works to the entire estate is complex and inefficient. This problem is compounded by the failure of the legislation to specify the types of organisations that can act as managing agents for condominiums, the qualifications of staff employed in this area and the responsibilities of the home owner associations who are charged with supervising the activities of these agents. Blunt and Muziol-Weclawowicz’s (1998) study of Poland reveals similarly poor condominium governance arrangements but attributes them, not to shortcomings in the legislation, but rather to the failure of local authorities to enforce its provisions and the unwillingness of many home owners to participate in home owner associations. Shigehiro (2005) also relates the underdevelopment of the mortgage market in many of the new EU members to governance issues. In particular she singles out implementation of regulations for this sector as the key problem in most of the countries in question, since in her assessment, the basic legal infrastructure for mortgage lending and funding and regulatory and supervisory systems are relatively well established. In addition, she criticises poorly targeted government home purchase support schemes in many new member states, on the grounds that because they are open to the vast majority of households, they are insufficiently generous to enable low-income earners to access the market. Figure 11.2 above revealed that although the mortgage debt to GDP ratio was much lower in the NMS than in the EU-15 in 2004, mortgage debt grew at a much faster rate in the former group of countries (by 44.4 per cent compared to 9.6 per cent). However, the extent of this growth varies significantly among the countries in question, which indicates that national-level factors are also impeding the development of the mortgage-lending market. Merrill and Kozlowski (2001) ascribe modest mortgage-lending growth in Poland to several factors. For instance, banking legislation has constrained the growth of mortgage banks by limiting the proportion of the market value of dwellings they can lend. Consequently, the dominance of the universal banks in mortgage lending has gone unchallenged. This problem is compounded by the large number of small, inefficiently run mortgage banks currently in operation, which have little capacity for growth. Foreclosure of mortgage loans in Poland also is time consuming and difficult and valuation and credit approval skills are underdeveloped. This last factor explains why mortgage loan approval criteria in Poland are relatively conservative and loan terms much shorter than is the norm in the EU-15.
Institutional drivers of housing inequalities
271
5. Conclusions Differences in housing quality across Europe reflect a range of social and economic factors including income levels, historic patterns of urbanisation and the extent of war damage and reconstruction (Maclennan et al. 1997). However, the preceding discussion has identified a number of institutional factors – summarised in Table 11.4 – which have also had a significant impact on these international variations. For instance, historical differences in the availability of public and private finance for housing have contributed to poorer housing standards in the southern compared to the northern states of the EU-15. Furthermore, in recent years, convergence of housing supply systems in these two groups of countries, as a result of the contraction of the social rented sector in the north and the greater availability of mortgage credit in the south, have stimulated convergence in housing standards (Doling 1997). Domanski’s chapter in this volume (10) demonstrates that in housing quality terms the former communist new EU member states and candidate countries form a distinct group, characterised by low housing standards arising from institutional aspects of the East European housing model that dominated in these countries until the early 1990s. Following the political and economic reforms of the early 1990s, housing standards in these countries worsened even further, as a result of the privatization of the formerly state owned housing stock, the sharp reduction of state funding for new house building and the underdevelopment of the mortgage lending market. Consequently, the gap in housing standards between the west and east of Europe has grown since the 1980s. Poorer quality housing in the Central and Eastern European NMS obviously has negative implications for the quality of life of occupants, and in extreme cases for morbidity and life expectancy (Bonnefoy et al. 2003). The literature also reveals other social and economic problems arising from these poor housing conditions. For instance, according to Ball (2005), lack of rented housing, coupled with difficulties in access to owner-occupied housing, has reduced labour mobility. Tosics (2003) reports that underinvestment in the housing infrastructure inspired excessive use of cheaper solutions such as septic tanks which have had negative long-term environmental effects. Pinchler-Milanovich (2001) reveals that as higher income households move out of poor quality estates, increasing socio-spatial segregation is an emerging problem. Therefore, these East/West housing inequalities also have negative implications for economic and social cohesion within the European Union (Stephens 1999). The significance of these housing inequalities in turn raises the issue of how they should be addressed and whether the European Union as well as national and local governments should play a role in this regard. EU involvement in this area has traditionally been constrained by the fact that housing is not an EU competency and consequently housing-related projects have been proscribed from receipt of direct aid from the European Structural Funds (Stephens 1999). However, following a campaign by the EU wide representative bodies for providers of homelessness services and social housing, this is about to change (FEANTSA 2003; CECODHAS, undated). In December 2005, agreement was reached that the next round of structural funding may contribute to the financing of housing projects in the new member states and in Romania and Bulgaria, although the detailed eligibility criteria for access to this finance have yet to be decided.
Central and Eastern European NMS and CC
South EU-15 and Malta and Cyprus
Housing tenure system
North EU-15
Governance
Housing construction systems
Housing finance and subsidy systems
Housing tenure system
Housing construction systems Governance
Housing finance and subsidy systems
Housing tenure system
Housing construction systems Governance
Housing finance and subsidy systems
Category
Groups of countries Historically – high social housing output associated mainly with unitary tenure systems Recently – low social housing output associated with withdrawal of state subsidies for new build social housing Historically – high housing investment levels associated with high social housing output and well developed mortgage systems Recently – withdrawal of state subsidies for social house building has been mitigated by the availability of private finance High proportion of the housing stock is of recent construction Low proportion of the housing stock is in high-rise dwellings Historically – rent controls in the private rented sector Historically and recently – rent controls were in place for a limited time and were often accompanied by measures to improve housing standards Historically – low social housing output associated mainly with dual tenure systems Recently – high output of housing for owner occupation and private renting Historically – low housing investment levels associated with low social housing output and underdeveloped mortgage systems Recently – high housing investment associated with the wider availability of mortgage credit High proportion of the housing stock is of recent construction High proportion of the housing stock is in high rise dwellings Historically – rent controls in the private rented sector Historically and recently– poor management of social housing Historically – state provision of most housing associated with East European housing model Recently – privatisation of formerly state-owned housing associated with collapse of the East European housing model Historically – low housing investment levels associated with reliance on state funding and prioritisation of investment in other areas and low rents Recently – low investment associated with the withdrawal of state subsidies housing construction subsidies and the underdevelopment of commercial mortgage market Low proportion of the housing stock is of recent construction High proportion of the housing stock is in high-rise dwellings Historically – monopolistic construction systems and materials shortages associated with East European housing model Recently – inadequate arrangements for governance of condominiums, mortgage lending and social and private rented housing
Specific drivers
Table 11.4 Institutional drivers of housing inequalities in an enlarged Europe
Negative
Negative Negative Negative
Negative
Negative
Positive Positive Negative Negative Negative Negative Negative
Negative Positive Negative
Positive Positive Negative Positive
Neutral
Positive
Positive Neutral
Impact
Institutional drivers of housing inequalities
273
The campaign to have housing projects designated as eligible for structural funding was predicated principally on the concern that neither governments nor households in this part of Europe will be able to shoulder the cost of improving the high-rise housing stock. Research commissioned by the Netherlands Ministry of Housing Spatial Planning and the Environment (2004) costs this task at Ä350 billion. It estimates that at current rates of progress it would take 40 years to complete, which is problematic because many of these dwellings urgently require remedial works. Difficulties mentioned earlier in raising social housing rents in order to pay for upgrading of social housing in those new EU members where this tenure is extremely small and residualised suggests that social housing should be prioritised for receipt of structural funding. The arguments for extending eligibility to owner occupied and private housing are less compelling, because the available evidence suggests that intervention by national governments could provide a more sustainable solution to quality problems in these tenures at relatively small cost. For instance, the United Nations Economic Commission for Europe (undated) has produced detailed guidelines on improving condominium governance in this part of Europe, which if implemented would help to ensure that high-rise dwellings are adequately maintained. Furthermore, research on the NMS has identified a number of national level reforms which could support housing market development. Merrill and Kozlowski (2001) argue that introducing mortgage default insurance would enable the liberalisation of lending criteria in countries where mortgage lending is currently constrained. Shinozaki (2005) argues that better targeting of existing state home purchase supports at low and middleincome households would invigorate these sections of the housing market which are currently dormant in many of the NMS. By enabling increased output of new, higherquality dwellings and facilitating the sale of dwellings by owners who cannot afford maintenance costs, improved functioning of the housing market could play a key role in improving housing standards in the new and candidate EU members.
References Allen, J., Barlow, J., Leal, J., Maloutas, T. and Padovani, L. (2004) Housing and Welfare in Southern Europe, London: Blackwell. Andrews, K. and Sendi, R. (2001) ‘Large housing estates in Slovenia: a framework for renewal’, European Journal of Housing Policy, 1, 2: 233–255. Ball, M. (2005) RICS European Housing Review 2005, London: Royal Institute of Chartered Surveyors. Barlow, J. and Duncan, S. (1994) Success and Failure in Housing Provision: European Systems Compared, Oxford: Pergamon. Blunt, A. and Muziol-Weclawowicz, A. (1998) ‘Improved management of the existing stock: the case of Poland’, Housing Studies, 13, 5: 697–711. Bonnefoy, X., Braubach, M., Krapavickaite, D., Ormand, D. and Zurlyte, I. (2003) ‘Housing conditions and self reported health status: a study in panel block buildings in three cities of Eastern Europe’, Journal of Housing and the Built Environment, 18, 3: 329–352. CECODHAS (undated), ‘Urban matters working group (undated) towards a review of cohesion policy, 2007–2013: proposed contribution by CECODHAS, unpublished position paper, Brussels. Doling, J. (1997) Comparative Housing Policy: Government and Housing in Advanced Industrialized Countries, London: Macmillan. Domanski, ´ H. and Ostrowska, A. (2004) ‘Housing and the local environment’, pp. 15–22, in European Foundation for the Improvement of Living and Working Conditions (eds), Quality
274
Michelle Norris
of life in Europe, Luxembourg: Office for Official Publications of the European Communities. Dunleavy, P. (1981) The Politics of Mass Housing in Britain, 1945–1975: A Study of Corporate Power and Professional Influence in the Welfare State, Oxford: Clarendon. European Mortgage Federation (2004) Hypostat 2004, Brussels: European Mortgage Federation. Fahey, T., Nolan, B. and M‚itre, B. (2004) ‘Housing expenditures and income poverty in EU countries’, Journal of Social Policy, 33, 3: 437–454. FEANTSA (2003) The Revision of the Structural Funds, Brussels: FEANTSA. Forrest, R. and Murie, A. (1988) Selling the Welfare State: The Privatisation of Public Housing, London: Routledge. Gibb, K. (2002) ‘Trends and change in social housing finance and provision within the European Union’, Housing Studies, 17, 2: 325–336. Harloe, M. (1995) The People’s Home? Social Rented Housing in Europe and America, Oxford: Blackwell. Hegedüs, J. and Tosics, I. (1996) ‘The disintegration of the East European housing model’, in D. Clapham, J. Hegedüs, K. Kintrea and I. Tosics (eds), Housing Privatization in Eastern Europe, London: Greenwood. Hegedüs, J. and Tosics, I. (1998) ‘Rent reform: issues for the countries of Eastern Europe and the newly independent states’, Housing Studies, 13, 5: 657–678. Jacobs, J. (1961) The Death and Life of Great American Cities, Harmondsworth: Penguin. Kemeny, J. (1995) From Public Housing to the Social Market: Rental Policy Strategies in Comparative Perspective, London: Routledge. Lux, M. (2003) ‘Efficiency and effectiveness of housing policies in the Central and Eastern European countries’, European Journal of Housing Policy, 3, 3: 243–265. Maclennan, D., Stephens, M. and Kemp, P. (1997) Housing Policy in EU Member States, Luxembourg: European Parliament, Directorate General for Research. Meijer, F. (2002) Building regulations in Europe, Part 1: Comparison of the Systems of Building Control in Eight European Countries, Delft: Delft University Press. Merrill, S. and Kozlowski, E. (2001) ‘Developing housing finance in a transition economy: the case of Poland’, Journal of Housing Economics, 10, 3: 363–392. National Board of Housing Building and Planning, Sweden and Ministry for Regional Development of the Czech Republic (2004) Housing Statistics in the European Union, 2004, Karlskrona: Boverkert. Netherlands Ministry of Housing, Spatial Planning, and the Environment (2004) Sustainable Refurbishment of High-rise Residential Buildings and Restructuring of Surrounding Areas, The Hague: Ministry of Housing, Spatial Planning and the Environment. Newman, O. (1972) Defensible Space: People and Design in the Violent City, London: Architectural Press. Norris, M. and Shiels, S. (2004) Regular National Report on Housing Developments in European Countries: Synthesis Report, Dublin: Stationery Office. Oxley, M. and Smith, J. (1996) Housing Policy and Rented Housing in Europe, London: Spon. Pinchler-Milanovich, N. (1994) ‘The role of housing policy in the transformation process of Central-East European Cities’, Urban Studies, 31, 7: 1097–1115. Pinchler-Milanovich, N. (2001) ‘Urban housing markets in Central and Eastern Europe: convergence, divergence or policy “collapse”’, European Journal of Housing Policy, 1, 2: 145–187. Power, A. (1997) Estates on the Edge: The Social Consequences of Mass Housing in Northern Europe, Basingstoke: Macmillan. Priemus, H. and Boelhouwer, P. (1999) ‘Social housing finance in Europe: trends and opportunities’, Urban Studies, 36, 4: 633–645. Roberts, A. (2003) ‘Privatization and rent deregulation in Eastern Europe’, pp. 45-63, in S. Lowe and S. Tsenkova (eds), Housing Change in East and Central Europe: Integration or Fragmentation?, Aldershot: Ashgate.
Institutional drivers of housing inequalities
275
Sheridan, L. (2003) Building Regulations in Europe – Part II: Comparison of Technical Requirements in Eight European Countries, Delft: Delft University Press. Shigehiro, S. (2005) A Comparative Assessment of Housing Finance Markets in Transition Economies, Housing Finance Markets in Transition Economies, Trends and Challenges, Paris: OECD Publishing. Shinozaki, S. (2005) A Comparative Assessment of Housing Finance Markets in Transition Economies, Housing Finance Markets in Transition Economies, Trends and Challenges, Paris: OECD Publishing. Sillince, J. (1990) ‘Housing policy in Eastern Europe and the Soviet Union’, pp. 6–58, in J. Sillince (ed.), Housing Policies in Eastern Europe and the Soviet Union, London: Routledge. Stephens, S. (1999) ‘The fiscal role of the European Union: the case of housing and the European Structural Funds’, Urban Studies, 36, 4: 715–735. Taylor, N. (1998) Urban Planning Theory since 1945, London: Sage. Tosics, I. (1998) ‘European integration and the East-Central European “outsiders”’, in M. Klienman, W. Matznetter and M. Stephens (eds), European Integration and Housing Policy, London: Routledge. Tosics, I. (2003) ‘Comparative perspectives on urban housing conditions’, pp. 73–80, in S. Lowe and S. Tsenkova (eds), Housing Change in East and Central Europe: Integration or Fragmentation? Aldershot: Ashgate. Tsenkova, S. and Turner, B. (2004) ‘The future of social housing in Eastern Europe: reforms in Latvia and Ukraine’, European Journal of Housing Policy, 4, 2: 133–149. United Nations Economic Commission for Europe (2005) Housing Finance Systems for Countries in Transition: Principles and Examples, Geneva: United Nations Economic Commission for Europe. United Nations Economic Commission for Europe (undated) Guidelines On Condominium Ownership of Housing for Countries in Transition, Geneva: United Nations Economic Commission for Europe. van der Heijden, H. (2002) ‘Social rented housing in Western Europe: developments and expectations’, Urban Studies, 39, 2: 327–340. Wassenberg, F., Turkington, R. and van Kempen, R. (2004) ‘Prospects for high-rise estates’, pp. 15–30, in F. Wassenberg, R. Turkington, R. van Kempen (eds), High Rise Housing in Europe, Delft: Delft University Press. World Bank (1990) Proceedings from a Seminar on Housing Reforms in Socialist Economies, Washington, DC: World Bank.
Part IV
Social capital and social cohesion
12 Patterns of sociability in the enlarged EU Manuela Olagnero, Paola Torrioni and Chiara Saraceno
1. Spheres of sociability: substitution or integration? Sociability is a multi-faceted concept. It identifies distinct dimensions and levels of social relations in which individuals may be involved. According, for instance, to Paugam and Russel (2000), there are at least three spheres, or levels, of sociability: primary, secondary and tertiary. The primary sphere involves immediate family and household relations (co-residence). The secondary sphere concerns interactions (contacts) with neighbours, friends and relatives (not co-resident) and social and family support from people outside the household. The tertiary sphere relates to social participation (activity in political and associative domains). Depending on circumstances, these three spheres may re-enforce each other; or, on the contrary, high involvement in one sphere may weaken participation in another. In the literature on social movements, much attention has been paid to the role of secondary with respect to the tertiary sphere of sociability; several studies have found a relationship between the two dimensions. This finding may be interpreted in different ways. Diani (2003a: 6–7) argues that personal friends, relatives, colleagues and neighbours may all affect individual decisions to become involved in a social movement. Networks may provide opportunities for action through the circulation of information about ongoing activities, existing organisations, people to contact and reduction of the practical costs attached to participation. But also, the possibility of cross-pressure and of opposing mechanisms (where social participation can promote or re-enforce friendship) has been acknowledged. In the literature on patterns of civicness, strong involvement in one’s own family has been mainly interpreted as an obstacle to developing trust in a wider set of individuals and groups. Hence, it has often been directly taken as an indicator (or predictor) of lack of engagement in broader forms of social participation and in civic values (see e.g. Ginsborg 1998; Putnam 2002). In comparative welfare state studies as well, family solidarity has often been perceived either as a cause or a consequence of weak collective solidarity. At the same time, strong forms of collective solidarity have been perceived as a threat to, or a cause of weakening (‘crowding out’) of interpersonal, primary or even secondary solidarity and relationships. Recent research on intergenerational solidarity, however, offers evidence contrary to the crowding out hypothesis (e.g. Kohli 2005; Künemund and Rein 1999; Van Orschoot and Arts 2005). The aim of this chapter is therefore to assess (1) whether distinct patterns of sociability – i.e. of linking and balancing the spheres of social relations indicated above – exist in enlarged Europe; (2) whether they overlap either with political
280
Olagnero, Torrioni and Saraceno
(particularly that between old and new democracies) or welfare regime divides, or whether there are other – cultural, economic – dimensions to be taken into account in order to explain cross-country differences and similarities. The issue of divergence and convergence of European countries is at the centre of an ongoing debate (see for instance Mayer 2001 for the convergence theory; Heinz 2001; Mahon 2005; Tomka 2003 for the divergence thesis). Diversity, if not outright divergence, emerges as a feature not only of old EU-15 but also of the former socialist countries, both during communist rule1 and even more so after the fall of the communist regimes (see e.g. Sztompka 1999: 208–209; Illner 1999: 242–243, Ferrera and Rhodes 2000; Tomka 2003; Manning; 2004). Manning (2004), in particular, suggests that distinct welfare regime patterns may be identified not only within EU-15, but also within the former socialist group of countries.2 The empirical basis is constituted mainly by micro-data from EQLS (European Quality of Life survey). Compared with other data sets that provide information on social networks (ISSP, EVS, ESS)3 EQLS data are less detailed. Missing information, however, is counterbalanced by the extension of the study to 28 countries, and by the more recent timeframe in which observations were made (spring 2003) (see also Kohler, in this volume). Comparison with findings from other surveys will be made when relevant. The focus will be on the secondary and tertiary sphere of sociability, since household patterns are the object of another chapter (Saraceno, this volume, ch. 2). The main analytical question concerns the relationship between involvement in the secondary and tertiary sphere of sociability, that is, in some kind of community or collective participation. Are they somehow related or do they depend on different mechanisms? Do they form distinct mixes and balances that individuate countries and/or groups of countries? For the secondary sphere of sociability, two dimensions have been selected: contacts and support. The indicators concerning contacts are based on frequency and type (face to face or at a distance) of contacts with non co-resident family members and others (friends or neighbours). The indicators regarding support are based on the experience of having received support in case of extreme need and on expectations of receiving support in the case of specific emergencies. Social support is an informal resource, as it does not involve professional or institutional intervention (Pierce et al. 1990; Thoits 1984). The strategic relevance of support in the analysis of social cohesion is apparent, considering that it taps both the availability of a network and the ability of individuals to activate it in case of need. The experience of support testifies that a network is available and operates. Expectation of being able to receive support in case of an emergency (when urgently needing a sum of money, when ill, when in psychological distress) is an indicator of trust in one’s own network, thus of the feeling to be socially included. At the same time, the subjects from whom support is expected (family, friends, others) represent an indirect indicator of the variety of the informal network that is available to the individual. Receiving and giving support are thus embedded in the same process of social exchange. With regard to tertiary sociability, indicators are focused on participation in social activities, the target of which goes beyond the individual’s or household’s immediate interest. The indicators available are: participation in volunteer work, participation in organised political activities, personal mobilisation towards a politician in order to achieve a specific goal (what has been defined as ‘particularized contacting’: Sharp 1984). These indicators partly cover what, according to Diani (2003b: 302), is the full
Patterns of sociability
281
range of possible involvement: individual interaction with a politician, coordinated efforts towards achieving a collective aim, structured forms of participation based on formal membership in collective bodies regulated by some mechanisms of internal regulation. In the next section, a brief overview of the distribution and intensity of the different spheres of sociability is presented at the aggregate level.
2. A cross-country overview of sociability patterns 2.1 The distribution of the three spheres of sociability across Europe Table 12.1, based on an aggregation of countries by date of entrance in the EU, gives an overall picture of sociability and its components. It emerges that more people live alone in EU-15 than in the other two-country clusters. This is partly due to the higher incidence of extended households in the latter, which reduces the likelihood that the elderly live alone when their partner dies, and partly to the different patterns of entering adulthood and forming a family (Saraceno et al. 2005; Saraceno, this volume, Ch. 2). A higher incidence of living alone, however, does not result in fewer contacts with family outside the household. Actually, these contacts are less frequent in CC-3 than in both EU-15 and NMS. Overall, the family emerges as the main sphere of sociability throughout Europe. Yet friends play an important role as well, although more in EU-15 and less in CC-3. Social participation, as defined through the two indicators of participation in voluntary and political activities, on the contrary, appears reduced throughout Europe, and more so in NMS and CC-3 than in EU-15, although, as will be shown in the following sections, cross-country differences are substantial. Throughout Europe, both the experience of social isolation (absence or scarcity of social and/or family contacts) and that of lack of support in case of an emergency seem very limited. The only clear divide is apparent between EU-15 on the one hand (with the highest percentage of socially integrated respondents) and CC-3 on the other hand (with the highest incidence of respondents who are socially isolated). 2.2 The secondary sphere of sociability and the role of the family Family remains in Europe an important source of support at all levels: financial, practical, emotional (see also Saraceno et al. 2005). The large majority of Europeans feel they can rely on their families beyond the household boundaries when they are sick, or urgently need money, or psychological support. The extended family plays a particularly important role in NMS in buffering economic risks and overall in socially integrating individuals – such as the un- or under-employed and lone parents – who would otherwise risk isolation. Differences concern the relative prominence, or availability, of other kinds of support. Thus, family networks appear more exclusive in CC-3 than in EU-15 and NMS. In the latter two clusters, the role of non-family networks in providing support is greater, although not an alternative to the family network. Other differences emerge when focusing on the sources of support and direction of contacts at country level (Table 12.2). They seem to defy any clustering of countries either by welfare regime or by political history.4 A Scandinavian model can be roughly identified through the relative greater incidence of non-family rather than family networks in providing support. This pattern emerges
282
Olagnero, Torrioni and Saraceno
Table 12.1 Dimensions of sociability, by country clusters (%) Spheres
Dimensions
Indicators
Primary
Co-residence
Living with others Keeping in contact with family (children and/or parents) Keeping in contact with friends With experience of support received from anyone not living in the household With experience of support provided to anyone not living in the household Support expected in prevalence from family Support expected in prevalence from others Participating in voluntary activities Participating in political activities
Contacts Secondary Support
Tertiary
Social participation
EU-15 NMS-10 CC-3 74 65
85 65
91 47
78 9
65 18
59 21
19
26
39
64
72
68
19
13
15
17 13
9 7
4 6
Source: EQLS 2003 Notes: Question HH1: ‘Including yourself, can you please tell me how many people live in this household? Question Q34: On average, thinking of people living outside your household, how often do you have direct (face-to-face) contact with a) any of your children, b) your mother or father, c) any of your friends or neighbours? Question Q35: On average, how often do you have contact with friends or family by phone, e-mail or by post? Question Q23: Over the past month, have you: a) attended a meeting of a charitable or voluntary organisation, b) served on a committee or done voluntary work for a voluntary organisation? Question Q24a: Over the past year, have you attended a meeting of a trade union, a political party or political action group, attended a protest or demonstration, or signed a petition? Question Q36: From whom would you get support in each of the following situations? a) If you needed help around the house when ill? b) If you needed advice about a serious personal or family matter? c) If you were feeling a bit depressed and wanting someone to talk to? d) If you urgently needed to raise 1000 euros (in CC-3 and NMS countries: 500 euros) to face an emergency? Question Q62: In the past year, did your household give regular help in the form of either money or food to a person you know not living in your household (e.g. parents, grown-up children, other relatives, or someone not related)? Question Q63: In the past year, did your household receive regular help in the form of either money or food from a person you know not living in your household?’ ‘Keeping in contact’ combines high frequency of face-to-face contacts with family or friends (includes the modalities of Q34: ‘More than once a day’, ‘Every day, or almost every day’, ‘At least once a week’) with high frequency of contact at distance with family or friends (includes the modalities of Q35: ‘More than once a day’, ‘Every day, or almost every day’, ‘At least once a week’); ‘Support expected in prevalence from family’ and ‘support expected in prevalence from others’ refer to those who: a) indicate family or others as source for support (Q36) three times out of four; b) indicate family or others as an alternative to ‘nobody’ at least two times out of four. ‘With experience of support received from anyone not living in the household’: respondents answering yes to Question 63. ‘With experience of support provided to anyone not living in the household’: respondents answering yes to Question 62. ‘Participating in political activities’: respondents answering yes to item ‘a’ of Question 24; ‘Participating in voluntary activities’: respondents answering yes to item ‘a’ or ‘b’ of Question 23.
Denmark Finland Sweden Austria Belgium France Germany Luxembourg Netherlands Ireland United Kingdom Greece Italy Portugal Spain Czech Republic Hungary Poland Slovakia Slovenia Cyprus
28.7 26.7 21.8 17.8 20.3 18.4 15.0 16.8 21.0 17.2 22.2 22.1 15.0 16.7 24.2 22.7 31.4 25.2 22.5 24.4 48.6
Experience of giving material (food) and financial (money) support 10.0 12.6 7.3 13.0 6.7 10.1 7.9 6.4 9.7 8.2 10.1 19.5 6.5 11.6 12.0 14.8 19.9 17.2 16.6 9.4 10.1
Experience of receiving material (food) and financial (money) support 55.3 48.6 59.2 64.0 64.1 51.3 65.0 64.9 66.4 62.4 63.9 72.2 68.8 76.1 73.7 65.4 77.2 76.0 71.4 66.9 79.2
Expecting support from family
21.8 24.7 16.7 16.8 19.6 28.6 17.4 15.2 17.2 20.2 20.2 13.2 10.3 13.1 9.8 15.7 11.1 10.5 13.4 12.8 10.3
Expecting support from others
69.8 69.5 60.6 82.1 86.1 59.1 75.2 76.1 82.5 86.1 73.4 68.4 83.5 74.7 74.1 76.7 87.0 94.0 82.9 80.7 90.8
High frequency of face-to-face contacts with family
Table 12.2 The secondary sphere of sociability (type of support and contacts), by country (%)
85.7 89.7 82.7 89.4 82.8 79.5 87.8 84.7 89.8 96.7 87.1 89.7 90.4 96.2 90.4 79.9 84.5 89.2 88.8 88.4 80.4
High frequency of face-toface contacts with friends or neighbours
Continued
57.7 63.7 62.6 51.2 45.6 41.8 40.8 37.6 50.5 64.8 55.9 57.5 70.5 35.3 48.7 32.5 43.0 31.1 39.9 66.9 63.7
High frequency of contacts at distance
43.9 29.5 36.5 37.5 34.3 39.4 39.4 26.4
5.7 20.6 29.9 34.2 34.4 24.8 18.8 14.6
Experience of receiving material (food) and financial (money) support 76.3 53.0 47.3 62.2 58.7 73.7 66.4 65.3
Expecting support from family
4.9 26.1 32.5 20.5 27.1 12.8 14.4 17.0
Expecting support from others
95.0 82.8 75.4 62.0 71.3 85.9 65.6 77.5
High frequency of face-to-face contacts with family
90.0 86.7 83.9 88.6 90.8 88.6 89.0 87.6
High frequency of face-toface contacts with friends or neighbours
65.3 48.9 44.6 31.0 41.6 20.7 33.4 46.1
High frequency of contacts at distance
Notes: Giving material/financial support (Q62): % of people with experience of giving material (food) or financial (money) support to a person not living in the household; Receiving material/financial support (Q63): % of people with experience of receiving material (food) or financial (money) support from a person not living in the household; High face-to-face contacts with family (Q34): % of people with high frequency (more than once a day, every day or almost every day, at least once a week) of face-to-face contacts with family; High frequency of face-to-face contacts with friends or neighbours (Q34): % of people with high frequency (more than once a day, every day or almost every day, et least once a week) of face-to-face contacts with friends; High frequency of contacts at distance with family or friends (Q35): % of people with high frequency (more than once a day, every day or almost every day, at least once a week) of contacts with family or friends by phone, mail or post; Expecting support in prevalence from family (Q36): % of people who: a) indicate family as source for support three times out of four; b) indicate family as an alternative to ‘nobody’ at least two times out of four; Expecting support in prevalence from others (work colleagues; friends; neighbours, someone else) (Q36): % of people who a) indicate others as source for support three times out of four; b) indicate others as an alternative to ‘nobody’ at least two times out of four. The four areas of support are a) help around the house when ill; b) a serious personal or family matter; c) feeling a bit depressed and wanting someone to talk to; d) urgently raise 1000 euros to face an emergency. (In NMS and CC-3 the reference is 500 Euros).
Source: EQLS, 2003
Malta Estonia Latvia Lithuania Bulgaria Romania Turkey Total
Experience of giving material (food) and financial (money) support
Table 12.2 The secondary sphere of sociability (type of support and contacts), by country (%) – cont’d
Patterns of sociability
285
more clearly in Denmark and Finland (especially among women: Saraceno and Olagnero 2005), to a lower degree in Sweden, Latvia and Estonia. A non-Scandinavian country, Bulgaria, shows a similar pattern. A balance between family and non-family networks seems to identify the countries that are usually included in the continental model. Yet, some exceptions should be mentioned. France, for example, seems nearer to the Scandinavian pattern of support in showing a strong relevance of non-family social ties and in the somewhat reduced role of the family. Finally, a strong presence of family support identifies the Mediterranean countries, including Cyprus and Malta. But it is also found in some of the former socialist countries, such as Hungary, Poland, Slovakia as well as Romania and Turkey. Ireland, in its turn, appears to be closer to the Mediterranean model than to the UK. The analysis shows that also with regard to contacts with family and friends or neighbours it is quite difficult to identify specific and stable patterns at an aggregated level, except for the Scandinavian countries. Social isolation (limited contacts) is generally higher in some of the former socialist countries such as Lithuania, Czech Republic, Bulgaria and Romania. Yet Portugal and France show similar percentages. Intra-cluster differences emerge also with regard to the incidence of face-to-face contacts. 2.3 The third sphere of sociability and its domains Social participation, much higher in EU-15 than in NMS and CC-3, also shows substantial intra-country differences within EU-15. Table 12.3 offers an overview respectively of volunteer and political activities comparing EQLS with EVS, ISSP and ESS data. According to the EQLS source, respondents living in the Scandinavian and continental countries such as Denmark, Finland, Sweden, Luxembourg, the Netherlands, Austria, France, but also in Ireland and Cyprus, exhibit the highest degree of participation in volunteer activities (serving on a committee or doing voluntary work for voluntary organisations). According to EVS data, the United Kingdom, Sweden, Netherlands, Greece, Finland and Luxembourg are in the first rank with regard to voluntary work specifically in social welfare services. The leading position of the United Kingdom could be explained by the traditional role played by volunteer organisations at the community level in the UK (see e.g. Hall 2002). Respondents in some of the Scandinavian and Continental countries, such as Sweden and the Netherlands, are also the most active in religious groups. But they are joined by respondents living in countries with quite different religious and political traditions, such as Greece and Slovakia, as well as Austria. France lags behind Scandinavian countries, but also behind Germany, Belgium and United Kingdom. With respect to political activity, and according to aggregated information provided by EQLS,5 Scandinavian countries and Austria are again the most active. Findings about Sweden seem to disprove the presence of the ‘democratic discontent’, together with a slackening participation in politics, that in the 1990s some researchers noted in this country, especially among young people (Rothstein 2002).6 Germany appears to be a complex case of selective disengagement: a low rank concerning engagement in social welfare services is coupled with a remarkable position concerning activities in religious organisations and a modest incidence of membership in political organisations. Data from ISSP allow us to disentangle the different aspects of political participation and to observe their different distribution in the various countries.
286
Olagnero, Torrioni and Saraceno
Table 12.3 The tertiary sphere of sociability (activity in political and associative domains) Currently doing unpaid voluntary work for
Denmark Finland Sweden Austria Belgium France Germany Luxembourg Netherlands Ireland United Kingdom Greece Italy Portugal Spain Czech Republic Hungary Poland Slovakia Slovenia Cyprus Malta Estonia Latvia Lithuania Bulgaria Romania Turkey
Over the past month
In the past
Social welfare services for elderly, handicapped or deprived people
Religious or church organisation
Attended a meeting charitable or voluntary organisation
Served on a committee or done voluntary work for voluntary organisation
Participated in activities of trade unions or professional associations
EVS (1999/2000)
EVS (1999/2000)
EQLS (2003)
EQLS (2003)
ISSP (2001)
4.0 7.2 9.2 2.3 5.9 4.1 1.9 6.9 9.1 3.8 13.5 7.5 5.1 0.8 2.6 3.4 2.5 2.2 6.1 4.9 — 5.0 2.6 1.8 0.6 1.5 1.0 —
3.4 7.8 23.5 7.1 6.1 3.2 5.6 6.1 11.9 7.7 6.3 9.3 6.7 2.6 3.7 2.8 5.4 3.7 13.1 4.5 6.1 13.1 2.8 3.8 4.2 1.8 3.6 —
17.8 12.4 18.9 15.4 14.7 12.9 17.4 23.1 18.7 18.4 14.4 4.6 14.0 3.7 9.0 10.2 7.5 4.9 8.1 9.3 21.6 15.0 8.9 4.4 4.9 2.9 3.0 3.7
17.4 16.1 15.5 17.5 13.1 15.2 14.1 19.7 19.2 19.7 14.2 4.9 13.8 4.8 7.0 8.7 6.1 4.7 6.9 14.3 18.1 14.8 9.9 7.4 6.4 2.7 2.7 2.7
16.8 21.4 — 9.2 — 13.5 9.4 — — 7.3 — 9.5 — 3.8 10.8 5.5 6.3 — 9.9 18.1 — — 3.3 — — — —
Sources: EVS 1999/2000: ‘Please look carefully at the following list of voluntary organizations and activities and say which, if any, are you currently doing unpaid voluntary work for?’; ISSP 2001: ‘People sometimes belong to different kinds of groups or associations. The list below contains different types of groups. For each type of group, please tick a box to say whether you have participated in the activities of this group in the past 12 months. The table shows the % for respondents that have participated once, twice or more than twice in the activities of trade unions or professional associations or political parties or groups; ESS 2002/2003: ‘There are different ways of trying to improve things in [country] or help
Patterns of sociability
287
12 months Participated in activities of political parties or groups
Worked in political party or action group
Signed petition
Taken part in lawful public demonstration
Contacted politician or government official
Political activities
ISSP (2001)
ESS (2002/2003)
ESS (2002/2003)
ESS (2002/2003)
EQLS (2003)
EQLS (2003)
4.1 3.5 5.0 10.3 5.4 4.9 3.9 3.7 3.3 4.7 3.4 4.8 3.0 4.2 6.1 4.7 2.9 2.9 — 3.5 — — — — — — — —
28.2 24 40.8 27.3 33.9 34.8 30.4 28.9 22.4 27.6 40 4.7 17.4 7.3 24.2 16.1 4.2 6.9 — 11.8 — — — — — — — —
8.3 2 6.4 9.6 8.4 17.9 10.6 20.9 2.9 7.1 4.4 4.5 11 4.3 17.5 4.6 3.7 1.3 — 2.7 — — — — — — — —
5.5 6 — 7.9 — 7.4 6 — — 5.6 — 8.7 — 3.6 6.2 1.2 0.6 — 3.7 6.3 — — 0.6 — — — —
12 13.7 17 12.2 11.2 5.9 7.1 11.4 13.5 13.6 9 7.3 6 3.5 5.6 4.2 4 8.4 4.8 4.7 11.6 20.9 5.6 19.2 13.7 4.4 3.5 7.2
27.5 18.3 36.4 18.4 12.6 14.3 9.4 16.7 11.3 15.6 11.1 9.8 14.2 5.7 11.7 10.9 4.3 5.4 14.8 10.4 12.3 33.5 7.8 7.9 7.4 4.3 6.9 6.5
prevent things from going wrong. During the last 12 months, have you done any of the following?’ Included in the table are respondents that have done the following activities: Worked in political party or action group, signed a petition, taken part in lawful public demonstrations; EQLS 2003: ‘Over the past year, have you attended a meeting of a trade union, a political party or political action group, attended a protest or demonstration, or signed a petition?’ ‘Over the past year, have you contacted a politician or government official?’
288
Olagnero, Torrioni and Saraceno
The Scandinavian countries, together with the Netherlands, are near the top also in the case of individualised political contacts (measured by the action of contacting a politician). But the highest percentages are found in Malta and in the Baltic states Latvia and Lithuania. According to Sharp (1984: 656), two quite different mechanisms may encourage individualised political contacts: that of concern for community well-being and that of need. The former is positively associated with social status and is higher in richer areas. The latter is negatively associated with social status. Survey data do not, however, allow us to discover whether these two mechanisms coexist or are mutually exclusive and whether one or the other prevails within a given country. Austrian, Spanish, Belgian, Swedish, French, Greek, Irish and Czech respondents are the most engaged in political parties. Active participation in trade unions is a distinctive trait of Finland and Denmark, but also of the Czech Republic, Slovenia and France. The Eastern European countries (with the exception of Slovenia and the Czech Republic) show a clearly lower rate of participation in every domain of political engagement (trade unions and political parties) than all other countries. ESS data also allow us to single out some non-conventional forms of participation different from participation in organised groups. Signing a petition, an initiative that implies taking a public stance on some controversial issue, is a practice that involves a substantial minority of respondents only across Northern/Western Europe (with Sweden and the UK, who are in pole position, followed by France, Belgium, Germany, Luxemburg, Denmark and Ireland). In the Mediterranean group, only Spanish respondents emerge as politically active, particularly with regard to taking part in a public demonstration. Overall, participation in associative and political activity – as defined by the various surveys - seems to involve a substantial percentage of the population in a few countries only: the Scandinavian nations, but also Malta, constantly rank the highest in participation not only in social organisations but also in political parties and trade unions. The overview of the various forms of social participation proposed above provides a picture of Europe where traditional (political, religious, or welfare state) boundaries are somewhat blurred. The continental countries are similar to the insular countries. The Eastern European countries on average show the lowest degree of political participation. But they are also quite differentiated, possibly partly because of their pre-socialist political traditions. Only the Scandinavian countries seem to keep a systematic distinctiveness, although with meaningful internal differences. Cross-country differences in the degree and patterns of political participation may depend not only on different levels of individual political interest or engagement. They may also depend on different political and professional cultures and organisations, which cannot be considered here, that offer distinct options to participate in the various countries.
3. How European countries cluster around patterns of sociability Boundaries become even more blurred if pairs of indicators regarding both secondary and tertiary spheres of sociability are considered: countries of old and new Europe can move closer, ‘divorce’, or join, sometimes breaking the boundaries of well-settled typologies (e.g. that described by Esping-Andersen in 1990).
Patterns of sociability
289
If one relates two of the indicators of the secondary sphere (contacts with family and expectations of receiving support from family) that are commonly thought to specifically distinguish Mediterranean and continental countries from Scandinavian ones (Figure 12.1, left side), the expected similarities within the group of Mediterranean countries (Portugal, Spain and Greece) are confirmed. In all other cases, however, there are some unexpected divergences and convergences. France (low level both of family support and contacts) is distant from the other continental countries; Sweden is near Lithuania in showing a pattern in which both contacts and expectations of support from family are scarce. Denmark is close to Bulgaria, and Finland to Latvia, in showing a pattern in which family contacts are more relevant than family support. Poland, clearly separated from the other Eastern countries, is close to Malta and Cyprus in showing the highest scores for family support and contacts. Some Mediterranean and Eastern European countries appear interestingly similar, while two Baltic countries (Latvia and Lithuania) show similarities with Scandinavian nations. If participation in volunteer activities is related to participation in political activities (see Figure 12.1, right side), some other interesting displacements can be observed. Portugal is near to some Eastern European countries (low level of participation in volunteer and political activities), Finland is close to continental countries (middle level of participation in both fields of participation), whereas Sweden, isolated, holds the top position both in volunteer and political activities.
EUROPEAN CLUSTER
40
CY HU
PT
EU 15 NMS 10 CC3
MT RO PL
ES EL SK
IT
70
SI NL
TR
LT
60
SW
DE LU AT BE CZ IR UK BG DK EE
FR
50
FI
% People engaged in political activities
% People expecting support by family
80
EUROPEAN CLUSTER SW
EU 15 NMS 10 CC3
MT
30 DK
20 FI
IR FR BE NL CY ES OUK CZ EL SI DE EE RO LT LV SK
10
TR
LV
AT LU
IT
PT BG
HU
0 40 50 60 70 80 90 100 % High frequency of contacts with family
10 15 20 25 30 0 5 % People engaged in voluntary activity
Figure 12.1 Distribution of countries along two dimensions of the secondary and tertiary spheres of sociability Source: EQLS, 2003
290
Olagnero, Torrioni and Saraceno
These kinds of ‘crossing borders’ movements, and the related effect of reshuffling conventional boundaries, have already been observed in comparative welfare state research that goes below the aggregate level to look at specific policies. It is the case, for instance, of patterns of family and state provision of care in different countries (e.g. Alber 1995; Pfau-Effinger 2005; Tomka 2003; Bahle in this volume). In order to better understand whether and how different patterns of sociability in the secondary and tertiary sphere combine within each country, and whether they identify both specific patterns of overall sociability and distinct country groups, a cluster analysis was performed. The rationale of clustering is ‘to organise countries into categories that are assumed to have some underlying similarity. This is attractive partly because it allows a greater degree of simplicity and parsimony when examining inter-country differences, partly because it provides a theoretical underpinning for categorising and interpreting these differences’ (Berthoud and Iacovu 2004: 14–15). The results are shown in Figure 12.2. The analysis indicates that European countries are in effect clearly enough divided into distinct groups or clusters.7 One of them overlaps with Esping-Andersen’s socialdemocratic welfare regime. Another partly overlaps with the continental welfare regime, but it also includes the UK. The Mediterranean countries are however separated into different clusters and the same occurs with the former socialist countries. The order of presentation of clusters (and their naming) substantially follows their rank along the scale of intensity of social participation. 1
2
3
Wide range includes Denmark, Finland, and Sweden. The cluster is characterised by a collectively oriented sociability, channelled both through participation in associations and political parties and in informal networks, with high solidaristic expectations, and aptitudes for wide range contacts. The countries of this cluster all belong to the cluster that in comparative welfare state analysis is agreed as being universal, comprehensive and redistributive (Ferrera and Rhodes 2000). Their citizens appear characterised by wide range sociability. Commitment (to the community) includes most of the countries of EspingAndersen’s Continental type – the Netherlands, Austria, Belgium, Germany, Luxembourg, France – as well as the UK. This cluster has in common with the wide range group a high degree of associative participation, but it exhibits less political engagement. Furthermore, low incidences of giving support, but fairly high expectations of support are observable. Respondents in the countries of this cluster place great importance on face-to-face contacts. In case of need, support expected from others is also noticeable. On the basis of these characteristics, the overall pattern of sociability can be defined as oriented towards the community, including both family and the wider community of friends and groups one belongs to. Convergence includes Greece, Spain, Italy, Slovenia and Ireland. Some traits of a ‘complex sociability’ pattern seem to be common to the countries of this heterogeneous group: contacts both face to face and at a distance, importance of others in social contacts, and frequent expectations of being helped by family in case of need. Overall, this pattern of sociability can be defined as convergent, since it depends both on family and non-family networks, it relies on both the public and private sphere, and includes both proximity and distance as dimensions of relationships. In contrast to the commitment cluster, there is little role for membership in associations.
Patterns of sociability
291
Rescaled Distance Cluster Combine CASE Label
0
5
15
25
Num
Netherlands Austria Belgium United Kingdom Germany Luxembourg France Estonia Latvia Bulgaria Hungary Poland Czech Republic Slovakia Portugal Lithuania Turkey Romania Denmark Finland Sweden Greece Spain Italy Slovenia Ireland
Figure 12.2 Patterns of sociability in Europe (country clusters) Source: EQLS 2003 Notes: In the dendrogram, the length of the horizontal lines (branches) indicates the distance between the subgroups that are joined together; the longer the branch, the greater the distance between the countries that come together in the cluster. The cut-off point identifying the clusters is represented by the black vertical line: we have chosen to identify clusters at a distance as close as possible to the 0–5 range.
4
5
Proximity includes Hungary, Poland, Czech Republic, Slovakia and Portugal. The countries included in this group share a pattern of sociability built mainly on face to face contacts, particularly with family, which is also the main source of experienced and expected support: a sociability characterised by public disengagement, private exchanges and family cooperation. and 6 Mutual Help (including Estonia, Latvia, Bulgaria), and closeness (including Lithuania, Turkey and Romania). These two groups share the same low degree of involvement in politics, a routine experience of mutual help, a traditional (face to face) pattern of being in contact with friends. Nevertheless, a clear difference can be found between the first group and the second with regard to some aspects
292
Olagnero, Torrioni and Saraceno of secondary sociability and its internal composition. In the mutual help cluster, the relevance of non-family networks as sources of support, the widespread ability to maintain contact also at a distance, the well-balanced experience of giving/receiving, shape a pattern of sociability based on mutual cooperation also outside the family. In the closeness cluster, the relevance of family networks as sources of support, the strong relevance of face-to-face contacts, a quite unbalanced flux of giving/receiving material/financial help, point to a pattern of sociability mostly restricted to the family.
Country clusters based on patterns of sociability, therefore, only partly overlap with welfare regimes. But are there other social variables that may account for both differences and country clustering? These variables may be macro, or structural. That is, they may be related to the overall level of living of a country. They may also be micro. They may refer either to the distribution and country specific relevance of individual characteristics or to the distribution and relevance of specific characteristics of local contexts. At the macro level, three ‘structural’ indicators have been selected. The first, GDP, concerns the general economic performance of a country and the degree of well-being at the aggregate level. The second, the employment rate, indicates access both to income and to non-family relationships. The third, percentage of internet users, indicates not only the degree of technological literacy, but also the degree to which individuals have the means to keep in contact, and also develop participation, beyond face-to-face situations and even beyond the use of the telephone. Figure 12.3 (a, b, c) shows that clusters corresponding to distinct patterns of sociability are quite consistent also with respect to the selected variables. Two clear blocs, internally homogeneous, emerge in Figure 12.3a. The wide range and commitment clusters are clearly located near to the right side of the table, in which the purchasing power is higher. On the left side, where poorer countries are located, we also find the mutual help and closeness clusters. The proximity and the convergence clusters are the two most internally heterogeneous, as indicated by the high standard deviation. They point to a transitional area, in which socio-economic inequalities translate also in differences, if not inequalities, in access to a variety of social networks. Figure 12.3b shows a less polarised configuration in terms of both the intensity of the indicator and its variance. Generalised participation in paid work is a precondition for an articulated pattern of sociability, while low employment rates are associated with a more restricted pattern of sociability. In Figure 12.3c, the position of clusters is very similar to that of the previous one, but the closeness cluster further increases its distinctiveness and isolation. A low standard of living as indicated by the GDP, a low employment rate, a scarce diffusion of internet use (all these traits are more frequent in the mutual help and closeness clusters), illustrate some structural conditions that constrain sociability within the boundaries of family and close community to a larger degree than in other countries of Europe. As for the proximity cluster, with its intermediate position and relatively high degree of internal heterogeneity, it emerges like a ‘bridging’ cluster, between Western and Eastern Europe. In order to verify whether cross-cluster differences in patterns of sociability may depend on a different distribution of micro and meso characteristics, two sets of indicators have been selected. One includes individual properties pointing to different
Patterns of sociability
293
GDP per capita (PPS) (2002)
standard deviation
3000 Proximity
2500
Convergence
2000 Commitment
1500 Mutual help 1000
Wide
Closeness 500 0 0
5000
10000
15000 Mean GDP
(a)
20000
25000
30000
Total employment rate (age 15–64) (2002) 7.0 standard deviation
Proximity 6.0 Closeness 5.0
Mutual help
Commitment
4.0 Wide 3.0 Convergence 2.0 1.0 45
50
55
60
(b)
65
70
75
Mean employment rate total
standard devittion
Internet user rate (per 100 cit.) (2001) 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0
Mutual help Commitment Proximity
Wide
Closeness 0
(c)
Convergence
5
10
15
20 25 30 Mean internet user rate
35
40
45
50
Figure 12.3 Distribution of sociability clusters with regard to selected structural indicators Source: For GDP per capita: Eurostat, 2004; for employment rate Eurostat, 2004; for internet users (EU 15 countries): Human Development Report 2002/2003 (UNPD, 2003); (NMS AND CC3), Eurostat 2003
abilities, opportunities and normative obligations towards sociability (gender, age, educational level, employment condition, professional category, income quartile, religious affiliation8). The other includes resources and constraints related to physical and organised contexts of living and working (size of living area, economic precariousness,9 number of paid working hours of employed people, internet usage). These sets of indicators have been inserted, as independent variables, in two different regression models (through a multiple OLS regression) in order to analyse their impact on the degree of engagement, respectively in public and private sociability (Table 12.4a and 12.4b).10
Lone parent (in nuclear or extended household) Gender Living in a medium/large city Number of hours worked per week
Highest quartile
−0.060 (*)
Employed
−0.061 (*)
Living in a medium/ large city
Highest quartile
Living with parents 2
Self employed
Professionals, Managerial University education Other nonmanual professionals Attendance at religious services
Mutual help
0.035 (**)
0.048 (*)
0.050 (*)
0.065 (*)
0.114 (*)
0.146 (*)
0.180 (*)
Beta coeff.
−0.038 (*)
Living in a medium/ large city
Attendance at religious services
Employed
University education Professionals, Managerial Other nonmanual professionals Gender
Proximity
0.064 (*)
0.069 (*)
0.078 (*)
0.106 (*)
0.164 (*)
0.175 (*)
Beta coeff.
Number of hours worked per week
Number of hours worked per week)
Gender
Secondary education
Professionals, Managerial University education Other nonmanual professionals Attendance at religious services
Convergence
0.028 (**) −0.032 (*)
0.034 (*)
0.037 (*)
0.050 (*)
0.054 (*)
0.093 (*)
0.111 (*)
0.128 (*)
0.175 (*)
Beta coeff.
−0.052 (*) −0.087 (*)
0.043 (**)
0.044 (**)
0.065 (*)
0.070 (*)
0.092 (*)
0.093 (*)
0.132 (*)
0.156 (*)
Beta coeff.
Lone parent (in nuclear or extended household) Age
Self-employed
Gender
Medium quartiles
Other non-manual professionals
Professionals, Managerial Highest quartile
Closeness
−0.085 (*)
0.057 (*)
0.063 (*)
0.066 (*)
0.072 (*)
0.100 (*)
0.115 (*)
0.147 (*)
Beta coeff.
Notes: * p = <0.05; ** p = <0.10. Multiple OLS regression model; dependent variable: index of engagement in public sphere; model summary (wide): constant = 0.221; R-adj = 0.077; N (valid) = 2305; model summary (commitment): constant=0.281; R-adj = 0.086; N (valid) = 4051; model summary (convergence): constant = 0.249; R-adj = 0.087; N (valid) = 2023; model summary (proximity): constant = 0.041; R-adj = 0.10; N (valid) = 2966; model summary (mutual help): constant = 0.194; R-adj = 0.087; N (valid) = 1770; model summary (closeness): constant = 0.175; R-adj = 0.043; N(valid)1703; Reference categories: education (none and primary together); gender (woman); income level (lowest quartile); occupational status (unemployed); occupational class (skilled and not skilled workers); living size area (open country and small village). Frequency religious services: max = more than once a week; min = less than once a year. The table shows the standardized coefficients. All coefficients are significant.
Source: EQLS 2003
0.034 (**)
0.047 (*)
Living as a couple with children (in nuclear or extended household) Other non manual professionals
0.051 (*)
Lone parent (in nuclear or extended household) Living with parents
Highest quartile
0.064 (*)
0.050 (*)
Self-employed
0.071 (*)
Medium quartiles
Living as a couple with children (in nuclear or extended household) Living as a childless couple (in nuclear or extended household) Professionals, Managerial
0.096(*)
0.165 (*)
0.200 (*)
University education Attendance at religious services Secondary education
Attendance at religious services University education Professionals, Managerial
Beta coeff. Commitment
Wide range
Table 12.4a Involvement in public sociability, by cross-cluster
−0.078 (*)
−0.100 (*)
−0.212 (*)
Employed
University education
Age
Age Living in a medium/ large city Employed Gender (man)
Lone parent (in nuclear or extended household) Living as a couple with children (in nuclear or extended household) Number of hours work(ed) per week
−0.075 (*) −0.092 (*)
−0.047 (*) −0.060 (*)
−0.031 (**)
0.031 (*)
0.034 (**)
0.064 (*)
Living with parents
Age
Employed
Living in a medium/large city
Other non-manual professionals Secondary education
Professionals, managerial
Living as a childless couple (in nuclear or extended household) Self-employed
Convergence
−0.225 (*)
−0.080 (*)
−0.072 (*)
0.048 (*)
0.050 (*)
0.051 (**)
0.074 (*)
0.087 (*)
0.099 (*)
Beta coeff.
Age Gender (man)
Being able to make ends meet
Living as a childless couple (in nuclear or extended household) University education Attendance at religious services Lone parent (in nuclear or extended household) Self-employed
Living as a couple with children (in nuclear or extended household) Secondary education
Proximity
−0.050 (*) −0.081 (*)
0.110 (*) Living as a childless couple (in nuclear or extended household) 0.092 (*) Living as a couple with children (in nuclear or extended household) 0.089 (*) Attendance at religious services
Closeness
Internet usage
−0.106 (*) Lone parent (in nuclear or extended household) Employed Gender (man) Medium quartiles Being able to make ends meet Age
0.054 (*) Secondary education
0.084 (*) Professionals, Employed managerial 0.080 (*) Living with Other nonparents manual professionals Professionals. 0.071 (*) Other nonManagerial manual professionals
University education
Medium quartiles
Secondary education
Beta Mutual help coeff.
−0.036 (**) Being able to make ends meet
0.033 (**)
0.055 (*)
0.068 (*)
0.069 (*)
0.074 (*)
0.093 (*)
0.138 (*)
Beta coeff.
−0.128 (*) −0.129 (*)
−0.047 (**) −0.075 (*) −0.083 (*)
0.061 (*)
0.065 (*)
0.072 (*)
0.091 (*)
0.107 (*)
0.107 (*)
0.136 (*)
0.185 (*)
Beta coeff.
Notes: *→ p = <0.05; **→ p = <0.10. Multiple OLS regression model; dependent variable: Index of engagement in private sphere; model summary (wide): constant = 9.672; R-adj = 0.060; N (valid) = 2292; model summary (commitment): constant = 8.567; R-adj = 0.04; N (valid) = 4045; model summary (convergence): constant = 8.583; R-adj = 0.070; N (valid) = 2009; model summary (proximity): constant = 8.359; R-adj = 0.041; N (valid) = 2933; model summary (mutual help): constant = 9.209; R-adj = 0.070; N (valid) = 1746; model summary (closeness): constant = 10.197; R-adj = 0.080; N(valid)1612; reference categories: education (none and primary together); gender (woman); income level (lowest quartile); occupational status (unemployed); occupational class (skilled and not skilled workers); living size area (open country and small village). Frequency religious services: max = more than once a week; min = less than once a year; Able to make ends meet (not being able to make ends meet). The table shows the standardized coefficients. All coefficients are significant.
Source: EQLS 2003
−0.045 (*)
Self-employed
Living with parents
Being able to make 0.057 (*) ends meet Medium 0.047(**) quartiles
Living in a medium/ large city
0.047 (*) 0.037 (*)
Attendance at religious services
0.069 (*)
0.069 (*)
Secondary education
0.081 (*)
0.089 (*)
Beta coeff.
Lone parent (in nuclear or extended household) Highest quartile
Internet usage
Commitment
0.090 (*)
Beta coeff.
Attendance at religious services
Wide range
Table 12.4b Involvement in private sociability, by country cluster
296
Olagnero, Torrioni and Saraceno
Table 12.4a shows that in all clusters, individual, rather than meso, properties have a positive impact on participation in tertiary sociability. But the relevant properties differ somewhat between clusters. High education plays a crucial role in favouring tertiary sociability in all clusters, but less in mutual help and closeness than in the others. Being integrated into a religious community is important everywhere except in the closeness cluster, but especially so in wide range and commitment. Belonging to a high or middle professional category counts remarkably in commitment, convergence, and in all former socialist country clusters. It has low relevance in wide range. Employment status is positively associated with participation in tertiary sociability particularly in proximity and mutual help. In the closeness cluster, having a good economic position also favours engagement in tertiary sociability. Clear obstacles to public engagement are, on the contrary, represented more often by meso factors (size of living area, number of hours worked, especially in mutual help and proximity, but also in convergence) than by individual properties. Only in closeness might the negative relationship between (older) age and public engagement point to a social divide between young and old people. The background conditions of engagement in secondary sociability also seem to differ across clusters (Table 12.4b), although less clearly. In the wide range and (less clearly) in the commitment clusters, risky life course transitions or conditions (such as being unemployed or being a lone parent) seem to have an incentivating effect on engagement in family and other private informal relationships. In proximity, the strength of private networks is particularly related to the family routine (being a couple with children), but also to the presence of some difficulties in making ends meet. In mutual help, having a middle/high level of education and income, being employed but also facing great difficulties to make ends meet, increase the likelihood of being included in strong ties. This is even truer in closeness. Overall, throughout Europe belonging to the high-middle social strata seems to favour involvement in secondary sociability. But in mutual help and closeness, that is in the clusters that comprise most Eastern European countries, material/financial need has a further important role in the activation of these relationships.
4. Different private/public balances: neither crowding out nor amoral familism The patterns of sociability identified above represent specific, and in some case divergent, balances between involvement in secondary and tertiary sociability. While individual and collective returns of social networks can interlace,11 the aims of secondary and tertiary sociability are clearly distinct. Secondary sociability concerns private aims and resources, involving individuals and close communities of families, kinships and friends. Tertiary sociability concerns involvement in public activities and in institutionalised groups, either to fulfil individual interests through collective means, or to achieve collective aims in the public interest. As indicated above, the two combine quite differently in the various country clusters. The spectrum of combinations is wide, but the two extreme poles are not symmetrical. At one pole of the spectrum, in fact, which identifies mainly the Eastern European countries, there is a strong asymmetry in favour of the secondary, private and mostly circumscribed to family, sphere. At the other pole, which identifies the Scandinavian countries, the situation is
Patterns of sociability
297
not reversed. On the contrary, strong involvement in a more diversified, private sphere (family, but also friends and others) combines with a comparatively strong involvement in the public sphere. Involvement in relations outside the family and in the public sphere does not weaken relationships within the family and close community. This finding may be related to the evidence emerging from studies on the impact of welfare state provision and generosity on family solidarity. These studies (e.g. Kohli 1999; Arts et al. 2003; Kohli et al. 2005) show that the proportion of persons who provide support to family members is higher in the Northern European and Scandinavian countries than in the Mediterranean ones, although the intensity of the support provided is greater in the latter. Together, these two types of findings, therefore, seem to disconfirm the ‘crowding out’ hypothesis.12 Rather, they suggest that exclusive involvement in family contacts and dependence on family solidarity may weaken the availability to be engaged in a wider range of relationships. This exclusivity, in turn, may be a consequence of lack of resources for and in the public sphere. These findings may also offer a new perspective in the long-standing debate over so called ‘amoral familism’ (Banfield 1958; Altan 1986) and to the critiques that have been levied against this concept and particularly to the causal mechanism it postulates. According to the supporters of the amoral familism thesis, strong involvement and loyalty to the close family on the one hand and engagement in social participation on the other hand are opposite behaviours and belong to opposite cultural complexes. The former is an indicator of particularism, the second of civicness. Critics of this theory have mainly argued against the supposed causal priority of family solidarity in producing a low degree of engagement in the public sphere. The issue is whether the absence or weakness of civicness is a consequence of the exclusiveness of family loyalties and obligations or, as Pizzorno (1976) argued, this exclusiveness is a consequence of lack of other options, of a weak public associative sphere that does not solicit information, reciprocity, and thus trust (see also Negri and Sciolla 1996; Sciolla, 1997). Exclusive reliance on family, however, may also be the negative output of institutional inefficiency in providing and organising collective resources and services (Rothstein and Stolle 2003; Antoci et al. 2006). This thesis receives support in the finding that a strong asymmetry in the balance between involvement in the private and public sphere of sociability and a restriction of the private sphere to family and face-to-face contacts is more easily found in the poorest ‘old EU’ countries and in the Eastern European countries. Disaffiliation, rather than ‘familism’, might be the underlying process, at least in the latter countries, particularly in the light of ‘the civil unrest that had been growing throughout the late 1980s in socialist societies’ (Manning 2004: 215). The wide social mobilisation of those years, which involved a large share of the population in spontaneous and organised political activities, seems to have vanished, for the time being, following the fall of the regimes and the difficulties of the transition period. It is not possible to explore here the phases of development and crisis of former socialist countries in this respect. A possible explanation could lie in the ambivalence embedded in the coexistence of participation, increasing ‘commercialism’, and declining attention to social needs. To this, one should add a ‘new gender inequality, by which while the majority of new actors were men, women have had to continue to cope with double burdens of paid work and family responsibilities that
298
Olagnero, Torrioni and Saraceno
have been a mark of state socialism, but under more trying circumstances’ (Manning 2004: 215).
5. Conclusions Our initial questions were: do distinct patterns of sociability exist in the enlarged Europe? Are European countries converging in institutional, behavioural and cultural patterns or are they instead diverging? Furthermore, do the observed patterns of sociability overlap with political or welfare regime divides, or do the European countries group together and divide according to other criteria? The country aggregations emerging from the cluster analysis based on indicators of sociability clearly indicate that distinct patterns of sociability exist and that they may not be easily derived either from political or welfare state traditions. Nevertheless, some overlapping with the clusters of Esping-Andersen’s typology can be found. This overlapping concerns Scandinavian countries, and partly continental nations. The former all belong to the same sociability cluster – wide range. The latter also belong to a sociability cluster of their own, commitment. But in this case, they share this pattern with a country, the UK, which has a largely different welfare state framework and political culture. All the Southern European countries, with the exception of Portugal, share a specific sociability pattern – convergence – indirectly confirming the thesis of those who, disagreeing with Esping-Andersen, argue that the Mediterranean countries constitute a specific welfare regime of their own, distinct from the continental one. However, the Southern European countries share this pattern of sociability with Ireland and Slovenia as well: a liberal (but Catholic) and a former socialist country, respectively. Former socialist countries are distributed across four different patterns of sociability: compatibility, proximity, mutual help and closeness. This seems to confirm that former socialist countries are as differentiated as old EU-15 countries. The concept of convergence to diversity (Boh 1989) seems to fit better with our findings than the categories of divergence and convergence. Comparing the two groups of countries (EU-15 and former socialist) as if they were internally homogeneous, therefore, makes little sense. At the same time, even the established method of clustering countries on the basis of certain welfare state categories does not appear to provide a satisfactory explanation as regards this area of behaviour. High standards of living, high rates of employment and high levels of technological literacy, seem to account for the way the countries cluster with regard to patterns of sociability and more precisely with regard to patterns of balancing, or unbalancing, involvement in secondary and tertiary sociability and in private and public relationships. Particularly, a strong asymmetry in favour of the private, mostly family sphere has been observed in the clusters that comprise exclusively former socialist countries. But at the other pole, which identifies the Scandinavian countries, the situation is not reversed. On the contrary, strong involvement in a more diversified private sphere combines with a comparatively strong involvement in the public one. The compatibility of engagement in both spheres of sociability, observed especially in the wide range, but also in the commitment and convergence clusters, does not offer ground for supporting the hypothesis that a strongly solidaristic welfare state weakens private solidarities, or that involvement in public life and relations
Patterns of sociability
299
undermines close, private relationships. At the same time, there is no evidence that a strong involvement in close relationships, particularly with family, may be interpreted as a form of ‘amoral familism’, according to which loyalty towards family weakens loyalty to, and interest in, the collective good. The hypothesis that a balance of sociability tipped towards relationships with near kin may be an indicator of disaffiliation does not find support as well in the data. It is not possible to say that it is refuted, however, since other information would be necessary to test it, both at the micro and macro level. A final word of caution is necessary. Differences between European countries are underpinned by such a great range of unobserved factors (not only concerning sociability), that these findings can only be considered as conjectures (Blekesaune and Quadagno 2003; Delhey and Newton 2005). Whether the balance between private and public sociability depends on individual traits of citizens, or on collective properties (shared cultural traditions, inheritance of role models, availability of opportunities) of a local, and sometimes path dependent, sort, or even on overall institutional properties of political and economic regimes, must therefore still remain an open question.
Notes 1 For example Estonia had for years the highest standard of living of the USSR, old Czechoslovakia had been on the verge of west European economic and social compatibility in the 1930. Slovenia had been the most liberal and wealthiest of the regions in Yugoslavia. According to different typologies of social policy and reform considered by Manning, also Czech Republic, Hungary and Poland enter in the group of eastern countries closer to western ones (more numerous and efficacious reforms, social programmes approaching western welfare state patterns, a more liberal oriented approach in economic development) (see Manning 2004: 219). 2 According to Manning (2004), former socialist societies had in common the allocation of substantial welfare resources through the enterprise, which made employment a key passport to a variety of social support resources (even food). 3 ISSP (International Social Survey Program): for the analysis we comment the data of the module ‘Social Networks II’ performed in 2001. EVS (European value Survey): data reported have been collected in the third wave 1999/2000 (for more details see Halman 2001). ESS (European Social Survey): for the analysis we use the data of the first round, performed in 2002/2003 (for more details see European Social Survey: Round 1, End Of Grant Report, 2004). 4 In the present analysis the geo-political country aggregations are named in the following ways: Scandinavian countries (Denmark, Finland, Sweden), Mediterranean countries (Italy, Greece, Portugal, Spain), Insular E15 countries (Ireland, United Kingdom), Insular NMS countries (Malta and Cyprus), Continental European countries (Austria, Belgium, France, Germany, Luxembourg, the Netherlands), Eastern European countries (Czech Republic, Hungary, Poland, Slovakia, Slovenia), Baltic States (Estonia, Latvia, Lithuania), candidate countries (Bulgaria, Romania, Turkey). 5 In the EQLS questionnaire, it is not possible to distinguish between non-conventional and conventional participation (signing a petition, taking part in a protest or demonstration are put together with being active in political structures or in voluntary organisations), or to detect different domains of conventional participation (trade unions are classified together with political parties). 6 According to EQLS data, the Swedish respondents show a constantly high degree of political engagement, even in the young age brackets (the participation rate for people aged 18–24 is 38 per cent, it amounts to 43 per cent for those aged 25–34, and to 42 per cent for those aged 35–44).
300
Olagnero, Torrioni and Saraceno
7 We used the complete linkage method: the distances between clusters are determined by the greatest distance between any two objects in the different clusters (‘furthest-neighbour method’). This technique tends to produce consistent and compact clusters with strong internal homogeneity. The distances between countries are Euclidean distances (see Lance and Williams 1967a, 1967b; Blashfield 1980). We have chosen to identify clusters at a distance as close as possible to the 0–5 range. The choice of a strict standard has allowed us to isolate clusters in which the countries are in most cases homogeneous. In a first phase we considered all 28 countries. Different controls were performed, resulting in a steady cluster configuration for 26 countries out of 28. For Malta and Cyprus it was not possible to identify a clear position: in some controls each of them constituted its own cluster, while in other cases the two countries formed together an autonomous cluster, but the distance between the two countries was considerable (higher than the range chosen as the level of cluster identification). Since the focus of the analysis is to discover groups having the highest possible internal homogeneity, we decided to eliminate Malta and Cyprus from the analysis altogether. We chose the solution that would represent the best rapport between ‘parsimony’ of the solution (few cluster of countries), semantic interpretation of the clusters, and minimum loss of information overall. The cophenetic correlation coefficient (Sokal and Rohlf 1962) between the original distances and the reproduced distances of the countries of our sample is 0.70. The output of the cluster analysis has been controlled by a discriminant analysis. This technique is used for the cross-validation of the classification of cases. The findings are coherent in 98 per cent of cases: only for Malta and Cyprus there is no convergence. 8 In this work, attendance to religious activities is used as an indicator not only of belonging to a specific religious community, but of overall of integration in a community. We are however aware that belonging to a specific religion might have a substantial and differential impact on sociability, as Durkheim’s work on suicide suggested over a century ago. 9 The indicator of precariousness is subjectively defined as great or very great difficulty to make ends meet. It is assumed as a given constraint, rather than an outcome or an input from and towards specific actions. Economic precariousness is here measured by scaling the answers to the question (in EQLS) concerning the ability of people to make ends meet. 10 The index of engagement in the secondary/private sphere has been constructed summing up the different relationships with family (outside the household) and non-family members of one’s network, taking account both of the mode (face-to-face or at a distance) and of intensity. The index of engagement in the tertiary/public sphere has been constructed summing up, with different weights, the various activities respondents are involved in. 11 As Putnam (2002) remarks, private returns can develop from public networks. 12 The crowding out hypothesis states that the stronger the welfare state, thus collective solidarity, the weaker family solidarity (see e.g. Van Orschoot and Arts 2005).
References Alber J. (1995) ‘A framework for the comparative study of social services’, Journal of European Social Policy, 5, 2: 131–149. Altan T. (1986) La nostra Italia, Milano: Feltrinelli. Antoci, A., Sacco, P. and Vanin, P. (2007) ‘Social capital accumulation and the evolution of social participation’, Journal of Socio-Economics, 36: 128–143. Arts, W., Halman, L. and Van Oorschot, W. (2003) ‘The welfare state: villain or hero of the piece?’, pp. 275–310, in W. Arts, J. Hagenaars and L. Halman (eds), The Cultural Diversity of European Unity, Leiden: Brill. Bahle T. (2008) ‘Family policy patterns in the enlarged EU’, in this volume. Banfield, E. (1958) The Moral Basis of a Backward Society, Glencoe, IL: The Free Press. Berthoud, R. and Iacovu, M. (2004) ‘Introduction’, pp. 1–20, in R. Berthoud and M. Iacovu (eds), Social Europe: Living Standards and Welfare States, Cheltenham, UK, Northampton, MA: Edward Elgar.
Patterns of sociability
301
Blashfield, R.K. (1980) ‘The growth of cluster analysis: Tyron, Ward and Johnson’, Multivariate Behavioral Research, 15: 439–458. Blekesaune, M. and Quadagno, J. (2003) ‘Public attitudes toward welfare policies: a comparative analysis of 24 nations’, European Sociological Review, 19, 5: 415–427. Boh, K. (1989) ‘European family life patterns: a reappraisal’, pp. 265–298, in K. Boh, M. Bak, C. Clason, M. Pankratova, J. Qvortrup, G.B. Sgritta and K. Waerness (eds), Changing Patterns of European Family Life: A Comparative Analysis of 14 European Countries, London, New York: Routledge. Delhey, J. and Newton, K. (2005) ‘Predicting cross national levels of social trust: global pattern or Nordic exceptionalism?’, European Sociological Review, 21, 4: 311–327. Diani, M. (2003a) ‘Introduction. Social movements, contention actions and social networks: from metaphor to substance?’, pp. 1–20, in M. Diani and D. McAdam (eds) Social Movements and Networks: Relational Approaches to Collective Action, Oxford: Oxford University Press. Diani, M. (2003b) ‘Networks and social movements: a research programme’, pp. 299–319, in M. Diani, D. McAdam (eds), Social Movements and Networks: Relational Approaches to Collective Action, Oxford: Oxford University Press. Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism, Cambridge: Polity Press. European Social Survey (2004) ‘European Social Survey: round 1, end of grant report’, Online. Available http: (accessed 11 January 2007). Eurostat (1) (2003) Eurostat Yearbook 2003, Luxembourg. Eurostat (4) (2004) GDP per capita in Purchasing Power Standards for EU, Candidate Countries and EFIA, Statistics in Focus, Economy and Finance, no. 27. Ferrera, M. and Rhodes, M. (2000) ‘Recasting European welfare states: an introduction’, pp. 1–10, in M. Ferrera and M. Rhodes (eds) Recasting European Welfare States, London: Frank Cass. Ginsborg, P. (1998) L’Italia del tempo presente. Famiglia, società civile, Stato 1980–1996, Torino: Einaudi. Hall, P. (2002) ‘The role of government and the distribution of social capital’, pp. 21–58, in R. Putnam (ed.), Democracies in Flux: The Evolution of Social Capital in Contemporary Society, Oxford: Oxford University Press Halman, L. (2001) The European Values Survey: A Third Wave. Source Book of 1999/2000 European Values Study Surveys, Tilburg: Tilburg University Press. Heinz, W. (2001) ‘Il lavoro e il corso della vita. Prospettive comparative’, pp. 333–352, in C. Saraceno (ed.), Età e corso della vita, Bologna: Il Mulino. Illner, M. (1999) ‘Second thoughts on the transformation in Eastern and Central Europe’, pp. 234–248, in T. Boje, B. van Steenbergen and S. Walby (eds), European Societies: Fusion or Fission?, London: Routledge. ISSP (2004) Social Networks II, Köln: Zentralarchiv für Empirische Sozialforschung. Kohli, M. (1999) ‘Private and public transfers between generations: linking the family and the state’, European Societies, 1: 81–104. Kohli, M. (2005) ‘Intergenerational transfers and inheritance: a comparative view’, pp. 266–289, in M. Silverstein, R. Giarrusso and V.L. Bengtson (eds), Intergenerational Relations across Time and Place, New York: Springer. Kohli, M., Künemund, H. and Vogel, C. (2005) ‘Intergenerational family transfers in Europe: a comparative analysis’, paper for the Research Network on Ageing at the 7th European Sociological Association (ESA) Conference, Torun, September 9–12. Künemund, H. and Rein, M. (1999) ‘There is more to receiving than needing: theoretical arguments and empirical explorations of crowding in and crowding out’, Aging and Society, 19, 1: 93–121.
302
Olagnero, Torrioni and Saraceno
Lance, G.N. and Williams, W.T. (1967a) ‘A general theory of classificatory sorting strategies I: hierarchical systems’, Computer Journal, 9, 4: 373–380. Lance, G.N. and Williams, W.T. (1967b) ‘A general theory of classificatory sorting strategies II: clustering systems’, Computer Journal, 10, 3: 271–277. Mahon, R. (2005) ‘Child care: toward what kind of social Europe?’, Social Politics, 9, 3: 343–379. Manning, N. (2004), ‘Diversity and change in pre-accession Central and Eastern Europe since 1989’, Journal of European Social Policy, 14, 3: 211–232. Mayer, U. (2001) ‘The paradox of global social change and national path dependencies: life course patterns in advanced societies’, pp. 89–110, in A. Woodward and M. Kohli (eds), Inclusions and exclusions in European societies, London: Routledge. Negri, N. and Sciolla, L. (eds) (1996) Il paese dei paradossi. Le basi sociali della politica in Italia, Roma: Carocci. Paugam, S. and Russell, H. (2000) ‘The effects of employment precarity and unemployment on social isolation’, pp. 243–263, in D. Gallie and S. Paugam (eds), Welfare Regimes and the Experience of Unemployment in Europe, Oxford: Oxford University Press. Pfau-Effinger, B. (2005) ‘Welfare state policies and development of care arrangements’, European Societies, 7, 2: 321–347. Pierce, G.R., Sarason, B.R. and Sarason, I.G. (1990) ‘Integrating social support perspectives: working models, personal relationships and situational factors’, pp. 173–215, in S. Duck (ed.), Personal Relationships and Social Support, London: Sage. Pizzorno A. (1976) [1958] ‘Familismo amorale e marginalità storica ovvero perchà non c’è niente da fare a Montegrano’, pp. 237–252, in E. Banfield (ed.), Le basi morali di una società arretrata, Bologna: Il Mulino. Putnam R. (ed.) (2002) Democracies in Flux: The Evolution of Social Capital in Contemporary Society, Oxford: Oxford University Press. Rothstein B. (2002) ‘Social capital in the Social Democratic state’, pp. 289–331, in R. Putnam (ed.), Democracies in Flux: The Evolution of Social Capital in Contemporary Society, Oxford: Oxford University Press. Rothstein, B. and Stolle, D. (2003) ‘Social capital, impartiality and the welfare state: an institutional approach’, pp. 191–210, in M. Hooghe and D. Stolle (eds), Generating Social Capital: Civil Society and Institutions in Comparative Perspective, New York: Palgrave. Saraceno, C. (2008) ‘Patterns of family living in the enlarged EU’, in this volume. Saraceno, C. and Olagnero, M. (2005) ‘Social networks and patterns of informal support. Do welfare regime patterns make a difference?’, paper presented at ECSR Conference on ‘Comparative European Studies’, Paris, 25–26 November. Saraceno, C., Olagnero, M. and Torrioni, P. (2005) Family, Work and Social Networks in Europe, Luxembourg: Office for Official Publications of the European Communities. Sciolla, L. (1997) Italiani. Stereotipi di casa nostra, Bologna: Il Mulino. Sharp, E. (1984) ‘Citizen-demand making in the Urban Context’, American Journal of Political Science, 28, 4: 654–670. Sokal, R., Rohlf, F. (1962) ‘The comparison of dendrograms by objectives methods’, Taxon, 11: 33–40. Sztompka P. (1999) ‘The cultural core of post-communist transformations’, pp. 205–214, in T. Boje, B. van Steenbergen and S. Walby (eds), European Societies. Fusion or Fission? London: Routledge. Thoits, P.A. (1984) ‘Explaining distributions of psychological vulnerability: lack of social support in the face of life stress’, Social Forces, 62, 2: 453–481. Tomka B. (2003) ‘Western European welfare states in the 20th century: Convergences and divergences in a long-run perspective’, International Journal of Social Welfare, 12, 4: 249–260.
Patterns of sociability
303
UNDP (2003) Human Development Report: Millennium Development Goals. A Compact Among Nations to End Human Poverty, New York, Oxford: Oxford University Press. Van Orschoot W. and Arts W. (2005) ‘The social capital of European welfare states: the crowding out hypothesis revisited’, Journal of European Social Policy, 15, 1: 5–26. Worms, J.-P. (2002) ‘Old and new civic and social ties in France’, pp. 137–188, in R. Putnam (ed.), Democracies in Flux: The Evolution of Social Capital in Contemporary Society, Oxford: Oxford University Press.
13 Feeling left out Patterns of social integration and exclusion Petra Böhnke
Introduction Ever since the Maastricht Treaty prioritised the fight against poverty and social exclusion, European policy discussions have paid much attention to social inequality and to the convergence of living conditions in the European Union. The challenge to preserve social stability and to promote inclusion throughout Europe has even become greater when the new member states joined the EU in the eastern enlargements of 2004 and 2007. Policy debates now make frequent reference to the term ‘social exclusion’ even though there is little consensus on how to define the concept or how to operationalise it for empirical measurement. Whereas some scholars consider the concept as a new paradigm in social inequality research that leaves behind the conventional restriction to distributional issues by emphasising instead the importance of social integration, others vigorously doubt its innovative potential, considering it a politically driven change in terminology without any theoretical value added. Social exclusion is indeed an iridescent term with manifold meanings. Poverty, unemployment and the multidimensionality of social disadvantages are commonly understood as bases of or even synonyms for social exclusion, taking it for granted that they impede social integration and that a decent standard of living decides who is ‘in’ or ‘out’ in society. Little is known, however, about the conditions under which people perceive themselves as included members of mainstream society in the sense of sharing not only a common standard of living but also of participating actively in social, political and cultural life. Although there is consensus that social support and family back-up are important factors that contribute to a decent living, this life domain has hitherto been strongly neglected in the social exclusion debate. In a similar vein, welfare state arrangements, the level of social protection, and a country’s overall prosperity or political culture which presumably all influence the sense of belonging or of being left out, are given scant attention in public debates on social exclusion. Departing from such under-discussed aspects of social inclusion or exclusion, this article provides a novel perspective. Instead of taking the Laeken indicators as a starting point, which monitor objective dimensions of deprivation and social exclusion in official EU-social reporting, the focus is here on individuals’ own perceptions, on how they evaluate their chances of belonging and of being a part of their society.
Social integration and exclusion
305
Social exclusion is here understood as an aspect of perceived social relations, as a concept capturing a sense of subjective marginalisation. The empirical question then is, how widespread feelings of marginalisation are in the enlarged European Union and how nation-specific differences are to be explained. A further aim is to identify objective risks which give rise to subjective perceptions of marginalisation and to disentangle the role which material living standards, social network support and collective country characteristics play in fostering a sense of social cohesion. Given the heterogeneity of living conditions in the enlarged EU, it should be of high concern to policy makers if similar risks and disadvantages bring about identical perceptions of marginalisation everywhere, even if political, economic, social and cultural circumstances are different. The analysis is based on a representative Eurobarometer survey carried out in the new member states (NMS) as well as in the (former) candidate countries, Bulgaria, Romania, and Turkey (CC-3), in spring 2002. This was harmonised with standard Eurobarometer surveys for the 15 old EU member states (EU-15) in previous years.1
1. From inequality to social exclusion Supplementing the focus on income in traditional social inequality research, the social exclusion debate refers not only to a lack of resources but also to limited opportunities to participate in social life. This expansion corresponds with the fact that in more recent poverty research the multidimensionality of poverty, its long-term development and its subjective perception have become growing concerns. The consequences which being socially disadvantaged have for individual participation chances as well as for overall social cohesion are now moving to the fore of attention. This is also reflected in theoretical developments which emphasise the capacity of individuals to live a decent life and draw attention to the political circumstances, institutional structures and welfare state arrangements that enable people to live respectable, affluent and secure lives (Sen 1993, 1999). In the analysis of social inequality this implied a shift in focus from the distribution of material resources to equal opportunities for social participation, integration and the realisation of social rights. Apart from a concern with basic needs and the guarantee of a minimum standard of living, the social exclusion approach has drawn attention to the process of becoming detached from the moral, social and political order of a community (Room 1995: 5). In the understanding of the EU, social exclusion is a process whereby certain individuals are pushed to the edge of society and prevented from participating fully by virtue of their poverty, or lack of basic competencies and lifelong learning opportunities, or as a result of discrimination. This distances them from job, income and education opportunities as well as social and economic networks and activities. They have little access to power and decisionmaking bodies and thus often feel powerless and unable to take control over the decisions that affect their day to day lives. (European Commission 2004: 8) Less politically obliged attempts to define social exclusion usually result in three main characteristics: multidimensionality, interdependence and dynamic with an emphasis on
306
Petra Böhnke
the relational nature of social disadvantages (Silver 1994; Abrahamson 1998; Kronauer 1998; Littlewood and Herkommer 1999; Barnes et al. 2002). The emerging consensus has so far not been given much attention in empirical research, however (exceptions are Paugam 1996 and Burchardt 2000 with regard to the multidimensionality of social exclusion). Income and employment continue to serve as the most important indicators in social reporting activities dealing with social exclusion risks, completed by scant information on education, housing and health (Atkinson et al. 2002, 2005). Most scientific studies start with the assumption that an analysis of poverty, unemployment or multiple deprivation captures social exclusion in all its dimensions and, of course, there are arguments that sustain this position. Labour market access, for example, significantly increases not only a household’s income, but also its social integration. Marginalisation and exclusion should, however, also be experienced subjectively by individuals as a sense of not belonging to society or as feeling of being left out. Asking people directly how they judge their personal opportunities to participate and to what extent they feel recognised and integrated offers the possibility of scrutinising empirically which circumstances give rise to such views. Do unemployment and poverty cover the whole story or are other factors, like for example social networks, family integration or the political culture of a country relevant in fostering a sense of marginalisation? To what extent are there buffers or mediating aspects that prevent disadvantages from resulting in perceived exclusion? Pursuing such questions also means taking subjective well-being seriously as an important dimension of social inequality. Such a subjective approach becomes even more suitable when a comparative analysis of social exclusion across several countries with different historical, cultural, social and economic developments is aimed at. In its fight against social exclusion since the 1990s, the EU has focused on growing difficulties in labour market access and on increasing poverty in countries accustomed to a relatively high standard of living. In the new member states – which are coping with considerably lower standards of living, widespread poverty even in absolute terms, and profound system changes in the most recent past – the understanding of social exclusion might vary, however, reflecting country-specific problems, experiences and aspiration levels. With an eye on 28 European countries, the analysis of the data will proceed in the following steps. First the distribution of perceived social exclusion – measured as a lack of recognition, feeling useless, inferior and left out of society – in old and new member states of the European Union is described. Second, the determinants of perceived social exclusion are explored within and across countries. On the micro-level the hypothesis underlying EU social policy that poverty and the individual experience of unemployment are the crucial factors in shaping social exclusion is investigated. Are there country-specific risk factors that condition the perception to be at the margins of society, or are there more general risk groups to be found similarly across all country borders? Furthermore, the analysis seeks to shed light on the role of social network support in fostering a sense of belonging. The focus is on the connection between social ties and the perception of social inclusion as well as on the power of social network support to mitigate the consequences of material deprivation. Third, a central question is whether there are country-specific patterns in the interplay between socio-economic precariousness, social isolation and feelings of marginalisation that mirror national welfare state arrangements and cultural traditions.
Social integration and exclusion
307
In a last step a multivariate statistical analysis seeks to disentangle the complex interplay of country characteristics and individual experiences which shape the perception of social exclusion. The basic assumption here is that the prevalence of perceived social exclusion will differ between countries according to their overall level of welfare and social security. In addition, we will examine the hypothesis that vulnerable groups are likely to feel more marginalised in countries where the overall level of welfare is high, because stigmatising effects of hardship are greater the less common the experience of objective deprivation is.
2. The prevalence of belonging and marginalisation Two Eurobarometer surveys provide measures of perceived social exclusion. People were asked whether they strongly agree, agree, disagree or strongly disagree with the following statements: • • • •
I don’t feel the value of what I do is recognised by the people I meet. I feel left out of society. I don’t feel that I have the chance to play an useful part in society. Some people look down on me because of my income or job situation.
Agreement with these items reveals perceived worth – and uselessness, feelings of inferiority and a sense of lacking recognition and acceptance. Summing up information from these single indicators, an index on belonging and marginalisation was constructed that ranges from −8 (marginalisation: strong agreement with all four items) to +8 (belonging: strong disagreement with all four items; Cronbach’s alpha for the index is 0.746). Figure 13.1 shows the country-specific distribution of the mean index values. Table 13.1 gives the distribution of the single indicators within single countries and country groups. In order to condense the information, the data are also given for country groups which represent families of nations according to cultural, historical and political similarity. What evidence is there about perceived social exclusion in various European countries? First it should be noted that in all 28 countries the majority of the population perceives themselves as socially included, as the mean index values are positive throughout Europe. There are striking differences in the degree to which people have a sense of belonging to society, however. Comparing the new member states and (former) candidate countries with the old EU-15, a sense of belonging is strikingly more prevalent in the old member states. Only the Mediterranean new member states and the old Southern European member states disrupt the image of a clear east–west divide. The former stand out among the new member states for their relatively high sense of belonging, the latter are a very heterogeneous group in which Portugal and to a lesser extent also Italy stand out with relatively low levels of perceived social inclusion. The feeling to be respected and socially included is most widespread in the Nordic countries, particularly so in Denmark and Sweden. The Baltic States, the Czech Republic, Slovakia and the three candidate countries are at the end of the European rank order indicating that they have relatively large shares of the population who feel marginalised. Instead, Cyprus as well as Slovenia perform comparatively well and range close to the Scandinavian and continental region.
308
Petra Böhnke 5,12
Scandinavian Countries Mediterranean NMS Western European Countries Continental Europe South Europe EU-15 Visegrad Baltic States Candidate Countries Denmark Sweden Cyprus Slovenia Finland Austria Luxembourg Greece Netherlands Spain France Ireland Malta Great Britain Belgium Germany Italy Poland Hungary Portugal Latvia Estonia Romania Lithuania Turkey Czech Rep. Bulgaria Slovakia 0,00
4,10 3,52 3,49 3,32 2,59 1,86 1,67 6,10 5,12 4,22 4,18 4,11 4,06 3,99 3,90 3,79 3,75 3,60 3,45 3,44 3,40 3,34 3,15 3,02 3,02 2,79 2,74 2,00 1,94 1,78 1,74 1,67 1,65 1,40 1,02 1,00
2,00
3,00
4,00
5,00
6,00
7,00
Figure 13.1 The prevalence of belonging and marginalisation in the EU, mean index values Source: Eurobarometer CC EB 2002, EB 51.1 Sept–Oct 2001 Notes: The index on belonging and marginalisation ranges from −8 (marginalisation) to +8 (belonging). The assignment of single nations to country groups follows the following families of nation logic: Scandinavian countries (Denmark, Sweden, Finland), Mediterranean NMS (Cyprus, Slovenia, Malta), Western Europe (Netherlands, Ireland, Great Britain), Continental Europe (Austria, Luxembourg, France, Belgium, Germany), Southern Europe EU-15 (Greece, Spain, Italy, Portugal), Visegrad countries (Poland, Hungary, Czech Republic, Slovakia), Baltic States (Latvia, Estonia, Lithuania), (former) Candidate Countries (Bulgaria and Romania, as well as Turkey,). Grey bars indicate EU-15-nations, black bars NMS and CC-3.
Table 13.1 provides more detailed information showing how widespread various kinds of felt marginalisation and perceived uselessness are. Ranging from 9.6 per cent in Slovenia to 43.9 per cent in Turkey the variation in felt lack of recognition is particularly striking. The perception to be excluded from society is most prevalent in Slovakia (40.6 per cent) and Bulgaria (23.5 per cent), whereas large majorities of 90 per cent or more of the respondents feel well-included in the Mediterranean and Southern European countries as well as in Denmark and Sweden and some other countries. By and large, we find the pattern that perceived marginalisation in all its peculiarities is most prevalent in the Baltic and the candidate countries. Feelings of uselessness and inferiority are also fairly widespread in the continental and Western European countries, however: around 16 per cent of French and Germans do not feel that they have the chance to play a useful part in society, and 14 per cent of the British do not feel accepted because of their income or job situation. There are nine European nations where more than half of the population can be classified as successfully integrated, because the respondents express disagreement
5.12 4.1 3.52 3.49 3.32 2.59 1.86 1.67
6.1 5.12 4.11
4.22 4.18 3.44
3.79 3.45 3.4
Scandinavia Mediterran. NMS Western Europe Continental South Europe EU-15 Visegrad Baltic countries Candidates
Denmark Sweden Finland
Cyprus Slovenia Malta
Netherlands Ireland Great Britain
Belonging (mean index value)
21.4 18.8 17
18.6 9.6 19.4
13.4 18.7 14.7
16.2 12.8 17.7 18.6 21 19.6 20.3 37.9
I don’t feel the value of what I do is recognised by the people I meet (% of population)
4 6.9 8.1
5.7 3 8.1
4.6 5.7 12.1
7.1 4.3 7.7 7 5.5 10.3 15.2 13.7
I feel left out of society (%)
7.5 10.5 12.1
12.7 7.7 9.8
7.4 11.3 15.6
11.3 8.9 11.5 14.5 12.8 17.8 20.2 29.4
I don’t feel that I have the chance to play a useful part in society (%)
8.5 9.2 13.7
4.8 7.1 8.2
6.6 8.8 15.4
9.9 6.5 13.5 11.2 10.5 10.9 18.2 24.8
Some people look down on me because of my income or job situation (%)
48.7 56 48.8
58.2 52.8 54.4
70.4 50.3 53.6
56.9 53.9 48.6 44.2 44.8 35.1 25.2 22.8
Continued
Successful integration (strong disagreement or disagreement with all four items, %)
Table 13.1 Perceived social exclusion, mean index value and distribution of single indicators per country (agree and strongly agree summarised)
3.9 3.75 3.02 2.74
3.02 2.79 1.65 1.02
2.0 1.94 1.74
1.78 1.67 1.4
Greece Spain Italy Portugal
Poland Hungary Czech Republic Slovakia
Latvia Estonia Lithuania
Romania Turkey Bulgaria
23.5 43.9 29.7
16.9 14.7 24.5
18 12.5 32.4 24.5
22.3 13 26.4 21.2
16.7 23.7 21.5 21.4 17.4
I don’t feel the value of what I do is recognised by the people I meet (% of population)
13.8 12.2 23.5
15.6 14.9 15
4.2 13.6 12.6 40.6
8 3.6 5.6 9.5
5.2 3.5 7.6 7.1 7.7
I feel left out of society (%)
20.6 31.2 39
28 23.3 13.7
15.8 17.9 21.8 25.3
12.8 6.9 16.4 16.2
8.4 6.9 16.1 13 16.8
I don’t feel that I have the chance to play a useful part in society (%)
19.3 27.7 17.2
19.6 17.5 17.6
11.6 8.2 12 9.2
6.7 11 10.6 12.5
10 11.2 11.2 11.5 11.7
Some people look down on me because of my income or job situation (%)
24.9 22.6 18.7
27 27 23.3
38.5 41.5 18.2 14.9
49.4 56.8 35.9 41.6
53.2 46.9 42.4 43.4 42
Successful integration (strong disagreement or disagreement with all four items, %)
Notes: See Figure 13.1 for index explanation. Single countries are sorted by the mean index value of belonging within each country group.
Source: Eurobarometer CC EB 2002; EB 56.1 Sept.–Oct. 2001
4.06 3.99 3.6 3.34 3.15
Austria Luxembourg France Belgium Germany
Belonging (mean index value)
Table 13.1 Perceived social exclusion, mean index value and distribution of single indicators per country (agree and strongly agree summarised) – cont’d
Social integration and exclusion
311
with all statements pertaining to social exclusion. These are the Scandinavian countries and the Mediterranean new member states, as well as Ireland, Austria and Spain. On the other hand, a complete sense of belonging is realised for less than one-third of the population in the Czech Republic, in Slovakia, in the three Baltic countries, and in Bulgaria, Romania and Turkey. By and large, our descriptive overview discerns a clear pattern: the Scandinavian population feels successfully socially integrated to a large extent. On the other hand, the Baltic states are confronted with the largest integration deficits. Within the EU25, perceptions of marginalisation are most widespread in Latvia, Lithuania and Estonia. Bulgaria, Romania and Turkey stand out with similarly high levels of perceived social exclusion and worthlessness as the Baltic nations. The vast heterogeneity in country-specific levels of welfare and social protection certainly goes a long way in explaining these strong differences between European nations. Where overall prosperity is low and social protection is weak, perceived marginalisation is widespread. Such an interpretation fails, however, to explain the relatively high levels of felt inclusion in the Mediterranean and Southern European countries despite their lower levels of welfare. Perhaps the strong role of the family with a high degree of integration into family support networks serves as a buffer preventing a strong sense of social exclusion. The next sections deal with these issues in more detail.
3. Who is at risk of social exclusion? In order to identify the factors which affect the sense of belonging or marginalisation, this section explores how perceived social integration varies among different population groups and to what extent we find identical risk groups in all European countries. Figure 13.2 reports the mean index value of belonging for three different population groups in each nation: for people with high income, people with low income and for those who experience serious solvency problems. The latter indicator was chosen because the data set did not offer a conventional poverty indicator or a convincing measure of economic hardship. Serious solvency problems are here supposed if respondents indicate difficulties during the last year in either paying the rent or mortgage, the water, gas or electricity bills or in repaying loans. In all European countries higher incomes clearly go together with a sense of successful social inclusion, whereas feelings of belonging are weaker on lower income levels. Obviously material resources and a decent standard of living help people to feel integrated and well-respected. This general pattern prevails in all countries. The degree to which people feel excluded or included at a given level of income varies significantly across countries, however. In general the sense of belonging is more widespread in more affluent countries. Thus, in some of the rich old member states even people with low incomes have a stronger or at least similarly strong sense of belonging as people with high incomes in the poor new member states. Yet, whilst the level of felt inclusion is generally higher in the more affluent old member states, the gap separating the economically privileged from the poor in their sense of belonging is lower in the poorer countries. These differences become particularly striking if we look at the subjective social inclusion of people with solvency problems. This group is relatively small in the old member states, but relatively big in the new member states and candidate countries.
7,50 7,00 6,50 6,00 5,50 5,00 4,50 4,00 3,50 3,00 2,50 2,00 1,50 1,00 0,50 0,00 −0,50 −1,00 −1,50
highest income quartile
lowest income quartile
serious solvency problems
Baltic States Candidate Countries Visegrad Western European countries Southern Europe EU-15 Mediterranean countries NMS Continental Europe Scandinavian countries
Petra Böhnke
Slovakia Czech Rep. Lithuania Estonia Turkey Rumania Latvia Bulgaria Portugal Poland Great Britain Italy Hungary Malta Belgium Spain Luxembourg Netherlands Slovenia Germany Greece Ireland Austria France Cyprus Finland Sweden Denmark
312
Figure 13.2 Mean index value of belonging in three population groups Source: Eurobarometer CC EB 2002; EB 56.1 Sept–Oct 2001 Notes: When there is no value reported for those experiencing serious solvency problems, the number of cases is below 30. For index-information see Figure 13.1.
As indicated by the length of bars, the gap separating those in economic hardship from the economically more fortunate groups is comparatively small in the groups of new member states, but large in Scandinavia, continental Europe and the Western European countries. Moreover, those with solvency problems in the more affluent countries tend to feel even more left out of society than their peers in the poorer new member states. A comparison between the Baltic nations and the candidate countries on the one side, and the Western European and continental countries illustrates the point strikingly.2 The fact that only small minorities suffer from such poor living conditions in the prosperous countries makes the gap separating them from mainstream society much more incisive than in poorer nations where solvency problems are widespread. Hence we find the socio-economic polarisation in the sense of belonging to be more pronounced in the well-off EU-15 nations. The next step expands the search for determinants of perceived social exclusion by incorporating also labour market attachment, the employment status as well as the experience of long-term financial difficulties into the analysis (Table 13.2). The independent effects of these factors are explored with the help of multiple regression models that are calculated for eight families of nations which were formed with respect to their assumed similarity in institutional and cultural background. Age, gender as well as individual countries within each cluster of nations are statistically
Actual or former employment status (skilled worker) Service class Routine nonmanual Self-employed Farmers Non-skilled worker
Labour market attachment (reference group: employed) Unemployed Retired
.205 .272* .181 .982** −.527***
.220 −.470 −.069
−1.48*** .020
Continental Europe
.235 .245
−1.76*** −.835***
Scandinavian countries
.468 −.010 −.511**
.631*** .085
−1.50*** −.376*
Western Europe
Table 13.2 What determines perceived marginalisation and belonging?
.287 .294 −.049
.281 .255
−.305 .160
Southern Europe EU-15
.590 −.389 −.899***
−.161 −.133
−.669* −.490*
Mediterranean NMS
.027 −.394* −.505***
.100 .028
−.764*** −.140
Visegrad
.451 −.207 −.422**
.500*** .275*
−.754*** −.303
Baltic states
Continued
−.508 .206 −.455*
.548** .157
−.865*** −.165
Candidate countries (Romania, Bulgaria)
15
−1.45*** −1.66***
−.348** −1.03*** −1.59***
Continental Europe
15
−1.43*** −1.99***
−.299 −.067 −.505**
Western Europe
10
−1.43*** −.762***
−.107 −.250 −1.07***
Southern Europe EU-15
10
−.393 −1.22***
.163 −.149 −.962***
Mediterranean NMS
15
−.886*** −.971***
−.226* −.293* −.611***
Visegrad
10
−.997*** −1.04***
−.365** −.436** −.560**
Baltic states
13
−1.32*** −1.75***
−.880*** −1.25*** −1.82***
Candidate countries (Romania, Bulgaria)
Notes: The dependent variable is the index on belonging and marginalisation (see Figure 13.1, ranging from −8 (marginalised) to +8 (belonging). The models are controlled for age, sex and country. In the Candidate Countries’ group Turkey is missing, because of lacking information on the employment status. Significance level: * p < 0.05, ** p < 0.01, *** p < 0.001. The table shows fixed effects regression models for country groups’ (unstandardised regression coefficient and level of significance).
Source: Eurobarometer CC EB 2002, EB 56.1 Sept.–Oct. 2001
21
−1.43*** −2.65***
Long term poverty (no financial difficulties) Less than 2 years > 2 years
% of explained variance
−.329* −.742*** −1.54***
Income (highest income quartile) Third Second Lowest
Scandinavian countries
Table 13.2 What determines perceived marginalisation and belonging? – cont’d
Social integration and exclusion
315
controlled, so that the results give an impression of the net impact of socio-economic determinants independently of demographic influences or country-specific peculiarities. The results clearly identify unemployment as a major determinant of perceived social exclusion (see Table 13.2). In this sense, the most widely cited mechanism in the social exclusion debate is confirmed in the empirical analysis which shows significant negative effects of unemployment in all country groups (except for the Southern European old member states where the effect is also negative but below the level of statistical significance). Exclusion from the labour market weakens social integration and is strongly connected to the perception of marginalisation even when we control for income. This underlines that employment is not merely a source of income but also a major mechanism of social inclusion. Retirement has less uniform consequences for belonging, once the effects of age, income and former labour market attachment are controlled. In most country groups it produces a weak and negative effect without being statistically significant in the majority of cases. This suggests that retirement – while similarly interrupting social ties formed at the workplace as unemployment – is free of the social stigma and felt humiliation that accompanies the unsuccessful search for a vacancy in the job market. Having a job is more important for the sense of belonging than having a particular kind of employment. This is suggested by the fact that the employment status has only rarely statistically significant effects on subjectively felt inclusion. In general people with better jobs tend to feel more successfully integrated, but the respective effects are not very strong. Only non-skilled workers display an over-proportionate tendency to feel excluded in most country groups. This suggests that a minimum quality of working conditions is needed to foster a sense of social inclusion. The access to material resources is a similarly important determinant of social integration as having a job. The lower the income, the more likely it is that people feel at the margins of society. The longer financial difficulties constrain living conditions, the more likely people are to feel socially excluded. Thes results confirm the general hypothesis that social exclusion is closely related to unemployment and poverty. However, there are differences concerning the impact of these explanatory variables in different country groups. Especially the relatively weak explanatory power of the model in the new member states and in Southern Europe suggests that additional mechanisms must be at work to foster a sense of belonging or exclusion. The next section therefore specifically explores the role of social network support.
4. The role of social back-up The debate about risks that determine social exclusion has been mostly restricted to socio-economic aspects. However, it is well known how important family integration and reliable social support networks are for subjective well-being in general (Argyle 1987; Headey and Wearing 1992; Diener et al. 1999; Layard 2005). Hence the question is to what extent the perception of belonging is also influenced by the availability and the quality of social network support. Earlier research has shown that unemployed and poor people face a higher risk of losing social network support (Böhnke 2005). Given the multidimensional nature of social exclusion processes, it is therefore very likely that lack of family back-up and limited access to social support are detrimental to self-confidence and perceived social inclusion especially when living in socio-economically precarious conditions.
316
Petra Böhnke
The nature of the harmonised data set offers only a limited potential for the analysis of social support systems, because information on the size of households and the number of children in households is not available for all countries. Therefore, the analysis is confined to the following survey questions: is social support from outside the household available in case of depression and financial need?; how satisfied is the respondent with his family life?; does the respondent feel left out of his family?; is he married, divorced, separated or widowed? Three research questions guide the analysis of this information: is there a connection between the availability of social support and the feeling of belonging to society?; are social networks able to protect those who have to cope with economic strain from feeling excluded?; how do socioeconomic determinants and social support interact to shape perceptions of marginalisation and what effects do they produce independently of each other? As Figure 13.3 shows, access to social network support is clearly associated with a strong sense of belonging in all European countries.3 People who cannot rely on someone from outside their household when they feel depressed, need someone to talk to or need short-term financial help, are more likely to feel excluded from society, inferior and useless. In most countries the feeling of belonging becomes particularly precarious when family life is disrupted. Although the level of felt social inclusion is different from country to country, the general pattern that family integration and the availability of support network foster a sense of belonging is valid in all nations. As indicated by the length of the bars, the availability of social support is connected most strongly to a sense of belonging in the Scandinavian region. It has the weakest impact in the Baltic states as well as in the candidate countries where the
no social support
dissatisfied with family life
Candidate Countries Baltic States Visegrad Southern Europe EU15 Western European Countries Continental Europe Mediterranean NMS Scandinavian Countries
social support
Slovakia Bulgaria Lithuania Turkey Czech Rep. Latvia Estonia Romania Portugal Hungary Italy Poland Germany Great Britain Belgium Malta Ireland France Spain Netherlands Luxembourg Greece Slovenia Austria Finland Cyprus Sweden Denmark
7 6,5 6 5,5 5 4,5 4 3,5 3 2,5 2 1,5 1 0,5 0 −0,5 −1
Figure 13.3 Perceptions of belonging and the availability of social network support (mean index values) Source: Eurobarometer CC EB 2002, EB 56.1 Sept–Oct 2001 Note: For index information see Figure 13.1.
Social integration and exclusion
317
sense of belonging is generally lower. The differences with regard to single countries are striking. In Denmark, Sweden and Greece, even people who cannot rely on social support perceive themselves nevertheless as more belonging than people with social back-up in Slovakia, Bulgaria, Lithuania, Turkey, Czech Republic, Latvia, Estonia and Romania. This emphasises once again that although a general mechanism is active in all countries – the more reliable social support is, the more people will see themselves as a respected member of society – the general level of belonging varies vastly by country. Presumably, a lack of social support networks is best compensated for in countries where developed welfare states offer extended social services that can step in when other support mechanisms fail. Although these results suggest a connection between access to social relationships and a general feeling of belonging, the question remains in what sense do social backup and control over material resources interact to foster a sense of social inclusion? Do poverty or unemployment on the one hand and social isolation on the other determine the perception of belonging independently from each other or are they intertwined? Previous research showed the cumulative nature of respective disadvantages across Europe. Contrary to the idea that poverty breeds solidarity and that the poor may be rich in social relations, those with low incomes cannot rely on more extended social network support but, rather, lose social contacts and feel more isolated (Gallie et al. 2003; Crow 2004; Böhnke 2006). The protective function of reliable social back-up becomes particularly important when socio-economic deprivation makes it most needed. Hence the question is to what extent the integration into social support networks can buffer the effects of economic hardship. Looking only at respondents who experience long-term financial difficulties (for more than two years), we find a clear pattern: in all country clusters the availability of social support from outside the household or from family members helps to foster the perception of belonging when a lack of economic resources restricts participation chances. As shown in Figures 13.4a and 13.4b, the mean index values of belonging are always higher when people can rely on social support. Longterm poverty without social back-up results in a stronger sense of marginalisation (index values are low). Beyond this general mechanism, we once more find remarkable country-specific effects: people experiencing long-term economic strain feel even more marginalised in more prosperous nations. This suggests that the smaller the group of the long-term poor is, the more incisive is the experience to be marginal among its members. The fact that the gap between the long-term poor with and without social support is largest in the Scandinavian countries as well as in continental Europe suggests that stigmatisation processes are at work. Suffering from permanent financial difficulties without social or family networks one can rely upon promotes social exclusion feelings everywhere. But in prosperous countries where only small minorities experience such shortcomings, people are more likely to see personal failure rather than structural constraints as the explanation for their problems, so that the feeling of being an outsider is nourished. A multiple regression analysis for country clusters including the availability of material resources as well as the social back-up variables helps in understanding these interaction processes more clearly. To what extent do unemployment and poverty lose their impact on perceived social exclusion when social back-up is integrated into the model? And are socio-economic aspects or social back-up more strongly connected
318
Petra Böhnke
Southern Europe EU-15 Mediterranean NMS Western Europe Baltic States Visegrad Candidate Countries Continental Europe Scandinavian Countries −0,5
0
0,5
1
1,5
2
2,5
3
3,5
4
Index of belonging, mean value long term financial difficulties and no social support long term financial difficulties and social support
(a) Mediterranean NMS Visegrad Baltic States Southern Europe EU-15 Candidate Countries Scandinavian Countries Continental Europe Western Europe −1
−0,5
0
0,5
1
1,5
2
2,5
3
3,5
4
Index of belonging, mean value long term financial difficulties and dissatisfied with family life long term financial difficulties and satisfied with family life
(b) Figure 13.4(a) Social support outside the household as a buffer? (b) Family back-up as a buffer? Source: Eurobarometer CC EB 2002, EB 56.1 Sept–Oct 2001 Reading example: In the Southern European countries of the EU-15 people who suffer from long-term financial difficulties and cannot rely on social support from outside the household feel less belonging than poor people who are integrated in a reliable social network (index value 2.1 vs. 2.8). The lower the mean index value the more people perceive themselves as marginalised. The index ranges from −8 (marginalised) to +8 (belonging). The sequence of country clusters is here sorted by the mean index value of people who experience both long-term poverty and lack of social back-up.
Social integration and exclusion
319
with the risk of perceiving oneself as an outsider? Which patterns arise in the different country groups? Looking first at the percentages of explained variance at the bottom of Table 13.3, it is evident that social back-up enlarges the explanatory power of these models across all country groups. In particular, this impact is relatively strong in the prosperous old member states (Scandinavia, Continental and Western Europe). Some more specific results are also worth noting. Among the social support variables, feeling left out of the family is by far the strongest predictor of subjective social exclusion. The high explanatory power of this variable is particularly noteworthy since its impact is here measured independently of age, widowhood or retirement. Even the more moderate indicator of weak family bonds – dissatisfaction with family life – produces significant effects on the feeling of belonging. The lack of social support from outside the household is slightly smaller in its effect, but nevertheless also important in explaining perceived marginalisation. Marital status by itself has no strong impact. The fact that the effects of divorce or widowhood vanish when other variables are present suggests that it is not divorce or widowhood as such which cause feelings of marginalisation, but rather the factors that are associated with them which are here included in the model: low income and economic precariousness on the one hand and disrupted social ties on the other. The comparison with the results of Table 13.2 (see bottom line of Table 13.3) shows that the effects of the socio-economic determinants decline in most cases, once social back-up is taken into account as an additional explanation. Yet, even though the effects decline, unemployment, low-income and long-term financial difficulties remain important risk factors on their own that cause feelings of marginalisation. Whilst differences between the country groups persist, the general pattern is that lowincome and durable economic precariousness result in restricted participation chances and feelings of marginalisation independently from the social networks people are integrated in. In other words: both, socio-economic precariousness as well as weak social bonds diminish people’s subjective attachment to society.
5. Societal conditions and individually experienced social exclusion Individual living conditions like people’s attachment to the labour market, their income level and their social relationships are important factors that have to be considered when trying to understand the dynamic of social exclusion experiences. Even when these mechanisms are taken into account, the overall prevalence of belonging and marginalisation remains remarkably different across European countries. Hence we must also ask to what extent specific macro-characteristics of a given country shape the overall level of perceived marginalisation. The distribution in Figure 13.1 has already suggested that prosperous nations are more successful in establishing widespread feelings of belonging. In addition to the level of affluence, other characteristics related to the quality of society – the scope of its welfare state schemes, or the level of trust in its institutions and democratic culture – may have an impact as well. Bivariate macro-correlations (not shown here) of aggregate country characteristics reveal close connections between the nation-specific levels of belonging and various country properties. The higher the GDP per capita in a country and the higher its quality of society as rated by citizens, the more people tend to feel included. The more
−.286 .000 −.318
−.167 −.670*** −.938*** −1.14*** −1.33***
Income (highest income quartile) Third −.289 Second −.524** Lowest −1.00***
Long-term poverty (no financial difficulties) Less than 2 years −1.06*** > 2 years −2.04***
−1.26*** −1.66***
.500** −.013 .317 .038 −.559**
Western Europe
Actual or former employment status (skilled worker) Service class .291 .235 Routine non-manual .288 .307* Self-employed .083 .106 Farmers −.371 .745* Non-skilled worker −.017 −.434**
Continental Europe
−1.25*** −.346*
Scandinavian countries
Labour market attachment (reference group: employed) Unemployed −1.51*** −1.32*** Retired −.645*** .030
Socio-economic determinants
−1.23*** −.683***
−.036 −.246 −.599**
.430* .277 .268 .423 .007
−.176 .240
Southern Europe EU-15
Table 13.3 Socio-economic precariousness and social support as determinants of belonging
−.306 −1.08***
.204 .000 −.570*
−.358 −.166 .595 −.409 −.850***
−.566* −.248
Mediterranean NMS
−.740*** −.861***
−.265* −.276* -.253
.010 −.020 −.050 −.398* −.382**
−.650*** −.167
Visegrad
−.768*** −.905***
−.401** −.392* −.205
.418** .264 .158 −.042 −.386*
−.710*** −.256
Baltic states
−1.24** −1.54***
−.893*** −1.22*** −1.33***
.480* .096 −.569 .110 −.313
−.675** −.120
Candidate countries (Romania, Bulgaria)
23 (+6)
22 (+9)
−.010 −.056 −.106
−3.04*** −1.18***
−.421
22 (+2)
−.292* −.447 −.076
−3.62*** −1.22***
−.614***
18 (+5)
−.215 −.466 −.546*
−2.88*** −.840**
−.857***
23 (+2)
.412** −.102 −.330*
−2.82*** −.624***
−1.09***
14 (+4)
.129 .288 −.467*
−2.09*** −.652***
−.447**
18 (+5)
.278 −.260 −.135
−2.37*** −.884***
−1.28***
Notes: The dependent variable is the index on belonging and marginalisation (ranging from −8 (marginalised) to +8 (belonging). The models are controlled for age, sex and country. In the Candidate Countries group Turkey is missing, because of lacking information on the employment status. Numbers in brackets indicate the increase of the explained variance compared with the regression models calculated in Table 13.2, which have been restricted to socio-economic variables only Significance level: * p < 0.05, ** p < 0.01, *** p < 0.001. The table shows fixed effects regression models for country groups, (unstandardised regression coefficient and level of significance).
Source: Eurobarometer CC EB 2002, EB 56.1 Sept.-Oct. 2001
31 (+7)
% of explained variance
−2.87*** −1.24***
−3.98*** −1.38***
−.072 −.512** −.134
−.838***
−1.43***
Marital status (Reference group: married) Unmarried −.029 Divorced/separated −.024 Widowed −.233
Social relations Lack of social network support Feeling left out of family Dissatisfied with family life
322
Petra Böhnke
a government invests in social security, the more people tend to perceive themselves as belonging. This suggests that developed social protection schemes contribute to fostering a sense of cohesion. On the other hand, high rates of relative income poverty decrease the average level of subjective attachment to society. The same is true for the degree of social inequality. Countries with widespread poverty and high degrees of inequality typically have higher proportions of people who feel at the margins of society. There is reason to assume, then, that the macro-characteristics of the home country have an impact on people’s sense of belonging independently of or in addition to the specific circumstances in which they live. The impact of such macro-characteristics is widely neglected in current debates on social exclusion and not adequately considered in official monitoring systems of social inclusion. Yet, society provides individuals with a wide range of economic, institutional and cultural frames which shape everyone’s opportunities to participate. The question then is to what extent the societal context which people live shapes the perception of marginalisation in addition to individual life circumstances such as poverty, unemployment or lack of social support, and how micro and macro aspects of living conditions interact. In order to asses the relative impact of macro-contexts and micro-conditions, several random intercept models have been calculated.4 The effects shown in Table 13.4 are calculated in separate models testing the effect of one macro indicator at a time in order to avoid multicollinearity problems. They include the whole European member state population and enclose individual socio-economic determinants as well as the variables on social support which were included in the models discussed above (Tables 13.2 and 13.3). The question is whether societal conditions have an impact on individually experienced marginalisation even if socio-economic determinants like
Table 13.4 The independent effect of country characteristics on the individual experience of social exclusion (standardised regression coefficients) Country characteristics GDP per capita1 Poverty rate2 Income Inequality3 Social Security Expenditure4 Quality of Society5 Unemployment Rate6
.323*** −.319*** −.247*** .363*** .427*** −.266***
Interaction effect Low income * GDP per capita Long term financial difficulties * GDP per capita
−.063** −.191***
Source: Eurobarometer CC EB 2002, EB 56.1 Sept–Oct 2001 Notes: 1 Eurostat 2002; 2 Eurostat 2004a, b; 3 CIA 2002; 4 Eurostat, structural indicators; 5 index summarising trust in social protection system, trust in other people, the perception of tensions and the quality of public services, see Böhnke 2005; 6 European Commission 2002, Eurostat 2002. For each macro indicator a separate random intercept model has been calculated. The dependent variable in each model is the index of belonging and marginalisation (ranging from −8 (marginalised) to +8 (belonging). On the individual level the models include socio-economic, support and demographic variables (see Table 13.3). Countries are included as an additional hierarchical level. The sample consists of respondents from 25 EU member states (N = 19406). Significance level: * p < 0.05, ** p < 0.01, *** p < 0.001.
Social integration and exclusion
323
being unemployed, living with low income or lacking social support are taken into account. The effects for the socio-economic and support variables are not reported once again, because they confirm what has already been discussed above. Instead I want to highlight the important message that results from the significant findings of the macro variables in Table 13.4: the experience of social exclusion is a multi-dimensional process. It is a mixture of individual living conditions – especially long-term poverty and lack of social support – and the societal context into which these living conditions are embedded that determines if citizens are likely to feel included or excluded. Apart from individual living conditions, the quality of society at large also influences the extent to which people feel part of social life. Hence it matters a lot if the country someone lives in has a high level of affluence, good governance and generous social security schemes. Such macro-conditions effectively promote feelings of belonging and social inclusion. Moreover, low levels of unemployment, poverty and inequality also foster a social climate where people feel socially included. The connection between macro-characteristics and individual living conditions is illustrated by the statistical model considering interaction effects. It shows that economic precariousness hampers feelings of belonging especially in rich countries. As the negative interaction effects show, low-income and long-term financial difficulties promote strong feelings of marginalisation especially when the persons concerned live in relatively prosperous nations. Even though the overall level of perceived social exclusion is higher in poorer countries, those who live in dire circumstances are more likely to feel marginalised in rich countries where poverty is less widespread. Being less numerous in prosperous countries, people living in severe poverty are more likely to feel personally responsible and blame themselves rather than society. In short, poverty is more likely to come with a stigma in rich countries. Figure 13.5 illustrates the different impact which long term poverty has on perceived social exclusion in different contexts by showing the strength of the regression coefficients in European countries ranked by the level of GDP per capita. It is evident that the impact is comparatively low in less affluent countries such as Poland, Estonia, Latvia or Slovakia, but relatively high in richer countries such as Germany, Denmark, Austria, Ireland, Sweden, Belgium or the UK. These results confirm that personal economic precariousness fosters feelings of social exclusion everywhere in Europe, but its impact is even worse in prosperous countries, where durable economic strain only affects a minority of the population and thus has more stigmatising effects.
6. Conclusion Society matters – this is a key message that can be derived from this chapter. In order to understand why people feel marginalised, individually experienced social disadvantages like unemployment, poverty and especially long-term difficulties provide convincing explanations. However, it matters a lot in which country people experience economic strain and joblessness. Being poor in a rich country is connected with limited participation chances, but compared to the poor in less prosperous countries the poor in rich countries are on average better off. In this absolute sense, it is better to be poor in a rich country. In relative terms, however, it is worse to be poor in rich countries, because the gap to mainstream society is larger and feelings of social exclusion
Petra Böhnke
B (regression coefficient, impact of long term financial difficulties on perceptions of belonging, under control of several socio-economic, support and demographic variables)
324
0 5000
−0,5
10000
15000
20000
25000
30000
35000 40000
45000
50000
Spain Czech Rep Portugal
Poland Estonia Latvia Slovakia
−1
Cyprus Greece Malta Slovenia
Italy Luxembourg
France Netherlands Germany
−1,5
Lithuania Hungary
Denmark Ireland Austria Sweden Belgium UK
2 R = 0,1057
−2
Finland
−2,5 GDP per capita
Figure 13.5 The impact of long-term poverty on perceived social exclusion across countries according to their welfare level Source: Eurobarometer CC EB 2002, EB 56.1 Sept–Oct 2001, Eurostat 2002
are more likely to be incisive as the poor feel more excluded in comparison with their well-off compatriots than the poor in less affluent countries. The smaller the group of disadvantaged people is and the more they are separated from average living conditions in their country, the more likely it is that they perceive themselves at the margins of society. Consequently, severe poverty which is relatively often experienced in poorer nations, while affecting only a small minority in richer countries, is a more forceful driver of social exclusion feelings in rich nations. Even the role social networks play in fostering a sense of belonging varies with the societal context; their general impact, however, independently from socio-economic disadvantage or in interaction with them, is strongly confirmed by this analysis. There is no doubt that the EU social policy strategy to fight social exclusion by promoting employment and by reducing poverty rates meets the demands of the people concerned. But the notion that social exclusion is predominantly dependent on unemployment and economic strain must be qualified. The role family back-up and social network support play in preventing marginalisation has been severely underestimated in the recent debate. Policy interventions cannot concentrate solely on the unskilled, the unemployed and on poor people. They must be co-ordinated with family policy and attempts to strengthen civil society networks.
Social integration and exclusion
325
Moreover, the country-group specific results ask for more differentiated policy interventions. The concentration on unemployment seems to fit best in the old EU-15 member states. The post-communist new member states need a more encompassing approach that helps them increase the overall level of economic prosperity, political stability, and social security. Combating social exclusion in the new member states means assuring a basic standard of living and rising prosperity in general as well as the urgent fight against acute poverty. In the well-off member states of the old EU-15 prevention has to be more fine-tuned to the needs of specific groups: even though the risk groups are small in size, they suffer a lot, and feel even more cut-off from mainstream society than their peers in the new member states. People perceive themselves as integrated members of society when their individual circumstances allow them to live a decent, economically stable life and when they have confidence in back-up through family and friends. These aspects do not cover the whole story, however, because living conditions are also influenced by a country’s general political and economic performance. A generally high level of prosperity and social security as well as general trust in institutions and fellow men belong to the factors which foster feelings of belonging and social cohesion. In order to understand the different patterns of social exclusion across Europe, the notable impact of socio-economic conditions must be taken into account together with the societal context into which they are embedded and which frames individual experiences. What do the results suggest with regard to the appropriateness of the indicators the EU encourages its members to monitor social exclusion with? Apart from the selfperceived health status, life expectancy and low education, all Laeken-indicators focus on poverty and unemployment. A reasonable expansion should focus on social support and participation chances, explicitly emphasising the interdependence of access to resources and integration deficits. Lastly, one must consider that identical levels of deprivation may have different bearing on felt social exclusion depending upon the macrocontext in which it is experienced. Hence, more context-related indicators would broaden our understanding of exclusion processes in European countries substantially. The core insight of this chapter is that an understanding of perceived social exclusion is dependent on personal as well as on societal conditions: richer countries have lower levels of perceived social exclusion, nevertheless polarisation is higher and risk groups feel more marginalised relative to their privileged counterparts within their own country. Finally, we must not forget that if we were to apply pan-European standards, even those who are relatively well-off in the mainstream of poorer countries would belong to relatively marginal groups compared to the standard of living in the richer member states.
Notes 1 In order to monitor living conditions in the accession countries the European Foundation for the Improvement of Living and Working conditions in Dublin funded a project in 2002/2003, in which these data have been harmonised. The project was coordinated by the Social Science Research Centre in Berlin (WZB) with the Economic and Social Research Institute Dublin (ESRI) and the Demographic Research Institute in Budapest (DRI) as partners (Alber and Fahey 2004, Böhnke 2004, Nauenburg et al. 2003). 2 In Scandinavia, those who are insolvent are similarly prone to feel left out of society as their peers in the Baltic or candidate countries, but this is mostly due to the unusually high level of belonging felt by those with solvency problems in Sweden.
326
Petra Böhnke
3 The objection may be raised that feelings of being excluded from society and lack of private social back-up, although theoretically meant to be different, are actually measuring the same latent construct. However, the correlations between belonging and different measures of social back up are relatively small, reaching a maximum of −0.30 in the case of family exclusion. Additionally, factor analysis has shown that only the items of belonging are clearly extracted to one factor. 4 Many thanks to my colleague Ulrich Kohler, whose support and expertise have been essential to solve methodological and statistical difficulties in this section.
References Abrahamson, P. (1998) ‘Combating poverty and social exclusion in Europe’, pp. 145–176, in W. Beck, L. van der Maesen and A. Walker (eds), The Social Quality of Europe, Bristol: Policy Press. Alber, J. and Fahey, T. (2004) Perceptions of Living Conditions in an Enlarged Europe, European Foundation for Living and Working Conditions, Luxembourg: Office for Official Publications for the European Commission. Argyle, M. (1987) The Psychology of Happiness, London: Methuen. Atkinson, A.B., Cantillon, B., Marlier, E. and Nolan, B. (2005) ‘Taking forward the EU social inclusion process’, independent report commissioned by the Luxembourg Presidency of the Council of the European Union, manuscript. Atkinson, T., Cantillon, B., Marlier, E. and Nolan, B. (2002) Social Indicators: The EU and Social Inclusion, Oxford: Oxford University Press. Barnes, M., Heady, C., Middleton, S., Millar, J., Papadopoulos, F., Room, G. and Tsakloglou, P. (2002) Poverty and Social Exclusion in Europe, Cheltenham: Edward Elgar Publishing. Böhnke, P. (2004) Perceptions of Social Integration and Exclusion in an Enlarged Europe, Dublin: European Foundation for the Improvement of Living and Working Conditions. Böhnke, P. (2005) First European Quality of Life Survey: Life Satisfaction, Happiness and a Sense of Belonging, European Foundation for the Improvement of Living and Working Conditions, Luxembourg: Office for Official Publications for the European Commission. Böhnke, P. (2006) ‘Einkommensarm, aber beziehungsreich? Zum Zusammenhang von Armut und sozialen Beziehungen in der erweiterten EU’, pp. 107–134, in J. Alber, and W. Merkel (eds), Europas Osterweiterung: Das Ende der Vertiefung? Berlin: edition sigma. Burchardt, T. (2000) ‘Social exclusion: concepts and evidence’, pp. 385–406, in D. Gordon, and P. Townsend (eds), Breadline Europe: The Measurement of Poverty, Bristol: Policy Press. CIA (2002) ‘The World Factbook’. Online. Available Http: (accessed 16 January 2007). Crow, G. (2004) ‘Social networks and social exclusion: an overview of the debate’, pp. 7–19, in C. Phillipson, G. Allan, and D. Morgan (eds), Social Networks and Social Exclusion: Sociological and Policy Perspectives, Aldershot: Ashgate. Diener, E., Suh, E.M., Lucas, R.E. and Smith, H.L. (1999) ‘Subjective well-being: three decades of progress’, Psychological Bulletin, 125: 276–302. European Commission (2002) Employment in Europe 2002: Recent Trends and Prospects, Luxembourg: Office for Official Publications for the European Commissions. European Commission (2004) Joint Report on Social Inclusion. Online. Available http: (accessed 16 January 2007). Eurostat (2002) Statistical Yearbook 2002: The Statistical Guide to Europe, Luxembourg. Eurostat (2004a) ‘Poverty and social exclusion in the EU’, Statistics in Focus, Population and Social Conditions, 16/2004. Eurostat (2004b) ‘Monetary poverty in new member states and candidate countries’ Statistics in Focus, Population and Social Conditions, 12/2004.
Social integration and exclusion
327
Eurostat, Structural Indicators. Online. Available Http: (accessed 16 January 2007). Gallie, D., Paugam, S. and Jacobs, S. (2003) ‘Unemployment, poverty and social isolation: Is there a vicious circle of social exclusion?’, European Societies 5, 1: 1–32. Headey, B.A. and Wearing, A.J. (1992) Understanding Happiness: A Theory of Subjective Well-being, Melbourne: Longman Cheshire. Kronauer, M. (1998) ‘“Social exclusion” and “underclass”: new concepts for the analysis of poverty’, pp. 51–75, in H.-J. Andreß (ed.), Empirical Poverty Research in a Comparative Perspective, Aldershot: Ashgate. Layard, R. (2005) Happiness: Lessons from a New Science, London/New York: Penguin. Littlewood, P. and Herkommer, S. (1999) ‘Identifying social exclusion: some problems of meaning’, pp. 1–21, in P. Littlewood, I. Glorieux, S. Herkommer and I. Jönsson (eds), Social Exclusion in Europe, Aldershot: Ashgate. Nauenburg, R., Böhnke, P., Delhey, J., Keck, W. and Fliegner, F. (2003) ‘Quality of life in the European Union and the acceding and candidate countries: description of the data’, Unpublished manuscript, Berlin: Social Science Research Centre Berlin. Paugam, S. (1996) ‘Poverty and social disqualification: a comparative analysis of cumulative social disadvantages in Europe’, Journal of European Social Policy, 6, 4: 287–303. Room, G. (1995) ‘Poverty and social exclusion: the new European Agenda for policy and research’, pp. 1–10, in Graham Room (ed.), Beyond the Threshold: The Measurement and Analysis of Social Exclusion, Bristol: Policy Press. Sen, A. (1993) ‘Capability and well-being’, pp. 30–53, in M.C. Nussbaum and A. Sen (eds), The Quality of Life, Oxford: Clarendon Press. Sen, A. (1999). Development as Freedom, Oxford: Oxford University Press. Silver, H. (1994) ‘Social exclusion and social solidarity: three paradigms’, International Labour Review, 133, 5–6: 531–578.
14 The perception of group conflicts Different challenges for social cohesion in new and old member states1 Jan Delhey and Wolfgang Keck
Introduction In autumn 2004, the Netherlands were shocked when film-maker Theo Van Gogh, author of an Islam-critical movie, was shot on the street by a Muslim immigrant. In autumn 2005, riots broke out in Paris after the unsettled death of two young immigrants. Over ten nights in a row, thousands of cars and dozens of buildings burned down, and immigrants, most of them male, jobless, and with a Magrebh background (but a French passport), street battled against the police. Seen in conjunction with everyday discrimination against minorities and ethnically motivated offences, these occurrences put the self-image of European societies as places where different groups live peacefully together into question. For the EU, this is particularly alarming, since it aims at increasing levels of social cohesion in its member states. This contribution looks at levels of perceived tensions between social groups as indicating problems of societal cohesion. ‘Perceived tensions’ refers to the level of tensions in a society, as reported by its citizens. We do not attempt to make statements about the ‘real’ level of conflicts, if this is actually possible. Instead we focus on public perception of two types of conflicts: on the one hand on ‘traditional’ class or stratification related tensions between status groups such as rich and poor, and management and workers; on the other hand on tensions between different ethnic and racial groups, which relate to immigration and the emergence of multicultural societies. More specifically, we are interested in two main issues: first, from the EU enlargement perspective we examine if citizens in the new member states perceive more tensions in their societies, and if their conflict agenda differs from the agenda in the 15 old member states. Because of their communist heritage and recent economic transformation, especially in the post-communist countries strong distributional conflicts would come as no surprise. Second, we are interested in the degree of consensus over social tensions prevailing within single nations. Marxist theory in particular claims that conflict consciousness is highly class related (cf. Kelley and Evans 1999), and realistic conflict theory makes a similar suggestion for ethnic relations
Social cohesion in new and old member states
329
(Stephan and Stephan 2000). Again from an EU-enlargement perspective, we investigate the extent to which status position translates into conflict ratings in old and new member states. We proceed in three steps. In the next section, group conflicts are put into the broader conceptual context of social cohesion, and the association between conflict perception and social cohesion is tested. Then, levels of conflicts as perceived by the citizens of the 25 EU member states and the 3 candidate countries are described. Finally, differences in conflict perception are explained, first between countries, then within countries. Our database is the European Quality of Life Survey (EQLS), which was conducted in 2003 in 28 European countries.
1. Conflicts: a neglected dimension of social cohesion research Issues of social cohesion have gained growing attention within the last years, both from academia and policymakers. The basic concern is that immigration, economic restructuring and globalisation pose a threat to the integration of modern societies (Council of Europe 1998). The EU defines social cohesion as relating ‘to the degree to which individuals and groups within a particular society are bound by common feelings of consensus, share common values and goals and relate to one another on a co-operative basis’ (Eurostat 2001). Despite this broad definition, EU policy initiatives have been heavily concentrated on issues of poverty, unemployment, and social exclusion, and sociological research has adopted this narrow focus. Problems of poverty, deprivation and exclusion are graver and more widespread in the new member states, especially in post-communist countries (Böhnke 2004; Russell and Whelan 2004). But there is considerable variation within the camp of EU-15, too, and some welfare states, especially the Nordic ones, are more effective than others in combating social problems. However, issues of poverty and social exclusion tell only part of the cohesion story. They highlight how well individuals or certain risk groups are integrated into the labor market, into social security systems, or society at large. Unfortunately, little can be learned about how different social groups such as rich and poor, managers and workers, or nationals and immigrants relate to each other. Whether their interrelations are basically co-operative or hostile is another but yet important aspect of social cohesion (Berger-Schmitt 2000). Strong group tensions indicate fragmentation of society and difficulties in mediating conflicting interests. This is not to argue for harmony. From proponents of conflict theory (Coser 1965, Dahrendorf 1959) one can learn that conflicts are built into social order; hence in open societies group tensions are normal aspects of social life rather than dysfunctional occurrences (cf. Crouch 2000). As Simmel (1950) has observed, conflicts are actually functional for social relationships, because they integrate the antagonists through a common framework of rules and negotiations. In a similar vein, Dahrendorf (1959) has argued, against Marx, that conflicts maintain social order rather than revolutionise it. Nevertheless, it can be argued that milder forms of conflicts are more likely to bring the benefit of drawing society together than strong conflicts. Especially if institutions for effectively resolving conflicting interests are lacking, conflicts are likely to fuel solidarity within contending groups rather than between them. Hence, on theoretical grounds there is an ambivalent relationship between group tensions and social cohesion.
330
Jan Delhey and Wolfgang Keck
We cannot clarify in this small chapter whether conflicts are ‘good’ (functional) or ‘bad’ (dysfunctional) for societies and their cohesion. But since we are concerned with group tensions as perceived by ordinary citizens, our objective is to clarify if the perception of conflicts indicates a lack of social cohesion in the eyes of the citizens. The EQLS data which we use in this chapter give us the opportunity to scrutinise this by relating tension perceptions to generalised social trust. Trust, as the belief that others will behave in a friendly and reliable manner (Inglehart 1991), can be read as the respondents’ estimation of the trustworthiness of the people around them (Delhey and Newton 2003). This collective property makes social trust a generally accepted indicator of social cohesion (Berger-Schmitt 2000; Berman and Philips 2004; Chiesi 2002) – and opens up the opportunity to investigate whether tension perception and distrust go hand in hand. In particular, we expect those individuals perceiving a lot of tensions as least trustful. Separately for each country, Figure 14.1 shows the individual-level association between perceived conflicts between poor and rich people and social trust.2 More precisely, it is tested whether the perception of ‘some tensions’ or ‘a lot of tensions’ is detrimental to trust, compared to the perception that there are ‘no tensions’ at all. The key result is that especially strong tension perceptions matter. In all countries except Lithuania, people who think that there are a lot of tensions between rich and poor show a lower level of interpersonal trust. This tendency proved to be significant in 17 of 28 countries. The perception of ‘some’ tensions reduces generalised trust less consistently, and in only three countries in a statistically significant way. We take this as an indication that people do not want to live in a high-conflict environment, whereas ‘some’ group tensions are regarded as ‘normal’ and unproblematic contentions in pluralistic societies.3 This interpretation is reinforced by the fact that most respondents in the EQLS survey chose the ‘some tensions’ option, when asked for rating tensions between rich and poor. Strong tensions, in contrast, evoke distrust and undermine social cohesion. Hence we feel justified to restrict our analysis to those respondents seeing ‘a lot of tensions’ (in the following also referred to as strong tensions or strong conflicts). What kind of group tensions might be characteristic of the EU member states? Traditionally, the sociological interest in industrial societies has focused mainly on class and/or distributional conflicts. But already in the 1950s, Dahrendorf (1959) has argued that social conflicts are multi-faceted and do not congeal around one central issue like class. Manifold conflicts can release societies from being preoccupied with one single fundamental cleavage. Moreover, conflicts may go through phases or lifecycles. They develop, rise, and decline. A frequent assumption for advanced postindustrial societies is that ‘traditional’ distributional conflicts, especially class-related ones, are in decline, while new conflicts arise. Beck (1986) highlights ecological risks, Inglehart (1989) stresses conflicts around post-materialistic issues, and Castells (2003) highlights the power of cultural identity. Hondrich and Caplow (1994) list several new conflicts – among them gender conflicts – that are said to have replaced traditional ‘vertical’ conflicts. In particular, they conceive of immigrants vs. nativeborn as the coming major divide in (West) European societies. In view of strained pension schemes, others have announced the clash of generations. The European Quality of Life Survey gives us a set of 28 European countries, East and West, at very different levels of modernisation, hence it is particularly suited for testing the idea that conflict agendas depend on societal modernisation. We expect to find the lowest
Social cohesion in new and old member states
331
Denmark Finland Sweden Ireland United Kingdom Netherlands Austria Belgium France Germany Luxembourg Greece Italy Portugal Spain Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania Bulgaria Romania Turkey −2.5 −2.25 −2 −1.75 −1.5 −1.25 Beta Coefficient
−1
−0.75 −0.5 −0.25
0
0.25
0.5
Some tensions, significant
Some tension, not significant
A lot of tensions, significant
A lot of tensions, not significant
Figure 14.1 Impact of perceived tensions between rich and poor people on social trust Source: EQLS 2003, own calculations Notes: Reference is no tensions perceived; coefficients are from an OLS-regression controlled for sex, age, income, occupational status, and education. Significance is reported at a 5% level.
332
Jan Delhey and Wolfgang Keck
perceptions of vertical conflicts in the post-industrial societies of Western Europe. A different line of reasoning can be found in Schumpeter’s classical essay ‘Die sozialen Klassen im ethnisch homogenen Milieu’ (1927), where the argument is developed that ethnic homogeneity and class conflict are related to each other. Rapid social change in ethnically homogenous societies boosts class conflict because social mobility is higher and status positions are in flux. Therefore, the transformation processes in the post-communist countries should facilitate class conflicts. In ethnically heterogeneous countries, where ethnic origin determines life chances and cultural orientations to some degree, class conflicts should become less dominant, especially if there are cross-cutting cleavages between classes and ethnic groups.
2. Levels of perceived tensions in European societies 2.1 The data Our analysis is based on data from the European Quality of Life Survey, which has been carried out in summer 2003 in the 15 EU member states, the 10 accession countries, and the 3 candidate countries at that time. As described in the introduction, the surveys are representative of the population aged 18 and over in each country with sample sizes around 1,000 respondents, except for the 5 smallest countries, Luxembourg, Cyprus, Estonia, Malta and Slovenia, which have sample sizes of 600. The EQLS inquired how respondents assess the intensity of tensions between social groups in their country by asking the following question: In all countries there sometimes exists tension between social groups. In your opinion, how much tension is there between each of the following groups in [this country]: (1) (2) (3) (4) (5)
Poor and rich people Management and workers Men and women Old people and young people Different racial and ethnic groups?
The response options were ‘a lot of tensions’; ‘some tensions’ or ‘no tensions’. From our list of five conflicts one can easily imagine that they do not congeal around one central issue. Poor vs. rich and management vs. workers can both be classified as vertical conflicts, since the antagonists occupy advantageous or disadvantageous positions in the social hierarchy. The former cleavage centres on the controversial issue of who gets what and how much. The items to be distributed are usually tangible – money, wealth, property, for example. The latter type of tension centres on conflicting interests and power relations at the workplace. In contrast, old vs. young and men vs. women are horizontal conflicts, based on ascriptive characteristics. They may overlap with vertical conflicts, but do not necessarily do so. Lastly, relations between racial and ethnic groups cut across the dichotomy of vertical and horizontal. They certainly entail an economic dimension, since often immigrants live at the margins of their host society (Hoffmann-Novotny 1987; Park et al. 1925). But issues of culture, identity and political rights are also at stake; hence we treat ethnic
Social cohesion in new and old member states
333
conflicts as a third type. These conceptual types are confirmed empirically by factor analysis. Therefore, we combine the conflict items poor vs. rich and management vs. workers into one factor, vertical tensions. Similarly, old vs. young and men vs. women can be combined to horizontal tensions. Finally, ethnic tensions perception constitutes a third dimension, which is statistically associated with both vertical and horizontal tensions. A weakness of our indicator of ethnic tensions is that we do not know to which particular racial and ethnic groups the answers refer. 2.2 Tension perceptions across Europe Figure 14.2 provides the information about how strong tensions are in European societies according to their citizens. The countries are presented in descending order of vertical tension. At the bottom of Figure 14.2, the country group averages are displayed. Denmark Luxembourg Finland Portugal Italy Sweden Germany Ireland Austria United Kingdom Spain Netherlands Belgium France Greece Cyprus Estonia Lativa Lithuania Poland Slovakia Malta Slovenia Czech Republic Hungary Bulgaria Romaina Turkey EU-15 NMS CC-3 EU-25 0
20 Horizontal
40 Vertical
60 Ethnic
Figure 14.2 Intensity of group tensions as perceived by the population (% saying ‘a lot of tensions’) Vertical tensions: simple average of ‘poor and rich’ and ‘management and workers’. Horizontal tensions: simple average of ‘old people and young people’ and ‘men and women’. Country group averages are simple country averages and do not consider different country population sizes.
334
Jan Delhey and Wolfgang Keck
In the EU-25, on average 42 per cent perceive a lot of tensions between ethnic or racial groups, 34 per cent perceive strong vertical tensions, and 14 per cent strong tensions between age groups and men and women. Hence, citizens are much more concerned with either clashes between ethnic groups or with ‘traditional’ vertical cleavages, than with horizontal ones. Moreover, the rating of horizontal tensions is rather homogeneous across the enlarged Europe. In most countries, between 10 and 20 per cent say that tensions between men and women and between age groups are strong. Only the Greeks (27 per cent) and Turks (33 per cent) exceed this figure, two culturally rather traditional societies (Gerhards 2005; Inglehart 2002). These findings induced us to concentrate exclusively on vertical and ethnic tensions for the remaining sections of the chapter.4 Although we have no time-series data at hand, it is clear that distributional issues have not completely faded from the agenda. If it is true that they have been replaced to some extent by new cleavages, then ethnicity rather than generations or gender have gained importance, which may reflect the fact that many member states, especially in Western Europe, have become immigration societies. From an enlargement perspective, the crucial finding is that the conflict agendas of old and new member states differ. Whereas in the new member states, distributional issues are seen as a more important cleavage than ethnic tensions (43 per cent vs. 34 per cent), the ranking in the old member states is the other way around (29 per cent see strong vertical tensions, 46 per cent strong ethnic tensions). Thus the last wave of enlargement has not so much brought in societies with greater cohesion problems in general, but rather countries with a different conflict agenda. Possible future enlargements with Bulgaria, Romania and Turkey will even reinforce the shift towards distributional conflicts. The severity of tensions varies widely among individual nations. Vertical tensions are seen as especially strong in some post-communist countries of Central-East Europe and in Turkey. The post-communist citizens, drilled over decades to cherish egalitarian ideals, had to accommodate to rising levels of inequality during system transformation (Milanovic 1998). Among the EU-15, only Greek and French respondents hold the majority view that society is shaken by vertical conflicts. In contrast, the Nordic states are perceived as largely conflict-free by their citizens. This is consistent with the view that universal welfare states foster social cohesion. A low level of tensions is also reported for Luxembourg and Cyprus. The other countries are located in a middle position between these two extreme groups. Italy, Malta, the UK and the Netherlands lean more towards the Scandinavian pattern, the Czech Republic, Slovenia, and Slovakia more towards the pattern of the other post-communist countries. The perceived severity of ethnic tensions varies to a considerable extent, too. France and to a lesser extent also the Netherlands, Belgium, Greece, the Czech Republic and Hungary stand out as countries where majorities of the respondents see a lot of tensions. In most of the new member states ethnic relations are perceived as much less conflictual, with the Baltic countries and Cyprus, as well as Bulgaria, distinguished by their low levels of perceived tensions. These latter findings are counterintuitive given that in Bulgaria the integration of the Turkish minority is far from being easy, whilst the Baltic Republics have the problem of integrating the large Russian-speaking minority.5 The low level in Cyprus is probably related to the fact that the two potentially hostile national groups were divided into separate parts of the country and that the survey was administered only in the Greek part. In Figure 14.3, the intensity of vertical and ethnic conflicts is displayed simultaneously. The circles mark off countries with similar patterns, which have been identified
% perceive a lot of tension between ethnic groups
Social cohesion in new and old member states
335
70 4 NL
60
FR
CZ
50 40
IR IT
DK
FI
1
PT
GR
HU
MT
UK SE
8
BE
TR
SK AT
ES
SI
DE
7
3
RO
30 LU PL
2
20
LV CY EE
5 BG
10
6
LT
0 0
10
20 30 40 50 % perceive a lot of vertical tensions
60
70
Figure 14.3 Intensity of perceived vertical and ethnic tensions, country clusters Source: EQLS 2003, hierarchical cluster analysis
by means of a statistical cluster analysis. The most cohesive societies are Denmark (cluster 1, almost no vertical tensions, and average ethnic tensions) and Luxembourg and Cyprus (cluster 2, below-average in both dimensions). France, Hungary and Greece stand out as the least cohesive societies where both kinds of tensions are perceived as rather strong by a majority of people (cluster 8). Lithuania (cluster 6) is unique in scoring high on vertical tensions, but very low on ethnic tensions. The Netherlands, Belgium, and the Czech Republic (cluster 4) stand out with the combination of under-average or average vertical tensions and of strong tensions between ethnic groups. The largest cluster (cluster 3) is characterised by under-average or average vertical tensions, and average ethnic tensions. Except for Malta this cluster exclusively consists of old member states. The distribution of countries is far from resembling any of the well-known families-of-nations typologies. The most obvious pattern is the almost perfect distinction between EU-15 member states and post-communist countries. The difficult transformation towards democracy and market society in countries with a socialist tradition has obviously nurtured a climate of strong vertical tensions. Among the EU-15, only France and Greece are located in the righthand side of the diagram, indicating strong distributional conflicts.
3. Country differences in levels of conflicts 3.1 Composition effects? The results that have been presented so far take no account of nation-specific differences in the socio-demographic composition of the population. Yet people living in each nation differ in some respects: some countries have older populations and some younger, some have more educated populations and some less educated.
336
Jan Delhey and Wolfgang Keck
If socio-demographic groups differ in their views on conflicts, then these differences between individuals will confound comparisons between nations (cf. Kelley and Evans 1999). To test for composition effects we used regression standardisation procedures. In particular, we adjusted for age, sex, region, income and employment status. This procedure revealed that by and large, individual-level differences in perception of tensions are rather small. The key message here is that composition effects are not at work. When the standardised social composition is applied, conflict perceptions usually change only slightly, by less than 1 percentage point. There are only two cases (out of 56 cases tested)6, where the standardised conflict measure deviates more than 3 percentage points from the reported value.7 Country differences are not caused by composition effects of the populations, but either by social, economic and political conditions, or by less tangible forces like culture or national history. 3.2 Explaining vertical tensions In this section we probe deeper into the question why social cohesion is lower in some places than in others. Since vertical and ethnic conflicts relate to different strands of sociological thinking, we have to distinguish between the two, beginning with vertical conflicts. The level of inequalities seems to be a self-evident explanation for cross-national differences. In fact we would expect inegalitarian societies to breed more tensions than egalitarian societies. By and large, this rule of thumb holds. Our first measure is the Gini-Index, which measures the degree of income inequality between individuals.8 Indeed, tension perceptions rise with income inequality, albeit the association is not very strong (correlation 0.33, p = 0.09, 27 countries). Moreover, at the same level of inequality, we find a great diversity of conflict ratings, e.g. between Hungary and Denmark. This suggests that the levels of inequality regarded as acceptable vary widely across Europe (Arts et al. 1995; Delhey 2001; Haller et al. 1995; Kluegel et al. 1995). A second, more indirect measure of inequality is the unemployment rate. High unemployment rates seem to fuel awareness of vertical tensions more than the overall degree of inequality does (r = 0.58, p = 0.002, 27 countries). This indicates that the citizens of various European countries may have discrepant views regarding acceptable levels of inequality, but are rather similar in regarding high levels of unemployment as intolerable. Moreover, high unemployment weakens the position of workers against managers, and the fear of becoming unemployed infects larger segments of the population. Both developments can be expected to sensitise public opinion for distributional conflicts. The weakness of the two measures of inequality employed here is that they do not measure directly the incidence of vertical conflicts, but more indirectly the potential for their emergence. The emergence of conflicts also depends on additional factors, among them above all the level of economic development. The richer a society is, the less likely groups presumably are to clash over distributional issues because even though inequalities still exist, they lose their sting (Beck 1986; Schulze 1992). The data strongly support this idea: in richer societies fewer respondents perceive strong vertical tensions. No other country characteristic shows such a strong association (r = −0.65, p = 0.000, 28 countries). This suggests that up to a certain threshold of national wealth, societies are highly concerned with who gets what, but such concerns take a back seat once higher levels of prosperity are reached. Trade unions may also contribute to a climate of social peace – or conflict. There are two rival assumptions: in the tradition of Tilly’s resource mobilisation school
Social cohesion in new and old member states
337
(McAdam et al. 2001), one could argue that unions fuel conflicts in mobilising workers and enhancing respective awareness. In contrast, drawing on Dahrendorf’s conflict sociology, there are equally good reasons to expect that unions channel and pacify distributional conflicts. We use union density rates for testing these two rival assumptions. As many have discovered before us, it is not easy to compare union density rates across nations. We use two sources, the website of the European Foundation for the Improvement of Living and Working Conditions, and Visser (2006). No matter which of the two sources one employs, strong unions make for less intense tension perceptions. However, the association is much stronger for the Visser data (r = −0.71, p = 0.002, 16 countries) than for the European Foundation’s data (r = −0.33, p = 0.119, 24 countries). We interpret the negative coefficients as a pacification effect, which is more in line with Dahrendorf’s theory than with Tilly’s. Strong trade unions help channelling industrial conflict, and fostering a tradition of social dialogue they function as a safeguard of social peace. In this sense, trade unions work for, not against (perceived) social cohesion in the European social model. Welfare state proponents also see embracive social policies as an effective antidote to social conflicts. This idea of a pacifying effect of the welfare state is supported by the data (r = −0.34, p = 0.107, 24 countries), although the association is not particularly strong. But as a rule of thumb, the more a country spends on social issues, relative to GDP, the less its citizens are prone to perceive strong distributional conflicts. Finally, social capital theorists claim that ‘features of social organisation such as networks, norms, and social trust … facilitate coordination and cooperation for mutual benefit’ (Putnam 1995). Hence, the larger a society’s stock of social capital, the more smoothly can interest conflicts be solved, whereas a lack of social capital can block co-operation and fuel antagonisms. Empirically, our measure of social capital, civic participation, is negatively associated with tension levels (r = −0.54, p = 0.005, 25 countries). This confirms that strong civil societies are low-conflict societies and that social capital helps in bridging differences between economic groups. We found strong support for Schumpeter’s classical idea that vertical conflicts are more dominant in societies undergoing rapid social change – which are, namely, the post-communist countries. In contrast, we found no support for his second thought that ethnic homogeneity enhances the awareness of the class cleavage. Across the 28 countries, there is virtually no relationship between measures of ethnic composition on the one hand, like (Alesina et al. 2003) ethnic fractionalisation measure (r = 0.18, p = 0.378) or the proportion of foreigners (−0.16, p = 0.487), and perceived vertical tensions on the other. Obviously, vertical tensions can emerge in both homogeneous and more heterogeneous settings (Figure 14.4). It is worth mentioning, however, that in all our countries except Luxembourg, the proportion of foreigners is below 20 per cent, and that we have no information at hand to what extent social positions are allocated along ethnic lines. To sum up, social stability and continuity, prosperity, low unemployment, and a strong civil society seem to be the most important factors contributing to a climate of low vertical tensions. Strong unions and a developed welfare state may have also pacifying effects. (However, to single out the influence of these factors is impossible, since to some degree they are all intertwined.) In this light we are better able to understand why reported tensions are so strong in most post-communist nations. While their ideological heritage predisposes the citizens to be critical against inequalities, since the collapse of state socialism many got poorer, and only few got richer. Large segments of the population have suffered from periods of unemployment or stay in
338
Jan Delhey and Wolfgang Keck Unemployment % perceiving strong vertical tensions
% perceiving strong vertical tensions
Income inequality (GINI) 80 70 60
HU FR SI BG
50
PL RO
GR LT TR
CZ LV EE DE BE ES MT AT IR NL IT UK PT SE LU FI CY
40 30 20 10
DK
0 10
20
30
40
80 70 GR
60
HU RO
50 40
AT IR NL UK PT LU SE CY
30 20 10
0
2
GR LT TR PL HU RO FR SI BG SK CZ LV EE ES DEBE MT AT IR IT UK PT NL SE FI CY
10
LU
DK
0 5000
15000
25000
% perceiving strong vertical tensions
% perceiving strong vertical tensions
20
4
6
70
50 40 30
LU
10
DK
0 15
30 20 10
LV EE ES
HU
RO BG
DE UK NL
BE IR PT
AT
MT IT LU
CY
20
25
30
35
Union density (Visser)
SE FI
DK
0 10 20 30 40 50 60 70 80 90 100 Union density, % of eligible workforce
% perceiving strong vertical tensions
% perceiving strong vertical tensions
40
BE DE AT PTIT UK NL SE FI
Social spending as % of GDP
GR SI
FR
SI
20
10
70
SK
HUPL
SK CZ LV EE ES IR MT
Union density (EF)
50
GR
LT
60
35000
80
PL
8 10 12 14 16 18 20
80
GDP per capita in PPP
60
FI
Social spending
70
30
CZ EE LV BE DE ES MT IT
Unemployment rate, in %
GDP level
40
SK
0
50
80
50
PL BG
DK
Income inequality, Gini index
60
LT
FR SI
80 70 60 50 40 30
FR
PLHU SK CZ ES DE
BE
AT IR NL UK IT FISE
20 10
DK
0 10 20 30 40 50 60 70 80 90 100 Union density, % of eligible workforce
Figure 14.4 Cross-national correlates of vertical tension perception
Social cohesion in new and old member states Proportion of foreigners
% perceiving strong vertical tensions
% perceiving strong vertical tensions
Social capital (civic membership) 80 70 GR
60
HU
50
LT PL FR SK SI
CZ LV EEBE DE ES AT MT IR PT IT UK LU CY
40 30 20
NL SE
FI
10
339
DK
0 10 20 30 40 50 60 70 80 90 100 Association membership, % of population
80 70 GR
60
HU PL
50
FR SI CZ LV DE ES BE IT IR AT PTNL UK FI SE
40 30 20 10
SK
LU
DK
0 0
10 20 30 Proportion foreigners, in %
40
% perceiving strong vertical tensions
Ethnic fractionalisation 80 70 GR
60
LT
PL HU FR SI SK
50
CZ
40
DE MT ATIR IT PT NL UK SE FI CY
30 20 10
ES
EE
BE LV
LU
DK
0 0
.1
.2
.3
.4
.5
.6
.7
.8
Ethnic fractionalisation, Index 0–1
Figure 14.4, cont’d.
insecure job situations. The dismantled socialist welfare systems were unable to compensate for income losses. Despite the return to economic growth since the mid-1990s, levels of living are still far below the European Community average. Moreover, in a climate of distrust (Sztompka 1999), rich people are often perceived as dishonest thieves rather than smart entrepreneurs (Franzen et al. 2000). All this makes citizens in the post-communist countries more inclined to see vertical tensions. The Nordic countries and Luxembourg, in contrast, combine fortunate conditions such as prosperity, low inequalities and low unemployment, a generous welfare system, and a strong civil society. 3.3 Explaining ethnic tensions As for vertical tensions, we have no comparative information on the ‘real’ magnitude of ethnic conflicts for the 28 countries at hand. But we can borrow some ideas from
340
Jan Delhey and Wolfgang Keck
social-psychological theories about prejudice and attitudes towards foreigners. Although primarily these are individual-level theories, we apply them here at the country-level. Classic realistic group conflict theory (Levine and Campbell 1972; Sherif 1966) holds that competition for scare resources like wealth or territory is the main driver for prejudice. Recent theories emphasise the subjective perception of threats rather than objective competition (Stephan and Stephan 2000). Moreover, in this ‘integrated threat theory’, a distinction is drawn between realistic threats – perceived threats to the power and the well-being of the ingroup – and symbolic threats – perceived threats to the worldview of the ingroup. Both can drive ethnic conflicts. It is not easy to operationalise these ideas for our 28 countries, but one can, first, imagine that the level of ethnic competition, whether real or perceived, increases with mere ethnic diversity (cf. Quillian 1995; Taylor 1998). We use three ethnic composition measures, two of them describing the magnitude (ethnic fractionalisation, proportion of foreign population), the third describing population changes (net migration). The scatter diagrams reveal, however, that ethnic conflicts are not a simple function of ethnic composition. On the one hand, tension perception is somewhat more widespread in populations with a larger share of foreigners, if one excludes Luxembourg, which is an exceptional case both with respect to the number and the composition of foreigners – for the most part, highly educated West European Eurocrats or bank employees.9 On the other hand, contrary to this finding there is a mild tendency that tension perception decreases with ethnic fractionalisation. This surprising relationship, however, is primarily an effect of the unusual pattern of the three Baltic states and Luxembourg, which are highly fractionalised, but without reporting strong tensions. Hence it would be premature to conclude that the climate of ethnic relations is totally unrelated to the ethnic composition of the population. But obviously European citizenries have very different capabilities to tolerate ethnic diversity. Another caveat must be made with respect to our measures of ethnic composition, which are crude at best. The well-known example of naturalised Frenchmen from African origins, who are counted as ‘nationals’ in the census, might illustrate the problem of measuring the proportion of ‘foreigners’ or ethnic diversity. With respect to immigration, we find again a mild but non-significant tendency for levels of reported tensions to be higher in countries with a large inflow of migrants, although the picture is very complex and more of a reversed U-curve than a steady line (for a similar finding: Rosar 2004). Second, poor economic conditions might fuel ethnic competition. If jobs are scarce and the economy is flagging, foreigners might be perceived as competitors over scarce positions and goods in the first line. Contrary to this expectation, poor labour market conditions do not breed ethnic tensions in Europe. Since it is predominantly the postcommunist new member states and candidate countries which deviate from the expected pattern, caution in generalising the finding is in place, however. A stagnant economy, does in fact breed ethnic tensions (r = −0.37, p = 0.07, 28 countries), but this pattern is heavily influenced by the three Baltic tigers. Attitudes about immigrants can serve as more direct measures of perceived threats to the ingroup. For our purpose we utilise opinions on whether there are too many immigrants in the country. The proportion of citizens saying that the country hosts ‘too many immigrants’ turns out to be the best single correlate of ethnic tensions (Figure 14.5). The more widespread this opinion is, the more tensions between ethnic and racial groups are reported (r = 0.55, p = 0.002, 28 countries). Taken together
Social cohesion in new and old member states
Proportion of foreigners (ex LU)
80 70 60 50 40
FR NL BE HU CZ GR UK IR SI ES SE IT AT FI DK PT DE
30
SK
LU
PL
20 LV
10 0 0
10
20
30
% perceiving strong ethnic tensions
% perceiving strong ethnic tensions
Proportion of foreigners 80 70 60 50 40
FR NL BE HUCZ GR UKIR SI SE IT ES AT FI DK PT DE
30
SK
LU
PL
20 LV
10 0
40
0
10
Proportion foreigners, in %
20
FR NL
50 40
BE
GR
CZ
MT UK HU SE IR SK ATIT SI PT DK FI DE
30
ES
LU
PL
20
LV
CY
EE
LT
10 0 0
.1
.2
.3
.4
.5
.6
.7
80 70
FR
60
CZ HU
50
UK SE SK SI AT IT FI DK DE
40 30
10
30
0
GDP growth 1998–2002
SK
LU
20
PL LV
CY
EE
10
BG LT
0 0
2
4
6
8 10 12 14 16 18 20
Unemployment rate, in %
CY
−2 −1 0 1 2 3 4 5 6 7 8
% perceiving strong ethnic tensions
% perceiving strong ethnic tensions
40
LU
Net migration rate per 1000 inhabitants
NL
50
ES
PT
EE
LT
Unemployment
FR BE HU CZ GR MT UK IR SE ES SI AT DK IT DE PT FI RO
MT IR
PL
.8
80
60
BE NL GR
LV
20
Ethnic fractionalisation, Index 0–1
70
40
Net migration
% perceiving strong ethnic tensions
% perceiving strong ethnic tensions
Ethnic fractionalisation
60
30
Proportion foreigners, in %
80 70
341
80 70
FR NL BE CZ GR
60 MT
50
HU
UK SE SI ES SK IT AT FI DE DK PT
40 30
IR
LU
20
CY
10
LV EE LT
0 0
5 10 15 20 25 30 35 40 45 50 GDP growth in %, 1998–2002
Figure 14.5 Cross-national correlates of ethnic tension perception
342
Jan Delhey and Wolfgang Keck
% perceiving strong ethnic tensions
Too many immigrants in country 80 70
FR
60 50
SE
40
FI
SK DK
RO
30
PL
BE GR NL CZ HU MT UK IR TR ES IT SI ATDE PT LU
LV
20
BG
10
EE LT
CY
0 30
40 50 60 70 80 90 100 Too many immigrants, % agree
Figure 14.5, cont’d.
with the questionable effect of ethnic composition of the population, this shows that the evaluation of ethnic relations are more a matter of worries and perceived threat than of objective ‘problems’ and realistic threat. This makes it both difficult to explain cross-country differences in ethnic tension perceptions, and to mitigate them through public policies.
4. Individual-level differences: ‘class consciousness’ or consensus? How much within-country dissent is there about the level of conflict? Departing from Karl Marx’s famous dictum ‘Being determines consciousness’, structural theories postulate that attitudes are a function of an individuals’ life situation and positions in the social structure. Groups in distinctive positions have opposing interests and therefore can be expected to have different attitudes (Kiecolt 1988). We assume that those in disadvantaged social positions perceive particularly strong vertical tensions, whereas those in advantaged positions tend to downplay them. Table 14.1 shows if this assumption holds true. The entries reveal the extent to which certain social groups deviate in their awareness of tensions from the population at large. Advantaged position was measured as top-income quartile (but this does not necessarily mean rich people), by managerial occupational position, and by tertiary education. Disadvantaged position was measured as bottom income-quartile (not necessarily poor people); workers; and primary education. At first glance, structural position predicts quite well the general tendency of deviations from average opinion. In almost all countries, higher-status groups see fewer tensions than the population average, while low-status groups see more. There are only very few exceptional cases where the disadvantaged are slightly less conflict-conscious than the population average. However, the deviations are usually small. Moreover, a multivariate analysis showed the difference between low- and high-status
Social cohesion in new and old member states
343
Table 14.1 Vertical tension perception in different social groups (deviations from national mean in % points) Occupational class Country
A lot of Managerial Worker tensions(%) Mean
Income position
Educational level
Highest quartile
High
Lowest quartile
Low
Deviation from mean in % points
Denmark Finland Sweden Ireland United Kingdom Netherlands Austria Belgium France Germany Luxembourg Greece Italy Portugal Spain
5 19 20 27 24 24 29 34 49 35 21 59 25 25 34
0 −3 −7 −7 −3 −5 −1 −9 −4 −1 −2 −3 −4 −8 −5
2 2 4 2 1 6 2 6 3 5 0 5 5 2 3
−2 5 −6 −2 −3 −8 4 −1 −4 −10 −10 −3 −1 −12 −11
1 6 4 10 12 4 8 12 20 26 0 5 12 8 2
−1 −3 −3 −3 −4 −4 0 −4 −3 −4 −7 −5 −2 −6 −2
−2 6 4 10 −1 8 0 5 −1 13 −4 5 −1 3 0
Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania
18 29 46 39 54 52 45 36 36 57
−3 −10 −8 −2 −3 −2 −5 −2 −4 1
6 18 1 9 2 5 4 2 5 3
−4 −6 −9 −7 −5 −2 −2 −9 −15 −11
−14 15 −8 17 15 7 6 −1 −2 10
−7 −3 −3 −5 −2 −2 −1 0 −4 −4
5 2 1 8 3 −4 2 13 6 8
Bulgaria Romania Turkey
45 50 54
−5 4 0
4 0 0
−14 4 −6
3 −1 −4
−5 1 −9
1 −8 3
EU 15 NMS CC 3 EU 25 EU 28
29 43 50 34 36
−5 −5 −1 −5 −5
3 7 1 5 5
−8 −12 −7 −6 −6
4 2 −3 7 5
−3 −3 −5 −4 −5
1 2 2 0 2
Source: EQLS 2003, own calculations Notes: Grey shaded fields express a significant negative effect on tension perception if the privileged social group is compared to the less privileged reference group. Black shaded fields express a significant positive effect on tension perception if the privileged social group is compared to the less privileged reference group.
344
Jan Delhey and Wolfgang Keck
groups to be significant in only few countries, which suggests that worldviews are only weakly related to economic interests or class consciousness.10 This holds true for new and old member states alike. With respect to income position, in six countries the low-income group clearly differs from the rich group in their tension perception, if other characteristics such as sex, age, region, income and employment status are controlled for. Among the new member states, differences between top- and bottom-income groups are largest in Malta, the Czech Republic, Hungary and Lithuania. Among the EU-15, they are biggest in France, Germany, Italy and Portugal. But in most countries, the similarities between socio-economic groups are more striking than the differences. The most consensual opinions can be found in the Nordic countries. Differences between occupational groups are less marked than income differences. In most countries, people in managerial professions on the one hand, and workers on the other, neither deviate much from the population mean, nor from each other. This is especially interesting because these two groups happen to be directly addressed in one of our questions on vertical conflicts: tensions between management and workers. In none of the countries do managers and workers differ significantly in their perception of tensions. Among the new member states only Maltese workers stand out as seeing many more tensions than the average citizen. In the EU-15, occupational groups differ most strongly in Belgium. By and large, however, occupational position is far from determining tension awareness in Europe. Education tells a similar story. In a few countries only (Ireland, Luxembourg, Cyprus and Turkey) conflict perceptions differ substantially between highly and poorly educated. To sum up, tension perceptions are stratified in some countries to some extent, but not in a strong way. The Nordic countries show almost no stratification of attitudes at all. We also found no evidence that tension perceptions are more strongly stratified in the new member states. Whilst citizens of post-communist countries stand out for their particular awareness of distributional conflicts, this awareness is not stronger related to socio-economic position than elsewhere in Europe. Also with respect to ethnic tensions, people in disadvantaged positions can be expected to be more conscious of conflicts. The argument of realistic conflict theory goes that some segments of the population compete more than others with immigrants for scarce jobs, social benefits and affordable housing – namely the low-educated, blue-collar workers, the poor and unemployed people. This vulnerability to competition from immigrants might elevate the awareness of ethnic tensions. By contrast, fishing in different waters, the economically well-off can be much more relaxed about foreigners and enjoy the cultural enrichment which they bring to our societies without feeling threatened by them. Kunovich (2002) found blue-collar workers and the unemployed to be more prejudiced against immigrants, while education and income decrease prejudice. He also found social structural variables to play a more important role in the development of attitudes in Western Europe than in Eastern Europe. In contrast, our results do not give substantial support to the claims of realistic conflict theory. Perceptions of ethnic conflicts are neither concentrated within the bottom income group, nor among workers and the low educated (Table 14.2). Only Portugal shows this pattern. In Finland, it is the high status groups which judge ethnic tensions as problematic, not the low ones. At first glance, this seems to contradict earlier research which finds stronger anti-immigrant prejudice and attitudes among
Social cohesion in new and old member states
345
Table 14.2 Ethnic tension perception in different social groups (deviations from national mean in % points) Occupational class Country
A lot of Managerial Worker tensions (%) Mean
Income position
Educational level
Highest quartile
High
Lowest quartile
Low
Deviation from mean in % points
Denmark Finland Sweden Ireland United Kingdom Netherlands Austria Belgium France Germany Luxembourg Greece Italy Portugal Spain
39 37 44 46 49 61 40 60 64 38 25 57 40 36 42
−7 −3 −10 −1 1 −3 −1 2 0 3 −4 −1 1 3 3
4 0 4 2 −2 −4 1 1 2 −2 −4 −2 5 0 0
−11 12 −8 1 3 −1 −2 −3 −6 −5 −16 −17 −17 −9 4
2 −10 3 −3 1 −3 −3 −1 2 −12 2 −7 −10 8 7
−2 −2 −4 0 −1 −2 2 2 −2 4 −5 −4 −2 −2 2
5 2 6 1 −3 2 3 0 −2 1 −1 −3 −2 −5 −1
Cyprus Malta Slovenia Czech Republic Hungary Poland Slovakia Estonia Latvia Lithuania
16 50 42 55 55 23 43 13 19 10
−3 −5 −3 5 −5 −5 −6 −1 −2 −1
6 17 −6 −1 −1 0 0 −4 −1 1
−6 15 −15 −3 12 −3 −1 −3 −10 −2
−5 14 −4 −5 3 11 −1 −10 −3 −1
−4 −13 −2 −2 5 −3 −5 −2 0 −3
−2 −8 1 −7 9 2 1 −1 2 −1
Bulgaria Romania Turkey
14 33 46
−9 3 4
0 −5 −2
−5 −1 −3
−2 −5 −15
1 −2 −4
10 −4 −8
EU 15 NMS CC 3 EU 25 EU 28
46 34 39 45 44
−2 −3 −2 −2 −2
−1 0 −5 −1 −2
−6 −4 −11 −8 −8
−2 0 −16 −4 −6
0 −5 −6 −1 −1
1 −4 −2 5 5
Source: EQLS, own calculations Notes: Grey shaded fields express a significant negative effect on tension perception if the privileged social group is compared to the less privileged reference group. Black shaded fields express a significant positive effect on tension perception if the privileged social group is compared to the less privileged reference group.
346
Jan Delhey and Wolfgang Keck
disadvantaged segments of the population (Kessler and Freeman 2005; Kunovich 2002). But whereas attitudes towards foreigners capture individual belief systems, the question we utilised here focuses on the perception of the objective situation in the home country rather than on individual opinions. Hence, it is perfectly possible that people report a lot of conflicts because they witness them in everyday life experience or from media exposure, without being themselves involved in these conflicts, or without themselves holding attitudes that are negatively stereotyped against specific ethnic and racial groups. More refined analyses that would allow a comparison of the tension perceptions of the ethnic majority and of minorities in each country are not possible with the EQLS data. To sum up, tension ratings of advantaged and disadvantaged groups are strikingly similar. In the large majority of countries, positions in the social structure and class consciousness do not shape worldviews in this respect. One reason might be the above-mentioned socio-tropic nature of the question which asks for an account of the situation in the country rather than for the degree of antagonism felt individually. This suggests that individual answers to our question are quite good approximations of the ‘real’ societal climate. Conflicts are neither downplayed nor ignored from one side, nor overdrawn from the other. This signals a general acceptance of tensions as an unavoidable feature of pluralistic societies, and hence also good chances of successfully dealing with conflicts and of accommodating rival interests. With respect to ethnic relations, however, a negative implication is also worth mentioning: since the perception of ethnic tensions is not concentrated in a small underclass vulnerable to ethnic competition, the potential for a political mobilisation of unrest is rather high, in case the established parties should fail to channel the respective concerns.
5. Outlook This chapter has looked at levels of perceived conflicts between social groups as a possible threat to societal cohesion. First we demonstrated that the perception of strong tensions indicates a lower level of social cohesion as indicated by low levels of general trust. We have further shown that within countries there is more consensus than dissent about the strength of tensions. In European societies tension perceptions are only weakly stratified, i.e. privileged and disadvantaged groups do not differ much in their view on conflicts. Huge differences can be found, however, when comparing countries. Whereas old and new member states do not differ in their overall level of perceived tensions, they are preoccupied with different kinds of conflicts. In the post-communist countries, traditional class conflicts between rich and poor and between management and workers are seen as most concerning (‘traditional’ conflicts from a western point of view). In contrast, in the EU-15, citizens are most worried about relations between different ethnic and racial groups – worries that might even have grown since the survey was conducted in 2003. Hence, with recent enlargement towards the East, the conflict agenda within the Community was partly rewritten by giving traditional vertical conflicts on who gets what a greater weight – conflicts which have been successfully toned down in many affluent, social-capital rich welfare states of Western Europe, particularly so in the Nordic countries. In order to achieve its goal of strengthening social cohesion in the enlarged Europe, the EU has to fight two fires at the same time. (1) In the post-communist countries (as well as in France and Greece), the priority is to mitigate interest conflicts about
Social cohesion in new and old member states
347
the distribution of living conditions and life chances. A top-down strategy would require improving social security systems to make them more effective in preventing poverty, which in some of the new member states is particularly prevalent among pensioners. A bottom-up strategy would be to strengthen civil society in order to accumulate a larger stock of social capital, but this, of course, will work only in the long run, and with only vague prospects for success. If modernisation theory is right, then the mitigation of vertical tensions will be a side effect of economic catch-up. For a few years, the post-communist region has now enjoyed rapidly growing economies, and the more affluent people get, the more likely it is that distributional issues become less important. However, this optimistic outlook presupposes also that the disadvantaged groups will participate in the growing cake, and that the ways of getting rich are no longer viewed by the majority as being paved mainly by corruption and insider relationships. As long as the rich are considered as dishonest, distribution conflicts will loom large. Moreover, the survey has shown that in most societies of Western Europe, vertical conflicts have not faded away, with the partial exception of the Nordic countries. But since they depict a historically unique combination of fortunate social, economic, political and cultural forces, they may not provide a realistic benchmark for the other European societies. (2) The second fire the EU has to fight against consists of ethnic tensions – the top priority in most old member states, as well as in Hungary and the Czech Republic. It is beyond our expertise to give any sound suggestions as to how policy makers can react to this challenge. The ethnic problems within the single member states are too diverse to lump them all together, especially as long as we do not know which particular ethnic relations people have in mind when answering the EQLS question. However, it seems clear that steps must be taken from both sides, minorities and majorities, foreigners and natives, non-nationals and nationals, to get along. Since it is neither possible nor desirable to re-build the European Community into a ‘Fortress Europe’, a more cosmopolitan-minded Europe is needed (Beck and Grande 2004). Certainly, the macro-constellation of economic globalisation, rampant unemployment and fears of social descent on the one hand and growing numbers of immigrants into the European heartlands on the other are an adverse mixture for expanding our taste for cosmopolitanism. But there is no alternative. Otherwise we eventually run in to the danger of witnessing more Van Goghs being shot down on the street, and more burning cars in our suburbs. Table A14.1 Documentation of indicators Indicator
Definition
Source
Gini index
The Gini index of income inequality, taken from Human Development Reports, ranges from 0 (absolute equality) to 1 (absolute inequality). The higher the Gini index, the more unequal the income distribution is Harmonised unemployment rates, share of unemployed persons of the total number of active persons in the labour market. Active persons are those who are either employed or unemployed Gross domestic product per capita in purchasing power parities
United Nations 2005
Unemployment rate
GDP per capita
Eurostat 2005
Eurostat 2005 Continued
348
Jan Delhey and Wolfgang Keck
Table A14.1 Documentation of indicators—cont’d Indicator
Definition
Size of welfare state Proportion of foreigners
Total social expenditure per capita in purchasing power parities (2002/03) Share of population with foreign citizenship
Migration rate
Average yearly net migration rate between 1998 and 2002 Probability that two randomly selected individuals from a population belonged to different ethnic groups defined by racial and linguistic characteristics. The higher the score, the more diverse (fractionalised) the population. In perfectly homogenous countries, the probability is 0 Share of people who completely agree or somewhat agree on the proposition: There are too many immigrants in our country
Ethnic fractionalisation
Too many immigrants
Civic participation Union density 1
Union density 2
Source Eurostat 2005 Eurostat 2005, European Migration Centre 2005 Eurostat 2005 Alesina et al. 2003
Standard Eurobarometer 59.2 (2003), Candidate Countries Eurobarometer 2003.3, own calculations Eurostat 2004
Percentage of people participating in at least one organised activity (charity, religious activity, cultural activity, trade union, sport, environmental, etc.) Total membership figures as percentage of European total number of employees as defined in Foundation for the national labour force surveys; 2003, or the Improvement of most recent year for which data are available. Living and Countries were group in nine density groups Working Conditions 2005 Adjusted trade union members measured Visser 2006; Table 3 against the size of the employed wage and salary earners
Notes 1 We would like to thank the editors, in particular Jens Alber and Jelle Visser, for helpful comments on an earlier version of this chapter. 2 Social trust is measured by the question: ‘Would you say that most people can be trusted, or that you can’t be too careful in dealing with people?’ The scale runs from one (‘you can’t be too careful’) up to ten (‘most people can be trusted’). 3 We have repeated this analysis for the remaining conflict dimensions of the EQLS (see next section) as well as for a summary index of tensions, with very robust results. 4 In their chapter, Wolfgang Keck, and Agnes Blome (this volume, Ch. 3) deal with the potential of a generation cleavage. 5 In Latvia, 25 per cent of the population are non-nationals, in Estonia 20 per cent. These figures are made up almost exclusively of citizens from the former Soviet Union, especially Russians (Eurostat 2004). 6 The number of 56 tests is a product of having 28 countries and 2 conflict dimensions, vertical and ethnic.
Social cohesion in new and old member states
349
7 In Belgium, vertical conflicts score slightly lower after standardisation (−4 percentage points), in Malta slightly higher (+5 percentage points). 8 Detailed information on the indicators utilised is given in the appendix (Table A14.1). We are aware that some of the indicators are not the most refined measures of the concepts we are discussing, and that 28 countries are few cases to test theoretical assumptions against empirical data. 9 This shows that the composition of the foreign/non-national population might also be important, since nationalities are ranked in ethnic hierarchies (Haagendorn 1993). 10 It should be considered that significance depends on the number of cases in the sample. This explains why some strong differences in tension perception between status groups turn out as not significant on a 5 per cent level, like in Malta, whereas elsewhere moderate differences turn out as significant.
References Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S. and Wacziarg, R. (2003) ‘Fractionalization’, Journal of Economic Growth, 8: 155–194. Arts, W., Hermkens, P. and van Wijck, P. (1995) ‘Anomie, distributive injustice and dissatisfaction with material well-being in Eastern Europe: a comparative study’, International Journal of Comparative Sociology, 26: 1–15. Beck, U. (1986) Risikogesellschaft. Auf dem Weg in eine andere Moderne, Frankfurt a. M.: Suhrkamp. Beck, U. and Grande, E. (2004) Das kosmopolitische Europa. Gesellschaft und Politik in der Zweiten Moderne, Frankfurt a. M.: Suhrkamp Verlag. Berger-Schmitt, R. (2000) Social Cohesion as an Aspect of the Quality of Societies: Concept and Measurement, EuReporting Working Paper, Mannheim: ZUMA. Berman, Y., and Philips, D. (2004) Indicators for Social Cohesion, Amsterdam: European Foundation on Social Quality. Böhnke, P. (2004) Perceptions of Social Integration and Exclusion in an Enlarged Europe, Dublin: European Foundation for the Improvement of Living and Working Conditions. Castells, M. (2003) Jartausendwende. Teil 3 der Trilogie ‘Das Informationszeitalter’, Opladen: Leske & Budrich. Chiesi, A.M. (2002) ‘Social cohesion and related concepts’, pp. 235–253, in N. Genov (ed.), Advances in Sociological Knowledge, Paris: International Social Science Council. Coser, L.A. (1965) Theorie sozialer Konflikte, Neuwied: Luchterhand. Council of Europe (1998) Fighting Social Exclusion and Strengthening Social Cohesion in Europe: Recommadation 1355. Online. Available Http: (accessed 28.08.2001). Crouch, C.J. (2000) ‘Conflict sociology’, pp. 2554–2559, in N.J. Smelser and P.B. Baltes (eds), International Encyclopedia of the Social & Behavioral Sciences, Amsterdam: Elsevier. Dahrendorf, R. (1959) Class and Class Conflict in Industrial Society, London: Routledge & Kegan Paul. Delhey, J. (2001) Osteuropa zwischen Marx und Markt. Soziale Ungleichheit und soziales Bewufltsein nach dem Kommunismus, Hamburg: Krämer. Delhey, J. and Newton, K. (2003) ‘Who trusts? The origins of social trust in seven societies’, European Societies, 5: 93–137. European Foundation for the Improvement of Living and Working Conditions (2005) Trade Union Membership 1993–2003, European Industrial Relations Observatory online (EIRO). Online. Available Http: (accessed 3 January 2007). European Migration Centre (2005) Statistic. Online. Available Http: (accessed 3 January 2007).
350
Jan Delhey and Wolfgang Keck
Eurostat (2001) The Social Situation in the European Union 2001, Luxembourg: Eurostat/European Commission Directorate-General for Employment and Social Affairs. Eurostat (2004) The Social Situation in the European Union 2004, Luxembourg: Eurostat/European Commission Directorate-General for Employment and Social Affairs. Eurostat (2005) European and national short term indicators. Online. Available Http: (accessed 3 January 2007). Franzen, W., Haarland, H.P. and Niessen, H.-J. (2000) Transformationsbarometer Osteuropa 2000, Frankfurt/New York: Campus. Gerhards, J. (2005) Kulturelle Unterschiede in der Europäischen Union. Ein Vergleich zwischen Mitgliedsländern, Beitrittskandidaten und der Türkei, Wiesbaden: Verlag für Sozialwissenschaften. Haagendorn, L. (1993) ‘Ethnic categorisation and outgroup exclusion: cultural values and social stereotypes in the construction of ethnic hierarchies’, Ethnic and Racial Studies, 16: 26–51. Haller, M., Mach, B. and Zwicky, H. (1995) ‘Egalitarismus und Antiegalitarismus zwischen gesellschaftlichen Interessen und kulturellen Leitbildern. Ergebnisse eines internationalen Vergleichs’, pp. 222–264, in H.-P. Müller and B. Wegener (eds), Soziale Ungleichheit und Gerechtigkeit, Opladen: Leske + Budrich. Hoffmann-Novotny, H.-J. (1987) ‘Gastarbeiterwanderung und soziale Spannungen’, pp. 46–66, in H. Reimann and H. Reimann (eds.), Gastarbeiter, Opladen: Westdeutscher Verlag. Hondrich, K.-O. and Caplow, T. (1994) ‘Conflicts and conflict regulation’, pp. 225–246 in S. Langlois (ed.), Convergence or Divergence? Comparing Recent Social Trends in Industrial Societies, Frankfurt a. M./Montreal: Campus Verlag/McGill-Queens University Press. Inglehart, R. (1989) Cultural Change, Princeton, NJ: Princeton University Press. Inglehart, R. (1991) ‘Trust between nations: primordial ties, societal learning and economic development’, pp. 145–186, in K. Reif and R. Inglehart (eds), Eurobarometer: The Dynamics of European Public Opinion. Essays in Honour of Jacques-René Rabier, Houndmills: Macmillan. Inglehart, R. (2002) ‘Cultural cleavages in the European Union: modernization and cultural persistence’, pp. 73–84, in D. Fuchs, E. Roller and B. Weflels (eds), Bürger und Demokratie in Ost und West. Studien zur politischen Kultur und zum politischen Prozess, Opladen: Westdeutscher Verlag. Kelley, J. and Evans, M.D.R. (1999) ‘Public perceptions of class conflict in 21 nations’, pp. 43–71, in N. Tos, P.Ph. Mohler and B. Malnar (eds), Modern Society and Values: A Comparative Analysis Based on ISSP Project, Ljubljana, Mannheim: University of Ljubljana, ZUMA. Kessler, A.E., and Freeman, G.P. (2005) ‘Public opinion in the EU on immigration from outside the community’, Journal of Common Market Studies, 43: 825–850. Kiecolt, K.J. (1988) ‘Recent developments in attitudes and social structure’, Annual Review of Sociology, 14: 381–403. Kluegel, J.R., Mason, D.S. and Wegener, B. (eds) (1995) Social Justice and Political Change. Public Opinion in Capitalist and Post-Communist States, Berlin, New York: de Gruyter. Kunovich, R.M. (2002) ‘Social structural sources of anti-immigrant prejudice in Europe’, International Journal of Sociology, 32: 39–57. Levine, R.A. and Campbell, D.T. (1972) Ethnocentrism: Theories of Conflict, Ethnic Attitudes, and Group Behavior, New York: Wiley. McAdam, D., Tilly, C. and Tarrow, S.G. (2001) Dynamics of Contention, Cambridge: Cambridge University Press. Milanovic, B. (1998) Income, Inequality and Poverty during the Transition from Planned to Market Economy, Washington, DC: World Bank.
Social cohesion in new and old member states
351
Park, R.E., Burgess, E.W and MacKenzie, R.D. (1925) The City: Suggestions for Investigations of Human Behavior in the Urban Environment, Chicago: University of Chicago Press. Putnam, R.D. (1995) ‘Bowling Alone: America’s Declining Social Capital’, Journal of Democracy, 6: 65–78. Quillian, L. (1995) ‘Prejudice as a response to perceived group threat: population composition and anti-immigrant and racial prejudice in Europe’, American Sociological Review, 60: 586–611. Rosar, U. (2004) ‘Ethnozentrismus und Immigration’, pp. 77–101, in J.W. van Deth (ed.), Deutschland in Europa. Ergebnisse des European Social Survey 2002–2003, Wiesbaden: VS Verlag für Sozialwissenschaften. Russell, H. and Whelan, B. (2004) Low Income and Deprivation in an Enlarged Europe, Dublin: European Foundation for the Improvement of Living and Working Conditions. Schulze, G. (1992) Die Erlebnisgesellschaft. Kultursoziologie der Gegenwart, Frankfurt a. M., New York: Campus. Schumpeter, J.A. (1927) ‘Die sozialen Klassen im ethnisch homogenen Milieu’, Archiv für Sozialwissenschaft und Sozialpolitik, 57: 1–67. Sherif, M. (1966) In Common Predicament, Boston: Houghton Mifflin. Simmel, G. (1950) The Sociology of Georg Simmel, tr. and ed. Kurt Wolff, Glencoe, IL: Free Press. Stephan, W.G. and Stephan, C.W. (2000) ‘An integrated threat theory of prejudice’, pp. 23–46, in S. Oskamp (ed.) Reducing Prejudice and Discrimination, Mahwah, NJ: Erlbaum. Sztompka, P. (1999) Trust: A Sociological Theory, Cambridge: Cambridge University Press. Taylor, M. C. 1998. ‘How white attitudes vary with the racial composition of local populations: numbers count’, American Sociological Review, 63: 512–535. United Nations (2005) Human Development Report 2005, New York: United Nations Development Programme. Visser, J. (2006) ‘Union membership statistics in 24 countries’, Monthly Labor Review, 129, 1: 38–49.
Part V
Processes of Europeanisation
15 Migration and mobility culture An analysis of mobility intentions1 Hubert Krieger
Introduction The gap between what EU elites prescribe and what EU citizens actually want is nowhere more evident than in regard to migration. In EU official doctrine, migration between member states is seen as a good and necessary thing. The European Commission treaty enshrines the free movement of labour as a basic principle and policy makers believe that a genuinely integrated labour market – one where workers move readily between regions, economic sectors and occupations – would make the European economy more competitive and flexible. Even from workers’ point of view, it can be argued that those who are not faring well in their home labour markets might have better careers, higher incomes, reduced risk of unemployment and generally better and more satisfying lives if they were willing to take the opportunities offered by migration. Migration could thus be said to have quality-of-life benefits. Emigrant workers might also even eventually benefit their home economies since they would represent an external pool of labour with enhanced skills that might be available to be drawn upon at home through return migration if economic conditions improved. European citizens themselves share these positive views of migration at least to some degree: when asked what the EU represents to them, 53 per cent point to ‘freedom to travel and work in the EU’ as a major benefit. This factor is rated ahead of other features such as the introduction of the Euro as a common currency (44 per cent) and safeguarding peace (36 per cent).2 Nonetheless, official enthusiasm for migration within and between member states is countered by two aspects of the response of EU citizens: one is that Europeans do not in fact move that much, either across national or regional boundaries, the other is that many are fearful and resentful of those who do move, at least where that involves large inflows of migrants from poorer countries. The latter of these two features is the more prominent, since social and political tensions over immigration have become a central issue for public debate and electoral politics in many European countries. Anti-immigrant feeling is a widespread feature of the social and political landscape in Europe – Europeans are ‘tough on migrants’, as Boeri and Brücker (2005) put it. National governments in the EU, in contravention of the spirit of EU integration, have tended as a result to adopt restrictive policies on immigration from other EU countries (Boeri and Brücker 2005).
356
Hubert Krieger
Although the immigrant as unwelcome outsider is now a common image in Europe, it could be argued that the other aspect of this situation – Europeans’ reluctance to move themselves – is an equally important part of the overall picture and is one that has generated less interest. The data on migration within the EU are deficient in a number of ways and that is part of the reason why the subject has received limited attention – we do not know exactly how much migration there is between countries and regions in Europe (European Commission 2006: 216). Yet, there is a consensus that the level of migration among EU citizens is low. Boeri, Hanson and McCormick (2002) estimate that less than half a percentage point of the European labour force changes region of residence every year, compared to 2.5 per cent per year moving across state borders within the United States. The European Commission puts the annual level of cross-border migration among the active working age population at only one in a thousand, though this conclusion depends on considerable extrapolation from limited data (European Commission 2006: 216). It could be said that cross-border migration is not the central issue in the EU. It has a large population (492 million) and relatively few states (27), so that there could be considerable movement between regions within national boundaries (the United States, with its population of 300 million spread across 50 states is thus likely to have more inter-state migration than the EU simply because its states are more numerous and smaller in average population size). Thus, when we look at movement across subnational regional units in the EU, we get much higher estimates of migration levels. In the case of the largest such regional units, for example – the NUTS 1 regions, of which there are 89 in the EU-25 – the European Commission estimates that migration is at least ten times higher than in the case of migration between states. Even at that, however, among the EU-15 states, migration measured in this way amounts to only 1 per cent of the active age population per year. The Commission compares these to estimates of corresponding migration between states in the US of between 2.8 per cent and 3.4 per cent per year in the period 2000–2004 (European Commission 2006: 220, 223). These figures lead the European Commission to the view that Europe lacks a ‘genuine mobility culture’ (European Commission 2006: 207), and this is the issue we wish to investigate further in this chapter. Actual geographic mobility is frequently explained by reference to the balance between the perceived benefits of moving (such as greater job availability, higher pay, career development, or non-economic factors such as family reunification) and the costs and barriers that inhibit people from taking up these benefits (such as social dislocation, the financial costs of moving, linguistic differences, legal barriers). A ‘mobility culture’ is thus made up of a complex amalgam of ingredients, each of which can evolve in different ways and be influenced by different factors. In the absence of good comprehensive data for the whole EU on actual movement and the many influences that might account for it, we focus here on a central ingredient of mobility culture – people’s mobility intentions as indicated by their stated views on whether they intend to move within a fixed future period such as five years, one year, or one month. Part of the reason for this focus is pragmatic – comprehensive EU-wide survey data are available on the subject and make analysis possible, to a degree that is not possible with actual mobility. In addition, however, intentions are worth focusing on in their own right as revealing aspects of mobility culture. This is so not only to the degree that intentions translate into action but also because intentions
Migration and mobility culture
357
reveal something of people’s perceptions of the desirability of moving, even if obstacles of various kinds may inhibit them from acting on those perceptions. Furthermore, intentions themselves operate at different levels of concreteness and firmness and thus vary in their relationship to action. Krieger (2004: 9–10), for example, distinguishes between three different levels of mobility intentions – ‘general inclination’, ‘basic intention’ and ‘firm intention’ to migrate.3 The former could be thought as reflecting people’s general orientation to the idea of moving, with the latter posited as that which is most likely to lead to actual migration.4 The interplay between these different levels of intentionality is also an aspect to take into account in exploring mobility culture. Against this background, this chapter presents new results of empirical research into mobility intentions in Europe, with reference in particular to intentions over a five-year time horizon in regard to long-distance geographical mobility between regions and across countries (LDGM). The analysis uses data from two sets of Eurobarometer surveys. The more recent is the Eurobarometer survey dedicated to geographic and labour market mobility (EB 64.1), conducted in September 2005.5 Overall, 24,642 people were interviewed in all 25 member states. The average sample size per country was 1,000 respondents. A larger sample was collected in Germany with 1,500 respondents, and in the UK, with 1,300. Smaller samples of 500 respondents were collected in Cyprus, Luxembourg and Malta. The second data set is based on a combination of two Eurobarometer surveys. One is Eurobarometer EB 54.2, conducted in 2001, which provides data for the EU-15 and has a sample for that area comparable to that for EB 64.1. The other is the Candidate Countries Eurobarometer survey 2002.1, which covers what are now the 10 new member states, the two present-day accession countries Bulgaria and Romania and one of the current candidate countries (Turkey). The sampling took place at the beginning of 2002 in all 13 countries. For the purposes of this analysis, only the results for the 10 new member states are included. Similar sample sizes were collected as for EB 64.1, apart from an over-sampling in Poland, with 2,000 respondents. The main focus of the chapter is on the explanation of future intentions regarding LDGM. Is a change visible in the demographic and socio-structural patterns of people who intend to move? If so, what are the possible push and pull factors behind this change? These two general research questions give rise to a number of secondary questions. The first question can be broken down as follows. • • • • • •
Can one observe a structural change or a structural stability in the intentions to move region or country between 2001 and 2005? What is the influence of demographic factors and of country effects; does their relative importance change over time? Which socio-economic groups show the most change, and which are relatively stable? To what extent is the demographic and country effect modified by socioeconomic structures? How does past geographical and labour market mobility over the life course influence a person’s intentions towards future mobility? Are motivational factors important intervening variables, which help to explain the effects of demographic, socioeconomic and biographical factors?
358 •
Hubert Krieger To what extent can these results contribute to the development of a more robust theory of long-distance mobility?
The second question in turn, can be broken down as follows. • • • • • •
How can the increase in mobility intentions, seen at the country level for a number of European countries between 2001 and 2005, be explained? Which labour market conditions in Europe engender more ‘dynamic’ attitudes and which more stable ‘attitudes’ towards future migratory intentions? To what extent does the change in national income (GDP) influence the intention to migrate? What is the influence of subjective factors, such as life satisfaction? How important is institutional change in Europe, e.g. the enlargement of the European Union in 2004? Which is more important: a ‘stock effect’ considering the situation in 2001 or a ‘dynamic effect’ relating the change in intentions to the change of socioeconomic conditions in each country between 2001 and 2005?
1. Trends in mobility intentions, 2001–2005 Before turning to the detailed answers to these questions, it is important to note that comparing the data for 2001 and 2005, the stated willingness of European citizens to move region or country in the future is on a clear upwards trend. In autumn 2005, between 2.7 per cent and 4.0 per cent of the working age population (those aged 18 to 64 years) had a firm intention to move to another region of Europe, while between 1.7 per cent and 2.6 per cent firmly intended to move to another country.6 In absolute terms, this is a small but noteworthy increase upon the equivalent 2001 figures of between 0.7 and 1.0 percentage points; in relative terms, it represents an increase of 25 per cent for inter-regional mobility and more than 30 per cent for inter-country mobility. This will be seen as good news for European policy makers, who associate levels of future mobility with greater flexibility and better economic performance, in turn leading to higher levels of employment and income. The data on basic intentions point to higher levels of intended mobility. In autumn 2005, 8.1 per cent of the EU working age population expressed an intention to move in the next five years to another region, an increase of 2 percentage points in comparison to 2001. The unadjusted figures for cross-border migration show a similar trend: on average, 5.1 per cent of respondents in 2005 had some intention (a ‘basic’ intention) of moving to another country. Given the costs and risks of migration, it is not surprising that this figure is lower than that for intended inter-regional mobility. It does, however, represent an increase of 2.1 percentage points over the equivalent figures for 2001. Table 15.1 gives a more detailed overview of the country differences in the EU-25 in 2001 and 2005. Notable increases in intentions towards interregional mobility in 2005 are shown in France, the Nordic countries and Estonia. The least mobile populations are to be found in Hungary, Malta and Austria. The extent to which intentions towards interregional mobility varied between the member states also changed significantly between 2001 and 2005. The intentions of citizens in terms of intercountry migration display a similar degree of heterogeneity. A number of countries – Ireland, the Baltic countries (Estonia, Latvia and Lithuania) and Poland – indicated a
Migration and mobility culture
359
Table 15.1 Mobility intentions in EU member states, 2001 and 2005 (population aged 18 to 64) (%) Country
DK FI NL DE AT FR BE LU SE UK IE ES PT EL IT MT CY EE LV LT HU PL CZ SK SI EU-25 EU-15 PC-8 NMS
Intention to move to another region within same country
Intention to move to another country (in- and outside Europe)
2001
2005
Difference
2001
9.1 10.7 7.8 3.5 6.9 12.5 5.7 8.3 9.7 7.1 5.5 2.5 1.7 6.0 7.0 3.1 1.6 5.4 3.7 2.8 2.2 4.6 3.8 2.6 3.0 6.1 6.6 3.8
11.7 12.2 6.5 6.4 2.9 20.0 5.0 6.0 14.1 9.9 6.9 4.7 4.0 9.3 5.9 2.4 2.9 10.7 6.7 4.1 2.5 5.1 2.8 3.6 9.3 8.1 8.9 4.5
2.6 1.5 −1.3 2.9 −4.0 7.5 −0.7 −2.3 4.4 2.8 1.4 2.2 2.3 3.3 −1.1 −0.7 1.3 5.3 3.0 1.3 0.3 0.5 −1.0 1.0 6.3 2.0 2.3 0.7
5.5 5.4 4.6 0.8 2.7 5.1 3.5 5.6 7.0 5.1 5.9 1.0 0.2 1.0 3.3 0.4 2.9 2.7 3.2 4.9 1.2 2.5 1.7 3.3 1.5 3.0 3.1 2.3
2005 8.5 5.9 5.0 3.4 3.0 5.9 4.8 6.2 7.4 7.5 9.7 2.7 4.5 4.1 3.4 7.4 4.0 9.3 9.2 12.6 2.8 9.5 1.4 5.0 4.0 5.1 4.7 7.0
Difference 3.0 0.5 0.4 2.6 0.3 0.8 1.3 0.6 0.4 2.4 3.8 1.7 4.3 3.1 0.1 7.0 1.1 6.6 6.0 7.7 1.6 7.0 −0.3 1.7 2.5 2.1 1.6 4.7
Sources: Eurobarometer 64.1 (2005); combined Eurobarometer 54.2 (2001); Candidate Countries Eurobarometer (2002)
high level of intended mobility in 2005 (raw figures of over 9 per cent). Meanwhile, some ‘low mobility’ countries such as the Czech Republic, Hungary and Spain had an intention to migrate of below 3 per cent. Given Europe’s diversity, it would be surprising if the 25 member states displayed a unified trend in terms of geographical mobility; indeed, the change in intentions towards geographical mobility varies between the member states. In terms of unadjusted figures for interregional mobility, France shows an increase of 7.7 percentage points, while Austria shows a decrease of 4.0 percentage points. Between 2001 and 2005, 18 countries displayed an increase in mobility intentions, while seven displayed a decrease (Austria, Belgium, Czech Republic, Italy, Luxembourg, Malta and the Netherlands). The dynamics in terms of unadjusted figures for future inter-country migration are nearly as diverse, ranging from an increase of 7.7 percentage points in Lithuania to a decrease of 0.3 percentage points in the Czech Republic. The six largest increases over five years are in five new member states and in Portugal. The enlargement of the EU
360
Hubert Krieger
in May 2004 seems to have had a significant effect in raising the intention to migrate in half of the new member states.
2. Structural patterns of long-distance mobility intentions The focus in this section is on the factors affecting mobility intentions in 2001 and 2005, and on the question of how these factors compare between the two years. Are there structural patterns explaining LDGM in Europe that are stable over time, or are there significant changes? One basic hypothesis is that four years is too short a period to provoke a significant structural change in mobility intentions. According to previous research, reasons for LDGM intentions show stable patterns (Krieger 2004: 33; Fassmann and Hintermann 1997; IOM 1999). LDGM is influenced by a specific set of demographic and socioeconomic variables: men, and younger people are more likely to move, as are better qualified individuals, students, and tenants (as opposed to owner-occupiers). It is also strongly related to previous experiences of mobility during the life course. It might be expected that such a pattern would be unlikely to change over a period of four years. The counter-hypothesis is that, in the period of time under scrutiny, important changes in institutional settings and EU policies have taken place – the enlargement of the EU, and the implementation of the Lisbon strategy emphasising the need for greater mobility. Taken together, these changes may have had an uneven effect on different groups in the labour market, leading to a changing structural pattern in future mobility intentions. Certain groups, already more open to the possibility of mobility, may have become even more willing to move, because of the better opportunity structure provided by institutional and policy change in the EU. The multivariate analysis at the micro level will use five different groups of explanatory factors: demographic variables; country effects; socio-economic variables; past geographical and labour market mobility as part of the individual biography; and motivation for LDGM. Each estimated model will have a specific set-up of explanatory variables. This is related to the availability and usefulness of specific information: for example, the datasets provide information on motives related only to inter-country migration, not to interregional mobility. In addition, there are important differences in the available data sets for 2001 and 2005. As a consequence, the main question of whether we are witnessing structural stability or structural difference in the factors associated with an intention to move can only be answered within these limits. The following micro-level analysis will use a logistic regression model to explain the extent of LDGM intentions. The coefficients in such a model have positive or negative signs indicating the direction of the relationship. Each coefficient has to be interpreted in relation to its reference group. The higher the value of the coefficient, the greater is the difference (although the interpretation is sometimes more cumbersome). This means, for example for model 1, step 1 (in regard to the effect of gender), that men have a greater intention to migrate than the reference group (women).7 In addition, the relationship between dependent and independent variables is tested within the given model, keeping all other variables constant. The analysis is presented in four models – two models each for both 2001 and 2005, the first dealing with inter-country migration intentions in each year and the
Migration and mobility culture
361
second with inter-regional migration intentions. The dependent variable is basic intention to migrate. Each model proceeds through four steps. The first step entails a basic model, using only the demographic factors and the country dummies; a second step then includes the socio-structural variables; a third step includes information on past mobility; the fourth and final step includes motivations for and against mobility. The reason for this step-by-step approach is to identify direct and indirect effects of the different sets of variables on mobility intentions. A second reason is to reveal partial effects of given independent variables controlling for the impact of other effects. 2.1 Intentions regarding cross-border migration in 2005 The attempt to identify why respondents express openness towards future migration will start with an analysis of inter-country migration in 2005 (model 1 in Table 15.2). The first step in model 1 includes only demographic variables and country dummies. The results are fairly consistent with previous research: education has an important positive association with the intention to migrate, in particular for students, but also for the highly educated. It is worth noting, however, that those with low education do not differ significantly from the reference category (those with medium education).8 Different age groups show also the expected results: the youngest age group (18–24 years) shows the highest positive marginal propensity to migrate in comparison to the reference group (those aged 35–44); the 25–34 years age group shows a slightly lower positive factor; and respondents over 54 years of age show a high negative marginal propensity. Looking at the country effects, the results of the descriptive analysis in section 1 are confirmed by the multivariate analysis. With a propensity for future migration almost equal to the average, the Netherlands can be taken as a reference country. Compared to respondents in the Netherlands, those in several of the new member states, such as the Czech Republic, Hungary and Slovenia display a statistically significant negative propensity to migrate in the future, while the three Baltic countries, Poland and Malta show a significant positive propensity. Finally, there is a positive and significant gender effect in the data. Men are more open to the possibility of future migration than women. Looking at the overall explanatory power of step one in model 1, we can see a relatively low pseudo r2 of 0.155. In the second step of model 1, socio-structural variables are included. Being unemployed has a positive (push) effect on intention to migrate – an important political result. Unemployed people have a significantly greater willingness to migrate than the reference group (the retired), whereas employed people do not. As far as household structure is concerned, couples with children are less open to migration, whereas single people – independent of age and gender – are more interested in future migration. In addition, the expected negative association with home ownership is confirmed. Finally, in general, non-nationals consider the option of migration more than do country nationals. This can be interpreted either as an interest in returning to their country of origin, or as an indication of a developed history of migratory practice, which continues as a higher propensity to migrate to other countries. What has changed between the first and second steps of model 1? Both the marginal gender and age effects have slightly reduced. This means that part of these effects can now be explained through the additional four socio-structural variables. On the other hand, the effects due to education have increased. This can be explained
Country dummies (ref: NL)
Age (ref: 35–44)
Gender = male Education (ref: average)
Low education High education Still studying 18–24 25–34 45–54 55–64 BE DK DE EL ES FI FR IE IT LU AT PT SE UK CY CZ EE
0.342* −0.282 0.532* 1.086* 1.121* 0.847* −0.252 −1.463* −0.158 0.203 −0.498* −0.498 −0.850* −0.076 0.020 0.624* −0.644* 0.091 −0.437 −0.122 0.277 0.471* −0.213 −1.247* 0.615*
0.309* −0.289 0.547* 1.250* 0.725* 0.672* −0.429* −1.425* −0.295 0.164 −0.730* −0.719* −0.905* −0.089 −0.132 0.429 −0.716* −0.308 −0.599* −0.232 0.305 0.317 −0.339 −1.228* 0.644*
Raw coefficient 0.397* −0.217 0.438* 1.472* 0.845* 0.652* −0.466* −1.595* −0.237 −0.202 −0.682* −0.622* −0.750* −0.374 −0.247 −0.074 −0.412 −0.186 −0.535 0.176 −0.038 0.276 −0.461 −0.845 0.782*
0.378* −0.132 0.370* 1.184* 0.757* 0.623* −0.292* −1.083* −0.257 −0.271 −0.853* −0.736* −0.740 −0.396 −0.478 0.095 −0.558 −0.197 −0.686* −0.006 −0.248 0.286 −0.517 −0.986* 0.480
0.079 −0.271 0.258* 0.906* 1.108* 0.583* −0.134 −1.461* −0.453 0.397 0.044 0.274 −0.419 0.538* 1.226* −0.129 −0.416 −0.147 −0.860* −0.607* 0.700* 0.288 −0.639 −1.248* 0.361
0.080 −0.286* 0.312* 1.444* 0.789* 0.446* −0.179 −1.088* −0.395 0.408 −0.106 0.316 −0.355 0.549* 1.160* −0.027 −0.313 0.013 −0.985* −0.593* 0.735* 0.333 −0.564 −1.137* 0.473*
Raw coefficient
Step 2
0.154 −0.234 0.115 1.432* 0.955* 0.473* −0.225 −1.173* −0.262 0.150 −0.281 0.458 −0.037 0.237 1.020* −0.062 −0.061 0.371 −1.018* −0.236 0.356 0.379 −0.930 −0.787* 0.554*
Step 3
Step 1
Step 4
Step 1
Step 3
Estimates from logit model for intended mobility across regions whithin same country total population aged 18–64
Estimates from logit model for intended mobility to another country total population aged 18–64 Step 2
Model 3 (2005)
Model 1 (2005)
Table 15.2 Mobility intentions (regression models for 2005)
Number of relocations since first moved out of parental houshold (ref: once) Reasons which might encourage respondend to migrate
Homeowner Non-nationals Last move (ref: within region)
Household type (ref: couple no child)
Employment status (ref: retired)
Better family/friends support To be closer to family/friends Higher hh income Better housing conditions Better local environment Better health care facilities Better working conditions Better schooling system Shorter commuting time
Never moved Last move – across regions Last move – across borders Never 2−4 times 5−9 times 10 times or more
HU LV LT MT PL SK SI Working Unemployed Housewife/man Couple, with child Single Single parent Divorced/separated Widowed −0.312* 0.013*
−0.610* 0.667* 0.923* 0.601* 0.494* −0.148 −0.795*
−0.567 0.625* 0.945* 0.630* 0.468* −0.108 −0.751* 0.198 0.651* 0.511 −0.569* 0.397* −0.126 0.159 −0.071 −0.245* 0.007*
−0.302 0.652* 1.106* 1.141* 0.822* 0.175 −0.226 0.139 0.576* 0.502 −0.562* 0.530* −0.593 0.110 −0.078 −0.201* 0.007* 0.917 0.288* 1.341* −0.997 0.249 0.567* 0.842* 0.715* 0.572* 0.874* 0.621* 0.694* 0.442* 0.734* 1.180* 0.265
0.792 0.199 1.190* −0.956 0.200 0.488* 0.702*
−0.572 0.505 0.855* 1.008* 0.540 −0.228 −0.643 0.037 0.452 0.489 −0.535* 0.460* −0.576 0.023 0.011
−1.145* −0.099 −0.856* −1.330* −0.541* −0.725* 0.087
−0.979* −0.029 −0.775* −1.135* −0.457 −0.557* 0.200 0.526* 0.797* 0.485 −0.253* 0.135 0.214 0.288* 0.092 −0.566* −0.002
Continued
−0.818* −0.195 −0.526 −0.714 −0.302 −0.206 0.426 0.576* 0.697* 0.551* −0.297* 0.182 −0.371 0.174 0.106 −0.664* −0.003 −0.055 0.874* 0.574* −0.232 0.121 0.472* 0.746*
Note: *p < 0.05.
Constant Pseudo r-sqr Number of observations
Reasons which might discourage respondend to migrate
Better public transport Meet new people Discover new environment Better weather Miss family/friends support Miss direct contact with family/friends Lose own or partner’s job Lower income Worse housing conditions Worse local environment Worse health care facilities Worse working conditions Different school system Longer commuting time/worse pub. trans. Having to learn a new language −3.683* 0.155 18381
−3.488* 0.180 18225
Raw coefficient
−4.006* 0.209 15571
−4.305* 0.262 15571
−0.663*
−0.438* −0.355* −0.440* −0.170 −0.346* −0.058 −0.533* −0.407
0.239 0.535* 1.024* 0.453* −0.584* −0.599*
−3.063* 0.151 18381
−3.180* 0.164 18225
Raw coefficient
Step 2
−3.537* 0.187 15571
Step 3
Step 1
Step 4
Step 1
Step 3
Estimates from logit model for intended mobility across regions within same country total population aged 18–64
Estimates from logit model for intended mobility to another country total population aged 18–64 Step 2
Model 3 (2005)
Model 1 (2005)
Table 15.2 Mobility intentions (regression models for 2005)—cont’d
Migration and mobility culture
365
by the significant negative effect of home ownership on intentions to migrate. As a higher proportion of better educated respondents are home owners, controlling for homeownership brings the independent positive effect of better education closer to the fore. Finally, looking at the overall explanatory power of the sub-model in the second step, it has increased to a pseudo r2 of 0.18. In the third step of model 1, the individual history of geographical mobility is included. It is assumed that the distance travelled in previous geographical mobility and its intensity (i.e. the number of moves made in the past) has a positive association with the future migration intentions. Both hypotheses are confirmed by the data: respondents who had previously moved to another country are more open to moving in the future. Having moved between regions within the same country also has a significant, albeit lesser, effect. As far as the intensity of past regional mobility is concerned, only a high intensity (i.e. more than five moves) has a significant effect, whereas – in a threshold effect – lesser intensity has no significant effect upon future migration intentions. Do we observe any important changes when comparing the second and third steps of model 1? First of all, the direct influence of being male and of being a student increases in comparison to the previous model, with the introduction of the data on the personal history of mobility. On the other hand, the influence of younger groups – relative to their reference group – diminishes further. This means that age does affect the intention to migrate, because people of different age groups have a specific mobility biography: for example, younger people have had less opportunity to establish a high intensity of past mobility than have older age groups. In addition, the influence of three of the four socio-economic variables (with the exception of a single-person household structure) is slightly reduced through the introduction of variables covering the history of individual mobility. This means that important sections of the group of unemployed, home owners and non-nationals have a biography of more long-distance and/or more intensive geographical mobility during their life course, which has a positive effect on the future intention to migrate. Finally, the overall explanatory power of the model increases to 21 per cent. In the fourth step of model 1, two batteries of motivational factors are introduced into the analysis. The assumption is that positive motivations are the main drivers of future migration, and negative motivations are important obstacles. The interesting question in this context is to what extent motivational factors are independent of underlying demographic, socio-structural and life-course effects or, alternatively, if there are independent structural effects from a motivational set-up to explain future migration? Looking at the results, one should first of all note that with the further additions, the overall explanatory power of the model reaches 26 per cent. This is an increase of 10 percentage points over the starting model in the first step. Second, most of the important structural factors remain statistically significant. An exception is employment status and in particular unemployment, which no longer has a significant influence on future intentions to migrate. Not surprisingly, other structural factors lose part of their explanatory power in comparison to the sub-models in the third step. The two indicators of past mobility biography are also reduced in importance. It is reasonable to assume that past experience is combined with a realistic perception of possible positive and negative effects, which influences in turn a motivational set-up to future migration.
366
Hubert Krieger
Taking the positive motives to migrate first, only the motivations of shorter commuting times and better public transport are not significant in explaining intentions for future migration. The most significant is the wish to discover a new environment and the desire to move to a country with a better school system. Economic motivation plays an important, but smaller, role. Social factors, in terms of maintaining established relationships or making new friends are of third-level importance. The figures confirm that there is a set of relevant positive motivations independent of demographic, socioeconomic and general country effects. As far as the disincentives are concerned, the strongest independent negative association is with the need to learn a new language; this confirms widespread public beliefs regarding barriers to migration. Of second-level importance is the worry of losing social contacts and support from family and friends. And of third-level importance is the need to adjust to a different school system in the receiving country. When the strength of independent effects of motivational factors and disincentives are compared with each other, the data identify a higher association with positive than with negative motivations for future migration. 2.2 Intentions for future cross-border migration: 2001 and 2005 compared The key question here is whether there are consistent structural patterns explaining LDGM or, instead, if there are significant changes over a four-year period. In order to answer this, the results of the multivariate analysis for 2005 are compared with the results for 2001, following a similar step-by-step approach as with the 2005 data. In the first step of model 2 (Table 15.3) the only difference for the demographic variables in comparison to 2005 is that effects for the less educated and for the older age groups are not significant. That means that in 2005, the results for ‘age’ and ‘education’ were more symmetrical, whereas the results in 2001 show only significant effects at one extreme. Not surprisingly, considering the descriptive results in section 1, over the period 2001–2005 more change can be observed at the country level. Four smaller countries – Poland, Malta, Latvia and Estonia – changed from being less likely to more likely to migrate than the reference country (the Netherlands). One country (Finland) changed in the opposite direction, though not to a statistically significant degree. This could be a first indication that institutional change or the changing macro-economic performance may have increased the propensity to migrate in certain countries over the four-year period. In addition, the number of significant effects on the country level has increased from eight countries in 2001 to 13 countries in 2005, which means that country-specific factors have gained in importance over our reference period. In the second step of model 2, one socioeconomic variable – employment status – is included (the three variables ‘household type’, ‘home ownership’ and ‘nationality’ included in model 1 were not available in 2001 and were left out). The only difference shown by the results is that in 2001 the effect of unemployment on the future intention to migrate was stronger than in 2005. This could reflect a real difference in the effect of unemployment or could be due to the omission of three other socioeconomic variables. The third step of model 2 includes the two variables on past mobility – the distance involved in the last move and the number of moves. The results confirm the finding
Country dummies (ref: NL)
Age (ref: 35−44)
Gender = male Education (ref: average)
Low education High education Still studying 18–24 25–34 45–54 55–64 BE DK DE EL ES FI FR IE IT LU AT
0.416* 0.011 0.833* 1.091* 1.553* 1.138* 0.076 −0.131 −0.107 0.021 −1.518* −1.500* −1.481* 0.004 0.098 0.199 −0.168 0.486 −0.523
0.368* 0.048 0.847* 1.606* 1.462* 1.098* 0.077 0.038 −0.201 −0.045 −1.631* −1.507* −1.553* −0.097 0.083 0.193 −0.233 0.480 −0.532
Raw coefficient 0.508* 0.133 0.702* 1.843* 1.175* 0.650* 0.623* 0.362 0.089 −0.197 −1.439* −1.357* −1.893* −0.372 0.006 0.060 0.049 0.356 −0.138 −0.436 0.084 0.408 −0.556 0.423 −0.228
0.472* 0.179 0.678* 1.919* 0.999* 0.561* 0.417 0.249 −0.043 −0.239 −1.127* −1.653*
0.031 −0.153 0.382* 0.832* 1.322* 0.727* −0.208 −0.536* −0.254 0.007 −0.587* −0.424 −1.258* 0.243 0.440* −0.343 −0.096 0.167 −0.134
0.006 −0.144 0.393* 0.802* 1.263* 0.702* −0.230 −0.580* −0.323 −0.039 −0.659* −0.421 −1.298* 0.171 0.440* −0.340 −0.136 0.164 −0.138
Raw coefficient
Step 2
0.001 −0.125 0.291* 0.873* 0.932* 0.439* 0.223 −0.279 −0.171 −0.135 −0.331 −0.098 −0.701 0.041 0.542* −0.685 0.205 −0.008 0.016
Step 3
Step 1
Step 4
Step 1
Step 3
Estimates from logit model for intended mobility across regions within same country
Estimates from logit model for intended mobility to another country total population aged 18–64 Step 2
Model 4 (2001)
Model 2 (2001)
Table 15.3 Mobility intentions (regression models for 2001)
Continued
0.027 −0.171 0.244 0.777* 0.910* 0.530* 0.196 −0.196 −0.036 −0.321 −0.322 0.009 −1.267* 0.087 0.613* −0.907 −0.092 0.234 −0.096
Step 4
Employment status (ref: retired)
PT SE UK CY CZ EE HU LV LT MT PL SK SI Working Unemployed Housewife/man
−0.528 0.312 −0.576 0.505 1.066* −0.422
−0.586 0.436 −0.566 0.552 1.215* −0.202
−1.365* 0.078 −0.167 −1.507* −0.942* −0.512* −0.996* −0.796* −1.130* −0.533 −0.758* −0.982* −1.214* −0.126 0.414* −0.383
−1.337* 0.120 −0.154 −1.506* −0.858* −0.467* −0.919* −0.737* −1.055* −0.600 −0.669* −0.903* −1.153*
−1.385 −0.294 0.317 −0.695 0.293 −0.467 −1.013 −0.254 0.488
−1.896 −0.130 0.370 −0.750 0.020 −0.460 −0.892 −0.123 0.556
−2.442* 0.342 0.332 −0.800 −0.626 −0.459 −1.024* −0.176 0.166 −1.650* −0.699* −0.268 −1.132*
−2.494* 0.242 0.308 −0.848 −0.767* −0.529 −1.118* −0.310 0.028 −1.536* −0.810* −0.376 −1.240* 0.419 1.136* −0.251
Raw coefficient
Raw coefficient
Step 2
−0.737 −0.091 −0.188 −1.280 −0.829 −1.044* −1.196* −0.316 −1.021* −0.933 −0.316 −0.583 −0.862* −0.024 0.382 0.057
Step 3
Step 1
Step 4
Step 1
Step 3
Estimates from logit model for intended mobility across regions within same country
Estimates from logit model for intended mobility to another country total population aged 18–64 Step 2
Model 4 (2001)
Model 2 (2001)
Table 15.3 Mobility intentions (regression models for 2001) – cont’d
−0.504 0.001 −0.290 −1.275 −1.123 −1.137* −1.214* −0.265 −0.889 −0.819 −0.105 −1.587* −0.616 −0.103 0.313 −0.093
Step 4
Note: *p < 0.05.
Never moved Last move – across regions Last move – across borders Number of relocations Never since first moved 2–4 times out of parental 5–9 times household 10 times or more Net Household Lowest income quartile equvialent Third income quartile income Highest income quartile Constant Pseudo r−sqr Number of observations
Last move (ref: within region)
−4.557* 0.133 18038
−4.909* 0.141 18026
1.933* 0.064 0.454* 0.789* 1.878* 0.362 0.531* 0.424 −5.738* 0.215 5140
2.012* 0.718 0.377* 0.775* 1.481*
−5.573* 0.214 6864
−0.095 0.534*
−0.468 0.667*
−3.151* 0.115 18038
−3.007* 0.119 18026
−3.543* 0.143 6941
1.180* 0.380* 0.930* 0.532
0.775*
−1.276 1.051*
1.155* 0.327* 0.868* 0.530 −0.040 0.106 −0.089 −3.406* 0.144 5335
0.776*
−0.968 0.993*
370
Hubert Krieger
for 2005 that past movement increases the tendency towards future movement. The country dummies show that in 2001, only three countries were significantly lower than the reference country (the Netherlands) in their average intention to move: Germany, Spain and Greece. None of the other country dummies were significant. In 2005 this changed radically in the new member states: the three Baltic countries, Poland and Malta have a significant positive effect on the future intention to migrate, whereas the significant negative association in the three ‘old’ member states remains, albeit on a lower level. In the fourth step of model 2, an additional socio-structural variable that was not available in 2005 – income – is added but shows no significant association with willingness to migrate. As is known from previous research, income has no effect on willingness to migrate in the lowest- and highest- income quartiles (Krieger 2004: 47). Those empirical results are in contrast with theoretical considerations, which assume that a certain level of income is needed to move in the first place, in order to manage the information and search costs which go hand-in-hand with mobility. To conclude: demographic, socioeconomic and past mobility variables show very little differences in their effects on of the intention to migrate in 2001 and 2005. In this respect, the two models in our analysis show a fairly consistent structural pattern. The real differences between the two years are the significant increases in the willingness to migrate in five of the new member states – the three Baltic states, Poland and Malta While this increase might be explained by the enlargement of the EU in May 2004, no similar increase is found in another group of new member states (mainly Hungary, Slovenia, Slovakia, the Czech Republic and Cyprus), where negative tendencies relative to the reference country continued over the period 2001–2005. The reasons for this divergence in attitudes to migration among the new member states will be explored further in section 3 of this chapter. As far as the ‘old’ member states are concerned, respondents in Germany, Spain and Greece have a consistent and statistically significant negative attitude to migration in comparison to the reference country (Netherlands). In policy terms, these results point towards a relatively stable medium-term pattern of structural factors influencing the future willingness to migrate; in turn, these support political concerns related to a possible youth- and brain-drain in half of the new member states. This structural pattern is exacerbated by the quantitative increase in the intention to migrate in those countries between 2001 and 2005. For the receiving countries, the profile of possible immigrants from other member states is good news, as they can look forward to the arrival of young, qualified and highly motivated persons. The migrants envisage a new, exciting and welcoming social environment, as well as favourable economic conditions; other items high on the new arrivals’ agenda are adequate and accessible public services, including schooling. We would not interpret this as an indication of ‘welfare tourism’, as those motives are only of secondary importance and are combined with a strong willingness to perform economically in the receiving countries. 2.3 Intentions for interregional mobility in 2005 Model 3, presented above in Table 15.2, estimates the association of demographic, socio-structural and biographical factors with the intention for interregional mobility in 2005. It uses the same technique and some of the same variables as model 1
Migration and mobility culture
371
(the variables on motivation and disincentives for mobility, which are not included in the dataset for 2005, are omitted). In the first step of model 3, no gender effect is found, which contrasts with the pronounced gender effect found earlier for inter-country migration intentions (with men more likely to intend to migrate than women). Whereas the intention to migrate between countries follows more the traditional behavioural difference between men and women, the probability of future interregional mobility indicates an absence of the gender difference; from a developmental perspective, this can be seen as a further indication of a growing gender balance of LDGM. Education and age show similar effects as in inter-country migration intentions, with higher education and youth having positive effects. The country dummies identify France, Finland and Sweden as having a more positive orientation towards interregional mobility than the reference country, while Hungary, the Czech Republic and Malta are more negative. Otherwise, the structure of the model and its overall explanatory power of 15 per cent are relatively similar to the first step of model 1 for inter-country migration, apart from the lack of any gender effect. In the second step of model 3, both being employed and being unemployed significantly enhance interregional mobility intentions relative to the reference group (the retired). Couples with children are less likely to intend to move than couples without children, as are home owners compared to those in other housing tenures. Finally, both nationals and non-nationals display the same willingness to move between regions, which contrasts with the greater willingness of non-nationals to move between countries. In the next and final step of model 3, two indicators of past mobility are included, both of which show positive effects. The inclusion of these two variables causes the impact of higher levels of education on intentions to move to another region found in previous steps to become insignificant. This means that better education has an effect on the intention to move between regions only when it is bound up with experience of previous mobility. It is also noteworthy that unemployed people, regardless of their past experience of mobility or of other demographic and socio-structural factors, have stronger intentions towards interregional mobility than the reference group. This is a difference compared to model 1, in which employment status has no effect on the intention for cross-border migration. The higher willingness of the unemployed to move at regional level may be taken into consideration in questioning the attempt to blame the unemployed themselves for their predicament. 2.4 Intentions for interregional mobility, 2001 and 2005 compared Model 4, presented earlier in Table 15.3, estimates the factors influencing intended mobility between regions for 2001. In the first step of this model, as in that for 2005, gender has no effect, while education and age have the same effects as in 2005. The country effects for 2001 show 13 countries with a statistically significant effect, whereas the 2005 model contains 11 countries. Overall, 7 countries show a statistically significant effect for both 2001 and 2005: 6 countries have a consistently negative attitude towards interregional mobility (Czech Republic, Hungary, Lithuania, Poland, Portugal and Slovenia), and 1 country (France) has a consistently positive attitude relative to the reference country. In addition, in both years, 8 countries have
372
Hubert Krieger
no statistically significant relationship with intentions towards interregional mobility. The remaining 10 countries show some changes over the 4-year period, having a combination of significant and insignificant relationships in 2001 or 2005. Five of these countries are ‘old’ member states, and 5 are new member states. In the second step of model 4, as in the model for 2005, being unemployed is positively related with intentions to move. But in contrast to 2005, being employed has no effect. The third step of model 4 for 2001 confirms, in general, the results from 2005: previous mobility has a positive influence on intentions for future mobility. The only difference between the two years is that the highest intensity of mobility has a significant influence in 2005, and not in 2001. The most important difference between 2001 and 2005 is related to the effect of the inclusion of past mobility on the demographic and socioeconomic variables. In 2005, the unemployed remain statistically more likely to migrate, despite the inclusion of the personal mobility biography; whereas this influence is lost in 2001. In addition, older age groups no longer show any significant negative effects on the intention for inter-regional mobility. As far as the country effects are concerned, the introduction of the socio-structural and past mobility variables has an important effect, as it reduces the number of countries with a significant relationship from 13 to 5. The latter 5 countries consist of one old member state (France) and 4 new member states in 2001. In 2005, we also find 5 member states with significant effects – 2 ‘old’ member states (France, Austria) and 3 new member states. Three of these countries are overlapping in both years (France, Hungary and Estonia). In both 2001 and 2005, and for both dimensions of mobility intentions, the statistical analysis provides fairly stable models. There are only small differences in the effects of demographic, socioeconomic and biographical variables on intentions towards interregional and inter-country migration, arising mainly in connection with the gender and unemployment effects. There are also some differences in relation to the country effects. Whereas positive attitudes to future migration changed significantly in five new member states in 2005 relative to 2001, more positive attitudes to interregional mobility have developed only in Slovenia over these four years. In contrast, in the Czech Republic and Lithuania this even turned, in 2005, into a significant negative effect.
3. ‘Push’ factors and change in mobility intentions between 2001 and 2005 Earlier in this chapter, the increase in the two dimensions of mobility intentions in Europe between 2001 and 2005 was outlined. Intentions towards inter-country migration and interregional mobility both showed an increase of around 2 percentage points over the 4-year period. At face value, these results seem to indicate a growing openness on the part of European citizens towards mobility. However, a closer look at the data shows that the increase does not occur evenly across the 25 member states. In the case of the basic intention towards interregional mobility, the range is between an increase in France of 7.5 percentage points and a decrease of 4.0 percentage points in Austria (a range of 11.5 percentage points). In the case of the basic intention towards inter-country migration, the corresponding range is 8.0 percentage points, running
Migration and mobility culture
373
from an increase of 7.7 percentage points for Lithuania to a decrease of 0.3 percentage points for the Czech Republic. What is the explanation for these differences? As no individual panel data are available, an analysis of changes on the individual level is not possible. Hence the analysis turns to the macro-level, explaining changes in mobility intentions using the member states as units of analysis. Three different sets of hypotheses are tested in this context. A first hypothesis is that changes in attitudes towards LDGM are influenced by general economic and labour market conditions in the sending and receiving areas. In order to test this hypothesis, the following indicators will be used: GDP per capita (in purchasing power parity); overall employment rate; overall unemployment rate; and the long-term unemployment rate. In order to cover both dimensions of LDGM intentions (interregional mobility and inter-country migration), different information has to be used: for interregional mobility, indicators have to be tested on the regional level, e.g. differences in regional employment and GDP rates; for migration, nationallevel information will be used. In order to limit the scope of the analysis, this section concentrates exclusively on change in intended inter-country migration. A second set of hypotheses relates mobility intentions to aggregated micro-level indicators on subjective income conditions (‘making ends meet’) and to an indicator of general life satisfaction. Accordingly, mobility is influenced not only by economic conditions, but also by a larger ensemble of objective and subjective living conditions, which manifest themselves in a certain level of general life satisfaction. A third hypothesis considers the changing institutional set-up between 2001 and 2005, particularly the enlargement of the EU with the accession of 10 new member states. It is assumed that attitudes towards future migration have changed more profoundly in these 10 new member states than in the EU-15. Results in the previous sections have already indicated that the institutional change had different effects among the new member states. In half of the new member states, it has significantly increased intentions towards migration whereas, in the other half, intentions have remained stable or have only marginally increased. These results suggest a more sophisticated hypothesis combining institutional change with specific socio-economic conditions as an explanation. Comparing the results for the EU-15 with the results for the EU-25 will test the effect of EU enlargement. 3.1 Macro-economic ‘push’ factors of change in the sending countries The first part of this section looks at the association between existing macroeconomic conditions in the member states for the year 2001 and the change in attitude towards migration (the figures for 2005 minus those for 2001). The basic rationale is that macro-economic conditions at the start of the observation period may influence the extent to which willingness to migrate changed over the four-year period. In regard to the four indicators, the following hypothesis will be tested (based on established micro- and macro-economic thinking). •
The lower the stock in GDP and the lower the total employment rate in 2001 in a member state, the greater is the push for migration and the higher is the change in aggregate intended migration in this country.
374 •
Hubert Krieger The higher the overall unemployment and long-term unemployment rate in a member state in 2001, the higher is the push for migration and the higher the willingness to increase future outward migration in this country.
Considering the empirical results shown in Figures 15.1 to 15.4, all hypotheses are confirmed in relation to the association between existing macroeconomic conditions in 2001, and the change in intention to migrate between 2001 and 2005. Looking at the correlation coefficients, the strongest association is with GDP per capita (a value for r2 of 0.325); the weakest association is with the overall employment rate (a value for r2 of 0.169). Despite this statistical association, there are significant differences between countries. For example, as far as national income is concerned, Lithuania, Latvia and Slovakia had a similar level of GDP per capita in 2001. However, Lithuania shows an increase of nearly 8 percentage points in the intention to migrate, whereas Slovakia shows an increase of just under 2 percentage points. Similar differences can be observed between member states in the EU-15: in 2001, Ireland and the Netherlands had a similar level of GDP per capita; in Ireland, the willingness to migrate increased by nearly 4 percentage points, whereas in the Netherlands, it remained more or less stable. Considering the existing variations between the countries, it is fair to say that the relationship between macroeconomic income and employment figures at the start of the observation period and the observed changes in the intention to migrate between 2001 and 2005 confirm established economic thinking. The worse the national income and overall employment conditions, and the higher the national unemployment level in the sending country in 2001, the greater is the increase in willingness to migrate. The second part of the analysis in this section looks at the dynamic effects on the intention to migrate of changing macroeconomic conditions. The hypotheses can be stated as follows. •
•
The higher the increase in GDP and in the overall employment rate between 2001 and 2005 in a member state, the smaller is the increase in intended migration over the same period. The higher the increase in the overall unemployment rate and level of long-term unemployment between 2001 and 2005 in a member state, the greater is the increase in the willingness to migrate.
All four indicators displayed in Figures 15.5 to 15.8 show a consistent result, but one that contradicts the expectations of the hypothesis. According to these results, a greater increase in national income and employment, and a greater decrease in overall and long-term unemployment, in a member state between 2001 and 2005 are associated with a push for more migration resulting in a higher increase in the intention to migrate over the four-year period. With regard to the new member states, these results show that, despite a significant positive dynamic in those member states in relation to improved national income and employment conditions over the four-year period, low levels of national income and high levels of unemployment still dominate the dynamics of future intentions to migrate. In other words, the dynamic effects in the new member states are not yet
Migration and mobility culture
375
8
Change in intention to migrate 2001–2005
LT PL EE
MT
LV
6
PT
4
IE EL
DK DE UK
SI
2
SK
ES
HU
BE
CY
FR FI SE AT NL IT
0
LU
CZ
−2 100 150 200 Stock GDP (per capita, EU-25 = 100), 2001
50 R2 = .325
250
8 LT
Change in intention to migrate 2001–2005
PL
MT EE
6
LV
PT
4
IE EL
DK DE
SI
2
UK
SK ES HU
BE
CY
FR LU
0
FI AT
IT
NL SE
CZ
55
60
70 65 Stock total employment rate, 2001
75
R2 = .169
Figures 15.1–15.4 Stock of macro-indicators and change in intended migration 2001–2005, for EU-25
376
Hubert Krieger
8 LT PL
Change in intention to migrate 2001–2005
MT EE
6
LV
PT
4
IE EL
DK DE
SI
UK
2
SK
ES
HU BE
CY
FR
LU AT
NL
FI IT
SE
0 CZ
0
5
10
15
20
Stock total unemployment rate, 2001
R2 = .264 8
Change in intention to migrate 2001–2005
LT PL
MT EE LV
6
PT
4
IE EL
DK DE SI
UK
2 BE
CY LU NL AT SE
0
SK
ES
HU FR FI
IT CZ
0 R2 =
.26
5 Stock long-term unemployment (% active pop.), 2001
Figures 15.1–15.4, cont’d.
10
Migration and mobility culture
377
8
Change in intention to migrate 2001–2005
LT MT
PL EE
6
LV
4
PT IE EL
DK DE
2
SI
UK ES
SK HU
BE CY FR IT
0
LU
FI AT
NL
SE CZ
−10 R2 = .022
0
10
20
30
Change in GDP (per capita, EU-25 = 100) 2001–2005
8 LT
Change in intention to migrate 2001–2005
PL
MT EE LV
6
PT
4
IE EL
DK DE
SI
UK
2
SK HU CY FR SE
NL
AT
FI
ES BE
LU IT
0 CZ
−2 R2 = .107
0
2
4
Change in total employment rate 2001–2005
Figures 15.5–15.8 Change in macro-indicators and change in intended migration 2001–2005, for EU-25
6
378
Hubert Krieger
8 LT PL
Change in intention to migrate 2001–2005
MT EE LV
6
PT
4
IE EL
DK UK
2
SK
DE
SI
ES
HU BE CY FR FI
AT
IT
0
LU NL
SE
CZ
−10 R2 = .338
0 −5 Change in total unemployment rate 2001–2005
5
8
Change in intention to migrate 2001–2005
LT PL
MT EE LV
6
PT
4
IE EL SI
DK DE
UK
2
SK HU
ES
BE
CY
IT
0
LU AT
FI SE
FR NL
CZ
−6 R2 = .183
−4
−2
0
Change in long-term unemployment (% active pop.) 2001–2005
Figures 15.5–15.8, cont’d.
2
Migration and mobility culture
379
strong enough to dilute the trend of greater interest in migration, which is due to a significant gap relative to the richer member states. Can a turning point be identified, when a sufficiently positive dynamic has been attained that can compensate for a disadvantaged starting position on income, employment and unemployment levels? In conclusion: structural macroeconomic ‘push’ conditions related to income, employment and unemployment in the sending countries in 2001 are strongly associated with the change in intended future migration as predicted by standard economic thinking. The recent positive changes in the new member states in regard to income, employment and unemployment have resulted in little change in the intention to migrate: absolute stock conditions explain changes in the willingness to migrate, whereas – in a counter-intuitive result – the positive dynamics between 2001 and 2005 have no influence. 3.2 Subjective conditions in 2001 as possible ‘push’ factors Having identified the effects of macroeconomic ‘push’ factors on the recent change in intentions to migrate, the next part of this chapter turns to the analysis of subjective conditions in 2001 as a possible ‘push’ factor. Two important dimensions have been chosen: general life satisfaction, and the perception of subjective economic conditions. The first dimension is based on quality-of-life approaches in which general life satisfaction is regarded as a central indicator of overall individual well-being; the second dimension is well-founded in economic thinking as supplementing objective income conditions. Within this context, a subjective paucity of individual well-being in a society, and perceived difficulties in getting by on the household’s income, are seen as important ‘push’ factors for migration. Both effects will be tested with aggregated micro-data at the country level. As the data for both indicators are not available for 2005, the empirical analysis will be limited to absolute stock conditions in 2001 in comparison to 2005 and their possible effect on the change in the intention to migrate over the four-year period. The following hypotheses will be tested. • •
The lower the extent of overall life satisfaction in 2001 in a country, the greater is the increase in the intention to migrate, between 2001 and 2005. The greater the perceived difficulties in ‘getting by’ with the household income in 2001 in a country the greater is the increase in the willingness to move to another country.
The results are shown in Figures 15.9 and 15.10. Both indicators (general dissatisfaction with life, and greater subjective difficulties in achieving an acceptable standard of living on the available household income) have a strong positive association with the increased willingness to migrate. With an r2 of nearly 0.4, general life dissatisfaction has a stronger association with changing migration intentions than has GDP per capita (the strongest macro-economic indicator). In addition, the effect of the perceived subjective income situation is as strong as the objective income situation. In summary, the perception of reduced overall quality of life in a sending country constitutes an important ‘push’ factor for increased willingness to migrate. Dissatisfaction with household income confirms standard economic thinking regarding its importance for changing intentions to migrate. Overall, aggregated subjective
380
Hubert Krieger
8
Change in intention to migrate 2001–2005
LT PL
MT
EE
6
LV
PT
4
IE EL
DK DE
SI UK
2
ES BE CY LU NL AT SE
0
HU
SK
FR
FI IT CZ
0 R2 = .392
10 20 30 40 General life satisfaction (% not at all or not very satisfied) 2001
50
8
Change in intention to migrate 2001–2005
LT PL
MT EE
LV
6
PT
4
IE EL
DK DE
UK
SI
2
SK
ES BE FR
LU SE
0
HU
CY
AT
FI NL
IT CZ
20 R2 =
.321
40 60 80 HH gets by with income (% with (great) difficulty) 2001
100
Figures 15.9–15.10 Stock of subjective indicators, 2001, and change in intended migration, 2001–2005, for EU-25
Migration and mobility culture
381
conditions are stronger predictors of migration for the 25 member states than are objective macroeconomic indicators. In policy terms, these results are in line with the identified motivational set-up of migrants: a combination of social, economic and quality-of-life dimensions influence LDGM. Politicians in both receiving and sending areas have to take this into consideration when trying to retain or attract migrants. 3.3 Country groupings and their effect on the dynamic of the intention to migrate In this final part of the empirical analysis, the previous approach of section 3.2 is applied to the two different country groupings: the EU-25 and the EU-15. Are the observed results confirmed for both country groupings or are these results valid for only one of them? Based on the results of the comparison of structural patterns of migration in section 2, the expectation is that economic ‘push’ factors explain increased willingness to migrate mainly in the EU-25, and that these factors are of minor importance in the EU-15. In addition, it is assumed that quality-of-life aspects are important in both country groupings. In order to reduce the extent of data presentation, in the following table, the results for the EU-15 are compared with the already presented empirical results for the EU-25. The results in Table 15.4 are straightforward: static macro-economic and subjective ‘push’ factors in 2001 had an effect only on the changing willingness to migrate, if one considers the enlarged Europe of 25 member states as the unit of analysis. Looking at the explanation of changing attitudes to migration in the EU-15, differences in just three indicators (GDP per capita, general life satisfaction, subjective economic well-being) show a low level of statistical association with an increased willingness to move to another country. This means that, in 2001, only in the NMS did absolute structural stock conditions have a strong influence upon attitudes to inter-country migration. This is in line with the multivariate analysis on changing structural
Table 15.4 Comparison of the effect of macro-economic and subjective conditions on the dynamic intention to migrate between EU-25 and EU-15 (regression coefficient: T2) EU-25
EU-15
1. Macroeconomic push factors 1.1 Stock 2001 GDP EMP UEMP LUEMP
−0.325 −0.169 +0.264 +0.260
−0.145 −0.005 0.000 0.000
1.2 Dynamics 2001–2005 GDP EMP UEMP LUEMP
+0.022 +0.107 −0.338 −0.183
0.000 +0.030 0.000 +0.081
Subjective indicators: push factors (stock 2001) Life satisfaction +0.392 Subjective income conditions +0.321
+0.190 +0.179
2.
382
Hubert Krieger
patterns of the willingness to migrate in 2001 and 2005, which demonstrated the increasing influence of country-specific effects in five of the new member states. At the same time, mainly subjective conditions remain influential in both country groupings.
4. Conclusions According to the latest figures, the intention of European citizens to engage in LDGM in the future is on a slow upward trend. This suggests that European citizens are reacting positively to the new challenges of globalisation, to the demands for greater adaptability in all important life domains and to the enlargement of the EU. European policy-makers, who associate higher levels of future mobility with enhanced flexibility and better economic performance (in turn leading to higher levels of employment and income), will welcome this trend as good news. Whether these small increases herald the beginning of a genuine ‘mobility culture’ in Europe has to be considered with some caution. However, knowing the diversity of European labour markets and social welfare states, it is not surprising to see a very uneven development of this trend within the various member states. Europe cannot be said to have a uniform consistent shift towards greater openness to either inter-country or inter-regional migration given the range of variation across countries, with some showing a negative movement. Furthermore, this variation is so wide that a simple comparison of average stock figures of mobility on the European level with stock figures of future LDGM in other parts of the world (the USA, for example) is not useful. The dominating feature of the analysis of the positive dynamics in mobility intentions is the coexistence of stability and of limited change towards higher intentions to move. Structural stability, in terms of a persistent impact of demographic, social and biographical factors, is combined with a limited increase in willingness to be mobile. This dynamic is stronger for migration intentions at the inter-country level than at the interregional level. It is also much stronger as regards inter-country mobility in the new member states – in particular in Estonia, Lithuania, Latvia, Malta and Poland than in the rest of Europe. Structural macroeconomic ‘push’ conditions (income, employment and unemployment) in the sending countries in 2001 are strongly associated for EU-25 with the change in intended future migration as predicted by standard economic thinking. Yet, the recent positive changes in the new member states (EU-10) in regard to macroeconomic conditions have resulted in little change in the intention to migrate: absolute stock conditions explain changes in the willingness to migrate, whereas – in a counter-intuitive result – the positive dynamics between 2001 and 2005 have no influence. In addition to macroeconomic factors, the perception of reduced overall quality of life in a sending country constitutes an important ‘push’ factor for increased willingness to migrate. Dissatisfaction with household income confirms standard economic thinking regarding its importance for changing intentions to migrate. Discontent with individual life satisfaction has the strongest association with an increased attitude to migration. Overall, aggregated subjective conditions are stronger predictors of migration for the 25 member states than are objective macro-economic indicators.
Migration and mobility culture
383
In policy terms, these results point towards a relatively stable medium-term pattern of structural factors influencing the future willingness to migrate; this finding supports political concerns related to a possible ‘youth-drain’ and ‘brain-drain’ in several new member states. This structural pattern is exacerbated by the quantitative increase in the migration intentions in some of those countries between 2001 and 2005. For the receiving countries, the profile of possible new arrivals from other EU member states is good news, as they can look forward to young, qualified and highly motivated persons. The new arrivals have positive expectations, geared towards a new, exciting and welcoming social environment as well as towards favourable economic conditions. Factors that are also high on the on the agenda of the new arrivals are adequate and accessible public services (including schooling). In terms of political initiatives regarding European Year of Workers’ Mobility in 2006, the European Commission may reflect on the large variations in existing stocks and dynamics, as well as the differences in underlying motivations and structural conditions between the member states. Based on this, it could decide to discuss these variations within a systematic peer review between the member states process as part of the open method of coordination on employment policy.
Notes 1 Florian Fliegner of the Science Centre (WZB) in Berlin provided the data analysis and Aidan McKeown helped with the English language editing. 2 Eurobarometer 64.1 on geographical and labour market mobility (autumn 2005). 3 These results suggest that those with a firm intention to migrate amount to between a half and one-third of those with a basic intention to migrate. On the basis of the 2005 dataset a ‘general inclination’ cannot be measured. Krieger’s (2004) measurement of the general inclination to migrate in 2001 was based on three questions: ‘Do you intend to go to live and work – for a few months or for several years – in a current European Union country in the next five years?’ Responses to this question are combined with the results of two other questions on the same topic: ‘Do you intend to move in the next five years?’ and if the answer was positive the follow-up question: ‘In the next five years, do you intend to move to another city, town or village within the same region; to another region within the same country; to another country in Europe and to live in a country outside Europe?’ Positive responses on each of these three items were defined as indicating a general inclination to migrate. Those who identified ‘another country in Europe’ as their migration target in the last of these questions were considered as having a ‘basic intention to migrate’, while a fourth indicator – willingness to live in a country with a foreign language – was added to identify a ‘firm intention to migrate’ (Krieger 2004: 9–10). 4 The extent of the gap between intention and social action has been explained for example by Ajzen’s (1991) concept of planned behaviour which identifies subjectively perceived control of one’s own behaviour as an important factor. This includes two components: first, internal factors directly related to planned behaviour (information, abilities, skills, emotions) and under full personal control and, second, external factors defined by existing opportunities and dependency on others. The more behaviour is dependent on external control, the larger is the gap between observed intentions and actual behaviour. 5 An overview on the results on geographical and labour market mobility based on the Eurobarometer study is available in Vandenbrande (2006). An additional analysis on longdistance mobility is provided by the European Foundation (2006). 6 It is worth noting that these intentions over a five-year time horizon are not much higher than actual annual movements between states in the US noted earlier. It thus confirms the weak mobility culture in Europe, even though it might be on an upward trend. 7 As it is a logistic regression model, the probability is given as a logistic factor, which is difficult to interpret. It is however possible to transform this logistic factor into an odds ratio. Hence the probability is e0.342 = 1.4 times higher for men than for women.
384
Hubert Krieger
8 The level of education is defined by the age leaving school or university. Low level defines school leaving under 16 years of age, ‘medium’ leaving at under 20 years and ‘higher’ leaving school/university at more than 20 years of age.
References Ajzen, I. (1991) The Theory of Planned Behaviour: Organisational Behaviour and Human Decision Processes, Milton Keynes: Open University Press. Ajzen, I. and Fischbein, M. (1980) Understanding Attitudes and Predicting Social Behaviour, Engelwood Cliffs, NJ: Prentice-Hall. Boeri, T. and Brücker, H. (2005) ‘Why are Europeans so tough on migrants’, Economic Policy, 44, October: 629–677. Boeri, T., Hanson, G. and McCormick, B. (eds) (2002) Immigration Policy and the Welfare System, Oxford: Oxford University Press. Commission of the European Communities (2006) Employment in Europe 2006, Luxembourg: Office of Official Publications of the European Communities. European Foundation for the Improvement of Living and Working Conditions (2006) Longdistance Mobility in Europe: Getting the Balance Right, Dublin: European Foundation. Fassmann, H. and Hintermann, C. (1997) Migrationspotential Osteuropa – Struktur und Motivationen potentieller Migranten aus Polen, der Slowakei, Tschechien und Ungarn, ISR Forschungsbericht Heft 15, Vienna: Verlag der Oesterreichischen Akademie der Wissenschaften. IOM (1999) Migration Potential in Central and Eastern Europe, Geneva: International Organisation for Migration. Kalter, F. (2000) Theorien der Migration, in U. Mueller, B. Nauck and A. Diekmann (eds), Handbuch der Demographie 1 – Modelle und Methoden, Berlin, Heidelberg, New York: Springer Verlag. Krieger, H. (2004) Migration Trends in an Enlarged Europe, Luxembourg: Office for Official Publications of the European Communities. Vandenbrande, T. (ed.) (2006) Mobility in Europe: Analysis of the 2005 Eurobarometer Survey on Geographical and Labour Market Mobility, Luxembourg: Office for Official Publications of the European Communities.
16 Where we stand in Europe1 Citizen perceptions of European country rankings and their influence on subjective well-being Jan Delhey and Ulrich Kohler
Introduction German reunification was a large-scale experiment which has been often regarded as a test case for the recent enlargement of the European Union (EU) towards the East. ‘Transformation through unification’ (Zapf et al. 2002) certainly included strong advantages, among them ready-made political institutions and more than one trillion Euros of transfers from the richer part, which helped to improve living conditions of large sections of the East German population. However, these improvements were accompanied – and overshadowed – by skyrocketing unemployment and other social costs such as the debasement of biographies and the destruction of a familiar lifeworld. These social costs are often referred to in order to explain why levels of satisfaction of East Germans did not improve as fast as expected – despite being ultimately able to purchase new cars, to travel abroad and to equip their modernised flats with new furniture and consumer electronics. An alternative, but less frequently employed explanation is provided by reference group theory. This approach claims that in the course of unification, living conditions in West Germany became the ultimate yardstick for East Germans (Noll 1996; Delhey and Böhnke 2000). Progress and setbacks in catching-up were painstakingly reported, and the unexpected slowness of the catch-up process led to some disappointment. Many years after unification, the feeling of being second-class citizens is widespread, which prevents parts of the East German population from feeling well received in the unified Germany. The key message from the German experience is that once countries merge politically, people’s frames of reference can be expected to change, because they re-adjust their comparative standards for evaluating social conditions, away from co-nationals as the previous reference group to other groups living in the wider new political entity. This is exactly what might also happen – or has already happened – in the case of the European Union (EU), which since 2004 has assembled together 25 nationalities – and since 2007 as many as 27 – at very different levels of prosperity and quality of life. In political science, Haas has defined supranational integration as ‘the process whereby political actors in several distinct national settings are persuaded to
386
Jan Delhey and Ulrich Kohler
shift their loyalties, expectations and political activities toward a new centre, whose institutions possess or demand jurisdiction over pre-existing national states’ (Haas 1968: 168). For the purpose of this chapter, we define European social integration along these lines: as the process whereby citizens in the single EU member states shift their expectations and aspirations for a ‘good life’ towards the new centre, namely the EU institutions, and whereby they increasingly use the average living conditions within the Community as a yardstick for evaluating their own lot. The extent to which EU citizens use ‘EU-average living conditions’ as a comparison yardstick or ‘reference group’ is explored in this chapter. This issue is connected to the ongoing discussion about the possible Europeanisation of social inequalities (Beck and Grande 2004; Heidenreich 2006; Mau 2006; Delhey and Kohler 2006; Heidenreich 2003). In a nutshell, it is claimed that the process of political integration of Europe has not only created a unique political system of multi-level governance, but also a common social space in which living conditions and life chances are no longer determined by national forces alone. European-level policies and institutions such as EU citizenship and structural policy also play their part in structuring social inequalities across the EU; hence a panEuropean inequality regime is in the making. This emerging regime is not meant to replace national inequality regimes, but rather to be superimposed on them. Like some federal states – such as Germany, for example – the EU foremost conceives of steep territorial disparities between its member states as problematic, and addresses them within the framework of structural policy as a target for redistribution. Thus the kind of inequalities the EU focus public attention on are not class differences (which are the dominant concern in most European nation states), but the striking welfare gaps between richer and poorer member states. This policy emphasis on the territorial dimension of inequality is the main reason why scholars expect not only increasing awareness of EU citizens for welfare disparities within the Community, but also rising claims for transnational redistribution that arise from such cross-border comparisons (Beck and Grande 2004). However, whereas EU policies are well described in the literature, little is known about EU citizens’ frame of reference – do they really shift their expectations and aspirations for a ‘good life’ towards the new centre? Do they really compare their own lot, or that of their co-nationals, against the EU average? Or are they still nationally oriented, without looking beyond national boundaries for a comparison with relevant others? This chapter links to this debate. With the aim to provide some fresh insights grounded in empirical research: we address the following set of key questions, •
• • •
To what extent do EU citizens have a European frame of reference in the sense of rating economic and social conditions in their own country with respect to EUwide comparisons? In other words: does the perceived ‘European average’ constitute a yardstick of comparison and in this sense a new salient reference group? Where do Europeans place their own country vis-à-vis the EU average, and which rank order of countries can be derived? How accurate are public perceptions of their nation’s relative standing within the EU? Are collective comparisons of the home country with EU-wide standards relevant in the sense of having an impact on how people feel about their own quality of life?
Perceptions of well-being
387
Before dealing with these questions we start with some introductory remarks concerning the sociological reference group theory and its applicability to the EU.
1. The reference group concept applied to the EU level The basic idea of reference group theory is that people compare themselves with others when evaluating their own situation (Merton and Kitt 1950; Hyman 1968; Festinger 1968; Kelly 1968). As a consequence, it is possible to be satisfied with poor conditions, if the salient reference group lives in even poorer conditions. Likewise, it is possible to be dissatisfied at a very elevated level of living, if the salient reference group is the high society. Theoretically, there are many possible ‘others’, but traditionally it is assumed that the yardstick people most commonly use is their fellow citizens, especially their associates of the same status, namely the in-group of friends and associates. Hence, for the founders of the concept, it seemed almost natural that comparisons would be made within rather than between societies. Empirical research on subjective well-being has by and large adopted this nation-centred view (Easterlin 1974, Veenhoven 1991, Diener et al. 1992) in studying comparisons with friends, neighbours, colleagues and other social classes within the borders defined by nation states. Why, then, should the ‘EU-average’ – or a selected EU country – function as a salient reference group? First of all, globalisation processes suggest taking an international dimension in the definition of reference groups into account. Sociologists are observing the spread of consumer culture from the West to other parts of the world, driven by global media and marketing campaigns (Ger and Belk 1996), which may lead to a convergence of aspiration levels (Bruni and Stanca 2006). Likewise, tourism is one of the booming industries worldwide, and today ordinary citizens have the opportunity to visit more places around the globe than ever before in human history. In a world connected by television, the worldwide web and growing tourism, it is unlikely that people conceive of their country as an island which is totally isolated from the world outside. Yet comparisons and aspiration levels do not necessarily go global. The idea that Europeans pay more attention to Europe, than to the whole, wide world seems plausible on several grounds. Europe is an area of particularly dense cross-border transactions. The European countries are highly inter-connected to each other by flows of trade, communication, and tourism. Most Europeans have more ‘European’ experiences than global experiences, and ‘Europe’ or the EU is a category which is frequently referred to in the media, at least in the quality press (Trenz 2005). Institutionally, whereas the world at large constitutes a weak system at best, the European Union has undertaken various steps in political system building including the introduction of a joint currency in several member states (Bartolini 2005). There is presently no consensus on what kind of polity the EU constitutes – a network state (Castells 2003), or a cosmopolitan empire (Beck and Grande 2004) – but there is consensus that the EU establishes a distinct social space, overarching and linking together the societies of the 25 member states. On formal grounds, EU citizenship equips the member states’ citizenries with a body of equal rights; but in material terms the enlarged European Union is a diverse and heterogeneous body. In turning external disparities (between sovereign states) into internal ones (between its constituent member states), the European project makes welfare gaps more visible. The decisive
388
Jan Delhey and Ulrich Kohler
point is that EU bodies define precisely socio-economic differences between the member states – in stark contrast to cultural diversity! – as problematic for achieving a cohesive Community. Consequently, one of its major policy goals is to reduce crosscountry (and regional) disparities in levels of development, a task which has become even more challenging after expansion towards the East. The key here is that the EU addresses inequalities within the Community foremost as territorial disparities, not as class inequalities, which are traditionally a major concern of social policies at the national level (Heidenreich 2003). The EU expounds the problem of between-country disparities and sensitizes public agendas to this issue. Structural and cohesion policy subdivides member states and regions as eligible or non-eligible for EU money, increasing the awareness of the steep pan-European diversity in living conditions. In the Commission’s Cohesion Reports, member states are classified into four groups of prosperity, from super-standard to sub-standard, measured against the EU average. Eurostat’s social reporting activities collect a huge body of comparative information, and member states’ social and economic conditions are time and again ranked and compared vis-à-vis the EU average (weighted by each country’s population size, e.g. Eurostat 2004). The political rationale behind this system building via official EU-level statistics is to re-direct comparison standards to the Community as a whole and thereby strengthen its relevance and identity as a political entity. The national media increasingly refer to such European ranking lists, providing their readers and viewers with recurrent information on where their own country stands in Europe regarding various policy domains. In May 2006, the banner headline of a Berlin newspaper was: ‘Europe gets ahead of Germany’, reporting that Germany had fallen back economically during the past 15 years from one of the most prosperous to an average EU country. In the same vein, the recurrent debate about the nation-specific distribution of the EU budget lays open which countries belong to the (rich) club of net-contributors, and which form part of the (poor) group of net-receivers. In sum, the Community average, or other EU countries, particularly the big and/or prosperous ones, increasingly provide comparative yardsticks for assessing the social and economic conditions which are presumably not meaningless to ordinary citizens. This is, among others, suggested by Fahey and Smyth: In judging the adequacy of their personal situations, Europeans seem to have an uncanny grasp of where their societies stand in the international (or at least European) hierarchy of economic development and to take that standing into account in arriving at subjective evaluations of their personal circumstances. (Fahey and Smyth 2004: 24) In a similar vein Beck and Grande (2004) expect Europeans’ to extend their frame of reference well beyond the national realm, perceiving themselves, or their countries, as part of a larger European stratification system. As political integration proceeds, they claim, people will increasingly compare inequalities across countries, which will undermine the legitimacy of territorial disparities within the EU in the long run. Recent research suggests that such speculations are not without an empirical basis. In a three-country-study, cross-border comparisons with living conditions in other European countries were found to influence personal life satisfaction to a statistically
Perceptions of well-being
389
significant degree (Delhey and Kohler 2006). Feelings of collective deprivation against other countries in particular were found to make people less satisfied, suggesting that upward comparisons are more salient than downward comparisons. However, this study also revealed that Germans and Turks (less clearly the Hungarians) find it easier to assess the living conditions of friends, neighbours and co-nationals than those of average citizens in foreign countries, suggesting that the most frequently used standards of comparison are still national and local. Whereas this study was concerned with other European countries as potential reference groups, the present chapter studies to what extent the ‘EU-average’ serves as a comparative yardstick. In the remainder of this chapter we are concerned with how citizens rate social and economic conditions in their own country relative to the EU as a whole, how accurate these perceptions are, and how salient the ‘EU-average’ is as a comparison standard shaping people’s sense of well-being.
2. Data The analysis is based on data from the European Commission’s Eurobarometer survey. The Eurobarometer (EB) is a biannual set of public opinion polls conducted in the member states. Since it is a post-enlargement survey, we have information on old and new member states at hand.2 The surveys are representative of the population aged 15 and over in each country. Sample sizes are usually around 1,000 respondents, except for Luxembourg, Malta and Cyprus which has a sample size of 600, Germany, which has 2,000, and the UK, which has 1,300. We use EB 62.1 from late 2004. The strength of this survey for our purposes is that it provides questions on how respondents rate the state of affairs in their country in comparison to the average of the European Union. The question reads like this: For each of the following domains, would you say that the situation in (OUR COUNTRY) is better or less good than the average of the European Union countries? (1) (2) (3) (4) (5)
The The The The The
situation of the (NATIONALITY) economy; employment situation in (OUR COUNTRY); situation of the environment in (OUR COUNTRY); social welfare situation in (OUR COUNTRY); quality of life in (OUR COUNTRY)?
Answer categories were ‘much better’, ‘somewhat better’, ‘somewhat less good’ and ‘definitely less good’. The spontaneous answer ‘identical’ was additionally coded, yet not provided by the questionnaire. This means that respondents who think of their country as being on par with the EU average might feel forced to use either the ‘somewhat better’ or the ‘somewhat less good’ option, although neither of the two reflects their own opinion adequately. For lack of space, we will not deal with the policy domains ‘environmental situation’ and ‘social welfare situation’, but instead concentrate on the areas economy, employment, and quality of life. As a short form, we will refer to the questions stated above as country-EU-comparisons or collective comparisons.
390
Jan Delhey and Ulrich Kohler Economy
Employment
Quality of life
Belgium Finland Luxembourg Netherlands Slovakia Greece Hungary Denmark Poland Germany Sweden Slovenia Austria Czech Republic Lithuania Latvia Estonia Portugal Italy France Malta Ireland Cyprus Spain United Kingdom 0
5
10
15
0
5
10
15
0
5
10
15
Figure 16.1 Country–EU comparisons (share of missing answers by survey country) Source: Own calculations, Do-file: anmiss_2.do
3. Are citizens able to rate their country vis-à-vis EU average? Reference group theory tells us that comparisons with others require some information about, or at least an image of, the lot of those others (Merton and Kitt 1950). If respondents have no opinion about average social conditions within the Community, we assume that they cannot use the EU countries as a yardstick or ‘reference group’. By displaying the share of missing answers for each country, Figure 16.1 shows to what extent people actually make reference to EU-wide comparisons. The countries are presented in ascending order of refusals, from Belgium (with the smallest number of missing answers) to the UK (with the highest number). Regardless of the specific domain of comparison – economic situation, employment situation, or quality of life – on average some 5 per cent of EU-citizens are unable, for whatever reasons, to rate their country vis-à-vis EU average. The questions on EU-wide comparisons thus provide no bigger problems for respondents than other survey questions which have been repeatedly tried and tested and which also demand an evaluation of the external world, like generalised interpersonal trust (4 per cent missing on average, EQLS 2003). On the other hand, country-EU-comparisons are more difficult to answer than ratings of very personal states of affairs, like life satisfaction (only 1 per cent missing on average, EQLS 2003). It is also worth mentioning that Europeans obviously find it easier to compare their country with the EU average, than to assess the overall living conditions in other European countries. The latter questions result in some 10 to 25 per cent missing answers, depending on the country to be rated (Delhey and Kohler 2006). Hence, Europeans seem to have a more definite idea regarding how average life in the EU is, than regarding the living conditions in specific countries. New member states (marked by blacks dots in Figure 16.1) and old member states (white dots) do not differ in how often refusals occur. At first glance this may come
Perceptions of well-being
391
as a surprise: because the citizens of the old member states, particularly those in the founding members of the European project, may be expected to be more familiar with the Community. However, there are also good reasons to assume that the questions are easier to answer for citizens from new member states, because the striking socioeconomic contrast between West and East facilitates the task for new EU citizens to take a firm stand on the country-EU comparison. In contrast, the question is not plain sailing for many EU-15 citizens, because differences are less pronounced and because the questionnaire did not offer the option ‘identical’ as a response category.3 Five countries stand out with relative high fractions of missing answers across all three dimensions: United Kingdom, Spain, Cyprus, Ireland and Malta. These nations thus seem to be less oriented towards Europe than continental countries. It is noteworthy that four of these countries are islands, and Spain is a peninsula. Historically, the seaward empires England, Spain and Portugal were examples of early state-building with a comparatively early consolidation of national boundaries and less integration into the trade belt of continental European cities (Rokkan 1999).4 One should not overlook, however, that even in these countries the large majority of the respondents – nine out of ten – do have an impression of where their country stands in Europe. Within single countries, there are remarkable variations in the degree to which people apply European standards of comparison. Judging by the proportion of missing answers, the higher strata appear as more Europeanised – with lower proportions of missing answers – whereas people with little formal schooling, blue-collar workers and also women are more inward-looking and less likely to apply a European frame of reference. A more detailed description is given in Table 16.1, which presents the regression coefficients of logistic regression models for missing answers vs. non missing answers. The table shows that men have fewer difficulties in saying where their country stands in Europe than women. The likelihood of missing answers increases with age suggesting that the younger generation is more prone to look across national boundaries and to think European. In contrast people with low education are most likely to have national frames of reference. White collar workers and the self-employed think more European than blue collar workers or economically inactive people who seem to be more inward-looking, having greater difficulties in assessing where their country stands in Europe.
4. The ranking lists of countries: in the eyes of the citizens Where do Europeans place their country vis-à-vis the EU average? In Figure 16.2, shades of grey are used to display the differences in public opinion, ranging from dark grey (‘own country definitely less good than EU average’) to white (‘own country much better than EU average’). The darker a country’s bars are, the more people feel collectively deprived, whereas brighter bars indicate more people with a sense of collective gratification compared to the average member state. In the figure, the two groups of old and new member states are separated, and within both groups, countries are sorted from unfavourable to favourable ratings. Three results are particularly noteworthy. First, there is a steep east–west gap in comparative ratings. In most new member states, people rate the economic situation, the employment situation, as well as the quality of life in their own country as worse than the EU average – and a large proportion even perceive them as much worse.
0.09 9132 24713
0.11 8927 24713
0.07 9891 24713
Notes: *p < 0.05; coefficients for country dummies are omitted from the table.
Source: Own calculations. Do-file: anmiss_bygroups.do
0.08 9143 24713
0.07 9937 24713
0.07 9954 24713
0.09 9110 24713
0.08 9807 24713
−0.03 0.35* 0.28 0.47* −3.62*
Pseudo R2 BIC obs.
−2.73*
−0.06 0.42* 0.59* 0.95* −3.78*
−0.42* −0.72* 0.10
−0.84* −0.48* −1.55* −1.04* −0.70* 0.24 employment status (reference: self-employed) −0.44* −0.37* 0.34* 0.23 0.44* −0.13 1.10* 0.55* −4.08* −2.68* −3.86* −3.55* −3.80*
−0.72* −1.12* −0.49*
−0.54* 0.01*
−0.62* 0.01*
−0.58* 0.02*
−0.70* 0.02*
Employment
Men y/n Age Education (reference: low) Intermediate High Other, missing Combined occupational and White collar Blue collar Still at school Econ. inactive Constant
Economy
0.07 8599 24713
−4.22*
−0.55* 0.02*
0.07 8638 24713
−2.69*
−0.97* −1.30* −0.70*
0.07 8639 24713
−0.26 0.45* 0.51* 1.13* −3.95*
Quality of life
0.09 8482 24713
−0.19 0.41* 0.08 0.60* −3.86*
−0.59* −0.78* 0.21
−0.48* 0.01*
Table 16.1 Explaining difficulties of rating one’s own country (coefficients of logistic regressions of missing answer vs. non-missing answer)
Perceptions of well-being Economy
Employment
393
Quality of life
Luxembourg Denmark Ireland Austria Sweden Finland Netherlands Belgium United Kingdom France Germany Spain Italy Greece Portugal Cyprus Slovenia Malta Czech Republic Lithuania Estonia Slovakia Latvia Poland Hungary
0
.25
.5
.75
1
0
.25
.5
.75
1
Def. less good
Less good
Somewhat better
Much better
0
.25
.5
.75
1
Identical
Figure 16.2 Outcomes of country–EU comparisons (cumulated % of responses) Source: Own calculations. Do-file: ancomp_2.do
There is no big difference between the three dimensions in this respect, albeit the national quality of life tends to be painted in slightly less dark colours. At the bottom of the ranking list we find Hungarians who overwhelmingly (around 95 per cent) conceive of their country as lagging behind in all three dimensions. In the Baltics, Poland, and Slovakia, public opinion is similarly negative. The home country’s relative standing compared to the EU standard is seen more optimistically in Slovenia and Malta, especially with respect to quality of life. In these countries, similarly strong fractions of citizens perceive the collective situation as either below or above the EU average (once both options on the positive or negative side are lumped together). However, both Maltese and Slovenian citizens regard their country’s economy and employment situation as below-average. In Cyprus, a majority of people is convinced that their country offers better quality of life than the average member state. In stark contrast to the situation in most new member states, citizens of the old EU-15 predominantly perceive their countries as above average, particularly with respect to quality of life. Only Southern Europeans do not follow this pattern fully, as Portugal and Greece come close to the post-communist pattern of displaying a strong sense of collective deprivation across all three domains. Spaniards and Italians regard their country as performing worse with respect to the economy and the employment situation, while Germans and French rank their country as below-average only with respect to employment. In the remaining member states heading the table – Belgium, the Netherlands, Finland, Sweden, Austria, Ireland, Denmark, and particularly Luxembourg – most citizens perceive their country as better than the Community average. The predominantly positive response pattern in most West European countries suggests that people apply a pan-European yardstick which includes the new member states, rather than having only the old EU-15 in mind.
394
Jan Delhey and Ulrich Kohler
Second, Europeans tend to rank the overall quality of life as somewhat better than either the economic situation or the employment situation which tends to be seen in the most critical light. This is especially true for the Germans and the citizens of the Mediterranean countries Malta, Slovenia, Cyprus, Greece, Italy, Spain and France. This reiterates the recurrent finding that unemployment is the people’s dominant concern in almost all EU countries (Eurobarometer 64). Only the British are an exception to this rule by being more sceptical about the UK’s relative standing with respect to quality of life, than about the country’s employment situation. A third remarkable finding is that the country ranking derived from public perceptions fits quite well with ‘objective’ league tables. In Luxembourg, almost everybody believes that Luxembourg enjoys better-than-average economic conditions (95 per cent), better-than-average employment situation (86 per cent) and better-than-average quality of life (97 per cent). Objective data indeed reveal Luxembourg as by far the wealthiest member state, ranking among the most developed countries in terms of human development, and having a lower-than-average unemployment rate. A similar story may be told for the Belgium and the Netherlands, for the Nordic countries, and for Ireland. At the bottom of the ranking list, Poland, Slovakia, Hungary and the Baltic countries are indeed economically less well off and also occupy lower ranks on the human development index. Again much in line with the respondents’ subjective perceptions, countries like Malta, Cyprus and Slovenia are found in an intermediate position. This close correspondence between objective data and subjective rankings suggests that Europeans do look across national borders and have fairly clear concepts of where their country stands in European comparisons. The next section deals with the accuracy of people’s perceptions in more detail.
5. Perceptions or misperceptions? What exact properties do people have in mind when they assess the collective standing of their country in European comparisons? To answer this question we will look at the correlation between country rank orders in the citizens’ perceptions and in objective data on national performance. The EB survey questions do not fully clarify the issue because they do not clearly specify the bases for the judgements. Take, for example, the question on the relative perception of the home country’s economy. It leaves open whether respondents base their judgments on levels of living as reflected in GDP per capita, or on recent economic performance, as reflected in growth rates, or, on a combination of both aspects. Using growth rates or the level of living as the relevant yardstick makes a big difference, however. The Baltic countries are rather poor, compared to Western standards, but currently booming, while Italy and Germany are much wealthier, but have almost been stagnant for protracted years. So which of these aspects did European citizens have in mind when answering the question? In order to find out, we have calculated country-level correlations between the perceived relative standing of the country’s economy in the EU and the two objective measures, GDP per capita and 2000–2003 growth rates. The striking result is that correlations are moderate and negative for growth rates (−0.36), but high and positive for per capita income (0.91). The latter yielded an almost perfect correlation. Obviously, people use the economic level of living, and not recent economic performance, as a yardstick for comparisons. Additionally, we have computed correlations
Perceptions of well-being
10
20 30 40 50 GDP p. cap. (in PPS)
2 −2
0
CZ PL
HU
SK GR
2 4 6 Growth rates NMS
LT EE LV
8
IE
FI SE AT BE GB NL FR CY DE ES IT SI
0
0
MT PT
LU
DK
1
IE
−1
1
LU
FI ATSE BE GB NL FR CY DE ES IT SI
−1
CY DE ES IT
SI MT CZ LT EE SK PT GR PL LV HU
DK
MT PT
−2
LU
−2
1
DK IE FI SE AT BE GB NL FR
0 −1 −2
r = .02
2
r = .36
2
r = .91
395
−4
PL
CZ
SK HU
EE GR
−2 0 2 Part growth rates
LT LV
4
OMS
Figure 16.3 Cross-check between reality and perceptions (economic situation) Source: Own calculations. Do-file: anvalid_2.do
for controlled growth rates (controlled for level of GDP), since it is well-known from economic literature on convergence processes that it is easier for catching-up economies than for mature economies to achieve high growth rates (Barro 1991, 1996; Sachs and Warner 1995). This correlation, however, is close to zero, suggesting that it is simply the economic level of living which guides people’s perceptions of their countries’ economic situation. This is substantiated by three scatterplots in which the country’s relative standing in the eyes of its citizens, is plotted against our three objective measures of the economic situation (see Figure 16.3). Logically, there are four possibilities. The respondents may: 1 2 3 4
rightly perceive their country’s situation as worse than EU average; rightly perceive their country’s situation as better than EU average; wrongly perceive their country’s situation as worse than EU average; or wrongly perceive their country’s situation as better than EU average.
Judging by the assessment of economic wealth, the perceptions of the relative standing of the home country vis-à-vis EU standards must be considered quite accurate. Most countries are located either in the lower-left or upper-right quadrant formed by the horizontal and vertical lines, which represent the Community averages on both dimensions. Moreover, most countries find their place close to the regression line, although there are some places where the public mood is worse than the actual situation, particularly so in Greece and Italy. In the other two figures (perceptions plotted against growth rates and against controlled growth rates, respectively), the country dots are spread almost randomly. This confirms that it is actually the level of wealth that guides public opinions. How accurate are people’s perceptions of the employment situation? We use both employment rates and unemployment rates as objective measures of labour market performance. In both cases, people’s perceptions correspond to external reality to a considerable degree (see Figure 16.4). The majority of countries is located in those two quadrants that indicate accurate perceptions, rather than misperceptions (note that
396
Jan Delhey and Ulrich Kohler
LT LVEE
PT
PT HU
50
55 60 65 70 Employment rate
75
1 0
SI MT CZ
1
1
SI MT ITDE CZ
CY
ES
LT EE GR LV
SK
DK IE AT SE GB NL FIBE
LU
FR ES IT DE
LT PT LVEE SK PL HU GR
PL
2
SK GR HU
2
PL
DE CZ
LU DK IE ATSE NLGB BEFI CY FR
2
0 1
IT MT
FR SI
GBSE NL
0
FI CY
BE ES
DK AT
1
1
LU IE
2
r = .89
2
r = .68
2
r = .71
0
5 10 15 20 Unemployment rate NMS
10
20 30 40 50 GDP p. Cap. (in PPS)
OMS
Figure 16.4 Cross-check between reality and perceptions (employment situation) Source: Own calculations. Do-file: anvalid_2.do
unemployment is a reversely coded indicator, with low numbers being the policy goal). There are only very few countries where misperceptions of the ‘true’ ranking dominate. E.g. in Portugal, employment is far above the EU average, but nevertheless the Portuguese think that their employment situation is below the EU standard. In Hungary and Portugal, similar misperceptions can be found with respect to the unemployment rate (mood is worse than external reality). Luxemburg and Ireland have average employment rates only, but people nevertheless rate the employment situation as much better than EU average. A similar misperception can be found in Finland with respect to unemployment. One reason might be that for subjective evaluations of the employment situation, working conditions such as safety of workplace might also play a role. But by and large, external reality as reflected in rates of employment and unemployment is quite accurately reflected in public perceptions. Finally, we have plotted the perceived employment situation against GDP per capita. Surprisingly, this correlation is the strongest (0.89), suggesting that respondents use their assessment of economic wealth, relative to EU as a whole, as the main heuristic, even when ranking the area of employment. One remaining question then is which conditions shape the perception of the overall quality of life in a European perspective (see Figure 16.5). The United Nations’ Human development index (HDI) can be regarded as the best and most widely used composite indicator capturing the overall quality of life. Once again, we find people’s perceptions of their relative national quality of life, compared to the EU as a whole, to correspond quite closely to external reality as measured by HDI scores (r = 0.89). Next we split up the HDI in its three component parts, economic level of living, life expectancy, and educational attainment in order to determine in which realm we find the most realistic judgments.5 Subjctive perceptions correspond rather closely to external reality in the case of the two former measures, but more closely for economic level of living than for life expectancy. The accuracy of judgments is much lower with respect to the third component, educational attainment, where overt misperceptions occur. Remarkably the correlation with GDP per capita turns out to be even higher than the correlation with the compound HDI-Index. This suggests once again that when European citizens reflect upon the quality of life in their country, they first and
Perceptions of well-being 2
r = .33
2
r = .8
10
20 30 40 50 GDP p. cap. (in PPS)
70
PT
−2
EE LV HU
LTSK PL
1
1 0 −1
LT SK PT EE LV PL HU
LU
GR
CZ
−2
1 0 −1 −2
CZGR
LU BEFRSE AT DK FINL IE CY DE GB ES MTIT SI
0
LU FIBE AT DKIE FR NL SE CY DE GB ES MT SI IT
−1
2
r = .92
397
72 74 76 78 Life expectancy NMS
80
BE SE IE NLDK FR CY DE GB ES
AT
FI
SI
IT
GR
CZ PT SK PL
HU
LV
EE
LT
10 20 30 40 Perc attained tertiary educ.
OMS
Figure 16.5 Cross-check between reality and perceptions (overall quality of life) Source: Own calculations. Do-file: anvalid_2.do
foremost base their evaluations on their knowledge about the economic level of living. To sum up, public perceptions of where one’s country stands relative to the EU average are strikingly accurate. Misperceptions most often appear in countries that are close to the EU-average in objective terms, which might be partly explained by the restricted answer options, as there was no default option to answer ‘identical’. The results furthermore suggest that people use above all the economic level of living as their yardstick when rating their country in a European perspective. There is a clear tendency to rate the own country as above average in rich countries, and belowaverage in poor countries, regardless of the domain in question. This suggests that the most important ranking list of countries people have in mind is a ranking list of economic wealth.
6. Does the perception of one’s home country’s ranking in the European Union have an impact on individual well-being? Our last analysis is concerned with the consequences that collective comparisons have for personal well-being. Do Europeans who perceive their home country as collectively deprived, experience detrimental effects on their own life satisfaction? As mentioned before, collective comparisons clearly had an impact on personal well-being in the case of German unification. Improved living conditions did not translate into enhanced personal life satisfaction as expected, because there were widespread feelings of relative deprivation in east Germany which lowered the east Germans’ satisfaction levels and hampered the process of growing together of the two parts of Germany (Noll 1996, Delhey and Böhnke 2000). Recent comparative research also found evidence that comparisons with other countries often nurture dissatisfaction, if people think that living conditions are better in other places (Delhey and Kohler 2006). To what extent do perceptions of a country’s relative position vis-à-vis the EU average have a bearing on how people feel about their personal quality of life? The Eurobarometer survey allows us to pursue this matter, because it posed the following
398
Jan Delhey and Ulrich Kohler
question: ‘How would you judge the current situation in each of the following domains? [...] (6) Your quality of life: very good, rather good; rather bad; very bad?’ Individual quality of life is rated as high in the old member states (with the partial exception of the southern countries, and Portugal in particular), and as much lower in the new member states (with Cyprus, Malta and Slovenia as exceptions). This resembles the well-known picture obtained from other surveys which directly asked for general life satisfaction (Delhey 2004; Christoph and Noll 2003; Böhnke 2005). Hence, although the wording does not include the term ‘satisfaction’, we feel encouraged to interpret the question at hand as a measure of life satisfaction. We expect that the collective country vs. EU comparison influences how people assess their very personal quality of life. Those who believe that their country offers better quality of life (economic conditions, employment conditions) than the EU average, should be more satisfied; those who think quality of life is worse in their country should be less satisfied. In order to single out the influence of country–EU comparisons, it is necessary to cancel out the influence of individual life circumstances. This is imperative because people may be dissatisfied with their lives not only because they draw an unfavourable country–EU comparison, but simply because they live in poor personal conditions. A solution to this problem is provided by regression models6 which control for individual characteristics, namely the financial situation, occupational status, education level, gender, age7 and marital status – all of which are known to influence general life satisfaction (Argyle 1999; Headey and Wearing 1992). Moreover, we decided to compare only relatively deprived and gratified persons living within the same country. This takes into account that living conditions vary widely across countries, which would also confound the results of our analysis. Due to this double control, the remaining differences in personal life satisfaction are likely to reflect the different assessments of collective quality of life, compared to the EU average. Table 16.2 displays the results. The dependent variable is life satisfaction (rating of personal quality of life); independent variables are the mentioned control variables, supplemented by the country–EU comparison with respect to the economic situation. The first key result is that financial satisfaction is of paramount importance for life satisfaction. Money is a road to individual happiness everywhere, but the effect is stronger in some places than in others. Especially in the poorer post-communist member states, as well as in Portugal, money can buy individual happiness. This corroborates previous findings on the declining marginal utility of income on higher levels of wealth (Schyns 1998; Veenhoven 1999; Delhey 2004), and it also fits the idea that in the process of modernisation value priorities change from material to post-material concerns (Inglehart 1990). The second key result is that collective country–EU comparisons have a remarkable impact on individual well-being. This results from the first row of the Table 16.2. The coefficients are positive in all countries, indicating that the better the citizens rate national economic standing as compared to the EU average, the more satisfied they are with their personal quality of life. Moreover, in most countries effects are significant, even if we take the uncertainty of sampling into account.8 Across the EU-25, we find significant effects in 17 member states, 12 of them old members, and in five new members. This suggests that for West Europeans, the ‘EU average’ constitutes an even more salient reference point than for East Europeans, and in this sense they seem to be ‘Europeanised’ to a stronger degree. The computations also show that collective
Perceptions of well-being
399
country-EU comparisons explain a smaller part of life satisfaction than the individual financial situation, but are not really secondary to other factors.9 The same model can be applied to the other two dimensions of comparison, employment situation and overall quality of life. Due to lack of space, the results are not shown as full tables, but are selectively displayed in Figure 16.6. It results that collective country–EU comparisons with respect to the employment situation and quality of life in general are also important for how people evaluate their own life. With only one exception (employment situation, Poland), all coefficients are positive, and in most cases they are even significantly positive. Evidently, personal life satisfaction also has a collective dimension, reflecting that individual lives are embedded in social contexts. Furthermore, the EU average as a yardstick evidently has some relevance for ordinary citizens in most places. Comparing their own nation with the EU average can make people more happy or unhappy. Among the three collective comparisons we have employed here, the more embracive dimension of national quality of life is slightly more important than the two other ones, which might reflect that quality of life means more than being rich, although money seems to be the most important component.10
7. Conclusion To a large extent, this chapter has explored unknown territory – the EU as a possible yardstick for national living conditions. Coming back to our initial definition of social integration of Europe at large, it can be concluded that the EU is at least to some extent an integrated social space. Large segments of EU citizens are able to rate their country vis-à-vis the EU average, and collective country-EU comparisons are a salient feature of general life satisfaction. Perceptions of collective deprivation or gratification can make people more unhappy or more happy, although, of course, these kinds of comparisons are by no means the strongest satisfiers, or dissatisfiers. Offering only a snapshot of the situation at one point in time, our analyses cannot reveal if Europe has become a more salient reference group over time, as the concept of ‘Europeanisation’ as a process would require. We also do not know to what extent the EU-average might have replaced the nation-state standard as a meaningful reference group. It is well possible that Europe simply constitutes an additional layer of reference, in much the same way as it constitutes an additional layer of collective political identity (Bruter 2005). Finally, we are also in the dark whether the EU average is a more salient yardstick than the USA or images of a good life disseminated by world wide media, which would imply that processes of globalisation rather than Europeanisation are at work in producing new reference groups. By and large, people’s comparative assessments are quite accurate. Not everybody gets it right, but on average the respondents have a rather valid idea of their country’s relative position in the European hierarchy. The assessments do not differentiate very much between various domains of life. There is a clear tendency to rate the own country as above average in rich countries, and below-average in poor countries, regardless of the domain in question. The ranking list that shapes people’s subjective map of Europe is predominantly a ranking list of economic wealth. In this sense, the citizens’ rationale can be summarised by the slogan ‘it’s the economy, stupid’. It is possible that after accession, the EU average gains even increasing importance as a reference point shaping citizens’ aspirations in the new member states. This would
400
Jan Delhey and Ulrich Kohler
Table 16.2 Explaining personal life satisfaction: the influence of economic country–EU comparisons with various control variables (coefficients of ordered logit models) AT
BE
CY
CZ
DE
DK
EE
0.31* 2.35*
0.33* 2.55*
0.33* 1.59*
0.06 2.78*
0.25* 3.04*
0.32* 1.30*
0.13 3.40*
0.19 −0.03 0.00
0.11 −0.08* 0.00*
0.30 −0.05 0.00
−0.23 −0.12* 0.00*
−0.00 −0.06 0.00
−0.13 −0.03 0.00
0.08 0.11 −0.11* −0.03 0.00* 0.00
−0.34* −0.08 −0.07* −0.12 0.00 0.00
0.22 −0.07* 0.00*
0.05 −0.35 1.25
0.57* 0.30 −1.62*
0.26 0.48 0.70
−0.03 −0.20 −0.19
0.32 0.42 0.13
0.04 −0.06 −0.70
0.70* 0.58* −0.60
0.45 0.89 −0.80
−0.30 −0.11 −0.56
Combined employment and occupation (reference: self-employed) Managers 0.30 0.39 0.75 −0.33 0.32 Other white collar −0.35 0.21 0.56 −0.92 −0.19 Manual workers −0.20 0.20 0.13 −0.78 −0.79* House person −0.27 0.07 0.23 −0.97 −0.49 Unemployed −0.21 −0.32 −0.39 −0.77 −0.44 Retired 0.44 0.16 0.54 −1.46* −0.24 Students 2.20* 0.63 −1.37 1.49
−0.47 −0.78 −0.27 −0.34 −0.65 −0.63 0.17
0.36 −0.26 −0.52 −0.55 −0.82 −0.50 0.21
0.35 −0.21 −0.60 −0.59 0.11 −0.85* 1.05
0.46 0.01 0.41 0.98* 0.79* 0.15 1.45
0.07 −0.02 −0.06 −0.40 −0.21 −0.26 2.37
0.08 0.29 −0.01 −0.02 −0.12 −0.10 1.04
Marital status (reference: married) Unmarried −0.01 0.27 Divoreced −0.48* −0.48 −0.46 Widowed −0.41 Other/missing −0.70 −0.73
Comp. economy Own finan. situation Men y/n Age Age (squared)
Education (reference: low education) Intermediate −0.35 0.48 High −0.22 0.51 Other/missing −1.55* 0.43
Cutpoint 1 Cutpoint 2 Cutpoint 3 Obs.
−1.28 2.67 7.51* 938
0.08 3.42* 8.11* 979
ES 0.31* 2.80*
FI
FR
GB
0.11 2.27*
0.36 2.85
0.28* 1.56*
−0.25 0.87* 0.56 1.59*
0.05 −0.17 0.26 −2.10*
−0.20 −0.29 −0.66 −0.44
−0.23 −0.31 −0.16 −1.25*
0.20 0.19 −0.04 0.30
−0.13 −0.46 −0.17 −0.49
−0.27 −0.30 −0.11 −0.19
−0.17 −0.63 −1.14 −0.22
−0.12 0.04 −0.27 −0.35
−1.09 1.21 5.59* 447
−1.43 2.27 8.01* 907
0.54 4.18* 9.99* 1474
−2.5* −0.76 2.30* 1038
0.51 5.00* 11.6* 916
0.59 4.13* 9.51* 908
−0.97 1.33 5.78* 995
−1.23 2.59 8.88 907
−1.76 0.10 4.26* 1121
Economy situation
Employment situation
Quality of life
Portugal Greece Spain Italy Germany Finland France Sweden Belgium United Kingdom Netherlands Austria Denmark Ireland Luxembourg Latvia Poland Estonia Lithuania Slovakia Hungary Czech Republic Malta Slovenia Cyprus
−.5 0 .5 1 1.5 −.5 0 .5 1 1.5 .5 Regression coefficients from ordered logit models NMS 10
OMS 15
Figure 16.6 Impact of country–EU comparison on life satisfaction Source: Own calculations. Do-file: anvalid_2.do Note: Unstandardised coefficients on ordered logit models.
0
−.5
1
1.5
Perceptions of well-being
GR
HU
IE
IT
LT
LU
LV
0.19* 2.27*
0.37* 4.31*
0.41* 1.85*
0.40* 2.68*
0.04 −0.01 0.00
−0.19 −0.01 0.00
0.23 0.19
MT
NL
0.19* 3.71*
0.50* 3.08*
0.15 3.50*
0.30 0.19* 2.46* 2.39*
−0.03 0.02 −0.13* 0.02 0.00* −0.00
−0.20 −0.01 0.00
0.08 −0.01 0.00
0.12 0.12 −0.63
−0.11 −0.01 1.11
0.47 0.48 0.06
−0.33 0.18 −1.77
−0.38 0.14 0.11 −0.43 0.31 0.02 0.76
0.10 0.30 0.53 0.89 0.51 −0.02 2.02*
−0.20 −0.73* −0.08 −0.24 −1.04* −0.92* −0.12
0.12 0.16 0.26 0.01 0.97 0.04 0.89
−0.15 −0.59 −0.14 0.40
−0.19 −0.44 0.09 0.21
−0.14 −0.31 −0.18 −0.10
1.99* 4.39* 8.66* 979
4.35* 9.48* 16.2* 949
−2.5* −0.30 4.20* 886
PL
PT
SE
0.44* 4.57*
0.13 1.79*
0.27* 2.60*
−0.13 −0.08 0.00
0.42 0.03 −0.04 −0.22 0.04 −0.09* −0.08* −0.04 −0.00 0.00* 0.00* 0.00
−0.19 −0.04 0.00
0.01 −0.13 −0.10* −0.11* 0.00* 0.00*
0.88 1.04 0.25
0.09 0.28 −0.63
0.43 0.48
0.19 0.38 −0.14
−0.05 0.19 −0.22
0.88* 0.24 0.44
0.12 0.15 0.24
0.25 0.28 0.68* 0.45 −2.18* 0.40
−0.07 −0.50 0.21 −0.31 −0.18 −0.64 3.29*
−0.53 −0.69 −0.90 −0.38 −2.79* −0.90 1.52*
−0.47 −0.43 −0.51 −0.68 −0.53 −1.39* 0.87
−0.03 0.23 0.04 0.36 1.03 0.48 2.22*
−0.88* −0.98* −0.85* −1.18* −0.67 −1.53* −0.60
−0.14 −0.34 −0.28 −0.30 −0.43 −0.14
0.04 −0.77 −0.81 −0.08 −0.57 −0.26
0.38 −0.05 0.09 0.32 −0.21 0.16 1.00*
−0.72 −0.12 −0.75 −0.84 −0.25 −0.04 2.34*
0.57* −0.06 0.19 −0.36 0.17 0.11 0.81
0.06 0.32 0.00 0.36
−0.05 −1.11* −0.26 −0.19
−0.80 −1.18 −0.47 −0.65
−0.25 −0.08 −0.04 0.07
1.03* 38.03* 0.52 −0.02
−0.02 0.33 −0.55 0.33
−0.11 0.06 −0.45 0.30
0.16 −0.18 0.01 −0.53
−0.21 −0.51* 0.05 −0.14
0.16 −0.48 −0.48 0.51
−0.38 0.09 0.24 −0.17
2.92* 6.22* 11.7* 929
3.18* 7.57* 14.2* 894
1.27 5.27* 11.5* 489
2.11 6.19* 12.5* 931
2.93* 5.69* 11.1* 431
−1.36 1.05 5.47* 982
1.29 5.24* 11.4* 938
3.88* 9.01* 16.6* 901
−1.23 0.57 4.09* 963
−1.48 1.15 7.38* 955
−1.22 2.74* 7.86* 1114
0.08 3.37*
SI
401
SK 0.58* 2.69*
mean that economic convergence will not result in an equally strong convergence of subjective well-being. But it is also worthwhile to stress the main difference between German unification and EU enlargement towards the east. Germany is a redistributive nation-state, whereas the Community is mainly a regulative network of states. While Berlin is a ‘natural’ addressee for discontent and claims of redistribution of life chances, Brussels has only limited competence to improve living conditions in poorer areas. From this perspective it is much easier for a supranational Community to live with large territorial disparities than for nation-states – and even the latter have been able to cope with remarkable degrees of regional inequality as the examples of Italy or the United Kingdom illustrate.
Notes 1 We are grateful to the editors, particularly Jens Alber and Tony Fahey, for helpful comments on an earlier version of this contribution. 2 The dataset has been made available by the Central Archive for Empirical Social Research, University of Cologne (study number s4230). 3 It is in line with this reasoning that the spontaneous answer ‘identical’ was given more often in the old member states. 4 Interestingly enough, Spaniards also have more problems than other Europeans in rating the trustworthiness of people from other EU countries (Delhey 2005). 5 The latter was measured as a percentage of persons between 25 and 65 with tertiary education.
402
Jan Delhey and Ulrich Kohler
6 The dependent variable is measured on an ordinal scale with four categories. We have therefore used a so-called ordered logit model (also known as proportional-odds model) (Andersen 1997). 7 As it is common in regression models for life satisfaction, we have used age and agesquared as control variables. 8 Significance calculated by applying design-based standard errors, which are generally larger than those given by the standard formulas. Note that we were not able to calculate design-based standard errors for France because of limited information about the sample design. 9 The comparison variable is measured on a 5-point scale, while the satisfaction with the own financial situation is measured on a 4-point scale. Strictly speaking, the coefficients of the two variables are not comparable, therefore. However, the scales are similar enough to justify the conclusion above. To make sure, we have also calculated a ‘X-standardised solution’ (Long and Freese 2001), which brought up the same results. 10 Previous research on the determinants of individual well-being has shown that people tend to care less about gains than about losses (for a summary see Layard 2005). Our own fourcountry comparison of the impact of macro-perceptions on general life satisfaction similarly showed that the perceived relative deprivation of one’s home country impacts more on individual well-being than the perceived relative gratification (Delhey and Kohler 2006). Departing from these findings we expected the upward comparisons with the European mean in poorer countries to produce stronger effects on individual well-being than the downward comparisons in richer countries. In contrast, our 25-country comparison found a tendency for poorer countries to produce even slightly weaker impacts of collective comparisons on individual well-being. Since the respective country differences were not very marked with frequently overlapping confidence intervals of the regression coefficients as well as uncertain estimates of regression coefficients in countries with low variation on the independent variable of collective rankings, we would consider a substantive interpretation of this counter-intuitive tendency in our empirical results at this stage as pointless.
References Andersen, E.B. (1997) Introduction to the Statistical Analysis of Categorical Data, Berlin, Heidelberg: Springer. Argyle, M. (1999) ‘Causes and correlates of happiness’, pp. 353–373, in D.E. Kahnemann, E. Diener and N. Schwarz (eds), Well-being: The Foundations of Hedonic Psychology, New York: Russell Sage Foundation Publications. Barro, R. (1991) ‘Economic growth in a cross section of countries’, Quarterly Journal of Economics, 106: 407–444. Barro, R. (1996) Determinants of Economic Growth, NBER Working Paper, Bartolini, S. (2005) Restructuring Europe: Centre Formation, System Building and Political Structuring between the Nation-state and the European Union, Oxford: Oxford University Press. Beck, U. and Grande, E. (2004) Das kosmopolitische Europa. Gesellschaft und Politik in der Zweiten Moderne, Frankfurt a. M.: Suhrkamp. Böhnke, Petra (2005) Happy, Satisfied, Belonging: Subjective Well-being and Its Determinants in Europe, Dublin: European Foundation for the Improvement of Living and Working Conditions. Bruni, L. and Stanca, L. (2006) ‘Income aspirations, televisions and happiness: evidence from the World Value Survey’, Kyklos, 59, 2: 209–225. Bruter, M. (2005) Citizens of Europe? The Emergence of a Mass European Identity, Houndmills: Palgrave Macmillan. Castells, M. (2003) Jahrtausendwende. Teil 3 der Trilogie ‘Das Informationszeitalter’, Opladen: Leske & Budrich. Christoph, B. and Noll, H.-H. (2003) ‘Subjective well-being in the European Union during the 90s’, Social Indicators Research, 64, 3: 521–546.
Perceptions of well-being
403
Delhey, J. (2004) Life Satisfaction in an Enlarged Europe, Dublin: European Foundation for the Improvement of Living and Working Conditions. ¨ Delhey, J. (2005) Das Abenteuer der Europäisierung. Uberlegungen zu einem soziologischen Begriff europäischer Integration und zur Stellung der Soziologie in den Integration Studies, Soziologie. Forum der Deutschen Gesellschaft für Soziologie, 1, 34: 7–27. Delhey, J. and Böhnke, P. (2000) ‘Führt die materielle zur inneren Einheit? Zum Verhältnis von Wohlstandslage und subjektivem Wohlbefinden’, pp. 83–103, in H.-H. Noll and R. Habich (eds), Vom Zusammenwachsen einer Gesellschaft, Frankfurt, New York: Campus. Delhey, J. and Kohler, U. (2006) ‘From nationally bounded to pan-European inequalities? On the importance of foreign countries as reference groups’, European Sociological Review, 22, 2: 125–140. Diener, E., Sandvik, E., Seidlitz, L. and Diener, M. (1992) ‘The relationship between income and subjective well-being: relative or absolute?’, Social Indicators Research, 28, 3: 195–223. Easterlin, R.A. (1974) ‘Does economic growth improve the human lot? Some empirical evidence’, pp. 90–125, in Paul A. David and Melvin W. Reder (eds), Nations and Households in Economic Growth, New York, London: Stanford University Press. Eurostat (2004) The Social Situation in the European Union 2004, Luxembourg. Fahey, T. (2005) Rich and Poor in the Enlarged EU: An Expanded Approach to Measurement, Dublin: European Foundation for the Improvement of Living and Working Conditions. Fahey, T. and Smyth, E. (2004) ‘Do subjective indicators measure welfare?’, European Societies, 6, 1: 5–27. Festinger, L. (1968) ‘A theory of social comparisons’, pp. 123–146, in H. H. Hyman and E. Singer (eds), Readings in Reference Group Theory and Research, New York: Free Press. Ger, G. and Belk, R.W. (1996) ‘Cross-cultural differences in materialism’, Journal of economic Psychology, 17: 55–77. Haas, E. (1968) The Uniting of Europe, Stanford: Stanford University Press. Headey, B. and Wearing, A. (1992), Understanding Happiness: A Theory of Subjective Well-being, Melbourne: Longman Cheshire. Heidenreich, M. (2003) ‘Regional inequalities in the enlarged Euroland’, Journal of European Social Policy, 13, 4: 313–333. Heidenreich, M. (2006) ‘Die Europäisierung sozialer Ungleichheit zwischen nationaler Solidarität, europäischer Koordination und globalem Wettbewerb’, pp. 17–64, in M. Heidenreich (ed.), Die Europäisierung sozialer Ungleichheit. Zur transnationalen Klassenund Sozialstrukturanalyse, Frankfurt, New York: Campus. Hyman, H.H. (1968) ‘Introduction’, pp. 3–21, in H.H. Hyman and E. Singer (eds), Readings in Reference Group Theory and Research, New York: Free Press. Inglehart, R. (1990) Culture Shift in Advanced Industrial Society, Princeton, NJ: Princeton University Press. Kelly, H.H. (1968) ‘Two functions of reference groups’, pp. 199–206, in H.H. Hyman and E. Singer (eds), Readings in Reference Group Theory and Research, New York: Free Press. Layard, R. (2005) Happiness: Lessons from a New Science, Harmondsworth: Penguin. Long, J.S. and Freese, J. (2001) Regression Models for Categorical Dependent Variables Using Stata, College Station: State Press. Mau, S. (2006) ‘Grenzbildung, Homogenisierung, Strukturierung’, pp. 109–136, in Martin Heidenreich (ed., Die Europäisierung sozialer Ungleichheit. Zur transnationalen Klassenund Sozialstrukturanalyse, Frankfurt, New York: Campus. Merton, R.K. and Kitt, A.S. (1950) ‘Contributions to the theory of reference group behaviour’, pp. 40–106, in R.K. Merton and P.F. Lazarsfeld (eds), Studies in the Scope and Method of ‘The American Soldier’, Glencoe, IL: Free Press. Noll, H.-H. (1996) ‘Ungleichheit der Lebenslagen und ihre Legitimation im Transformationsprozess’, pp. 488–504, in L. Clausen (ed.), Gesellschaften im Umbruch.
404
Jan Delhey and Ulrich Kohler
Verhandlungen des 27. Kongresses der Deutschen Gesellschaft für Soziologie, Frankfurt, New York: Campus. Rokkan, S. (1999) State Formation, Nation-building, and Mass Politics in Europe: The Theory of Stein Rokkan. Based on His Collected Works, Selected and Rearranged by Peter Flora, with Stein Kuhnle and Derek Urwin, Oxford: Oxford University Press. Sachs, J.D. and Warner, A.M. (1995) Economic Convergence and Economic Policies, NBER Working Paper, Cambridge: National Bureau of Economic Research. Schyns, P. (1998) ‘Crossnational differences in happiness: economic and cultural factors explored’, Social Indicators Research, 43: 3–26. Trenz, H.-J. (2005) ‘Die mediale Ordnung des politischen Europa. Formen und Dynamiken der Europäisierung politischer Kommunikation in der Qualitätspresse’, Zeitschrift für Soziologie, 34, 3: 188–206. Veenhoven, R. (1991) ‘Is happiness relative?’, Social Indicators Research, 24: 1–34. Veenhoven, R. (1999) ‘Quality-of-life in individualistic society’, Social Indicators Research, 48: 157–186. Zapf, W., Habich, R., Bulmahn, Th. and Delhey, J. (2002) ‘The case of Germany: transformation through unification’, pp. 229–296, in W. Adamski, P. Machonin and W. Zapf (eds), Structural Change and Modernization in Post-socialist Societies, Hamburg: Krämer Verlag.
17 Assessing the quality of European surveys Towards an open method of coordination for survey data Ulrich Kohler Introduction Social scientists have discovered Europe. They increasingly investigate questions such as whether specific hypotheses are valid in all European nations, which national rules make countries more successful, and how far the goal of social cohesion between the European member states has been reached. One of the driving forces behind the increasing interest in Europe is the so-called ‘open method of coordination’. This is a process for evaluating social policies and their effects in combating social exclusion that has been agreed by the EU member states. It involves a set of guidelines and indicators, benchmarking and sharing of best practice. Its power as a coordination tool relies on the hypothesis that no member state wants to be seen as the worst in a given policy area. The open method of coordination needs data, and the EU goes to some lengths to get this data. They finance research projects and institutions which carry out European comparative research; and research projects and institutions carry out European comparative research in order to get the grants from the EU. As there is a lot of money involved, various procedures have been put in place to ensure the quality of the research. Research institutes and researchers are evaluated regularly, projects are only temporary, and funds have to be re-applied for from time to time. Many ‘results’ of these evaluation processes get published, and some of the strength of the evaluation process lies in the idea that no research institute wants to be seen as the worst. In this sense, the ‘open method of coordination’ is also applied to social research. Social research is done by condensing data. Nobody is interested in how satisfied Mr. Lopez from Tudela, Spain, is with his life, but we do want to know whether the Spanish are more satisfied with their lives than the Portuguese are. However, although we are not interested in the actual life satisfaction of Mr. Lopez, we will not get the correct answer to the condensed question, if we systematically fail to accurately observe the life satisfaction of Spain’s Mr. Lopezes. In other words, the answer to the research question heavily depends on the initial data quality. If the initial data lacks validity, one will not get valid answers to the research question, no matter how sound the statistical methods are. Unfortunately, we do not know much about the quality of European comparative datasets. That is not to say that we do not know what features good data should have, and how one can obtain such good data. On the contrary, we know quite a lot about that. But what is missing so far is information about which European comparative research datasets perform best on a certain set of criteria. That is, we lack an
406
Ulrich Kohler
‘open method of coordination’ for data quality. This chapter is a step in this direction. (Not necessarily ‘first’ – Eurostat might dispute that.) Some preliminaries are worth noting, however. I propose that comparative surveys should achieve four goals: First, obviously, the surveyed population should correspond to the population one wants to describe. Hence, if the focus is on the enlarged Europe, the surveys should cover all countries in the EU or on the threshold of EU membership. On the other hand, if the research interest is to test a general theory, then the number of countries is of secondary importance, as the hypothesised association should show up regardless of the countries included. Second, the surveyed population should be a representative sample of the target population. Third, the survey questions should clearly capture the substantive topics, i.e. measure what the researcher intends to measure. Finally, timing is often important; a survey should be as close as possible to events of substantive interest. There are several trade-offs between the four goals. Given a fixed budget, enlarging the country coverage of a survey will limit the ability to optimise the sampling method, and to improve the validity of the instruments, or vice versa. Other tradeoffs are already inherent in the goals themselves. Improving the survey instrument and insisting on interviewing even the difficult-to-reach respondents take time, and therefore conflict with timeliness requirements. It is also important to note that the high initial cost of a comparative survey will only be worthwhile if the survey ascertains data for different topics. The timing is therefore likely to be suboptimal for at least some of the ascertained data. Moreover, the goal of covering as many relevant topics as possible will be in conflict with the desire to have multiple indicators for each theoretical concept in order to reduce measurement error. Only one of these four goals is of specific interest when entire research programmes are assessed. The number of countries and the timeliness requirements both depend on the research topic, i.e. what is ideal for one research topic might be suboptimal for another. Also the quality of measurement can hardly be established in an absolute sense. A specific survey question might well measure a certain concept in one language, but not in another, and even if that question were a valid measurement, it would be only one question out of many. Hence, it is only the second goal, the sample quality that is at least conceptually independent of the research topic. In what follows I will therefore assess only this dimension of the survey quality. From what has been said before it should be clear that the quality of a sample is not equal to the quality of a survey as such. To some extent, the sample quality is a reflection of the decisions necessary for a specific research topic. When comparing the sample quality of comparative surveys one should therefore keep the initial research topics of the various surveys in mind. I will therefore start with a short description of the survey programmes to be compared. Afterwards the assessment will be done by giving answers to each of the following questions: ● ● ● ●
Is the entire process of drawing the sample known? How good are the sampling processes? Do the data match external criteria of representativity? Do the data match internal criteria of representativity?
1. Datasets The aim of this article is to compare the sample quality of the European Quality of Life Survey 2003 with other comparable surveys. The EQLS dataset has been used
Assessing the quality of European surveys
407
extensively in this book. The editors and some of the authors of this book have been involved with the survey since the beginning. It was carried out on behalf of the European Foundation for the Improvement of Living and Working Conditions. The major goal of the EQLS was to cover the entire EU at the time of the enlargement, including the candidate countries Bulgaria, Rumania and Turkey. Hence, country coverage and timeliness were crucial goals for the EQLS. The samples cover people aged 18 and above in the chosen countries. The sample sizes in each country were approximately 1,000 persons, apart from the smaller countries (Luxembourg, Cyprus, Malta, Slovenia and Estonia) where they were approximately 500.1 Population-representative datasets with a similar country coverage carried out in a similar time frame can be compared to the EQLS. I have therefore selected the Eurobarometer 62.1, which has been used by Delhey and Kohler (in this volume), the European Social Survey 2002, the European Value Study 1999, and the International Social Survey Program 2002.2 The Eurobarometer 62.1 (EB) is part of an ongoing cross-national research project, which has implemented at least two surveys in all EU member states since the early seventies. The goals of topicality and complete coverage of all EU member states have always outweighed the other goals as far as the Eurobarometer research programme is concerned. The strength of the Eurobarometer programme is the extremely narrow timetable; the period between the initial design of the questionnaire and the availability of the results is only a few weeks. It is therefore an ideal survey for questions about current topics, although the timeliness requirements make it difficult to deviate from a set of routine questions. The EB 62.1 was carried out in late 2004. The target population for the Eurobarometer surveys is generally citizens aged 15 and above. The sample sizes usually amount to at least 1,000 persons or 500 in smaller countries (Luxembourg, Cyprus and Malta). In the EB 62.1 the actual sample sizes vary between 500 in Malta and 1,561 in Germany (see Table 17.1). The European Social Survey 2002 (ESS) was the first round of a biennial multicountry survey funded jointly by the European Commission, the European Science Foundation and academic funding bodies in each participating country. The ESS puts a strong emphasis on data quality in terms of sample quality and validity of the instruments, while timeliness is only of secondary importance. There was more than two years between the official start of ESS 2002 and the data release, and the timetables for the subsequent waves look similar. Country coverage is also of secondary importance to the ESS. To take part, countries must follow the narrow methodological instructions of the central ESS administration, and must find the funding to cover the costs of fieldwork themselves. The 2002 survey covers 21 nations, 18 of which
Table 17.1 Target population, number of countries and number of observations, by survey programme Study EB 62.1 EQLS 2003 ESS 2002 EVS 1999 ISSP 2002
Target Citizens Residents Residents Citizens Residents
15+ 18+ 15+ 18+ 18+
Countries
EU
Min. Obs.
Max. Obs.
25 28 22 32 33
25 28 19 28 20
500 591 1,207 968 1,000
1,561 1,071 2,919 2,500 2,947
Source: Own calculations. Do-file: ansvydes02.do
408
Ulrich Kohler
are current EU member states. The samples cover the residential population aged 15 and above in each participating country. The ESS tries to achieve effective sample sizes of around 1,500, and 800 for small countries.3 The actual sample sizes in 2002 vary between 1,207 in Italy and 2,919 in Germany. The European Values Study 1999 (EVS) is the third round of a cross-national survey research programme which was started in the late 1970s by the European Value Systems Study Group. It has always been the major goal of the EVS to maximise country coverage, and the EVS succeeded in ascertaining data from all European countries, including those from Eastern Europe, as early as its second round in 1990. The third round now covers 32 countries, including all the members of the EU, Cyprus and all 4 candidate countries. The target population is defined as citizens aged 18 and over and the sample sizes vary between 1,000 and 2,000 in most countries. The lowest and highest sample sizes were found in Iceland and Russia respectively. The International Social Survey Programme 2002 (ISSP) is round 15 of a continuing programme of cross-national surveys. The ISSP is a group of research institutes, who agreed to jointly develop topical modules and to field these modules as a fifteenminute supplement in regular national surveys. To become a member of the ISSP, the institute has to agree to ascertain the ISSP module in accordance with relatively permissive ISSP requirements. Hence, the major goal of the ISSP can be seen as the validity of the instruments and the enlargement of the country coverage. Timeliness and sample quality by and large depend on the national surveys within which the ISSP modules are embedded. Surveys for the ISSP 2004 were carried out in 33 countries, 20 of which were EU members or EU candidates. The target population of the samples is residents aged 18 and above. Sample sizes vary between 1,000 in Latvia and 2,947 in the United Kingdom.
2. Is the entire process of drawing the sample known? The central criterion for objectivity of science is intersubjective traceability (Popper 1994: 18). This requires that every decision made during research is documented. In the context of sampling methods for face-to-face surveys this means that the following features of the sample should be published: ● ● ● ●
the the the the
sampling method; allowance for substitutions; components of the response rate; and contents of back-checking regulations
It is important to know whether the sampling method can be described as simple random sampling (SRS), as multistage register sampling, as multistage random-route sampling or as quota sampling. Among these types, SRS can be regarded as the gold standard.4 Unfortunately SRS is seldom applicable. Simple random sampling requires a complete list of all individuals in the target population, which is often not available. Therefore a multistage sample is usually used. Multistage samples involve the selection of geographical areas like communities as a first selection step. A list of individuals from the target population is more often available within these so called ‘primary sampling units’ (PSU), and such lists can therefore be used as sampling frames.5 Sometimes at least a list of households or addresses in the PSUs, which can than be
Assessing the quality of European surveys
409
used as an intermediate step, is available. A side effect of any multistage sampling method is that the statistics calculated from such data are usually slightly more insecure than those from SRS (Kish 1965). Another side effect is that the interviewer has to apply a methodological rule to select the respondent at a given household or address. This influence the interviewer has on the selection of respondents can be a source of errors. In many cases, however, a suitable sampling frame is still not available. In this situation a so called ‘random route procedure’ is applied, where addresses are selected by walking on a prescribed route, selecting every nth address. There are two versions of this idea; an elaborate version is to collect the addresses by random route, and to contact these addresses independently somewhat later. However, in what has been called ‘random route light’, the two versions are bound together and the interviewer makes immediate contact with the identified address. The major problem with this procedure is that it gives the interviewer a lot of influence on the selection of households. As a consequence, the decisions of the interviewers cannot be properly controlled. Finally, quota sampling means asking the interviewers to select respondents with specific properties. Quota samples do not meet the fundamental requirements of a scientific sampling method. During fieldwork, survey administrations have to deal with the problem of noncooperative or unreachable research units. An often applied workaround is to substitute such research units by others. Substitution, however, leads to over-representation of cooperative and easy-to-reach respondents, and tends to decrease the extent of efforts by interviewers to gain responses from the original research units. Hence, substitution is not defendable on theoretical grounds, and it is therefore important to know whether substitution was allowed. Response rates are the most commonly used measures of sample quality, and therefore documented (almost) habitually. An overview of the reported response rates of the five survey programmes is shown in Table 17.2. The table illustrates that three survey programmes achieved very high response rates in at least one country. The EQLS reports response rates of above 90 per cent for Germany and Malta, the ISSP achieved a response rate of 99 per cent for Spain, and the EVS has a similarly high value for Slovenia. These three survey programmes also have very low response rates in at least one other country. The EQLS had a minimum of 32 per cent in Ireland, the EVS 15 per cent in Spain, and the ISSP 20 per cent in France. The average response rates are very similar in three of the five survey programmes, but somewhat higher in the ESS. The response rates of the Eurobarometer are not documented. High response rates are often interpreted as a signal of good sample quality. In many cases, however, high response rates are merely indicators of insufficiently
Table 17.2 Minimum, average and maximum response rate, by survey programme
Minimum Average Maximum Missing (number of countries)
EB 62.1
EQLS ’03
ESS ’02
EVS ’99
ISSP ’02
n.a. n.a. n.a.
32 58 91
43 61 80
15 58 95
20 58 99
25
0
0
3
1
Source: Own calculations. Do-file: anresp01.do
410
Ulrich Kohler
controlled surveys. Consider the case of the random route light described above: in this case it is only the interviewer, who knows the person to be contacted. If he does not reach that person, he might contact a nearby substitute who is easily reachable. If an interviewer acts that way, it is unlikely that he would inform the survey administration about his ‘defective’ behaviour. Instead, he would list the easy-to-reach substitute as the correct eligible respondent. The result of this would be that the number of interviews completed and the number of eligible respondents will become closer to one another, which means that the response rate will be high. Hence, the more interviewers act in this way, and the less the survey administration can control the interviewers’ actions, the higher the response rate will be. Response rates can be only taken as quality indicators for well controlled surveys, and I will therefore not consider them any further. However one aspect of response rates is worth noting. As stated above, response rates are given by dividing the number of interviews carried out by the total number of eligible respondents. Although this general idea is pretty ubiquitous, the actual formulas for calculating the response rates are not. Differences exist regarding the definition of ‘non eligible respondents’ (Schnell 1997: 19–27). Therefore in order to achieve a meaningful interpretation of response rates, it is necessary to communicate the absolute numbers of non-eligible respondents broken down by the specific reason for non-eligibility. Hence, I will consider the full documentation of the response rate statistics as a quality indicator of a survey programme. Back-checking means re-contacting respondents to check whether the respondent really exists, whether the interview has really been conducted, or even whether interviewers have correctly recorded information. It should be clear from the outset that the ability to back-check decreases with the influence of the interviewer on the selection of respondents, and is very limited in the case of random route light. Besides this, survey organisers differ in their stipulations about the fraction of respondents to be re-contacted, and how this re-contact is done. The quality of a sample could rise with the proportion of respondents to be re-contacted. Hence, one should know if quality back-checks have been made and how large the fraction of selected respondents is. Table 17.3 displays our knowledge about the survey features described above. It should be read as follows: samples of 19 countries of the EU members and candidates have been drawn for the ESS. We have ‘complete knowledge’ (i.e. we are able to categorise the sampling method as SRS, multistage individual register, multistage address register, multistage random route or quota) about the sampling method for all of them. Likewise we know whether substitutions were allowed, and the components of the response rate for all countries. However, we do not know the back-checking regulations in 2 of the 19 ESS countries (Slovenia and Luxembourg).6 If one calculates the fraction of countries with known features for each of the columns, and adds these up, one will get a number between zero and four, whereby zero means that we know nothing, and four that we know everything. For the ESS this number is 3.9. Given this overall quality index the best survey-programme is the ESS, followed by the EVS and the EQLS. The figures presented in Table 17.3 need some clarification. I am not claiming here that the missing information is essentially unobtainable. It might well be that one can get the missing information by calling the survey administrator from the relevant survey organisation in each survey country. However, it is proposed that such information should be published in ‘readily available’ sources. As ‘readily available’ I have
Assessing the quality of European surveys
411
Table 17.3 Number of countries for which complete information about the specified component of the sampling process is available, by survey programme Study ESS 2002 EVS 1999 EQLS 2003 ISSP 2002 EB 62.1
Number of Sampling Substitution Resp. rate Back-check EU countries method regulations statistics regulations 19 28 28 20 25
19 20 28 10 25
19 28 28 20 25
19 22 28 0 0
17 26 0 16 0
Information quality 3.9 3.4 3.0 2.3 2.0
Source: Own calculations. Do-file: ansample01_1.do
considered all sources that are either available on the internet, or as printed technical reports, or as machine readable documents in the data delivery package.7 One should also note that the quality of the sources containing the necessary information is different. Generally, the information can be presented as a written report, like the EQLS, or by giving the material of a methodological questionnaire in raw format, as is the case with the EVS. Written reports make it easier the find the necessary information and involves some sort of information screening. It is fairly common to find that the raw material from methodological questionnaires is unclear, and the authors of the reports can often clarify any problems. These authors, however, have their own ideas about what is important, and this often leads to missing information. Finally one should note that it is easier to document a survey programme which uses the same process in all countries than a survey programme which takes the particular necessities of the respective country into account. The documentation of the Eurobarometer 62.1, for example, is very limited. However, as shown in Table 17.3, it documents the sampling methods of all countries as a multistage with random route light. The ISSP and the EVS, on the other hand, use different sampling methods, depending on the availability of sampling frames. We know that some countries in these two survey programmes applied a multistage sampling scheme, where individual registers are not available within the PSUs. However, we do not know whether households are selected from a list of addresses or by a random route technique. If one does not insist on this particular piece of information, the quality index of the ISSP and the EVS would be somewhat higher, although they would not get a different position in the rank order.
3. How good was the sampling process? In the previous section I have argued that the sampling method and the regulations concerning substitutions and back-checks are important components of the sampling process. I will now use the available knowledge about these components to compare the sample quality of the five survey programmes. Figure 17.1 illustrates our knowledge about the components of the sampling process. The first panel of the figure describes the sampling method. In particular it shows whether simple random sampling (SRS), or the several variants of multistage sampling, or quota sampling were used. The sampling methods are ordered along the horizontal axis by their respective quality. Hence, I regard SRS as having the best quality, and quota sampling as having the worst. Note that the term ‘unknown’ refers to samples that applied multistage probability sampling, but without documentation
412
Ulrich Kohler
of the technique used to sample the final observations. We do not know what kind of multistage sample we have, but we do know that it is at least better than quota sampling. SRS has been seldom used so far. It is fairly common in Denmark, Finland and Sweden, where suitable sampling frames for SRS exist.8 Simple random sampling has also been applied in Malta in the European Value Study. The sampling frame in Malta was a list of all registered voters, which is unsuitable for samples that include persons aged 18 and under. Multistage sampling with random route is the most often used sample method. Unavailable sampling frames might be one reason for the popularity of this technique. However, it is only in France, Cyprus, Lithuania, Bulgaria and Turkey that better sampling methods than random route have not yet been applied. The ESS in particular makes substantial efforts to bypass random route by applying registers as sampling frames. In the light of this, the figure reveals that random route is generally not the best available alternative for many countries.9 Quota sampling, which is very common in market research, has been seldomly applied, but it was used in several EVS participating countries. The heavy use of quota samples comes as a surprise, given the low reputation of quota sampling in the scientific realm and the rich academic literature growing from the EVS. However, the use of quota samples can be justified by budget restrictions, and by the primary goal of complete country coverage. The second panel in Figure 17.1 contains information on whether substitutions were allowed, and it will be treated as better if substitutions are not allowed. The figure reveals that substitutions were not allowed in the ESS and the EQLS, but always were in the Eurobarometer. Substitutions were forbidden in some of the participating countries in the ISSP and the EVS, but not in others. This somehow reflects
Sample method
Subst. allowed
Backchecks
Austria Belgium Denmark Finland France Germany Greece Ireland Italy Luxembourg Netherlands Portugal Spain Sweden United Kingdom Cyprus Czech Republic Estonia Hungary Latvia Lithuania Malta Poland Slovakia Slovenia Bulgaria Croatia Romania Turkey SRS
Indiv. reg.
EB
Add. reg. Random-route Unknown
EQLS
ESS
Quota
EVS
No
Yes
Yes
No
ISSP
Figure 17.1 Components of the sampling process, by country and survey programme Source: Own calculations. Do-file: ansample02_1.do
Assessing the quality of European surveys
413
the use of quota samples in the EVS. It is a basic property of quota sampling that the interviewers arbitrarily select persons with certain properties, and therefore can substitute every non-cooperative person by another person with the same properties.10 Finally, panel three in Figure 17.1 shows whether institutionalised back-checks have been applied. As mentioned above, this information is not available for the Eurobarometer, and the EQLS. The figure therefore only displays the back-checking regulations for the three other survey programmes, which have generally applied back-checks. However, one should keep in mind that back-checks cannot control whether an interviewer has interviewed the ‘right’ person in case of quota sampling and random route light. In the context of the sample quality, the back-checking regulations will only be of crucial importance when the interviewers have little influence on the selection process. To provide an overall assessment of the sample quality of the five survey programmes the information in Figure 17.1 needs to be summarised. One way to do this is to find out which survey programme has applied the best solution for a specific survey country. In Austria, for example, the EVS used the best sampling method, while in Belgium the ESS and the EVS ranks first. If we go on like this, we can count how many times each survey programme ranks first. Table 17.4 displays these numbers as a fraction of the number of countries in each survey programme. Hence, the 0.84 for the ESS in the column for the sampling method means that the ESS has applied the best sampling method in 84 per cent of its countries. Likewise the ESS has applied the best substitution practice in all countries and the best back-checking regulation in 94 per cent. The higher the numbers in Table 17.4 are, the better the particular component of the sampling process is. Following this logic, the ESS has the best sampling method, followed by the EQLS, the ISSP, the EVS, and the Eurobarometer. The substitution practice implies the same order, and the ISSP ranks above the ESS and the EVS as far as the back-checking regulations are concerned. It was not possible to calculate the figures for back-checking regulations for three of the survey programmes. Therefore, the overall quality index in the last column sums up only the figures for the sampling method and the substitution practice. This overall quality index confirms the exceptional position of the ESS as the best of the five. The assessment of quality shown in Table 17.4 can be considered as unfair in some respects. First of all, remember that the sampling methods of the ISSP and the EVS are not fully documented in some countries. It was therefore not possible to decide Table 17.4 Fraction of countries with best available sampling process, by survey programme Fraction of countries with best Study ESS 2002 EQLS 2003 ISSP 2002 EVS 1999 EB 62.1
Sampling method
Substitution practice
Back-check regulation
Quality index
0.84 0.50 0.40 0.25 0.16
1.00 1.00 0.60 0.39 0.00
0.94 n.a. 1.00 0.81 n.a.
1.8 1.5 1.0 0.6 0.2
Source: Own calculations. Do-file: ansvydes02_1.do
414
Ulrich Kohler
which form of multistage sampling was used. To construct Table 17.4 the unknown sampling method has been treated as less good than random route light. If these cases were in fact all multistage with individual registers, the figures for the ISSP and the EVS would have been 0.85 and 0.46 respectively. Hence the sampling method of the ISSP would have been ranked above the EQLS, but still below the ESS, and the position of the EVS would have been unchanged. Another point which is overlooked by the figures in the table is that there are quality differences within the different sampling methods. This is most obvious for the multistage random route samples. The ESS has applied random route in two countries, Austria and France. In both countries the ESS has divided the sampling of the addresses and the contacting of households by the interviewer.11 The random route collection of addresses in advance must be regarded as much better than random route light. Treating this elaborate version of random route like multistage sampling with address register, the figures get 0.95 for the ESS, 0.46 for the EQLS, and 0.12 for the Eurobarometer. If one applies both proposed alternative decisions, and recalculates the overall quality index, the ranking of the survey programmes will stay the same.
4. External criteria for representativity The most common way to assess the sample quality is to compare known distributions of the population with corresponding distributions of the sample. If we knew the proportion of people who are satisfied with their lives, we could calculate the corresponding proportion in the sample and compare the two figures. The higher the difference between both, the lower is the sample quality. In this section I will use this idea to assess the sample quality of the five survey programmes in question. However, before doing so, one should note some of the shortcomings of that attempt. Obviously comparisons are only possible for characteristics which are known for the population. Such representative checks usually rely on figures from statistical offices. However, these figures are often also ascertained by population samples, and hence vulnerable to sampling errors. A specific problem for the assessment of quality of samples from comparative studies is that it is not easy to evaluate the quality of the external source against which the samples are compared. Moreover, samples often do not cover specific parts of the population, such as people below or above a specific age, people not entitled to vote, people living in institutions, people living in a country illegally, and so on. It is not always easy to get figures from statistical offices that fit the specific sampling population. Last but not least, many characteristics are measured differently in different sources. It is therefore not easy to compare the frequencies of such characteristics in the different sources. It is, however, absolutely necessary to make the comparison only between comparable measured characteristics. In order to overcome some of the problems described here, I will base the following analyses on the fraction of women in each country. Gender is particularly well suited for our purpose because it is easy to measure, its distribution is relatively stable over time and it does not differ too much between subgroups of the population. The dots in Figure 17.2 show the difference between the fraction of women in the population aged over 15 as reported by the United Nations (United Nations Statistics Division 2002), and the fraction of women in the respective sample. A positive value indicates that women are over-represented in the sample, while negative values indicate under-representation. A value of 0.1 means that there are 10 percentage points
Assessing the quality of European surveys EB 62.1
EQLS 2003
ESS 2002
Austria Belgium Denmark Finland France Germany Greece Ireland
Italy
Luxembourg Netherlands Portugal Spain Sweden United Kingdom Cyprus Czech Republic Estonia Hungary Latvia Lithuania Malta Poland Slovakia Slovenia Bulgaria Croatia Romania Turkey
.1 EVS 1999
0
.1
.2
ISSP 2002
Austria Belgium Denmark Finland France Germany Greece Ireland Italy Luxembourg Netherlands Portugal Spain Sweden United Kingdom Cyprus Czech Republic Estonia Hungary Latvia Lithuania Malta Poland Slovakai Slovenia Bulgaria Croatia Romania Turkey
−.1 Random
0
.1
.2
.1
Quota
0
.1
.2 After weighting
Figure 17.2 Proportions of women (differences between survey and official sources) Source: Own calculations. Do-file: anextern01_1.do
415
416
Ulrich Kohler
more women in the sample than in the population. Hence, if the fraction of women was around 50 per cent, the observed fraction would be around 60 per cent. Deviations of that size are not uncommon. Overall, Figure 17.2 indicates a tendency towards over-representation of women. High deviation rates are particular visible for the Eurobarometer and the EQLS, but also for the other survey programmes. A plausible explanation for this overall pattern is the higher reachability of women in male-breadwinner households. If this explanation is true, the likely source for the deviations could be interviewers, who have not applied the sampling regulations correctly. Hence, there is some evidence of the importance of survey regulations which encourage interviewers to correctly apply the prescribed sampling method. The distance between the observed and the actual fractions are largest for the EQLS and the Eurobarometer, somewhat lower for the ISSP, and relatively low for the other survey programmes. It is reasonable to take these distances between the observed and the true values as an indicator of sample quality. It is however also reasonable to accept a certain amount of unavoidable random fluctuation for the differences. Whether one regards a specific difference as acceptable or not largely depends on the sample size. While high deviations from the true value are statistically likely for small samples, they are almost impossible for large ones. In other words, one might regard the same difference as unacceptable – i.e. probably due to non-random distortions – in a large survey, but as acceptable in a small one. To incorporate these ideas, the figure uses bars to show the 95 per cent confidence interval around the true value.12 The confidence intervals indicate the region within which one might regard a deviation as acceptable, and values that are outside the confidence intervals might be regarded as too big. Counting the number of countries that are outside the confidence intervals can be used as another indicator of sample quality. The more frequent values outside the confidence intervals are, the less good the sample quality of the survey programme is. A quick visual inspection of the figure gives the impression that the sample quality is relatively low for the EQLS and the Eurobarometer, and, yet again, relatively high for the ESS. I will, however, postpone a more thorough assessment of the sample quality after having explained one more detail of Figure 17.2: as shown in the last section, and as indicated by white dots in Figure 17.2, some of the observed fractions of woman are based on quota samples. As far as documented, all quota samples used gender as a quota criterion, and it seems plausible that gender was also used in the undocumented cases. It might be considered as ‘unfair’ to compare random samples with quota samples on the quota criterion. A ‘fairer’ mode of comparison might be to use the available representative weights. All survey programmes, except the ESS13 include representative weights for at least some of the countries. These weights are designed so that the sample matches specific national marginals, including the distribution of gender. To some extent, the theoretical ideas behind such weights are similar to quota sampling. In Figure 17.2, arrows are used to indicate the fraction of women after applying the representative weights. It can be seen that the representative weights change the fraction of women considerably in the direction of the true values, producing even smaller differences than quota samples. The weighted fractions are, however, essentially useless for assessing sample quality. It is easy to weight a sample so that it fits known population marginals, no matter how bad the sample is. Likewise it is easy to draw a quota sample so that it fits
Assessing the quality of European surveys
417
a known marginal. Hence, we can neither use the weighted results, nor the quota samples for a more thorough assessment of the sample quality. Table 17.5 summarises the information given in Figure 17.2 for the probability samples only. It displays the mean difference between the observed and true values, as well as the fraction of values outside the confidence bounds. To obtain an overall quality index, I have mirrored both quantities, so that high values indicate high sampling quality. Moreover, I have rescaled the average difference to have values between 0 and 1. The maximum value of the overall quality index would be 2, if a survey programme had no observed values outside the confidence bounds and the lowest average distance of all survey programmes. The ESS is once again the best survey programme. The EVS is second best, followed by the ISSP, the Eurobarometer and the EQLS. On average, the observed values of the EQLS deviate from the true values by about 5 percentage points; and almost 80 per cent (22 out of 28) of the EQLS countries show fractions of women that are too high. For the ESS, and also for the EVS, these figures are much lower. However, the fraction of countries outside the confidence bounds are too high even in the ESS, given the fact that only 5 per cent would have been expected from probability theory.
5. Internal criteria for representativity As mentioned above, the comparison between population marginals and observed quantities in a sample has some shortcomings. Internal criteria are one way to circumvent these shortcomings. The idea behind internal criteria of representativity is to define a subpopulation that has a known and fixed parameter, and to control whether this parameter is observable in a likewise defined subsample. Deviations from the known parameter can then be used as a criterion for the randomness of the subsample (Sodeur 1997). In what follows I will use the subpopulation of gender heterogenous couples who live together in a two-person household. Precisely 50 per cent of this subgroup are women and the other 50 per cent are men. I have selected this specific subgroup because it is possible to mark it out in all survey programmes. If only respondents who live in two person households, who are married, and who live together with their spouse, are selected it will be quite certain that they stem from the subpopulation of gender heterogenous couples in two-person households.14 Note that these respondents are older than the other respondents on average, but do not differ systematically regarding their household income or their location.
Table 17.5 Mean difference from the true value and fraction of countries with values outside the confidence bounds, by survey programme Study ESS 2002 EVS 1999 ISSP 2002 EB 62.1 EQLS 2003
Average difference
Fraction outside confidence bounds
Quality index
0.02 0.03 0.05 0.05 0.06
0.32 0.47 0.67 0.72 0.79
1.33 1.02 0.54 0.48 0.21
Source: Own calculations. Do-file: anextern01_1.do
418
Ulrich Kohler
Figure 17.3 shows the fraction of women among gender heterogenous couples. The more these fractions deviate from the true value of 0.5, the worse is the sample. Moreover, I will consider fractions of women outside the boundaries of the 95 per cent confidence interval around the true value as large. The results shown in Figure 17.3 are not as clear-cut as those for the external criteria of representativity. The overrepresentation of women in particular seems less strong when compared to Figure 17.2. A possible explanation for the somewhat smaller over-representation is that the traditional ‘male breadwinner model’ is less prevalent for two-person households than for families with children, so that men and women do not differ so much in how easily they can be reached. Besides the fact that the over-representation of women is less strong, the differences between the observed fraction of women and the true value of 0.5 are still remarkable. From the 125 values displayed in the figure, 39 (31 per cent) are outside the boundaries of the confidence intervals. Given that only 5 per cent of the values should be outside, this is quite a large amount. Moreover the differences are sometimes high in absolute terms, even though they do not fall outside the confidence bound. Comparing the results of the different survey programmes is again not as easy as it was for the external representativity checks. The first impression however confirms that the ESS and the EVS are of better quality than the other survey programmes, but I will return to this later on. First it should be remembered that the figures for the EVS, and to a lesser extent those for the ISSP, are somewhat biased due to the quota samples; the countries are again marked with a white dot in the figure. An interesting point about the quota samples is that they tend to differ substantially from the true value. This illustrates the fact that fixing the marginals for the entire sample does not guarantee correct distributions in any subset of the sample. A similar statement can also be made with respect to the representative weights. The arrows in the figure once again indicate that representative weights change the fraction of women considerably in the direction of the true value. But weighting can actually be harmful sometimes. A striking example of this can be found in the French ISSP 2002. In this case, the unweighted fraction of women among the gender heterogenous couples almost precisely matches the true value, while the weighted fraction is much too small. An explanation for this is that the selfselection mechanisms of gender-heterogenous couples are different from the self-selection mechanisms of other households. It is probable that the higher reachability of women in households with more than two persons (i.e. households with children) leads to an over-representation of women in the entire sample. The correction of that overrepresentation, however, also influences the fraction of women in two person households, where women are not easier to reach (see also Schnell 1993). Similar explanations might be made for quite a few other surveys, where the adjustment makes the subsample worse – albeit still within the confidence bounds around the true value. For a more thorough quality-assessment of the five European surveys, the information in Figure 17.3 needs to be summarised. Therefore the same logic was applied as for the external representative checks. Table 17.6 displays the mean difference between the observed and true values, as well as the fraction of countries with observed values outside the confidence bounds. For the overall quality index, both quantities have been mirrored, and the average differences were rescaled into values between 0 and 1. The maximum value of the overall quality index would be therefore again 2. Given the quality index, the ESS is the best survey programme. The ISSP, the EVS are somewhat less good, with similar values for the quality index. The EQLS and the
Assessing the quality of European surveys
EB 62.1
EQLS 2003
ESS 2002
Austria Belgium Denmark Finland France Germany Greece Ireland Italy Luxembourg Netherlands Portugal Spain Sweden United Kingdom Cyprus Czech Republic Estonia Hungary Latvia Lithuania Malta Poland Slovakia Slovenia Bulgaria Croatia Romania Turkey
.3 EVS 1999
.4
.5
ISSP 2002
Austria Belgium Denmark Finland France Germany Greece Ireland Italy Luxembourg Netherlands Portugal Spain Sweden United Kingdom Cyprus Czech Republic Estonia Hungary Latvia Lithuania Malta Poland Slovakia Slovenia Bulgaria Croatia Romania Turkey
.3 Random
.4
.5
.6
Quota
.7
.3
.4
.5
.6
.7
After weighting
Figure 17.3 Proportions of women among gender heterogenous couples Source: Own calculations. Do-file: anintern01_1.do
.6
.7
419
420
Ulrich Kohler
Table 17.6 Mean difference from the true value and fraction of countries with values outside the confidence bounds among gender heterogenous couples, by survey programme Study
Average difference
Fraction outside confidence bounds
Quality index
0.03 0.04 0.04 0.05 0.05
0.16 0.28 0.31 0.39 0.44
1.28 1.03 1.01 0.61 0.61
ESS 2002 ISSP 2002 EVS 1999 EQLS 2003 EB 62.1
Source: Own calculations. Do-file: ansummary_1.do
Eurobarometer lagged behind. The quality differences between the surveys are however less pronounced, as with the external representative checks. The average differences between the observed and the actual fractions of women vary between three and five percentage points, and the proportion of countries outside the confidence bounds is 44 per cent at most.
6. Summary I have reviewed the sampling quality of five European comparative survey programmes by assessing their documentation, their sampling process, and their conformability with external and internal criteria of representativity. Quality indices were developed for each of the four dimensions. Table 17.7 gives a summary of these four quality indices. In addition, the table also shows a measure of the overall quality of the sample, which summarises the four quality indices.15 The quality assessment has a clear winner, the European Social Survey. The ESS has the best documentation, uses the best sampling process, and shows the least deviations from internal and external criteria of representativity. The EVS is second best with good documentation, and a good performance on the criteria for representativity. The ISSP and the EQLS rank third and fourth, with very similar overall quality. The Eurobarometer lags behind. However, as far as the overall quality of the survey is concerned, the sample quality is only one dimension out of four. As pointed out in the introduction, it is usually not possible to raise the quality of any of those four dimensions at the same time. This is also the case for the European Social Survey. For the time being, one trade-off of the high ESS sample quality is the limited country coverage. Country coverage is not only limited as far as the number of countries is concerned, but also because of the unsystematic way in which countries are selected for participation. Since the countries themselves have to find funds for their participation, the composition of countries depends on a self-selection process. This self-selection resulted in an under-coverage of countries of the European Union as well as an over-coverage, due to the participation of Israel, Switzerland and Norway. In round 2 of the survey, the under-coverage has been molten somewhat, although it still exists, and in future rounds the self-selection mechanism may also lead to a loss of some European countries. Similar considerations can be made for all survey programmes. The EVS and the ISSP, for example, are quite complete regarding the country coverage, but both make painful compromises regarding the quality of the sampling process. Looking at the European Quality of Life Survey one has to concede a relatively strong over-representation of
Assessing the quality of European surveys
421
Table 17.7 Overall sampling quality of survey programmes
ESS 2002 EVS 1999 ISSP 2002 EQLS 2003 EB 62.1
Documentation
Sampling process
External criteria
Internal criteria
Sample quality
3.90 3.40 2.30 3.00 2.00
1.80 0.60 1.00 1.50 0.20
1.33 1.02 0.54 0.21 0.48
1.28 1.03 1.03 0.61 0.61
2.88 2.08 1.69 1.66 1.11
women, or, more generally, of easy-to-reach respondents. Its strengths in terms of the sampling process are outweighed by weaknesses revealed by the external and internal criteria of representativeness. This is unfortunate, as the EQLS has its special merits regarding the country coverage and timeliness. These latter features make the EQLS almost the natural choice for descriptive studies of the European Union, but at the same time such descriptive studies suffer most from the deviations from randomness. Analytical questions, like those presented in this volume, are probably less affected, as the fact that there are too many women in the sample does not automatically mean that observed differences in the behaviour or attitudes of men and women are also affected. For future rounds of the EQLS, the European Foundation might invest in improvements in the sample quality. A first glance at the overall measures relating to the quality of the sampling process reveals that the EQLS has in fact a relatively good sampling process. Does this mean that good sampling processes are useless? Not quite. The example of the ESS in particular shows that high investments in survey methodology payback. Moreover a closer look at the figures presented in the preceding sections shows that the internal criteria for representativity are better for the countries where the EQLS had used individual registers as sampling frames. Results not presented here indicate that sampling from individual registers generally produces better samples than those where the interviewers are involved in the selection of respondents (Kohler 2007). These results indicate, that the most important distinction is how far the interviewers are involved in the sampling process. For future rounds of the EQLS, efforts should be taken to diminish the influence of the interviewers on the sampling progress and to enlarge the power of the survey administration to control the interviewers. I would therefore propose the following changes to the general sampling design: ●
●
Bypass the random route light procedure wherever possible. As shown in Figure 17.1, this seems possible in the following countries, in addition to those where other advisable procedures are already in place: – Denmark, Finland, Sweden and Malta (Simple random samples); – Austria, Belgium, Germany and Slovenia (Multistage with individual registers); – Greece, Luxembourg, Netherlands, Portugal, Spain and the United Kingdom (multistage address samples) Hence, random route seems to be necessary only in France, Cyprus, Lithuania, Slovakia and Bulgaria. Address random route might be considered in these countries. Insist on institutionalised back-checks, which are promising especially for the non-random-route samples.
These changes would affect the other goals in the survey. Most obviously, timeliness will be harder to achieve, because it will take longer to reach the necessary number of
422
Ulrich Kohler
observations if the interviewers cannot evade the easy-to-reach respondents. However, as there is no comparable deadline, as was the case with the European enlargement for the first round of the EQLS survey, timeliness will be a less important goal in the next round. Given that the proposed sampling methods are more costly, the changes might also affect the goal of covering all EU members and candidates. Because this goal is sacrosanct, how far one can go in the proposed direction depends on the concrete costs of the respective sampling methods. This chapter is not the place to give advice about these necessary decisions. I would only like to remind all of us that survey data used to advise policy makers should be of very good quality. Questionable survey data does usually not hurt anybody, but if inaccurate social research influences practical policy, it will.
Notes 1 It is a common feature of European comparative survey programmes that sample sizes are lower in small countries. It should be noted that the reasons for this decision are financial. The sample size should generally be high enough to allow analyses with subgroups and to get estimates with reasonable confidence. It is obvious that the possibility of dividing the sample into subgroups is not related to the population sizes, and it is also well known that the population size does not influence the calculation of the standard error or confidence intervals for samples below 5 per cent of the population (Chochran 1953: 17; Kish 1965: 44). That is, even for the smallest European country – Malta, which has an overall population above the age of 14 of around 315,000 – a sample could have around 16,000 observations before the so-called ‘finite population correction’ makes any difference for the calculation of the confidence interval. 2 All datasets except the EQLS and the ESS are available at the Central Archive for Empirical Social Research, University of Cologne. The study numbers (ZA-Nr.) are: EB 62.1: s4230; ISSP 2002: s3880; EVS 1999: s3811. The EQLS 2003 is used by permission of the European Foundation for the Improvement of Living and Working Conditions, Dublin. The ESS is publicly available via the homepage of the European Social Survey on http://www.europeansocialsurvey.org. 3 In multistage or clustered samples the effective sample size is generally much smaller than the number of observations. The decreasing factor depends on the size of sampling units and the similarity between persons within a sampling unit (Kish 1965: 187–190; Schnell and Kreuter 2005). In effect the ESS therefore has to collect more observations if the sampling method is a multistage cluster sample than if it is a simple random sample from a population register. While it is quite worthwhile to follow the ESS approach concerning the effective sample sizes, there is still no reasonable justification for collecting less observations in smaller countries (see note 1). 4 I will not deal with stratified samples here. In practice, most samples involve microstratification, which, if correctly applied, will lead to ‘better’ samples. 5 The term ‘sampling frame’ is used for a list of research units from which a sample is drawn. 6 The percentage of respondents selected for quality back checks in Slovenia and Luxembourg is given as 100 per cent. I have treated this extraordinarily high percentage as false. 7 The sources of information about the survey organisation are as follows. EQLS: Ahrend (2003); ESS: European Social Survey (2004) EVS: Information from the methodological questionnaires, downloadable from http:www.za.uni-koeln.de/data/add_studies/kat50/EVS_1999_ 2000/ZA3811fb.pdf; ISSP: Klein and Harkness (2004). Eurobarometer 62.1: ebs_215_en.pdf, which is available in the archives at http://europa.eu.int/comm/public_opinion. 8 The Danish Central Person Register has a 99.9 per cent coverage of all persons who expect to stay in Denmark for at least 3 months. Similar registers also exist in Sweden and Finland. 9 The statement above rests on the assumption that the sampling frame is in fact well suited to drawing a sample. This assumption is questionable for sampling frames with a low coverage of the target populations. Sampling from registers is however somewhat more expensive than random route light, and so there is a natural barrier against applying unsuitable sampling frames instead of random-route. 10 For the EVS the information on substitution for Estonia is therefore questionable. 11 For the other survey programmes it is not known whether random route light or address random route were applied. The fieldwork documentation tends to document the prescribed numbers of contacts only, which is far less important information.
Assessing the quality of European surveys
423
12 The confidence interval has been calculated using the assumption that a simple random sample with the size of the respective sample has been drawn from the population. It should be noted that most samples of the survey programmes actually used a complex sample design, so that the effective sample size might be remarkably lower. A design-based calculation of the standard errors would lead to wider confidence intervals than those used in the figure. However, this effect can be considered as negligible when considering a pure demographic variable like gender. 13 The ESS contains so-called design weights. Design weights are weights that correct for unequal sampling probability given by the sampling design. Unlike representative weights they do not attempt to correct the self-(de)selection processes that commonly arise in samples for surveys. 14 The definition of the subgroup in the sample is, however, just a workaround. The true fraction of women in the subpopulation might deviate from 0.5 because: ●
●
Marriage between homosexual partners has become legal in some survey countries. It might be that homosexual marriage is gender specific, i.e. more frequent among men than among women (or vice versa). The drop-out from the sampling population might be gender specific. Some survey programmes, for example, restrict themselves on the eligible population or on ‘citizens’ of the respective countries. It might be that for gender heterogenous couples in 2-person households females are more often excluded from the sampling population (or vice versa).
All in all the effect of both mechanisms is believed to be small, however. 15 The quality indices were divided by their theoretical maximum and summed up to obtain the overall quality.
References Ahrend, D. (2003) The Quality of Life Survey: On Behalf of the European Foundation for the Improvement of Living and Working Conditions, Fieldwork, Hilversum: Technical Report of Intomart GfK. Chochran, W.G. (1953) Sampling Techniques, New York: Wiley. European Social Survey (2004) ‘ESS Documentation Report 2002/03: The ESS Data Archive, Ed. 5.1’. Online. Available http: (accessed February 2007). Kish, L. (1965) Survey Sampling, New York: Wiley. Klein, S. and Harkness, J. (2004) ‘ISSP study monitoring 2002: report to the ISSP General Assembly on monitoring work undertaken for the ISSP by ZUMA, Germany’, ZUMA Methodenbericht 2004/10. Kohler, U. (2007) ‘Surveys from inside: An assessment of unit nonresponse bias with internal criteria’, Survey Research Methods, 1, 2: 55–67. Popper, K.R. (1994) Logik der Forschung, 10. Auflage, Tübingen: J.C.B. Mohr. Schnell, R. (1993) ‘Die Homogenität sozialer Kategorien als Voraussetzung für “Repräsentativität” und Gewichtungsverfahren’, Zeitschrift für Soziologie, 22: 16–32. Schnell, R. (1997) Nonresponse in Bevölkerungsumfragen. Ausmaβ, Entwicklung, Ursachen, Opladen: Leske und Budrich. Schnell, R. and Kreuter, F. (2005) ‘Separating interviewer and sampling-point effects’, Journal of Official Statistics, 21, 3: 389–410. Sodeur, W. (1997) ‘Interne Kriterien zur Beurteilung von Wahrscheinlichkeitsauswahlen’, ZA-Information, 41: 58–82. United Nations Statistics Division (2002) Demographic Yearbook 2002. Online. Available http: http://unstats.un.org/unsd/demographic/products/dyb/dyb2.htm (accessed February 2007).
Index
accession process 3–4, 21, 131, 163 Accession Treaty (2003) 129–30, 176 acquis communautaire 9, 176, 192 Allen, J. 268 amoral familism 297, 299 Amsterdam Treaty (1998) 175, 201 Andrews, K. 259, 266, 270 anti-poverty policies 230–1 Atkinson, A. 210 Baldwin, P. 74 Ball, M. 261, 271 Barbagli, M. 47 Barbier, J.-C. 230 Baumol, W.J. 144 Beck, U. 330, 388 belonging: role of social support systems 315–19, 324; sense of 129, 213, 304, 306, 307–15 benefits 73, 82, 88, 106–7; income-tested 107–8, 109; maternity and parenting 108, 109, 111–13, 114–17, see also family allowances Billari, F. 29 birth rate 27, 28, 32, 105, 119, see also fertility; ideal-actual fertility gap; total fertility rate (TFR) Blair, Tony 5, 6 Blanchard, O. 153 Blunt, A. 269, 270 Boeri, T. 355, 356 border controls, elimination of 5 Brown, W. 190 Brücker, H. 355 budget negotiations, difficulties in consensus building 5–6 business services 142–3, 144 Caplow, T. 330 care provision (for elderly): age-specific perceptions 75, 76–87, 93–5; country-specific perceptions 80, 87; preferences on financing 83–7, 95; preferred arrangements 77–83
Castells, M. 330 Catholic church, family policies 12, 101–2, 103, 104, 109 Cavalli, A. 56–7 Chambers, R. 210 Chesnais, J.-C. 27, 28 child poverty 117–18, 119 childcare services 12, 13, 43, 106, 108–9, 113–15, 116, 119–20; communist countries 162; post-communist countries 163; preschools 109, 113 church–state relations 101–2 citizenship 6, 22, 162, 163, 230, 386; equal rights 6–7, 387–8 civicness 279, 297 cleavages see group conflicts Cleland, J. 31 collective bargaining 14, 178–84, 193; and customisation of agreements 181; and flexibility 179–82; and minimum wage regulation 183–4; and rising pay inequality 182–3; sectoral agreements 181, 182, 185–6, 187 collective solidarity 279 confidence see trust Common Agricultural Policy (CAP) 4 communism: collapse of 130, 139, 162–3, 173, 175, 235, 247, 252; and employment 162; and family policies 104–5, 162 compulsory altruism 83 conflict see group conflicts conflict theory 329, 340 Constitutional Treaty 3, 175, see also European Constitution construction systems 16, 254, 264–8; inefficiency 267–8 Consumer Price Index 222 consumption-related services 140, 142, 143–4 Copenhagen criteria 9, 21 country rankings 388; accuracy of citizens’ perceptions 394–7; citizens’ perceptions 391–4; impact of perceptions on individual well-being 397–9
426
Index
current life-style deprivation (CLSD) 203 Dahrendorf, R. 329, 330, 337 datasets, quality of 406–8 De Haan, A. 210 deprivation 129, 201–14; economic stress 203; income and 202, 203–6; and income poverty 207–9; life-style 203, 205–6; multidimensional approach 209–13; ten-item deprivation index 203, 205, see also poverty; social exclusion disadvantage see deprivation Diani, M. 279, 280–1 Docherty, P. 193 dual-career couples see two-worker couples Dunleavy, P. 258 Durkheim, E. 6 economic development: and employment 145–6; and group conflicts 336, 340; and quality of life 19–20 economic exclusion: measuring 209–13, see also social exclusion economic stress see deprivation education: effect on household patterns 69; effect on ideal-actual fertility gap 31–2; and intention to migrate 361, 365, 366, 371; and perceptions of group conflict 344 elderly: care provision 11, 62, 73, 74, 75, 76–87; household patterns 58–62, 64, 67, see also older workers; pension systems employment 12–16, 21, 129–57; agricultural sector 139–40; contracts 177–8, 182; and economic development 145–6; EU targets 132–3; full employment 13, 21, 73, 129, 144, 146–7, 149, 162; group-specific patterns 135–9, 146–50; impact of welfare policies 13, 147, 150–7; industrial sector 140, 142; informal sector 138; and job satisfaction 167–8, 169–72; job security/insecurity 167, 172, 182; levels and trends 130–4; part-time work 65, 147, 164, 178, 182, 190; promotion prospects 169, 172; regulation 163–4, 175–8, 187, 189, 192; sector-specific patterns 139–44; service sector 140, 142, 143–4; social pacts 185–6; stress 169; tripartite councils 186, 187; unemployment assistance 152–4; and voluntary social policy agreements 188–9; working conditions 13, 164–73; working hours 164–6, 178; youth unemployment 13, 138–9, see also labour markets; unemployment EMU (Economic and Monetary Union) 8, 175, 184, 187–8, see also single currency enlargement: consequences 3–9; and diversity in family policies 100; and employment 12; and labour institutions
175–93; and mobility intentions 373 Esping-Andersen, G. 16–17, 151, 290, 298 ethnic tensions 344–6, 348 Eurobarometer (EB) 7, 8, 9, 19; and care for elderly 76, 77–8; and ideal family size 30; mobility intentions 357; perceptions of well-being 389, 397–8; quality assessment of 407, 411, 413, 414, 416, 417, 420; and social exclusion 305, 307 European Commission Treaty 176 European Community Household Panel (ECHP) 62, 64 European Constitution 4; referendums on 7, 8 European Data Service 82 European Employment Strategy 163 European Environmental Agency 247–8 European Foundation for the Improvement of Living and Working Conditions 9, 30, 337 European Metalworkers’ Federation (EMF) 188 European Parliament, elections 7 European Quality of Life Survey (EQLS) 9; deprivation 202–3; generational tensions 76; group tensions 17; housing 235–6, 239, 241, 251; patterns of family living 49, 51, 52, 56, 62, 64, 65, 67; and pension systems 76, 87–8; quality assessment of 406–7, 410–11, 412, 413, 414, 416, 417, 418–21; sociability 280, 285; social cohesion 329, 330, 332, 346, 347; working conditions 164 European Social Model 8, 21, 163, 218, 232, 337; and employment 129, 150 European Social Survey (ESS) 19, 285; quality assessment of 407–8, 410, 412, 414, 416, 417, 418, 420, 421 European Trade Union Confederation 176 European Union, Europeans’ attitudes to 6–8 European Values Study (EVS) 6–7, 19, 30, 65, 285; quality assessment of 408, 410, 412–13, 414, 417, 418, 420 European Works Council (EWC) directive 192–3 Europeanization 18–19, 22 Eurostat 51, 167, 203, 209, 388 exclusion see social exclusion Fahey, T. 33, 208, 209, 211, 259, 388 family: formation 29, 31; size 27–8, 30–3, 36–7, 39–43 family allowances 108, 109–11, 116–17, see also maternity and parental benefits family life 10–11, 20, 47–70; communitarian families 101, 102, 105; and diversity 47–9; kinship ties 48, 52, 53, 58, 101, 104; and labour force participation 65–7, 69; nuclear families 48, 101, 102, 104, 105; patterns in EU 29, 49–53; solidarity
Index 75, 80, 83, 93; stem families 48, 101, 102, 104, 105; supportive role of family 281–5, 289–90, 297; young people 53–8, 67, see also generational relations; household patterns family policies 12, 100–20; autonomy model 104, 114; and birth rate 28, 42–3; and convergence 105–6; and extended family systems 120; families of nations of 100–6, 102, 104, 117–18; familism model 104, 117, 119; and family–work relationship 115–16, 117; hypothetical scenarios for NMS 105–6, 119; and individualisation of internal relationships 102–4; in new member states 104–6; in ‘old’ Europe 102–4; social expenditures 106–9; and state-church relations 101–2, 113, 119; state-provided family income 116–17; subsidiarity model 12, 104, 114, 116, 117, 119, 120; universality model 102, 104, 105, 116–17, 120; working family model 104–5, see also childcare services; family allowances; maternity and parenting benefits fertility 10, 27–44, 49, 50, 117–18, 119; aspirations 27–43; data 29–31; effect on pension systems 91; and pronatalist policies 28, 42–3; role of education 31–2, 37–40, see also ideal–actual fertility gap firms, and workers’ representation 192–3 Förster, M. 209 Fourastié, J. 144 Freedom House 4 Galland, O. 56–7 gender: differences 49, 53, 58–60, 62, 64, 80, 166–8, 171–2; and intention to migrate 361, 371, 372; relation 47–9; and representation in surveys 414–16, 417–18, see also women generational relations 11, 74, 76, 85–7, 92–3, 95–6; conflicts 74, 76, 85, 93, 95–6 Gerhards, J. 6 German reunification 181, 385, 397, 401 Gershuny, J. 144 Gibb, K. 260 Gogh, Theo Van 328 Goody, J. 101 Grande, E. 388 group conflicts 328–48; characteristic tensions 330–1; country differences in tensions 336–42; distributional 328, 330, 334, 335, 336–7, 344, 347; ethnic 328–9, 332–5, 339–42, 344–5, 346, 347; horizontal 332–4; levels of perceived tensions 332–5; and social change 330–2, 337, 347; and social cohesion 329–32, 346; and trust 330, 331, 339; vertical
427
330, 332–5, 336–9, 342–4, 346, 347; within-country differences in perception 342–6, see also social cohesion Haas, E. 385–6 Hajnal, J. 48, 52, 58, 60, 101, 104 Hanson, G. 356 Hantrais, L. 64 Health and Safety legislation 163 Hegedüs, J. 255, 257, 258, 267, 270 Hibbs, D.A. 182 home ownership 15–16, 235–9, 249, 250, 251, 252, 255; and intention to migrate 361, 365, 371 Hondrich, K.-O. 330 household patterns 51–3, 101; age at exiting parental household 49, 53; effect of education on 69; elderly 58–62, 64, 67; labour force participation 65–7, 69; in middle adulthood 62–5; neo-locality 67; single-person households 51, 52, 62–3, see also family life housing: age of housing stock 264–6; conditions 235–52; construction systems 16, 254, 264–8; East European model 16, 255–7, 258, 269; finance and subsidy systems 16, 254, 257, 258, 259–64; and general life satisfaction 250–1, 252; governance arrangements 16, 254, 268–70; high-rise dwellings 266–7, 270, 273; housing deficits 241–6, 252; institutional drivers of inequalities 254–73; living space 239–41, 242, 260; and local environment 247–9; mortgage lending 261–4, 270, 273; policy legacies 246, 252; privatisation 16, 238, 252, 255, 257, 258–9; and social trust 251–2; and structural funding 273; and subjective satisfaction 249–52; tenure 16, 235–9, 249, 250, 251, 254, 255–9, see also home ownership; social housing sector Human Development Index (HDI) 210, 394, 396 Huzzard, T. 193 Iacovou, M. 62 ideal–actual fertility gap 10, 27–8, 29–43; assessing 32–6; composition of 36–7; data 29–31; role of education 31–2, 37–40, 42; trends since 1981 40–1 immigration: and conflict/tensions 328, 340, 344–6, 347, 355; and restrictive policies 355; as threat to social cohesion 329 income: and deprivation 202–6; inequalities 207–9, 311; and perceptions of group conflict 344 income support 15, 58, 220, 221, 230, 232 Inglehart, R. 330
428
Index
integration 1, 3, 173, 175, 202; and citizens’ consent 7–9; differentiated 5; and immigration 329, 334, 335; political 386, 388; social 12, 119, 139, 202, 230, 304–25, 385–6, 399 International Institute of Labour Studies 153 International Labour Organisation (ILO) 178, 186, 188, 219 International Monetary Fund (IMF) 4, 210 International Social Survey Programme 2002 (ISSP) 19, 285; quality assessment of 407, 408, 411, 412, 413, 414, 416, 417, 418, 420 Iraq War 4 Iversen, T. 144, 182 Job Seekers Allowance (UK) 220 jobless growth 130, 145–6 Kaitz index 184 Kalniete, S. 8 Kemeny, J. 255, 260 Kertzer, D. 47 Kilpatrick, C. 189 Klaus, Vaclav 3 Kleinman, M. 7 Kozlowski, E. 260, 270, 273 Krieger, H. 357 Kunovich, R.M. 344 labour markets 129–57; access to 6–7, 12–13, 129, 306; changes in 176–8; and EU legislation 192–3; exclusion 137–9; free movement of workers 3, 13, 18, 176, 355; regulation 175–6, see also employment; unemployment Laeken indicators 210, 304 Lamers, J. 193 Land, H. 83 Laslett, P. 101 latent class analysis 14, 210–13 Le Play, F. 48 legitimacy crisis theory 251 Leisering, L. 96 life expectancy 80, 396; and deprivation 105, 210, 271; and patterns of family living 52, 59, 62, 64; and pension systems 73, 74, 91 life satisfaction 250–1, 252, 388, 390, 397–9 Lisbon European Council, agenda 21, 27, 129, 132, 149, 163, 201–2, 231, 360 Lisbon strategy 149 Locking, H. 182 long-distance geographical mobility (LDGM) 382 357, 360–72, 371, 373, 382 low-skilled workers 13, 135, 137, 138, 147, 149
Maastricht Treaty (1992) 175, 188, 201, 304 McCormick, B. 356 Maclennan, D. 259, 268 Maitre, B. 213 marginalisation 17, 20–1, 307–8, 311; feelings of 305, 306–7, 316, 317, 323–4; polarisation of perception of 325 Marginson, P. 187 marriage 47, 48, 49–51, 101; age at 49–50; break-up 50–1 Marx, K. 342 Marxism 328, 329 maternity leave 12, 43, 111 maternity and parenting benefits 108, 109, 111–13, 114–17 Meijer, F. 266 Merrill, S. 270, 273 migration 18, 138, 163, 176, 355–83, see also mobility intentions minimum income policies 15, 218–32 minimum wage 14, 154, 179, 182, 183–4 Mitterauer, M. 101 mobility intentions 355–83; changes in 366–70, 371–2; and country groupings 381–2; disincentives 366; inter-country 359, 360, 361–70, 372–3; inter-regional 359, 361, 370–2; levels of 357; mobility culture 18, 356–7, 382; and ‘push’ conditions in sending countries 373–9, 382; ‘push’ factors 365, 372–82; structural patterns 360–72; subjective conditions as ‘push’ factors 379–81; trends in 358–60 Moisio, P. 210 motherhood 50, 56, 65, 66–7 mothers, partnership situation 50, 53 multistage sampling 408–11, 412; randomroute sampling 408, 409, 410, 412, 421; register sampling 408–9, 410 Muziol-Weclawowicz, A. 269, 270 National Action Plans (NAPs) 231 neo-locality 67 OECD (Organisation for Economic Co-operation and Development) 113, 132, 151, 152, 154, 202, 223, 231 older workers 135, 137, 138, 147–9 Open Method of Coordination (OMC) 15, 219, 231, 232, 405 Padoa-Schioppa, Tommaso 8 parental leave 112, see also maternity leave Paugam, S. 279 pension systems 11, 73–6, 87–93, 95–6; age-specific perceptions 75–6, 87–93, 95; and demographic changes 73–4, 91, 92; and inter-generational tensions
Index 92–3; pay-as-you-go schemes 73, 129; trust in 11, 76, 87–93, 95, see also retirement perceptions of well-being 385–401; ability to make country–EU comparisons 390–1; data on 389–90; reference group theory 385–6, 387–9, 390 Persson, Goran 4 PHARE programme 186 Pinchler-Milanovich, N. 264, 269, 271 Pizzorno, A. 297 political participation 285–8 population: decline 27, 29; replacement 27, 29 post-communist countries: employment/unemployment 130–2, 138, 162–3; environmental concerns 247–8; family responsibility 87; group tensions 334, 337–9, 346–7; housing conditions 235; and housing sector governance 269; and housing tenure 258–9; and migration 163; and US as model 4 Posted Workers Directive 176 poverty 3, 4, 14–15, 73, 218, 220–1; and income 207–9; and social cohesion 329; and social exclusion 17, 20–1, 323–4, see also deprivation; social exclusion primary sampling units (PSUs) 408–9, 411 privatisation 4, 162, 163, 175, 187; housing 16, 238, 252, 255, 257, 258–9 production-related services 140–1, 142, 143 Protestantism, and family policies 102–4 Purchasing Power Standards (PPS) 203, 238 Putnam, R. 251 quality of work see working conditions quota sampling 408, 410, 411, 412, 418 redistribution 6, 7, 11, 218, 290, 386, 401 reference group theory 385–6, 387–9, 390 Reformation 102, 104 regulation see employment regulation Reher, D. 48 relative deprivation see deprivation relative income poverty see poverty retirement 76; early 147–9; and sense of belonging 315, see also pension systems Roberts, A. 258 Rose, H. 83 Rumsfeld, Donald 4 Russel, H. 279 safety nets see minimum income policies sampling process 408–22; back-checking 410, 421; comparing surveys 411–14; method 408–9; multistage sampling 408–11, 412; quota sampling 408, 410, 411, 412, 418; response rates 409–10; simple random sampling (SRS) 408, 410,
429
411, 412; substitution 409 Scharpf, F. 140, 144, 151 Schengen agreement 5 Schröder, Gerhard 4 Schumpeter, J.A. 332, 337 self-employed 166, 168, 171, 172, 177, 391 Sendi, R. 259, 266, 270 service sector: free movement of services 175–6; growth of 177 Services’ Directive (2006) 175–6 Sharp, E. 285 Sheridan, L. 266 Shigehiro, S. 270 Shinozaki, S. 261–4, 273 Sillince, J. 260, 267 Simmel, G. 329 simple random sampling (SRS) 408, 410, 411, 412 single currency 5, 175, 355, 387, see also EMU (Economic and Monetary Union) Single European Act 175 Sleebos, J.E. 28 Smyth, E. 388 sociability 279–99; disaffiliation 297, 299; patterns in country clusters 198, 288–96; political participation 285–8; private/public balances 296–8, 297, 298, 299; role of family 281–5, 289–90, 297; social participation 279, 281, 285–8, 289, 290, 305; spheres of 279–81 social assistance 152, 219–21; and activation 229–30; anti-poverty policies 230–1; benefit adjustments 222–3; categorical approach 219, 220; and decentralisation 222; diversity in programmes 221–3; eligibility 221–2; and entitlement 229; levels of generosity 223–9 social capital 16, 337, 347 Social Charter (1989) 201 social cohesion 6–7, 17–18, 202, 213, 328–48; definition of 329; threats to 328, 329–32, 346; and trust 213, see also group conflicts social dialogue 14, 163, 173, 186–7, 337 social dumping policies 13, 145, 152, 155 social exclusion 17, 201–2, 209–10, 304–25; definition of 305; and labour market attachment 312–15; and minimum income policies 218, 221, 230; risk groups 311–15; role of social back-up 316–19; and sense of belonging 307–11; and social cohesion 329; and societal conditions 319–23, 324, 325, see also deprivation; marginalisation social expenditure 82, 106–9, 150–1, 218 social housing sector 255, 257, 257–8, 259–60, 268–9; rent determination 270, 273
430
Index
social inclusion policies 230–1 social inequality 304–6, 322–3 social insurance 219, 220 social network support 305–6, 315–17, 324 social networks 306, 311, 315–19, 324 social pacts 14, 185–6 social participation 279, 281, 285–8, 290, 305 social policy 175, 231; and housing 254; implementation failure 189; and poverty/exclusion 201–2, 207, 208–9, 220, 306, 324; and voluntary agreements 188–9 social protection 218–19; and emergence of universal safety nets 219–21, see also social assistance Social Protocol 201 social security systems: age-specific perceptions of 75–96; and demographic changes 73–4 social tension see group conflicts social transfers 151, 152, 229 Spéder, Z. 33 state-church relations, and family policies 101–2 Stockholm European Council (2001) 132 Streeck, W. 189 Structural Funds 4, 5, 6, 133, 140, 271, 273 subjective well-being see life satisfaction subsidiarity 12, 75, 102, 104, 106, 114–16, 119, 120, 219 surveys: assessing quality of 405–22; country coverage 406, 407, 420–1; datasets 406–8; goals of comparative studies 406, 422; representativity 406, 414–20; sampling 408–22; timeliness 406, 422; topicality 406, 407 tax wedge 154, 155 tensions see group conflicts Therborn, G. 47 Tilly, C. 336–7 Todd, E. 101 Tosics, I. 255, 257, 258, 259, 260, 267, 270, 271 total fertility rate (TFR) 27, 32–3 trade unions 173, 180, 187–8, 288; and employment contracts 182; and group tensions 336–7; legislation 191; representation 189–93 Transparency International 4 tripartism 186, 187 trust 251–2, 279, 297, 330, 346, 390; and housing conditions 251–2; in institutions
319, 325; in pension systems 11, 76, 87–93, 95 Tsenkova, S. 254 Turner, B. 254 two-worker couples 65–7, 69 unemployment 73, 137–9, 147; and intention to migrate 361, 365, 366, 371, 372–9; jobless growth 130, 145–6; long-term 138 UNICEF 132 United Nations: Economic Commission for Europe 261, 273; Fertility surveys 50; Human Development Index (HDI) 210, 394, 396 unskilled workers: housing 242, 266; pay levels 166; and unemployment 147, 150, 151, 152, 154 Visser, J. 337 voluntary work 281, 285 vulnerability 210–14 wages 166–7, 172, 176–7, 182–3; wage coordination 187–8, see also collective bargaining; social pacts Washington consensus 4, 162 welfare regimes 280, 281, 290, 292, 298 welfare state 7, 15, 74–5, 218, 290; as antidote to group conflicts 337, 339; effect on attitudes to care for elderly 77, 82–3; and employment 147, 150–7 well-being see perceptions of well-being Whelan, T. 213 women: as care providers 77; and education 69; employment–fertility relationship 31–2; and labour markets 49, 65–7, 69, 74, 135, 138, 147, 162; representation in surveys 414–16, 417–18; wages 166; and working hours 164, see also gender; motherhood; mothers working conditions162–73 Working Tax Credit (UK) 220 World Bank 4, 157, 210, 258 Wren, A. 144 yardsticks of comparison see reference group theory young people: and employment 147–9; and labour market 135–7, 138–9; patterns of family formation 53–8, 67