AIEL Series in Labour Economics
For further volumes: http://www.springer.com/series/7370
Tindara Addabbo Giovanni Solinas •
Editors
Non-Standard Employment and Quality of Work The Case of Italy
Prof. Tindara Addabbo Department of Economics University of Modena and Reggio Emilia Viale Berengario 51 41121 Modena Italy
[email protected]
ISSN 1863-916X ISBN 978-3-7908-2105-5 DOI 10.1007/978-3-7908-2106-2
Prof. Giovanni Solinas Department of Economics University of Modena and Reggio Emilia Viale Berengario 51 41121 Modena Italy
[email protected]
e-ISBN 978-3-7908-2106-2
Springer Heidelberg Dordrecht London New York Library of Congress Control Number: 2011937432 Ó Springer-Verlag Berlin Heidelberg 2012 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Printed on acid-free paper Physica-Verlag is a brand of Springer Springer is part of Springer Science+Business Media (www.springer.com)
Acknowledgment
This edited volume has been made possible with the contribution of Associazione Italiana Economisti del Lavoro (AIEL).
v
Contents
1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tindara Addabbo and Giovanni Solinas
Part I
2
3
4
1
Non-Standard Employment, Labour Supply and Living Conditions
A Microeconometric Analysis of Female Labour Force Participation in Italy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Massimiliano Bratti and Stefano Staffolani
25
Balancing Work and Family: New Mothers’ Employment Decisions During Childbearing . . . . . . . . . . . . . . . . . . . . . . . . . . Andrea Neri, Martina Lo Conte and Piero Casadio
39
Part-Time and Temporary Employment: A Gender Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tindara Addabbo and Donata Favaro
53
Part II
Non-Standard Employment, Workers’ Mobility and Labour Market Structure
5
The Quality of Temporary Work . . . . . . . . . . . . . . . . . . . . . . . . Gianna Barbieri and Paolo Sestito
6
Temporary Help Workers in Italy. Where Do They Come From and Where Do They Go? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Federica Origo and Manuela Samek Lodovici
79
105
vii
viii
7
8
Contents
The Dynamics of Unemployment, Temporary and Permanent Employment in Italy. . . . . . . . . . . . . . . . . . . . . . Matteo Picchio
127
Labour Market Segmentation, Flexibility and Precariousness in the Italian North East. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Giuseppe Tattara and Marco Valentini
149
Part III
Part-Time Employment, Working Conditions and Job Attributes
9
Quality of Work: The Case of Part-Time Work in Italy . . . . . . . Brendan J. Burchell
175
10
Does Part-Time Mean Part-Satisfaction? . . . . . . . . . . . . . . . . . . . Sara Depedri
189
11
Workers’ Lifestyle Choices, Working Time and Job Attributes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Giovanni Russo and Edwin van Hooft
Part IV
12
13
205
Non-Standard Employment, Quality of Work and Firms’ Outcomes
Non-Standard Employment and Quality of Work: Towards New Forms of Measurement . . . . . . . . . . . . . . . . . . . . . Tindara Addabbo and Giovanni Solinas
233
Is There a Trade-Off Between Labour Flexibility and Productivity Growth? Some Evidence from Italian Firms . . . Federico Lucidi
261
About the Editors
Tindara Addabbo (e-mail:
[email protected]) is Associate Professor of Economic Policy at the University of Modena and Reggio Emilia (Italy). She is member of CAPP (Centre for the Analysis of Public Policies), CHILD (Center for Household, Income Labor and Demographic economics) RECent (Center for Economic Research) and of the Scientific Committee of Fondazione Marco Biagi. She has published in the areas of the gender impact of public and social policies, measurement of well-being in the capability approach, employment and wage discrimination by gender, income distribution and quality of work. Giovanni Solinas (e-mail:
[email protected]) is Full Professor of Economics at the University of Modena and Reggio Emilia (Italy). He is currently one of the directors of studies of the B.A. course in International Economics and Marketing (Faculty of Economics of Modena and Reggio Emilia), scientific director of the Library of Economics ‘Sebastiano Brusco’, member of the Scientific Committee of Fondazione Giacomo Brodolini (Roma), Fondazione Marco Biagi (Modena), ILO (Industrial Liaison Office, University of Modena and Reggio Emilia), and Capp (Centre for the Analysis of Public Policies). His main research interests are in industrial economics, labour economics and evaluation of public policies. In these fields he has been author and editor of books and had papers published in both national and international journals. He has been scientific director and researcher member in several European programs and projects.
ix
Contributors
Tindara Addabbo Department of Economics, University of Modena and Reggio Emilia, Italy, e-mail:
[email protected] Gianna Barbieri Ministry of Education, Statistical Unit, Rome, Italy, e-mail:
[email protected] Massimiliano Bratti Dipartimento di Scienze Economiche Aziendali e Statistiche, Università degli Studi di Milano, Milan, Italy, e-mail:
[email protected] Brendan J. Burchell Department of Sociology, University of Cambridge, Cambridge, UK, e-mail:
[email protected] Piero Casadio Economic Research Department, Bank of Italy, Rome, Italy, e-mail:
[email protected] Sara Depedri Department of Economics, University of Trento, Trento, Italy, e-mail:
[email protected] Donata Favaro Department of Economics M. Fanno, University of Padova, Padova, Italy, e-mail:
[email protected] Martina Lo Conte Social Statistics Department, Istat, Istituto Nazionale di Statistica, Rome, Italy, e-mail:
[email protected] Federico Lucidi Fondazione Giacomo Brodolini, Rome, Italy, e-mail: lucidi@ fondazionebrodolini.it Andrea Neri Economic Research Department, Bank of Italy, Rome, Italy, e-mail:
[email protected] Federica Origo Department of Economics ‘‘Hyman P. Minsky’’, University of Bergamo, Bergamo, Italy, e-mail:
[email protected] Matteo Picchio Department of Economics, CentER, ReflecT, Tilburg University, Tilburg, The Netherlands, e-mail:
[email protected]
xi
xii
Contributors
Giovanni Russo CEDEFOP, Finikas, Thessaloniki, Greece, e-mail: giovanni.
[email protected] Manuela Samek Lodovici Istituto per la Ricerca Sociale IRS, Milan, Italy, e-mail:
[email protected] Paolo Sestito Department for Structural Economic Analysis, Bank of Italy, Rome, Italy, e-mail:
[email protected] Giovanni Solinas Department of Economics, University of Modena and Reggio Emilia, Italy, e-mail:
[email protected] Stefano Staffolani Department of Economics, Marche Polytechnic University, Ancona, Italy, e-mail:
[email protected] Giuseppe Tattara Department of Economics, University Cà Foscari, Venezia, Italy, e-mail:
[email protected] Marco Valentini Tolomeo Studi e Ricerche srl, Treviso, Italy, e-mail: marco.
[email protected] Edwin A. J. van Hooft University of Amsterdam, Amsterdam, The Netherlands, e-mail:
[email protected]
Chapter 1
Introduction Tindara Addabbo and Giovanni Solinas
The recession following the worldwide economic crisis has also taken its toll on the Italian economy. National product has fallen by around 7 points: in 2009 product in terms of volume returned to the same level as it was the start of the millennium. This has had clear consequences on the labour market (Cingano et al. 2010). Between the second quarter of 2008 and the fourth quarter of 2009, the number of employed fell by 560,000 people, most of whom with fixed-term contracts or with forms of subordinate employment concealed as autonomous employment. The recovery in 2010 was marginal (+40,000 people). It is estimated that the number of hours of cassa integrazione (‘‘redundancy fund’’) claimed on an annual basis is equivalent to that of around 500,000 full-time workers. The unemployment rate has reached 8.7% of the labour force (10% among women), including workers without employment contracts, as estimated by Istat (the Italian national institute of statistics). If we make use of broader ranging indicators of the underuse of the labour force (which along with the unemployed, include workers drawing wages from the redundancy fund, those not actively seeking employment, being certain of not finding any), the underuse of the work force stands at over 11% of those potentially employable. This is less than France and Spain, but more than the UK and Germany (Bank of Italy 2011). Despite having gradually spread to all the forms of employment contract, the negative impact of the crisis on a number of non-standard positions has been particularly harsh: the Istat 2010 Labour Force Survey (LFS) shows that 63% of
T. Addabbo (&) G. Solinas (&) Department of Economics, University of Modena and Reggio Emilia, Viale Berengario 51, 41121 Modena, Italy e-mail:
[email protected] G. Solinas e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_1, Ó Springer-Verlag Berlin Heidelberg 2012
1
2
T. Addabbo and G. Solinas
the overall fall in employment may be attributed to fixed-term contracts and (autonomous) collaboration contracts: in 2009 alone, these forms of employment fell by 240,000 compared to a fall in the number of standard work contracts of 139,000. Compared to the average levels found in Europe during the crisis, the reduction of fixed-term employment appears to be more accentuated in Italy. Furthermore, during the crisis there appears to be a greater likelihood of leaving a non-typical job for unemployment or inactivity as well as a reduction in the job stabilisation rate (greater among men, despite the fact that the probability of moving into a permanent contract from a non-typical position is significantly lower among women). In this sense, even though for the first time after 4 years the incidence of nonstandard work on the employment total has fallen, the crisis tends to accentuate the dualism between those jobs with a high level of protection and the unprotected jobs to be found in the Italian labour market (Istat 2010). The unprotected segment in Italy is one of sizeable proportions. The most conservative estimates suggest that in Italy, the field of temporary labour in 2009 covered 12% of employees. If these are added to the irregular working positions— estimated by Istat at around 12% of the overall working contracts, the conclusion must be drawn that no fewer than one worker out of four belongs to the area of low or no protection (Istat 2011). There is an aspect in which the labour market conditions in Italy are among the worst in the highly developed nations: the concentration of precarious work and unemployment among the young. Today in Italy, with the crisis in full swing, little below one young adult out of three (28.9% in the fourth quarter of 2010) is unemployed. This is a young unemployment rate above that of Eurozone countries (which in November 2010 stood at 20.7% compared to 29.9% in Italy, Eurostat, 2011) and that, in a number of parts of the South, fluctuates around 40%. Only between 2007 and 2009, young unemployment rose by almost 10%: much more than in other advanced economies. This trend may not be put down to demography: the cohorts of young people entering the labour market, in relation to the active population, are far smaller in dimensional terms compared to other economies. And neither does it derive from the intensity with which the fall in recruitment levels has emerged during the recession. The insurgence of such a high level of young unemployment is down to the fact that the first jobs to go when companies need to make cuts are those with fixed-term contracts. Along with immigrant workers, young adults are the first to be made redundant. In other terms, just like in all the other economies with a twotier work market, the crisis is paid for by the weakest subjects and, among national workers, by the young. Fixed-term contracts, along with a wide range of forms of ‘bogus self-employment’, have become the most common entrance path to the labour market. Before the crisis, no less than 70% of recruitment took place using these forms of contract. Throughout the recession, in all likelihood this level has become even higher. Starting from the second half of the 1990s, the general diffusion of flexible contracts, supported by a number of legislative interventions, was certainly one of
1 Introduction
3
the factors that fostered the rise of employment at a rate which, in the previous decade, was greater than that of the main Eurozone countries. But in rather more favourable cyclical conditions than those currently to be found, it also ushered in what many came to see as a ‘normal’ situation of the alternation between work and non-work, i.e. between work and unemployment. Given than the system of social safety nets and redundancy pay varies on the basis of the previous employment position (in terms of company size and forms of contract), this alternation incurs higher costs for non-standard workers and people with intermittent working profiles. This is a phenomenon which has non-transitory effects on individual careers. For the very reason that the use of temporary forms of employment is so common (and levels of permanent employment are more or less static), the likelihood of moving from fixed-term to permanent contracts is relatively low: this passage concerns one contract out of 10 in the field of company employment, and much less in the sphere of para-subordinate work. Furthermore, the likelihood of stabilisation, which has fallen notably over the course of the crisis, particularly among males with fixed-term contracts, appears systematically lower among women (Istat 2010). At the same time, the likelihood of them losing their jobs is around eight times higher than that of other workers. But the risk is not simply that of a fragmented working career. If one considers that workers with fixed-term contracts, with similar individual characteristics, are paid on average 30% less than permanent workers, that many contractual institutions provide lower social security contributions, a low probability of transition from precarious work to stable work and ongoing experiences of unemployment outline a real danger of getting stuck in the mechanism for a substantial share of one’s working career. This may be found especially among low-income groups and those with very low protection standards, with obvious consequences on life choices, the possibility of starting one’s own family without depending on one’s family of origin, having or not having children and so on. What makes young people’s prospects even less bright is the effects on work training. Not only is the level of return of scholastic education lower among temporary workers than permanent workers, but they also receive less training on the workplace. For a young worker with a fixed-term contract, the likelihood of being trained on the job is 30% lower than that of a young worker employed on a permanent basis. Temporariness is a strong incentive not to invest in training, both for the company and for the worker. This worsens the situation for precariously employed young people on the labour market and contributes to reducing their perspectives of future income. The impact is strong not only on short and mid-term income expectations, but also on the expected pension income. The contributory system has stressed the weight of entrance conditions: a worker whose career is broken up over the course of a decade, even when at the end of his/her working life s/he manages to put together a good contributory seniority, his/her pension will be very low, often not far over the minimum pension (Berton et al. 2009). In the decade leading up to the crisis, as the increases in employment largely concerned women, the fragmentary career, the alternating between work and
4
T. Addabbo and G. Solinas
unemployment, less workplace training and the lowering of work and pension incomes are all phenomena that feature clear gender-based differences. The Istat LFS (2009 average) confirms that female employment is largely centred around non-standard work contracts. 14% of employed women, compared to 9% of men, have a fixed term or ongoing collaboration work contract (our own elaborations of Istat data, 2010, Table 3.3). Part-time work is to be found in particular among women (23% of employed women against 4% of employed men). When it is not desired, from many points of view, part-time work (along with undeclared work and fixed-term contracts) also constitutes one of the components of the secondary and largely unprotected market. During the crisis, the use of involuntary part-time has tangibly increased, while voluntary part-time has fallen (Istat 2010). In Italy, the greater likelihood of women being employed in non-standard working positions is associated with a lower likelihood of women contributing to the labour market at all. The female employment rate in 2003 stood at 42.7%, and despite increasing to reach 47.2% in 2008, it is still decidedly below the European average for the same year (59.1% in EU-27; European Commission, 2010). The crisis in Italy has led to a slower fall in male employment and a more rapid one in female occupation compared to the European average, thus contributing to further increase the gap between the Italian female employment rate and that in Europe. There is also a tangible territorial difference in female unemployment rates, ranging from a maximum of 57% in the North East to 30.6% in the South (2009 average). Furthermore, the issue of female inactivity has worsened, now standing at around 50%–13% more than the EU average (Istat 2010). Within the pool of inactive female workers, the crisis has led to an increase among the number of those most distant from the labour market, i.e. those not seeking employment and unwilling to start doing so. What’s more, the literature highlights the connection between the lower female labour supply, the scarce presence in our country of sectors that tend to offer female employment, the lack of childcare services and care for the elderly in the wake of the great inequality in the distribution of care work within the household. (Addabbo et al. 2010; Del Boca and Vuri 2007). The underlying element is the lack in Italy of those ways of working that make life and work time more reconcilable (such as voluntary part-time work), given the distribution of overall work time and the lack of childcare services and care for the elderly. As is known, greater female employment is associated with the greater spread of forms of contract such as part-time (Jaumotte 2003). In this sense, the rise in the demand for part-time work, which has increased over the last decade in Italy, may be considered an element which has contributed positively to the growth of female employment rates. In particular, the demand for part-time work may favour the provision of female work, in the presence of an ongoing mismatch between unpaid household working times by gender (Addabbo 2003), loosening the ties bound up with reconciliation difficulties that come to the fore more strongly where the time commitments of the main family carer increase. Nevertheless, an analysis of variation rates in part-time work between 1997 and 2007 shows how in Italy the rise of involuntary part-time work is dominant, even though
1 Introduction
5
in absolute terms the share of voluntary part-time work is greater (OECD 2010). Thus the risk remains that alongside the positive effect and the increased work supply, there may also be an increase in labour market segregation and inferior working conditions (Berton et al. 2009; Connolly and Gregory 2008a, b, 2009; Manning and Petrongolo 2008; Mumford and Smith 2009). This conclusion is in a certain sense strengthened by the fact that the lack of forms of work that encourage reconciliation between family and paid work time, especially among those subjects most heavily involved in care activities in certain stages of the life cycle, leads workers to exit the potential labour force. These are the themes that the volume will address.
1.1 Non-Standard Employment, Labour Supply and Living Conditions The volume picks up on the analysis of the conditions of labour supply, with particular reference to female labour supply and non-standard employment. The essays presented in the first part explicitly examine the factors that determine the likelihood of female employment (Bratti and Staffolani) highlighting the link between the phases of the life cycle in which there is the greatest concentration of family work (Casadio, Lo Conte and Neri) and part-time/fixed-term employment (Addabbo and Favaro). In general (and in line with the literature), the positive effect of institutional factors on the probability of female employment is confirmed, including the presence of childcare services and care for the elderly, allowing for the continuity of mothers’ working careers. Italy is characterised by one of the lowest birth rates in the world as well as a very low level of female employment. This defining feature of Italy is analysed by Neri, Lo Conte and Casadio, focusing on the ‘choices’ of those who find themselves in the most critical phase of their life cycle with regards both to the birth of their children and their own working careers, yet who have in fact already made a choice with regard to the birth rate issue: new mothers. Thus this is a selected sample (given the low birth rate and the difficulties inherent in reaching the desired number of children) of women mindful of their own working position and particularly sensitive to the occupational instability that young people have to face in the early years of their working lives (Addabbo 2005), instability which—as noted above—tends to continue for a relatively long time, and even more so in the wake of the current crisis. What are the working experiences of those who have just had a child? And what is the role, in terms of working continuity, of work-life balance policies? These are questions that can help to identify the policies and organisational schemes that, in the current framework of ever greater working instability during the period in which new families could be formed, thus raising the birth rate and narrowing the gap between the number of children desired and the actual number of children, and allowing for family planning choices which for many are currently out of reach. Neri, Lo Conte and Casadio use microdata from the Istat
6
T. Addabbo and G. Solinas
Italian Birth Sample Survey (IBSS) on mothers 18–21 months after the birth of their children. The comparison between the dataflow showing a trend towards inactivity among new mothers and those taken from the overall female population shows that new mothers are the most exposed to the risk of leaving the labour market. It thus becomes important to analyse the factors that positively influence the likelihood of mothers’ employment in a highly intense stage of the working life cycle, with fixed time costs due to the presence of children. In this analysis, the Authors deal with the issue of the non-random selection of the sample of new mothers and, given the non-availability in the dataset used of data regarding women without children, they make use of a broader sample extracted from the survey commissioned by the Bank of Italy on Italian families’ household income (Survey on Household Income and Wealth, SHIW) in order to carry out the test. Likewise, using the IBSS dataset, they carry out a test on the endogeneity between the decision to have more than one child and the mothers’ occupational choices. The factors that influence the likelihood of employment among new mothers are thus analysed using a bivariate probit model which estimates both the probability of new mothers’ not being in work 18–21 months after childbirth and the probability that the child attends preschool (a variable which is not included directly in the first model given the potential endogeneity between the occupational choices of the mother and the child’s attendance of preschool). The model confirms the existence of a strong negative correlation between the child attending preschool and likelihood of the mother being out of work. Furthermore, the estimate in this model confirms the positive role of the educational level of the mother as a means by which to significantly reduce the likelihood of being unemployed. An important factor in the likelihood of being out of work is also the mother’s previous working experience, guaranteeing a greater probability of finding employment after childbirth. This result is in keeping both with human capital literature and the greater fixed costs that employed women would be faced with in the wake of a career interruption. Furthermore, the probability of occupational continuity increases among women employed in the public sector, where there appears to be greater work life balance. The Authors therefore provide a second bivariate model to estimate the likelihood of a mother employed at the beginning of pregnancy who then left her job of her own accord and at the time of the interview is not employed, along with the likelihood of the child attending preschool. Among the results of the estimates, the greatest occupational instability may be noted among new mothers who before childbirth were employed with a fixed-term contract, as these appear to be more exposed to the risk of being out of work in the first model estimated, and to leave it of their own accord in the second. To what degree might public policies and the availability of part-time work contribute to the development of the supply of work for women? These issues are addressed by Bratti and Staffolani, who focus on Istat LFS microdata putting forward the results of an ‘ordered probit’ model on the likelihood of inactivity, as well as part-time and full-time occupation. The sample includes more than 40,000 women of a working age and allows for an analysis of the impact on the likelihood
1 Introduction
7
of their being in work and the occupational structure (part-time or full-time) of the socio-demographic structure of the family (the makeup of the nuclear family, the number and age of the children, the occupational structure, the presence of family members in need of care, and of inactive grandparents), of territorial variables (on the spread of part-time work and the state of the labour market) as well as individual characteristics (age and level of education). The great number of respondents in the sample and the significance of the data source used on a regional level also allows the Authors to include territorial variables on childcare and early care coverage ratios in order to check for the effect of the availability of services on a local level on female employment. Bratti and Staffolani confirm the positive effect of the presence of childcare services and assistance for the elderly on the female labour supply. The availability of part-time work increases the likelihood of women with small children participating in the labour market; furthermore, once territorial indicators on the state of the labour market and the availability of services are expressed within the model, the dummies on areas of residence do not appear to have any significant weight on the probability of participation. The territorial differences noted thus seem to be more linked to the state of the labour market (in this regard, note also the presence of a phenomenon discouraging female participation in the labour market in areas characterised by high levels of unemployment) and the presence of personal services clearly capable of reducing the fixed costs of entering and remaining in the labour market by those who in Italy appear to play the role of the main carer in the family. The presence of children under 15 or disabled people in fact greatly reduces the likelihood of mothers’ participation, particularly when there are children below 3 years of age—the age range in which there is the greatest lack of childcare services in Italy, and thus the greater the fixed costs are in terms of carer’s time. On the other hand, it is by virtue of this selfsame lack of care services outside the family in Italy that the presence of inactive grandparents appears to increase the likelihood of women’s participation in the labour market. It also shows that it is more likely for better educated women to work full time. Hence looking at part-time work as a way of working with a potentially positive influence on female participation in a country like Italy, in which—as mentioned above—the number of working women is particularly low and personal services able to encourage the supply of female labour are still few and far between, and distributed very heterogeneously across the territory, the article by Bratti and Staffolani confirms the possibility of intervening in terms of encouraging paid female work, particularly in phases of the family life cycle in which there are the greatest workloads linked to the presence of young children. Thus while the spread of part-time work potentially incentivates the supply of female labour, in the period analysed by Bratti and Staffolani (prior to the adoption of policies aimed at encouraging the adoption of part-time labour by companies, promoting its distribution) the presence of children under 15 is not associated with a greater supply of part-time labour on behalf of the mothers. As the Authors themselves underline, this could be due to the insufficient capacity of companies in that period to satisfy the requests from female workers for part-time contracts in the light of an
8
T. Addabbo and G. Solinas
increased family burden. Hence while the greater presence of part-time employment in a given area of residence incentivates women’s participation (potentially interested in exploiting this opportunity in periods of great domestic work) the very presence of those greater domestic workloads does not significantly influence the supply of part-time labour, thus calling for policies (which were later implemented in part) aimed at promoting the use of part-time as a working formula able to guarantee a better work-life balance. The greater incidence on the total of workers and the major role of part-time and fixed-term work for women lead Addabbo and Favaro to analyse the likelihood of part-time and fixed-term contracts from a gender perspective. To this end, the Authors make use of the IT SILC 2006 survey in the wider European Statistics on Income and Living Conditions Survey micro-data to produce bivariate probit models in which one equation estimates the probability of employment and the other the probability of being employed on a part-time or fixed-term contract. The multivariate analysis shows the existence of gender differences in the likelihood of working with fixed-term contracts: among men this probability is greater the younger they are, while the likelihood of being given a fixed-term contract increases for women as they get older. Furthermore, a greater exposure to the risk of working with fixed-term contracts may be noted among women with qualifications equal to or greater than the undergraduate degree. While the presence of children under 15 does not appear to influence the likelihood of either men or women working on fixed-term contracts, the likelihood of women working parttime increases when there are children under 15 to look after. As well as estimating the probability of being employed part-time or with a fixed-term contract separately for men and women (in order to capture the different weight by gender of the same factors), Addabbo and Favaro analyse the determinants of the probability of being employed part-time against the worker’s will. On average, the likelihood of being employed part-time involuntarily is greater among women; however, where there are children under 15, the likelihood of being unemployed involuntarily diminishes significantly and, vice versa, the likelihood of voluntary part-time employment increases. This result therefore shows a step forward in 2006 compared to the situation found by Bratti and Staffolani with reference to a previous period in the possibility that part-time work is a form of contract chosen to better address a greater domestic workload. However, while on one hand the greater likelihood of voluntarily opting for part-time work when there are young children—an option which, on the basis of the data analysed by Addabbo and Favaro, appears to be found more widely in the north of the country—it may be considered a positive sign with regard to female participation in the labour market and the possibility to reconcile paid work and family life; on the other hand, their greater presence in non-standard employment may entail costs that Addabbo and Favaro set out to quantify. The Authors underline the presence of specific costs linked to wage differentials, the lesser likelihood of accessing managerial positions and the lesser use of health services, showing in particular that passing up on health care treatment out of the lack of resources increases significantly among those working on fixed-term contracts.
1 Introduction
9
1.2 Non-Standard Employment, Workers’ Mobility and Labour Market Structure The articles in the second part of the volume examine the spread of non-standard work in Italy in greater detail, as well as the reasons behind it from a dynamic perspective, analysing the incoming and outgoing flows from non-standard employment. As already seen in the past, the spread of non-standard work is associated with the legislative changes that took on particular consistence over the course of the 1990s (cf. Sestito 2002, pp. 65–75). In the new regulatory framework—which in some aspects have undergone changes of a structural nature (such as the strengthening of the productive/manufacturing sectors etc. in which training on the workplace is not so relevant)—companies use the normative tools that the change in legislation has made available in order to cut costs and transform the quasi-fixed costs of labour into variable costs (simultaneously cutting costs on redundancy contributions, direct remuneration and social contributions). The first issue concerning the spread of the various forms of non-standard employment is how it is to be measured and the identification of the conditions of supply and demand. The essay by Barbieri and Sestito paints an overall picture of temporary employment in Italy through an estimate of the size of the phenomenon, an analysis of its composition (in relation to the characteristics of workers and jobs), offering a number of indications, which may be deduced from the official statistics, on the voluntary/involuntary nature of labour conditions. Estimating the dimensions of the phenomenon is not something that may be taken for granted. The Authors underline that there are size issues. By examining the data provided by administrative sources (Inps—the National Institute for Social Security—and the Ministry of Labour) and the Istat LFS data, 2004, it becomes clear (as also acknowledged in the following article by Tattara and Valentini) that the Istat LFS data underestimate (especially until 2004) the entity of temporary labour. Using the series compatible with the new edition of the Istat LFS data (used since 2004), they trace the rise of temporary labour over the last 15 years. It may be seen how the growth trend of temporary labour has come to the fore most constantly right from the early 1990s, with a particularly sharp acceleration over the last 2 years. Using this source, the stock indicators show that temporary labour in Italy concerns over 13% of workers. At the start of the 1990s, the percentage of temporary workers was roughly half. In terms of flow (annual recruitments in relationship to the overall total of workers), over this period, values slightly short of 40% are reached. Thus this is a phenomenon of great note as well as one which has shown no sign of slowing down over recent years, except the lesser incidence incurred by the shift towards inactivity or unemployment in the wake of the crisis (Istat 2010). The constant rise in the use of fixed-term contracts seems to suggest the predomination of factors of a structural nature. A further element highlighted in this article concerns the rise in the number of workers who state that they have accepted a temporary job out of lack of
10
T. Addabbo and G. Solinas
alternatives; i.e. the rise in the number of ‘involuntary’ temporary workers. The data available for 2006 indicate that more than 80% of temporary workers are ‘involuntary’. Analysing the make-up of temporary labour through the estimate of a probit model based on Istat LFS data for 2004 and 2006, the Authors show that temporary labour is the most commonplace among the weaker social groups, in the sectors and professions in which labour is least qualified. Compared to other groups of workers, temporary labour is found most commonly among women, students and young people first entering the labour market, as well as workers with the lowest levels of education and qualifications. Furthermore, the likelihood of having a temporary job is greater in the South of Italy and, in general, in the Italian provinces in which unemployment rates are higher. Given the absence of information on wages and the satisfaction levels of workers in the Istat LFS, as regards the aim of analysing the quality of labour in temporary working positions, the Authors choose three factors as indicators of the quality of labour: its involuntary nature, the worker seeking a different job, and training activities. In the latter two indicators, the Authors carry out a comparison between workers with fixed-term contracts and those with permanent contracts. As far as the involuntary nature of labour is concerned—an indicator of which the ambiguous elements are discussed by the Authors at length—the most interesting result we believe lies in their claim that while greater levels of involuntariness are to be found in the South, the local labour markets in which there is the greatest use of temporary labour are not necessarily those in which the highest levels of involuntariness are to be found. Levels of temporary labour and those of workers’ involuntary acceptance of it, in other words, are not correlated variables in terms of level nor in terms of time variations. What’s more, even though a contraction may be noted in the most recent part of the survey, job-hunting activities are undertaken more frequently by workers with fixed-term contracts than workers with permanent jobs: a datum which may reflect lesser job satisfaction, or which may be determined by the search for possible occupational alternatives once the current contract expires (and therefore, in some way, indicate a subjective perception of precariousness of one’s own working status). Many issues remain open, in fact as stressed at the beginning one of the main ones is whether (and to what extent) temporary labour constitutes a bridge towards stable employment. The three other articles that complete this part of the volume set out to illustrate this concept, from various different points of view and with a range of different analytical tools. The article by Picchio deals with this aspect of the analysis stressing the behaviour of companies. While, indeed, companies make use of temporary contracts exclusively as a means by which to limit labour costs and adapt to variations in product demand, temporary labour may become a segregated employment area which by no means favours the transition towards permanent employment. Vice versa, while work experience makes it possible to acquire some form of capacity and competence, temporary labour makes it possible to prevent the deterioration of
1 Introduction
11
human capital and can provide a signal which facilitates access to permanent posts. In this sense, the underlying hypothesis from which the Author sets out is complementary to the notion of temporary labour as a means of selecting the workers best suited to the company, as previously proposed in the volume by Barbieri and Sestito. The analysis focuses on the comparison between the probability of transition respectively from unemployment and temporary labour to fixed work with a permanent contract. To this end, three editions are used (2000, 2002 and 2004) of the Survey of Italian Households’ Income and Wealth (SHIW), carried out on a representative sample by the Bank of Italy every 2 years since 1989. The basis of the data makes it possible to construct a balanced panel made up of around 1,700 individuals aged between 15 and 64 in the first survey, and present in all three editions. On the basis of these data, dynamic probit models are created, the solidity of which is carefully tested. The results suggest that, given the observable and nonobservable characteristics of the worker, temporary contracts ease the transition towards permanent employment. Being employed with a temporary contract rather than unemployed in fact leads to an increase in the likelihood of being employed with a permanent contract by 13.5–16% points. The likelihood is lower among women and among workers in the South, and increases (not in a linear fashion) in relationship to education, work experience and family responsibilities. In this sense, the Author concludes that temporary labour facilitates the abandoning of the job-seeking circuit in favour of permanent employment. The essay by Origo and Samek Lodovici addresses the same theme on a subgroup of temporary workers. It in fact examines the probability of transition to stable employment among those employed with a temporary help contract, a tool introduced to Italian legislation in 1997 which since then, has been used increasingly by firms. It is estimated that there are now around 300,000 workers employed using this kind of contract. The essay explores a database especially put together on the basis of a sample survey, representative on a national level, referring to around 1,800 individuals with at least one temporary work experience between April 2003 and October 2004. The dataset contains detailed information on the characteristics of the workers and firms, and provides an adequate response to the issues linked to sample distortion. In the empirical analysis, using competing risk duration models which make it possible to bear in mind dependency on state, the probability of transition from temporary help work to other states in the labour market (occupation with fixedterm contracts, other forms of temporary work, job-seeking, or returning to the education system) is calculated. The results are only partially similar to those set forth by Picchio. Temporary help work is not itself a ‘‘trap’’, but whether or not there is a ‘‘stepping-stone’’ type effect, facilitating access to stable employment, depends on the characteristics of the worker, of the companies and the position occupied. In particular, the individual characteristics (gender, level of education, family background and makeup
12
T. Addabbo and G. Solinas
of the family) explain an important part of the transition towards unemployment and the (re-)entering of the education system, yet they have a less relevant effect with regard to the transition towards stable employment. The probability of accessing stable employment depends more on the conditions of the labour market, the characteristics of the company and the role held, with a lesser likelihood of stabilisation in the presence of unfavourable conditions. In this regard, we may consider the case of the public sector in which, in the period under study, the adoption of restrictive budget policies did not allow for new recruitment, and thus temporary labour did not in fact appear to promote stabilisation. Furthermore, for the group of temporary workers, previous work experience gains relevance. Access to stable employment is more common among (manual) jobs where there is previous experience of permanent employment (and temporary employment agencies appear to play a positive role here), while this is rather less the case for those jobs with a fragmented working career, characterised by a long series of temporary occupations and changes of workplace. Lastly, in determining the transition towards different states, the reasons why individuals and companies adopt these forms of contract are also important. The results of these forms of contract, in terms of status on the labour market, thus appear to be more heterogeneous than those put forward in the previous articles. The spread of temporary labour also has relevant consequences in terms of inequality and the segmentation of the labour market. Tattara and Valentini consider this theme. Their article draws on individual data from administrative sources (Inps) on wages and working careers with reference to the entire population of those working in manufacturing in two Venetian provinces (Vicenza and Treviso) throughout the 1975–1997 period. The dataset used allows for an accurate reconstruction of the working history, the succession and duration of the single workers and the wages received for each job. At this point, the Authors identify two categories: those who sooner or later will be given stable employment, greater protection and relatively high salaries (stayers) and those who, in the light of both their own individual characteristics and those of their job, have relatively low wages and appear to be trapped in a job which does not allow them any room for career development in the proper sense (movers). This component of workers who move in and out of the labour force, who frequently change occupation without ever stabilising their working position, characterised by a notably higher level of women and workers with a low level of formal education/qualifications grows to a large extent over the course of the period examined. The share of the movers, of the workers with chaotic career patterns or without any career perspectives, appears to grow from around 15–16% at the start of the 1980s to a little under 25% in the 1990s, with an even greater percentage of such workers employed in small and micro-sized companies (those under 50 employees). The result is important. By using a database limited to a specific area, yet solid and highly informative with regard to working careers, the article shows that temporary employment in general terms (regardless of the specific contractual forms that it might take on) is not confined to the period of entering the labour market. For a significant and growing share of employed workers, this may be
1 Introduction
13
looked upon as a permanent condition. In other words, Tattara and Valentini’s essay shows it is still possible in some ways for fixed-term contracts (at least for some elements of employment and to some degree) to facilitate the transition towards permanent contracts, but for many they do not give rise to working relationships with any significant continuity over time. And since the movers have an income profile significantly lower than that of the stayers, the long-term consequences are extremely important to consider.
1.3 Part-Time Employment, Working Conditions and Job Attributes As mentioned above, together with the rise in temporary labour in Italy, over the last decade we have witnessed a rapid spread of part-time employment. This is a complex phenomenon, involving somewhat diverse aspects such as the evolution of the family, attitudes to work, the evolution of company organisational models, with notable effects in terms of occupation, the structuring of the labour market and, more in general, of social organisation. There are companies which, on the basis of the type of product or service that they offer or by virtue of their internal organisational structure, discourage the use of part-time employment. On the contrary, there are also companies for which part-time employment provides organisational flexibility, increases in productivity, as well as the possibility of guaranteeing a continuous service at acceptable costs, while adequately dealing with peaks in demand. Also the range of part-time workers is extremely diverse. As shown previously, the use of part-time employment takes place in different ways between men and women. Among males, it is a phenomenon found largely in the phases of entering and exiting the labour market. If part-time work is examined only from the point of view of the life cycle, among women there are markedly different paths to be followed, also with regard to the willingness with which parttime work is accepted. Part-time employment may also be used by young people during the familiarisation phase of work, for whom part-time work constitutes one of the tools in the search and evaluation of various kinds of employment, or a tool with which to reconcile work and other learning activities. Similarly, there are components that by this means choose to occupy a borderline space between participation and non-participation in work, workers in search of a different job or elderly workers in the process of exiting the labour market. Thus it comes as no surprise that the evaluation of these forms of contract and their effects is controversial, and part-time is looked upon sometimes as a tool with largely positive connotations, making it possible at the same time to satisfy companies’ needs for flexibility and workers’ choices on time use, increasing levels of occupation. At other times, on the other hand, the negative aspects are emphasised, in terms of the role and duties of part-time workers within the company, their training and career opportunities, terms of pay, working commitment, involvement in organisational choices, sharing of company goals etc.
14
T. Addabbo and G. Solinas
In the third part of the volume, three essays are featured looking at part-time mainly from the point of view of working conditions, the wellbeing of workers with part-time contracts, their perception of the part-time relationship and their degree of satisfaction. The article by Burchell outlines the development of part-time employment in Italy in relation to the trends to be found in the main European countries, drawing on a number of sources including the European Working Conditions Survey (2002). It is argued that part-time employment in the main European countries is not a synonym of fixed-term, non-standard employment, nor does it coincide with low company seniority and working instability. Furthermore, it is not necessarily a second best for those who cannot find standard employment (and that on the contrary, involuntary part-time is less common in those very countries in which part-time as a whole is most widespread). The ‘partimers’ (and women in particular) are no less satisfied with their work than their full-time colleagues, most of all by virtue of the greater compatibility between work time and other activities. In this sense, it is not uncommon for part-time work to be preferred to full-time employment. Depedri’s essay reaches very similar conclusions in many respects. In this work the motivations, involvement and satisfaction with work as perceived by part-time workers are compared to full-time workers in the same companies. The data are taken from a study carried out on a set of around 2,000 workers occupied in the sector of social services both in profit-making and not-for-profit companies, many of which were small-sized. The choice of the sector is important. In personal care services activities in fact, a broad use of part-time contracts goes hand in hand with a clear feminisation of employment as well as the use of workers with a high level of education. Furthermore, given their relational character, even more than in other activities, motivation and commitment are key aspects of the quality of service offered. Adopting reasonable control indicators on the characteristics of the worker and the job, the essay highlights that in this sector, part-time workers enjoy a degree of job satisfaction entirely comparable with that of full-time workers. There are no major differences in (hourly) pay, in training opportunities, in the level of information and involvement received on the organisational and managerial choices of the company, nor in terms of career openings. This is the source of motivations, commitment and bonds no weaker than those of full-time workers. In Italy the social services sector as a whole (thus including both non-profit and for-profit companies) provides a good example of how part-time employment contracts may allow company needs for flexibility to coexist with the needs for greater time to dedicate to workers’ other activities. In this sense, the key message that comes across in this essay is that part-time work does not always imply partial satisfaction and/or a lower quality of work. That outlined is an interpretation which has major implications on the structuring of the normative framework and labour policy. Nevertheless, a number of notes of caution are necessary. The first is that there are potential long-term tradeoffs that should be born in mind with particular reference to Italy. Part-time
1 Introduction
15
employment may have been (and may still be) a means by which to make jobs available that otherwise would not have been so, as well as to attract groups otherwise confined to non-participation to take part in the labour market, yet the largely female connotation of the phenomenon risks increasing gender segregation. The second—and once more it is Burchell that reminds us—concerns the link between work and welfare: for many women, spending most of their working careers as partimers means receiving a low pension and increasing their risk of poverty. The third is that Depedri’s results are probably not generalisable to other economies: Burchell underlines how profound the differences are on the nature (and the extension) of part-time work only within the EU15. And neither are they applicable to the whole of the Italian economy. In various sectors (in which for example motivation and interpersonal relational skills are less relevant, and the use of highly-educated workers is less common), part-time workers tend to be concentrated in jobs where the tasks are less complex and demanding, with lesser potential for learning or career development. Russo and van Hooft give particular attention to this aspect, which already came to the fore in the essay by Addabbo and Favaro. The Authors examine a rather complex set of questions that reconnect the identity and image of one’s self, address workers’ time allocation choices, their preferences with regard to characteristic attributes of labour, and the behaviour of companies with regard to workers who express a preference for working less than the standard working hours. In the central part of the work, the identification and evaluation of the attributes of the work of those who opt for full-time employment and those who opt for parttime are reconstructed on the basis of a (subjective) assessment of a sample of 1,828 individual cases, of people both in work and job-seeking. The data were gathered in Holland, ‘‘the only part-time economy’’, as the Authors remind us, but the results concern aspects that may be extended to those employed in other advanced economies. Without going into detail, the most important result is that part-timers and full-time workers are not the same: be it in the role attributed to labour in relationship to other non-market activities, be it in considering work as a means of individual achievement, or to reach a satisfying level of income and social status. And nor are they the same in terms of the importance attributed to the career. They are not even the same in terms of the characteristic attributes of work (remuneration, benefits, security and stability, content and qualification of labour, working hours flexibility, etc.), and there are also great differences in each of the above-mentioned dimensions between men and women. From here, rather different identities emerge, constructed not in isolation, but in relation to the reference group’s system of personal and social relationships, and which may lead partimers to lose interest (and invest less) in their careers. However, the essay also proposes a different hypothesis which calls companies into question. The underlying idea is that the lesser career opportunities may derive not from the lack of motivation, nor—in general terms—from a lesser commitment or lower productivity of part-time workers, but from companies’ inherent difficulties in evaluating personal performances and, above all, from the
16
T. Addabbo and G. Solinas
inadequacy of company systems with regard to evaluating performance. Opting for part-time may in fact be interpreted as a signal of lesser commitment and lead to the company investing less in partimers, with a consequent exclusion from managerial positions and the opening of a wage gap in which they stand to lose out. In this sense, choosing to work part-time—albeit only for a fraction of one’s working life—may have long-term consequences.
1.4 Non-Standard Employment, Quality of Work and Firms’ Competitiveness The spread of the various forms of non-standard employment, in an age of rapid and deep-rooted technological and organisational change, and of a just as rapid growth of personal and business services (many of which are entirely new) poses two questions. The first is what the impact of non-standard employment is on the quality of work. The second is what the long-term consequences are on firms and their competitive capacity. A general response to these two queries is to be found in the concluding part of the volume. As part of the renewed attention to the themes of the quality of work and the relationship between the quality of work and the quality of living (Drobnicˇ et al. 2010), the contribution made by Addabbo and Solinas deals first of all with the issue of measuring quality according to the subjective perception that each single worker has of it. The article starts out from the definition put forward by Gallino (1993), who identifies four dimensions of the quality of labour (ergonomic, of complexity, of autonomy and of control), analysing it and integrating it with two other due dimensions: the relational/social dimension of work bound up in the need for social recognition and the dimension of compatibility between working/ living time/conditions. In the analysis of the social dimension, attention is placed on the network of relationships that a specific job manages to activate both within the workplace and outside. The reconciliation dimension makes it possible to verify the compatibility of working commitment with personal and family life. Thus in the essay, a measuring method is put forward, articulated across the six dimensions, which groups together more than 40 elementary indicators into synthetic indexes of various levels, exploiting fuzzy logic based on a previous model proposed in Addabbo, Facchinetti, Mastroleo and Solinas (2007). This tool makes it possible to clarify the interaction between the worker’s perception of the various dimensions that make up the quality and the remunerative/contractual characteristics of the job. Thus, Addabbo and Solinas illustrate the results of its application on a sample of 719 workers employed in companies in the metalwork, building, services, food, IT and textiles sectors in the province of Modena, drawing the data from a broaderbased study. By applying a statistical and econometric analysis to the output synthesis indicators of the fuzzy model, it may be seen how the various dimensions of the quality of work interact as well as how the different dimensions of quality of work differ on the basis of contract type.
1 Introduction
17
This approach makes it possible to identify low-quality labour and, with reference to non-standard work models, it shows how (in line with the literature and with the territorial context under analysis) only among those women who willingly work part-time does there seem to be any compensation between the reconciliation dimension and the economic one. In fact, women who happily work part-time show levels higher than the indicator measuring reconciliation between living and working hours, as well as lower values with regard to those indicators measuring the economic dimension (career, remuneration, protection and job security). With reference to the part-time or full-time forms of contract, the results of Addabbo and Solinas’ analysis show that more than the form of contract, it is its voluntary nature that determines higher levels of quality of work. While on one hand this result shows the importance of measuring the degree to which such a choice is taken freely in the studies on the quality of work, on the other hand, in correspondence with the increasing involuntary component in the rise of part-time work in Italy (OECD 2010), they focus on the effect of this increase on the quality of work itself. The greatest costs in terms of lower quality of work must also be measured against the different exposition of various groups of the population (women in particular, and those with a low level of education, resident in the South of Italy) to the likelihood of working part-time against their will, as outlined on various occasions over the previous pages. The evidence gathered does not show the existence of compensational differentials, but rather seems to indicate the presence of labour market segmentation between standard and non-standard positions, between work-based and qualification-based positions. The second issue—the impact of non-standard work on company performance—is dealt with in Lucidi’s essay. In recent years in Italy, there has been much debate on the slowing of productivity growth witnessed in the 1990s and even more sharply in the first decade of the new century. Most observers agree on a diagnosis of a structural type which first and foremost questions the structure and specialisation model of the Italian industrial sector. In this debate Lucidi’s essay introduces a new element: the relationship between the flexibilisation of the labour market (in terms of the process of determining wages and external flexibility) and its effects on productivity and, in particular, on the productivity of labour. This is a new element not only insofar as it is relatively unexplored, but also due to the survey perspective that the Author proposes, turning the traditional point of view on its head, according to which the rigidity of the labour market has negative effects on the productivity of labour (with an unfavourable effect on individual motivation and on allocative processes—from the companies and sectors in decline to those sectors enjoying growth). The hypothesis put forward is, on the contrary, that in a context of wage moderation, in the wake of the agreements of 1992 and 1993, the growing use of fixed-term contracts (and the ensuing rise in external flexibility) may be one of the factors that has led to determine the poor performance, in terms of production growth, of Italian manufacturing firms. The underlying idea is that a low wage pattern, coupled with the rise in external flexibility, may have a negative effect on the static and dynamic efficiency of the company in various ways: reducing the
18
T. Addabbo and G. Solinas
incentive to adopt new labour-saving technologies as well as the incentive (also among workers themselves) to invest in both general and specific human capital, mining the bases on which to build cooperative working relationships with individual employees, and lastly, leading to negative effects on aggregate demand. This idea is subjected to empirical verification by estimating the OLS equations on the growth of work productivity on a representative sample of over 3,000 manufacturing companies with over 10 employees, put together on the basis of microdata from the ‘‘Survey of Manufacturing Firms’’ carried out by Capitalia between 2001 and 2003, as well as from budget data supplied by the AIDA database (Bureau Van Dijk) containing specific information on the added value and the cost of labour. The model of reference for the estimation of the equations on the dynamics of productivity is that put forward in a range of different works by Sylos-Labini, integrated and modified as necessary with the aim of incorporating indicators of external mobility and to adapt it to the use of microdata. As is known, and as the Author underlines, the Sylos-Labini model links the growth of productivity to three components: the level of wages compared to the cost of investment goods as an element that leads firms to maintain (or eliminate) low productivity jobs and processes; investment spending, aimed at measuring the impact of the new technologies incorporated in newly purchased capital goods; and, lastly, the growth of aggregate market demand. Along with the regressors designed to measure these effects, there are also two indicators of external mobility (the share of employees in the firm with fixed-term contracts and the annual turnover rate of company employees). Details on the construction of the model, its various specifications and the results obtained may be found in Lucidi’s essay, though we may stress that all three of the components indentified by Sylos-Labini play the role predicted in determining the productivity of labour (even though the brief time span considered does not make it possible to distinguish between the existence of growing yields set off by the expansion of the markets due to factors that may simply be traced back to a greater use of the productive capacity). Rather more important, with regard to the theme discussed in this volume, is that the results convincingly testify there to be a trade-off between external flexibility and the growth of labour productivity. In particular, over the 3 years examined, the companies with a higher share of temporary workers and with a higher level of staff turnover show a lower growth rate in terms of added value per employee. This result holds true particularly for the most traditional firms and those which have no formalised R&D activities. The results are particularly relevant both in terms of labour policies and, in even more general terms, with regard to the reflection on the dynamics of productivity in Italy. On one hand, the long-excised theme of the role of aggregate demand rears its head, while on the other hand, it appears clear that the orientation towards wage moderation and the joint flexibilisation of the labour market in the long term poses issues of sustainability and the risk of tying down the economy to a spiral of low growth.
1 Introduction
19
1.5 A Provisional Conclusion Currently, it is unclear what the crisis exit strategies are. The recovery of the world economy follows an uneven path, just as monetary, taxation and management policies implemented by the various economies are uneven and uncoordinated. It appears evident that this is not a part of a cycle as seen before. The financial crisis comes to the fore against an international backdrop of profound and rapid change in world economic balances. Suffice to mention one single datum: according to the estimates of the International Monetary Fund (World Economic Outlook Databases), the share of the countries in the Eurozone on the world GDP, which in 2000 stood at 18% in terms of purchasing power, will fall to 13% by 2015. Over the same period, the share of the emerging Asian countries will double (from 15 to 29%), not so much due to population growth, but rather because of the product rise per inhabitant (worth only 8% in 2000 and estimated around 20% in 2015, compared to the Eurozone). In the middle of the last century, USA, Canada and Europe accounted for more than two thirds of the world GDP; today this figure is no more than 40–45%. The Italian economy is profoundly affected by these processes. As known, Italy has for many years shown an incapacity to grow at a sustained rate. Over the next 2 years, the increase in GDP will stand at around 1%. The growth of product per inhabitant has been falling for three decades, but this process has undergone a clear acceleration over the last few years. Over the last decade we have witnessed a clear loss of competitiveness compared to the main European economies, mainly induced by the changing dynamics of work productivity: suffice to say that with real wages more or less stable (+ 3%), between 1998 and 2008, in the private sector, productivity increased by 18% in France, 22% in Germany and only by 3% in Italy. This is not the right place to examine the structural problems of the Italian economy which have led to this trend. The factors cited in the vast literature are manifold and lay the blame with the productive structure itself, as well as the specialisation model in the manufacturing industry, the lack of liberalisation in the services sector, an inadequate infrastructure and shortcomings in secondary and tertiary training systems, and much more besides, underlining the negative impact on the adoption of new technologies and on the performance of the manufacturing sector as a whole. Of all these factors, this volume takes one in particular into consideration: in the end the common thread that ties together the various contributions concerns the dualism of the labour market, read in the light of the spread of the non-standard labour model. The measures adopted since the mid-1990s, encouraging the use of fixed-term contracts and part-time employment, as mentioned earlier, led to a significant rise in employment rates. Part-time employment derives from the motivations both of companies and workers in certain conditions. When part-time work is chosen on a voluntary basis, it may satisfy the needs of both parties. The same may not be said
20
T. Addabbo and G. Solinas
for temporary work: in certain conditions temporary labour may be transformed into a factor hindering development. In different ways and from different perspectives, the collection of essays included in this volume show that it is time to change tack. Without a perspective of the gradual stabilisation of precarious working relationships, the accumulation of human capital is compromised, particularly specific human capital, with effects of marginalisation and the exclusion of entire generations of young people, and just as serious knock-on effects on the productivity and profitability of firms. In the final analysis, on the very potential for the development of the economy as a whole. Reducing this dualism and the long-term effects of temporary labour on the working and living conditions of women and young people, insofar as they are more likely to be excluded from the current system of social safety nets because they are more likely to be affected by non-standard working conditions or spells of inactivity, means taking action in a number of directions. The entire system of social safety nets must be reformed. Along with the United States, among the developed nations Italy is the one that allocates the fewest resources to those who lose their jobs (OECD 2010). It is a question of creating forms of protection that also cover fixed-term workers, who in Italy today are all but excluded. In order to reduce the two-tier system of the Italian labour market, norms may also be introduced to make the use of fixed-term labour more costly, to set a maximum time scale for their adoption, and as regards remuneration, to introduce minimum pay levels. Among others, the proposal for a single permanent work contract including guarantees and progressive protection as company seniority increases, moves in this direction (Boeri and Garibaldi 2008). Work placement contracts should be completely rethought. Whatever the specific normative framework, the underlying problem remains that of making a clear distinction between the use of temporary employment contracts as a means of entering the labour market and the use of fixed-term and seasonal contracts as genuinely temporary employment. Finding a balancing point that guarantees companies a certain degree of flexibility during the hiring process, and that provides non-standard workers with a way out of the two-fold trap of a high level of precariousness and a high level of unemployment, and that allows those already employed to maintain reasonable labour standards will not be easy. And not only in Italy.
References Addabbo, T. (2003). Unpaid work by gender in Italy. In A. Picchio (Ed.), Unpaid work and the economy. London: Routledge. Addabbo, T. (2005). Genitorialità, lavoro e qualità della vita: Una conciliazione possibile? Riflessioni da un’indagine in provincia di Modena. Milano: Angeli. Addabbo, T., Caiumi, A., & Maccagnan, A. (2010). Unpaid work, well-being and the allocation of time in contemporary Italy. In T. Addabbo, M. P. Arrizabalaga, C. Borderias, & A. Owens (Eds.), Gender inequalities, households and the production of well-being in Modern Europe. Alderhot: Ashgate.
1 Introduction
21
Addabbo, T., Facchinetti, G., Mastroleo, G., & Solinas, G. (2007). A fuzzy way to measure quality of work in a multidimensional perspective. In J. Pejas & K. Saeed (Eds.), Advances in information processing and protection. Heidelberg: Springer. Bank of Italy (2011). Bollettino economico, 63, gennaio. Berton, F., Richiardi, M., & Sacchi, S. (2009). Flex-insecurity. Perchè in Italia la flessibilità diventa precarietà. Bologna: Il Mulino. Boeri, T., & Garibaldi, P. (2008). Un nuovo contratto per tutti. Milano: Chiarelettere. Cingano, F., Torrini, R., & Viviano, E. (2010). Il mercato del lavoro italiano durante la crisi, Questioni di Economia e finanza, Occasional paper, Bank of Italy, n. 68, giugno 2010. Connolly, S., & Gregory, M. (2008a). Feature: The price of reconciliation: Part-time work, families and women’s satisfaction. Economic Journal, 118(526), F1–F7. Connolly, S., & Gregory, M. (2008b). Moving down? Women’s part-time work and occupational change in Britain, 1991–2001. Economic Journal, 118(526), F52–F76. Connolly, S., & Gregory, M. (2009). The part-time pay penalty: Earnings trajectories of British women. Oxford Economic Papers, 61(Supplement 1), i76–i97. Del Boca, D., & Vuri, D. (2007). The mismatch between employment and child care in Italy: The impact of rationing. Journal of Population Economics, 20(4), 805–832. Drobnicˇ, S., Beham, B., & Prag, P. (2010). Good job, good life? Working conditions and quality of life in Europe. Social Indicators Research., 99, 205–225. European Commission. (2010). Report on Equality between Women and Men, 2010. European Commission, Directorate-General for Employment, Social Affaire and Equal Opportunities, http://europa.eu. Eurostat. (2011). Eurostat News Releases 5/2011 - January 7, 2011, from http://ec.europa.eu/ eurostat. Gallino, L. (1993). Dizionario di sociologia. Torino: Utet (I ed. 1978). Istat. (2010). Rapporto Annuale sulla situazione del paese. Roma: Istat. Istat (2011). Noi Italia. 100 statistiche per capire il Paese in cui viviamo. www.istat.it. Jaumotte, F. (2003). Female labour force participation: Past trends and main determinants in OECD countries. Economics department working paper, No. 376. Paris: OECD Publishing. Manning, A., & Petrongolo, B. (2008). The part-time pay penalty for women in Britain. Economic Journal, 118(526), F28–F51. Mumford, K., & Smith, P. N. (2009). What determines the part-time and gender earnings gap in Britain: Evidence from the workplace. Oxford Economic Papers, 61(Supplement 1), i56–i75. OECD. (2010). Employment Outlook: Moving beyond the jobs crisis. OECD: Paris. Sestito, P. (2002). Il mercato del lavoro in Italia. Com’è. Come sta cambiando. Roma-Bari: Laterza.
Part I
Non-Standard Employment, Labour Supply and Living Conditions
Chapter 2
A Microeconometric Analysis of Female Labour Force Participation in Italy Massimiliano Bratti and Stefano Staffolani
2.1 Introduction In March 2000 the European Council set out an ambitious target for female employment rates in Lisbon, which should reach the level of 60% by 2010. However, Italy is very far from reaching this target. Indeed, according to the Italian National Statistical Institute (Istat) 2003 official data, only 42% of women aged 14–64 were in employment and less than one in two participated in the labour force. One of the possible reasons for the low female labour force participation (LFP) is the incompatibility between work in the marketplace and childrearing, which in Italy could be exacerbated by the relatively low availability of public child-care services and part-time (PT hereafter) employment opportunities (Del Boca 2002). Hence, determining the factors associated with a high female labour force attachment appears to be a strong priority in order to implement policies aimed at increasing Italian women’s LFP and employment.
An early version of this chapter was presented at the 2004 Annual Conference of the Associazione Italiana Economisti del Lavoro (Aiel). We wish to thank the Aiel Conference participants and Paolo Sestito for useful comments. Funding from the Italian Ministry of Welfare and Social Policies is gratefully acknowledged. The usual disclaimer applies. M. Bratti (&) Dipartimento di Scienze Economiche Aziendali e Statistiche, Università degli Studi di Milano, Via Conservatorio 7, 20122 Milan, Italy e-mail:
[email protected] S. Staffolani (&) Department of Economics, Marche Polytechnic University, piazza Martelli 8, 60100 Ancona, Italy e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_2, Springer-Verlag Berlin Heidelberg 2012
25
26
M. Bratti and S. Staffolani
In this chapter we use microlevel individual data for Italy from the Quarterly Labour Force Survey (Rilevazione Trimestrale delle Forze di Lavoro, LFS hereafter) to undertake a microeconometric analysis of the factors affecting women’s LFP. We will give a particular emphasis on the factors that can be affected by public policies such as availability of public child-care and elderly-care services, local availability of PT jobs or the role of female education. The main research questions that we will attempt to answer are: (1) Does the availability of child care and elderly care raise female LFP? (2) Does PT employment represent a way for women to reconcile home and market work? (3) Do women with larger families prefer PT work? The structure of the chapter is as follows. Section 2.2 reports a brief survey of the empirical literature on female LFP in Italy and Sect. 2.3 deals with the description of our estimation sample. Section 2.4 describes the econometric model and Sect. 2.5 our main results. Section 2.6 concludes.
2.2 A Brief Overview of the Empirical Literature on Female Labour Force Participation in Italy A detailed survey of the empirical work that has analysed female LFP in Italy is beyond the scope of the present chapter.1 Hence, in this paragraph we simply report some of the main findings of this literature that will help us in the specification of our empirical model. According to the neoclassical model of labour supply, a primary factor affecting female LFP is female wage. Several studies have included wages in women’s LFP equations. Some examples include Colombino and Del Boca (1990), Colombino and Di Tommaso (1996), Aaberge et al. (1998), Aaberge et al. (1999, 2000) and Di Tommaso (1999) all of whom estimate significant positive effects of female wages on participation. By contrast, the estimated effect of the partner’s wage or income on women’s participation is usually negative, showing the existence of a negative non-labour income effect on participation (Colombino and Del Boca 1990; Colombino and Di Tommaso 1996; Aaberge et al. 1999; Di Tommaso 1999). A factor positively associated with female LFP is women’s education. Studies that have empirically found such an effect include Del Boca (1993), Tanda (1994), Colombino and Di Tommaso (1996), Di Tommaso (1999), Addabbo (1999), Chiuri (2000), Del Boca (2002) and Bratti (2003b). The effect of education might reflect both pecuniary factors, such as higher wages and a higher flow of future expected incomes, and non-pecuniary factors such as cultural influences. Some studies have investigated the presence of an added worker effect or a discouraged worker effect. In the first case women with unemployed partners should be more likely to participate to contribute to family finances, while in the second case women with unemployed partners should be 1
For a detailed survey see Bratti (2003a).
2 A Microeconometric Analysis of Female Labour Force Participation in Italy
27
discouraged from participating in the labour force, since they believe they will be very unlikely to find a job. There is some evidence supporting the prevalence of an added worker effect in Italy. Indeed, findings in this direction are offered by Borzaga and Contrini (1999), Prieto-Rodríguez and Rodríguez-Gutiérrez (2003). The role of institutional factors is investigated among others by Del Boca (2002) who finds positive effects of the availability of public child care and PT employment opportunities on the probability of participation. Marenzi and Pagani (2005) focus instead on the role that elder relatives have in women’s LFP. They find that women who can rely on informal child care by elder relatives are more likely to participate while those who have to take care of elder relatives have a lower probability of participation. Moreover, they consistently observe a positive effect of both the number of places in public elderly-care facilities and kindergartens at the regional level on female LFP. Bratti et al. (2005) study the LFP decisions of new mothers in the first three years after child birth and find a strong negative effect of the unavailability of child care (both formal and informal) on participation and positive effects of job characteristics associated with stronger levels of job protection and security, such as working in the public sector or with permanent contracts.
2.3 Data Description We use data from the fourth quarter of the 2002 LFS (October), carried out by the Italian National Statistical Institute (Istat). We select women aged between 15 and 64 who declared to be heads of the household or spouses2 and obtain a sample of 41,594 observations. As to the definition of the dependent variable in our econometric analyses, we define non participation (NP hereafter) following Istat. We consider all women who were employed full-time (FT hereafter) and did not declare to be searching for a new job as participating FT whereas those who were searching preferably for PT jobs as participating PT. Women employed PT voluntarily and involuntary were considered as participating PT and FT, respectively. Unemployed women were considered as participating PT if they declared to be prevalently searching for PT employment. We excluded 301 individuals who declared searching indifferently for PT and FT jobs. Therefore, our estimation sample consists of 19,922 women out of labour force (48.25%), 3,751 participating women preferring PT (9.08%) and 17,620 participating women preferring FT (42.67%). 2 In a previous version of this chapter (see the 3rd chapter of our report to the Italian Ministry of Welfare and Social Policies, downloadable at: http://www.dea.unian.it/staffolani/filespdf/ rapporto.pdf) we estimated a probit model for the probability of being ‘‘head of the household’’ or ‘‘spouse’’. Older women and women with low education showed higher probabilities to be in those states.
28
M. Bratti and S. Staffolani
Table 2.1 Means of territorial variables, by macro-areas Child care Part-time Unemployment rate
Elderly care
North Center South and Isles Total
3.42 1.16 1.08 2.09
10.28 10.14 3.55 7.61
9.82 8.24 6.94 8.41
4.07 6.43 17.42 9.74
In our sample, 84.7% of women have a partner; 60.6% have compulsory schooling or less whereas 30.9% have higher secondary schooling and 8.5% tertiary education. The average number of children is 0.82 in Northern Italy, 0.89 in the Centre, 1.33 in the South; single women have 0.5 children on average, whereas women with a partner have 1.13 children. In the following analyses we use territorial variables which may affect female LFP. In particular, we consider the number of places in public kindergartens per 100 children aged 0–2 by region in 2000, which will also be referred to as the child care coverage ratio, and the number of places in elderly-care public services per 100 seniors (aged more than 65) by region in 1999, which we will also refer to as the elderly care coverage ratio. Both variables have been taken from Marenzi and Pagani (2005). Other variables included in the econometric model are the percentage of PT employment out of total employment, as a proxy of PT work diffusion, and the unemployment rate, both measured at the provincial level and computed from the October 2002 LFS wave. Table 2.1 reports average values of these variables by macro-area. A clear gap between the South and the rest of Italy emerges, especially for child care availability and the unemployment rate. PT diffusion is more similar in the three macroareas. The elderly-care public service availability is very low all over Italy.
2.4 Econometric Issues Let us assume that a woman has to decide between three possible LFP states (j): NP (j = 0, NP), PT participation (j = 1, PT) and FT participation (j = 2, FT).3 We assume that the direct costs of choosing the state j given the characteristics of individual i denoted with xi are c(j|xi), which are weakly convex and increasing in j. For instance, these costs might reflect non-pecuniary costs due to the disutility produced by working time or pecuniary costs borne to buy external child care. The x’s being individual characteristics are the same across all alternative choices. Let us also assume that the discounted return to choose state j is R(j|xi, ei), concave and increasing in j, where ei is a person specific shifter of the return to state j. We define the utility for individual i in state j, i.e. Vij, as the difference between the 3
This economic rationalisation of ordinal response models was originally introduced in Cameron and Heckman (1998).
2 A Microeconometric Analysis of Female Labour Force Participation in Italy
29
return and the cost of choosing state j. The optimal LFP state is determined for each individual by solving the problem: Max Vij ½Rðjjxi ; ei Þ cðjjxi Þ j
ð2:1Þ
where j = 1, 2, 3 and j = 3 is the state with the highest economic return (i.e. FT participation, which does not necessarily imply that it is also the highest utility state for individual i). Let us assume that ei is stochastic, such that ei ? xi and: Rðjjxi ; ei Þ ¼ RðjÞuðxi Þei
ð2:2Þ
where Eðei Þ ¼ 1, ei 0, while cðjjxi Þ ¼ cðjÞ, i.e. participation costs do not depend on individual characteristics.4 If s is the optimal LFP state for individual i then: cðsÞ cðs 1Þ cðs þ 1Þ cðsÞ ei ½RðsÞ Rðs 1Þuðxi Þ ½Rðs þ 1Þ RðsÞuðxi Þ
ð2:3Þ
ei is therefore bounded by the ratios of marginal return to the marginal cost of the cðjÞcðj1Þ different states. If ei is continuously distributed and defining: exp½lðjÞ ¼ RðjÞRðj1Þ , then h i ð2:4Þ Prðj ¼ sjX ¼ xi Þ ¼ Pr expðlðsÞuðxi Þ1 ei expðlðs þ 1Þuðxi Þ1 : If we further assume that /ðxi Þ ¼ expðx0i bÞ and that ei is log-normally distributed, the expression above takes the more familiar form of the standard ordered probit model: Prðj ¼ sjX ¼ xi Þ ¼ Pr lðsÞ x0i b ui lðs þ 1Þ x0i b ð2:5Þ ¼ U½lðs þ 1Þ x0i b U lðsÞ x0i b where ui ¼ lnðei Þ and Uð:Þ is the standard normal distribution function. Eq. 2.5 gives the probability of choosing intermediate states, i.e. in our case j = s = 1 (PT). While for the two extreme states, NP and FT, the probabilities are given by: Prðj ¼ 1jX ¼ xi Þ ¼ Pr ui lð1Þ x0i b ¼ U lð1Þ x0i b ð2:6Þ ð2:7Þ Prðj ¼ 3jX ¼ xi Þ ¼ Pr ui lð2Þ x0i b ¼ 1 U lð2Þ x0i b respectively. 4 The specification can be adjusted so as to allow for both the return and costs of labour force participation to depend on individual and family characteristics (see, for instance, Lauer 2003 in the context of educational choices). However, this has no empirical relevance, since only the effect of the covariates on the ratio of the marginal return to the marginal cost of the LFP states, and not on the single components (return and cost), can be identified.
30
M. Bratti and S. Staffolani
The vector b and the two thresholds l(1) and l(2) are estimated using maximum likelihood. In this framework individual and household attributes increase or decrease the return to participation and therefore affect the probability of choosing the different states. In the specification of the econometric model we use the ‘‘standard approach’’ (see Browning 1992; Nakamura and Nakamura 1992), i.e. we include ‘‘child services’’ variables, such as the number and age of children, on the right-hand-side (RHS) of the female LFP equation. Although these variables are pre-determined with respect to LFP, they might be endogenous, i.e. they might be choice variables for an individual, and potentially jointly determined with female LFP.5 Unfortunately, the LFS does not contain retrospective data that could be useful to tackle the endogeneity problem and for this reason our estimates are likely to be affected by some endogeneity bias. However, we argue that the endogeneity problems are unlikely to affect all the RHS variables in the same way. For instance, educational choices are made much earlier than marital and fertility choices in Italy (where women generally get married and give birth after leaving FT education), and in this respect we could consider them predetermined and sufficiently exogenous with respect to future fertility and LFP decisions. Evidence in this direction is provided for instance by Bratti (2003b) who finds using data from the Bank of Italy’s Survey of Household Income and Wealth (SHIW) that the null hypothesis of weak exogeneity of women’s education with respect to both fertility and LFP could be rejected. Other variables that can be endogenous are the partner’s characteristics, such as his job qualification or education, in the case of positive assortative mating. However, also in this case partner’s characteristics are not perfectly controllable by a woman and the estimated correlations will partly reflect ‘‘causal effects’’. Some variables of particular interest for our analysis are those related to child care and elderly care availability and PT employment opportunities. Also in this case, however, there might be problems of endogeneity or reverse causality: women with higher labour market attachment might choose to reside in places where PT employment opportunities and child-care services are more diffused or the latter might be more diffused because female LFP is higher. Unfortunately, as anticipated, the LFS does not contain enough information to solve these problems. However, although it is not possible to exclude that some individuals will choose residence considering the opportunities of PT employment and the availability of child-care or elderly-care services, this would produce a substantial endogeneity bias only if these criteria are the main determinants of residential choices, which is very unlikely to be the case in the Italian context. In particular, because of the constraints of child-care or elderly-care services in terms of number of places (see Del Boca 2002 and Marenzi and Pagani 2005) and the high competition in their allocation, the decision of a household to move where public care services are
5 This is acknowledged by the so-called ‘‘purist approach’’ in which child-services variables are jointly modelled with labour force participation. For some examples of such approach in the Italian context see Di Tommaso (1999), Del Boca (2002) and Bratti (2003b).
2 A Microeconometric Analysis of Female Labour Force Participation in Italy
31
more developed will be a very risky one given the very low ex-ante probability of being assigned a place. Moreover, since we use public child care and elderly care availability at regional level, in our case a relevant endogeneity bias would only emerge if households prevalently choose their region of residence according to the local availability of public care services. Last but not least, geographical mobility is extremely low in Italy. For instance, in 1999 the percentage of the total population who resided in a different region (NUTS, level 2) the year before was only 0.9 and 1.2% when one considers the employed population (see European Commission 2002). As to the potential endogeneity of child-services variables, this is now widely acknowledged by labour economists. The issue is that unobserved individual factors might affect both LFP, i.e. the outcome of interest, and the explanatory variables (such as the number and age of progeny). Unfortunately, we do not have suitable data to instrument child-services variables, and given the cross-sectional nature of our data we cannot control for unobserved heterogeneity using individual fixed or random effects. However, we will use a very intuitive way for judging whether our estimated effects are likely to simply reflect spurious correlations or may partly capture causal effects: the degree of similarity of the effects of the number of children by age. If differences in the number of children simply reflect different time-invariant women’s unobserved characteristics, such as taste for market work, the effect of children is likely to be undifferentiated by children’s age. On the contrary a differentiated effect by age, in particular a decreasing effect with age, would suggest that the effect on female LFP is related to the higher timeintensity that the care of younger children requires.
2.5 Results In what follows, we will describe the results by considering different types of factors affecting female LFP: • territorial variables (Table 2.2), including macro-area dummies, local unemployment rates, child care availability interacted with the number of children by age, PT availability interacted with the number of children by age, elderly care availability6; • women’s personal variables (Table 2.3), including the highest educational qualification achieved and age by classes; • family composition (Table 2.4), including number of children by age, family size, number of employed children living in the family, number of employed family members older than 15, number of disabled family members, number of inactive grandfathers and grandmothers; 6 We also estimated the model with interaction terms between elderly care availability and presence of inactive grandparents and between inactive grandparents and young children but they did not turn out to be statistically significant.
32
M. Bratti and S. Staffolani
Table 2.2 Ordered probit estimates of female participation: territorial characteristics Variables coeff. z-stat. Marginal effects (%) Geographic area (ref. North) Centre South and Isles Unemployment rate Child care availability Child care: children 0–1 Child care: children 2–3 Child care: children 4 or more Part-time employment rate Part-time: children 0–1 Part-time: children 2–3 Part-time: children 4–6 Part-time: children 7–14 Part-time: children 15 or more Elderly care availability
0.049 0.100 -0.016 0.038 -0.004 -0.005 -0.011 -0.001 0.041 -0.001 -0.001 0.003 0.005 0.066
[0.85] [0.68] [1.75]* [7.84]*** [0.54] [0.90] [3.79]*** [0.05] [2.76]*** [0.09] [0.25] [0.61] [1.11] [2.51]***
NP
PT
FT
-1.96 -3.96 0.62 -1.52 0.17 0.18 0.43 0.02 -1.62 0.03 0.06 -0.12 -0.19 -2.61
0.05 0.07 -0.01 0.03 0.00 0.00 -0.01 0.00 0.03 0.00 0.00 0.00 0.00 0.05
1.92 3.89 -0.61 1.49 -0.17 -0.18 -0.42 -0.02 1.59 -0.03 -0.06 0.12 0.19 2.56
Note: No. of observations = 41,293 Pseudo R2 = 0.152 Wald chi2 (66) [p-value] = 8,333.78 [0.00] Robust z statistics in square brackets (absolute values). Standard errors are clustered by *significant at 10%; ***significant at 1% Table 2.3 Ordered probit estimates of female participation Variables coeff. z-stat. Marginal effects (%) Women’s education (ref. elementary or less) Lower secondary 0.422 Vocational 0.883 Upper secondary 1.082 Tertiary 1.611 Women’s age (ref. less than 30) 30–39 0.061 40–49 0.032 50 or more -0.415
NP
PT
FT
[15.71]*** [21.39]*** [23.81]*** [17.63]***
-15.84 -33.97 -41.14 -56.33
2.01 2.18 1.53 -1.60
13.83 31.79 39.60 57.93
[2.03]** [0.72] [5.20]***
-2.37 -1.24 16.44
-0.05 -0.02 -0.73
2.43 1.27 -15.71
Women’s characteristics Standard errors are clustered by **significant at 5%; ***significant at 1%
• partner’s characteristics (Table 2.5), including job qualification, age by classes and education. Although we report our estimates in different tables, all the estimated coefficients and marginal effects refer to the same ordered probit model including all the control variables listed above for which we report some statistics in the notes in
2 A Microeconometric Analysis of Female Labour Force Participation in Italy Table 2.4 Ordered probit estimates of female participation Variables coeff. z-stat. No. of children 0–1 No. of children 2–3 No. of children 4–6 No. of children 7–14 No. of children 15 or more Family size No. of employed children No. of employed family members ([15) No. of invalid family members No. of inactive grandfathers No. of inactive grandmothers
-0.557 -0.256 -0.196 -0.141 0.024 -0.035 -0.045 0.201 -0.192 -0.034 0.198
[4.06]*** [2.96]*** [3.05]*** [4.46]*** [0.70] [1.62] [2.02]** [4.51]*** [1.96]** [0.50] [3.70]***
33
Marginal effects (%) NP
PT
FT
22.13 10.17 7.81 5.60 -0.95 1.39 1.80 -7.99 7.64 1.36 -7.89
-0.41 -0.19 -0.14 -0.10 0.02 -0.03 -0.03 0.15 -0.34 -0.03 0.15
-21.72 -9.99 -7.67 -5.49 0.93 -1.36 -1.76 7.85 -7.30 -1.34 7.75
Family characteristics Standard errors are clustered by **significant at 5%; ***significant at 1%
Table 2.2. Marginal effects are computed at the sample means for continuous variables, and by calculating the difference in the probability of the three outcomes (NP, PT, FT) with respect to the reference category when changing the other dummies of the same group to 1 (e.g., women’s education) for categorical variables. With respect to the territorial variables (see Table 2.2), our results show a positive association between public child care availability and female participation in the labour market. A one point increase in the ratio of child care coverage is related to a decrease of 1.52 percentage points in the non-participation probability and to a 1.49 percentage points increase in the probability of participating fulltime. The interaction between child care availability and children by age shows that the positive effect of child care on participation decreases when children grow up. The availability of elderly-care services raises the attachment of women to the labour market: a one percentage point increase in the ratio of elderly-care coverage is associated to a 2.56 percentage points increase in FT participation and a decrease in the probability of NP of a similar magnitude. A discouraged worker effect seems to emerge: local areas with higher unemployment rates are characterised by a higher female non participation, although the effect is only significant at the 10% level. The ratio of PT employment in the local area significantly raises the FT participation probability only for women with children aged less than 2 (+1.59 percentage points). Contrary to the conventional wisdom about regional differences in female LFP in Italy, once the above local control variables have been included, women living in Centre, Northern and Southern Italy do not seem to behave differently in terms of LFP.
34
M. Bratti and S. Staffolani
Table 2.5 Ordered probit estimates of female participation Variables coeff. z-stat. Job (ref. manager, priv. services, typical) Single 0.634 Manager, industry, employee 0.062 Manager, industry, indep. 0.038 Manager, priv. services, atypical 0.203 Manager, priv. services, indep. -0.016 Manager, publ. services, typical 0.298 Manager, publ. services, atypical -0.044 Manager, publ. services, indep. -0.092 White collar, industry, typical 0.132 White collar, industry, atypical -0.045 White collar, priv. services, typical 0.131 White collar, priv. services, atypical 0.150 White collar, publ. services, typical 0.237 White collar, publ. services, atypical 0.313 Blue collar, industry, typical 0.097 Blue collar, industry, atypical 0.228 Blue collar, industry, indep. 0.360 Blue collar, priv. services, typical 0.071 Blue collar, priv. services, atypical 0.153 Blue collar, priv. services, indep. 0.432 Blue collar, publ. services, typical 0.158 Blue collar, publ. services, atypical 0.345 Blue collar, publ. services, indep. 0.416 Self-employed, industry 0.062 Self-employed, priv. services 0.157 Self-employed, publ. services 0.316 Not employed 0.032 Age (ref. 50 or more) \30 -0.063 30–39 0.068 40–49 0.018 Education (ref. tertiary) Elementary or less 0.064 Lower secondary 0.037 Vocational 0.056 Upper secondary 0.024
Marginal effects (%) NP
PT
FT
-24.36 -2.47 -1.52 -8.09 0.62 -11.84 1.75 3.61 -5.25 1.78 -5.21 -5.96 -9.43 -12.40 -3.84 -9.06 -14.24 -2.84 -6.11 -16.99 -6.29 -13.67 -16.38 -2.45 -6.25 -12.54 -1.27
-0.45 0.15 0.10 0.35 -0.05 0.35 -0.13 -0.30 0.28 -0.14 0.27 0.30 0.36 0.34 0.22 0.36 0.29 0.17 0.30 0.18 0.31 0.31 0.21 0.15 0.31 0.34 0.08
24.80 2.32 1.42 7.75 -0.58 11.49 -1.61 -3.32 4.97 -1.65 4.93 5.66 9.07 12.06 3.62 8.71 13.94 2.67 5.80 16.81 5.98 13.36 16.18 2.30 5.95 12.21 1.19
[0.78] [1.21] [0.55]
2.52 -2.70 -0.70
-0.08 0.04 0.01
-2.44 2.66 0.69
[1.19] [0.70] [1.25] [0.64]
-2.55 -1.49 -2.22 -0.94
0.05 0.03 0.04 0.02
2.50 1.46 2.17 0.92
[6.52]*** [0.69] [0.60] [1.29] [0.22] [2.95]*** [0.18] [1.04] [1.55] [0.45] [1.33] [0.73] [3.01]*** [2.23]** [1.67]* [2.50]** [4.50]*** [0.80] [1.01] [5.63]*** [2.19]** [2.25]** [2.09]** [0.88] [3.98]*** [3.50]*** [0.46]
Partner‘s characteristics Standard errors are clustered by *significant at 10%; **significant at 5%; ***significant at 1%
The marginal effects of the territorial variables on the probability of PT participation generally appear to be of limited size. As expected (see Table 2.3), highly educated women tend to participate fulltime; women with university education are less likely to participate PT than all the other categories. The marginal effects seem to be particularly high and are always
2 A Microeconometric Analysis of Female Labour Force Participation in Italy
35
statistically significant at the 1% level. Elder women are more likely not to participate. Family composition has a strong impact on female LFP. The presence of children aged less than 15 has a negative effect on both FT and PT participation, with a stronger effect on the former. The size of the effect tends to decline with children’s age: women with one child aged less than two have a 22.13 percentage points lower probability of NP than women without children while the ‘‘participation penalty’’ is only 5.6 percentage points for women with children aged 7–14. These differences in the effects of progeny by age suggest that our estimates are likely to measure causal effects of children on female participation, as we anticipated in Sect. 2.4, rather than simply reflect women’s unobserved heterogeneity in taste for children and LFP. Both family size and the number of employed children have negative effects, significant only at 10% level, on female LFP. Women living in families where other family members work are more likely to participate (probably because of family network effects). The presence of disabled members in the family is associated with a higher nonparticipation probability (+7.64 percentage points). When inactive grandmothers live in the family, women are more likely to participate (+7.75 percentage points FT and +0.15 percentage points PT). This effect can be explained in terms of informal child care that elder relatives may provide to participating women. Single women (see Table 2.5) are more likely to participate FT (+24.80 percentage points) and less likely to participate PT (-0.45 percentage points) with respect to the reference category (women with a partner working as a manager in the private services with a ‘‘typical’’ contract, i.e., a FT permanent position). Women with partners in the reference category are more likely not to participate with respect to all the other groups. It seems that an ‘‘income effect’’ operates: women with a partner in a high income job are less likely to participate. The other partner’s characteristics do not seem to matter.
2.6 Concluding Remarks In this chapter we have analysed female LFP using Italian Quarterly LFS data. A number of factors emerge as important determinants of female labour force attachment. Institutional factors such as child care and elderly care availability, which are highly differentiated across the country, and local unemployment rates all affect female behaviour. A higher provision of child-care and elderly-care services and a lower unemployment rate are positively associated with female LFP. These three variables apparently account for the generally observed lower participation rate in central and southern Italian regions. However, given that public care services are measured at regional level, these effects might partly capture other regional
36
M. Bratti and S. Staffolani
influences. PT employment opportunities at local level appear to increase female LFP only for women with very young children. In view of female employment rate increasing also in Italy, our analysis suggests that policy makers should improve care services, both for children and for elderly people. For instance, if our estimates fully reflect ‘‘causal effects’’, doubling the current child-care coverage ratio (in 2000, there were 7.62 public kindergarten seats every 100 children aged 0–2) would imply an increase of 11.35 percentage points in the female FT participation rate and of 0.23 percentage points in the female PT participation rate. Doing the same for the elderly-care coverage ratio (in 1999, there were 2.09 places in public elderly-care services per 100 seniors aged more than 65), would produce an increase of 5.4 percentage points in the female FT LFP. The presence of children aged less than 15 negatively affects female LFP, with a decreasing effect by children’s age. Our analysis does not support the idea that in Italy, in the period under study, PT was a way for women to reconcile home and market work, since the presence of young children rose women’s probability of inactivity, while having a negligible effect on PT participation. This may also be due to the period we have analysed (2002), since the diffusion of PT work is only a very recent phenomenon in Italy, and, therefore, it would be very interesting to replicate our analysis with current LFS data. The presence of inactive grandmothers in the family is positively correlated with participation and suggests the importance of informal child care. Single women and highly educated women are more likely to participate. Because of evident discouragement effects, unemployment reduction, which is an obvious way of increasing female employment, would also help to increase women employment by raising female participation. Last but not least, widening tertiary education, by increasing women’s participation in higher education and reducing drop-out rates, will improve female attachment to the labour force: the difference in the probability of FT participation between women with upper secondary schooling and university education, for instance, amounts to about 19 percentage points.
References Aaberge, R., Colombino, U., & Strøm, S. (1999). Labour supply in Italy: An empirical analysis of joint household decisions, with taxes and quantity constraints. Journal of Applied Econometrics, 14(4), 403–422. Aaberge, R., Colombino, U., & Strøm, S. (2000). Labour supply responses and welfare effects from replacing current tax rules by a flat tax: Empirical evidence from Italy, Norway and Sweden. Journal of Population Economics, 13(4), 595–621. Aaberge, R., Colombino, U., Strøm, S., & Wennemo, T. (1998). Evaluating alternative tax reforms in Italy with a model of joint labour supply of married couples. Structural Change and Economic Dynamics, 9(4), 415–433. Addabbo, T. (1999). Labour supply and employment probabilities in Italy: A gender analysis in a regional perspective. Economia e Lavoro, 33(3–4), 189–207.
2 A Microeconometric Analysis of Female Labour Force Participation in Italy
37
Borzaga, C., & Contrini, E. (1999). L’offerta di lavoro femminile tra partecipazione e squilibri Un’analisi su un campione di donne residenti nel comune di Trento. Economia e Lavoro, 33(3–4), 57–74. Bratti, M. (2003a). Partecipazione e orario di lavoro. In C. Lucifora (Ed.), Mercato, occupazione e salari: la ricerca sul lavoro in Italia. I. Capitale umano, occupazione e disoccupazione (pp. 30–57). Milano: Mondadori Università. Bratti, M. (2003b). Labour force participation and marital fertility of Italian women: The role of education. Journal of Population Economics, 16(3), 525–554. Bratti, M., Del Bono, E., & Vuri, D. (2005). New mothers’ labour force participation in Italy: The role of job characteristics. Labour, 19(s1), 79–121. Browning, M. (1992). Children and household economic behavior. Journal of Economic Literature, 30(3), 1434–1475. Cameron, S., & Heckman, J. (1998). Life cycle schooling and dynamic selection bias: Models and evidence for five cohorts of American males. Journal of Political Economy, 106(2), 262–333. Chiuri, M. C. (2000). Quality and demand of child care and female labour supply in Italy. Labour, 14(1), 97–118. Colombino, U., & Del Boca, D. (1990). The effect of taxes on labour supply in Italy. Journal of Human Resources, 25(3), 390–414. Colombino, U., & Di Tommaso, M. L. (1996). Is the preference for children so low or is the price of time so high? A simultaneous model of fertility and participation in Italy. Labour, 10(3), 475–493. Del Boca, D. (1993). Offerta di lavoro e politiche pubbliche. Roma: NIS. Del Boca, D. (2002). The effect of child care and part time opportunities on participation and fertility decisions in Italy. Journal of Population Economics, 15(3), 549–573. Di Tommaso, M. L. (1999). A trivariate model of participation, fertility and wages: The Italian case. Cambridge Journal of Economics, 23(5), 623–640. European Commission (2002) Comunicazione della Commissione al Consiglio, al Parlamento Europeo, al Comitato Economico e Sociale e al Comitato delle Regioni. Piano d’azione della Commissione per le competenze e la mobilità. European Commission, Bruxelles. Lauer, C. (2003). Family background, cohort and education. A French–German comparison based on a multivariate ordered probit model of educational attainment. Labour Economics, 10(2), 231–251. Marenzi, A., & Pagani, L. (2005). The impact of elderly parents on labour market participation of Italian women. Rivista di Politica Economica, 95(3–4), 155–189. Nakamura, A., & Nakamura, N. (1992). The econometrics of female labour supply and children. Econometric Reviews, 11(1), 1–71. Prieto-Rodríguez, J., & Rodríguez-Gutiérrez, C. (2003). Participation of married women in the European labour market and the ‘added worker effect’. Journal of Socio-Economics, 32(4), 429–446. Tanda, P. (1994). Partecipazione femminile in Italia: Evidenza empirica su dati individuali. Economia e Lavoro, 28(1), 123–134.
Chapter 3
Balancing Work and Family: New Mothers’ Employment Decisions During Childbearing Andrea Neri, Martina Lo Conte and Piero Casadio
3.1 Introduction A rise in female labour-market participation and a decline in birth rates have been observed in most European countries.1 Nevertheless, in the past decade, several countries, notably France, Spain and Germany, have experienced a joint increase in female participation and fertility, mainly because of national policies aimed at balancing work and family life. In Italy, by contrast, fertility has remained relatively stable (after a period of significant decline) while increases in female participation rates have been modest. Consequently, Italy today has one of the lowest fertility rates and one of the lowest levels of female employment.
1
The views expressed in this paper are those of the authors and do not necessarily reflect those of the Bank of Italy or of Istat. We are grateful to Renata Bottazi, Luigi Cannari, Giovanni D’Alessio and Federico Signorini for their helpful comments. Although the paper is a collective work, Sects. 3.1 and 3.2 are to be attributed to Piero Casadio; Sects. 3.3 and 3.7 to Martina Lo Conte; Sects. 3.4, 3.5 and 3.6 to Andrea Neri. A. Neri (&) P. Casadio Economic Research Department, Bank of Italy, Via Nazionale 91, 00184 Rome, Italy e-mail:
[email protected] P. Casadio (&) e-mail:
[email protected] M. L. Conte (&) Social Statistics Department, Istat, Istituto Nazionale di Statistica, Viale Liegi 13, 00198 Rome, Italy e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_3, Ó Springer-Verlag Berlin Heidelberg 2012
39
40
A. Neri et al.
Numerous explanations for this apparent anomaly have been advanced in the literature, and they generally cite social and cultural aspects, rigidities in the labour market, and inefficiencies in the child-care system. This paper studies the employment decisions of Italian mothers (and their determinants) during the childbearing period, using for the purpose the Italian Birth Sample Survey. The hypothesis is that this period is particularly critical, and that the difficulties faced by new mothers are among the factors that contribute to keeping their participation rate at a low level. The paper is organised as follows. Section 3.2 presents a brief review of the applied literature on female employment decisions. Section 3.3 describes the 2002 Italian Birth Sample Survey (IBSS) data. Section 3.4 discusses the problem of selection and endogeneity. Section 3.5 examines the determinants of the probability that mothers will not be in work about 18–21 months after childbirth. The focus then shifts to working mothers, examining their voluntary decision to quit work after delivery. Section 3.7 makes some concluding remarks.
3.2 The Main Determinants of Mothers’ Working Status The presence of children seems broadly to influence female employment patterns in most European countries. In 2003 the average European Union employment rate for women aged 20–49 without children under 12 was 75.1%. This rate decreases by about 23% points for mothers in the 2 years after childbirth. It subsequently increases with the age of the youngest child, rising by about 8% points with each change of age class from 0 to 2 years to 3–5, and then to 6–11 years (Eurostat 2005). The average European Union employment rate for women aged 20–49 with the youngest child aged over 12 seems to be about 5–10% points below that for women without children.2 In Italy, the employment pattern of mothers is quite different. The employment rate for women aged 20–49 without children under 12 is only 60.4%. The employment rate drops in the first 2 years after childbirth (by about 10% points), and there is no significant recovery when the youngest child becomes more than 3 or 6 years old. Work and fertility decisions jointly depend upon a broad set of factors, such as individual preferences, household composition, social habits, labour-market flexibility, childcare costs, and availability. We now present an overview of the main empirical results of previous studies (for more comprehensive surveys see, among others, Dex and Joshi 1999; Jaumotte 2003 and Boeri et al. 2005).
2
This time-pattern remains fairly similar for mothers with two children, while a third child negatively affects the probability of re-entering the labour market. In particular, for women with at least three children, the employment rate remains 10-20 percentage points below that for women without children, even when the youngest child is aged over 12.
3 Balancing Work and Family
41
Social and cultural aspects—In the vast majority of countries, mothers work longer hours than do fathers, and they cumulate paid jobs and unpaid duties. Different social habits strongly influence the division of caring tasks between partners, and therefore the distribution of paid working hours (Esveldt 2003; Sleebos 2003). The main determinant of a balanced distribution of duties and work within a couple is the woman’s human capital. A higher level of female education is associated with greater labour-market participation and a closer attachment to work (Gutiérrez-Domènech 2005; Jaumotte 2003). In Italy, too, the employment rates for women with tertiary educations are in line with the European average levels, for any number of children aged under 12 (Eurostat 2005; Table 3.2). Labour market flexibility and regulation—The diffusion of part-time contracts is also positively correlated with female participation (Bardasi and Gornick 2008; OECD 2003b). However, flexible contracts may increase female segregation in low-paid positions with reduced career opportunities and greater job insecurity, which in turn reduces labour-market attachment among new mothers (Addabbo 2003). Long or repeated spells of non-employment are likely to diminish the mother’s human capital and future wages (Beblo and Wolf 2002). Removing labour-market distortions could improve social welfare and reduce wage inequalities (Kamerman et al. 2003). Public support for families—Female participation is also influenced by government policies: family taxation, parental leave, childcare subsidies and child benefits (Palomba 2003; OECD 2003a). In particular, the availability of public childcare, and its flexibility in terms of hours provided per day, have a positive effect on women’s labour supply (Jaumotte 2003; Gornick et al. 1997; Del Boca 2003). In many European countries, the support given by relatives to the household, in terms of financial transfers and time, increases the opportunities for mothers to work and have children (Esveldt 2003). However, the presence of elderly relatives in the household may have a double role in explaining women’s work patterns: elderly relatives can provide household services such as childcare and domestic help, but they may also themselves require unpaid help, thereby discouraging the work participation of women (Marenzi and Pagani 2008).
3.3 The 2002 Italian Birth Sample Survey Our analysis uses data from the Italian Birth Sample Survey (IBSS), which was carried out for the first time in 2002 by the Italian National Statistics Office (Istat).3 The overall sample consisted of 50,408 births registered between the second half of 2000 and the first half of 2001, this being around 10% of total births in that period. Mothers were interviewed, using the CATI technique, about 18–21 months
3
The Survey’s structure and main results are described in Istat (2003, 2004, 2006a, b) and in CNEL (2003).
42
A. Neri et al.
after delivery. The questionnaire contained questions gathering general sociodemographic information on the new-born child, the delivery, and the parents (‘short’ form). The results presented in the next sections are based on one-third of the sample—16,788 mothers—interviewed with a ‘long’ form containing further sections about mothers’ working conditions before and after childbearing, the household composition, the formal and informal childcare networks, and the division of household chores. Employment before childbirth—Around 60% of the IBSS new mothers had been in employment when they discovered that they were pregnant. This percentage is about 9 points higher than that for all women reported by the 2001 Labour Force Survey (LFS). In the LFS, the female employment rate was about 47% between the ages of 15 and 49, and 51% between the ages of 20 and 40.4 The share of mothers working before childbirth was about 75% in the Northern regions of Italy, 65% in the Centre, and 43 and 38%, respectively in the Islands and the South. Also those percentages are much higher than the corresponding ones resulting from the 2001 LFS. Moreover, according to IBSS survey, a larger share of women worked in the public sector (17 vs. 14% in the 2001 LFS).5 All the above-mentioned differences suggest that a selection process is probably at work, given the correlation between fertility decisions and the current job status of women. Indeed, it seems likely that women with a higher degree of job instability prefer to postpone childbearing. We sought to control for this selection process in our estimation (see the following section). Changes in working status—About 47% of the IBSS new mothers were employed both before and after childbearing, thereby exhibiting a strong attachment to their jobs. By contrast, 37% of new mothers remained out of the labour force [inactive]: 16% had never worked before, while 21% had had at least one previous work experience. The remaining 16% of new mothers changed their working status: 4% started to work after delivery while 12% lost their jobs in the period around childbirth. The IBSS drop-out percentage for new mothers is far higher than that for all women: in the period 2001–2002, according to the Labour Force Survey,6 only about 6% of working women aged between 20 and 40 shifted to non-employment status, while in the IBSS about 20% of mothers in employment at the beginning of their pregnancy stopped work thereafter. In both surveys, drop-out transitions increase for women working on temporary contracts, or with no contract, and for part-time workers. In general, there is a negative balance between the flows of women entering the labour force after childbirth and those leaving it. A positive contribution is made
4 A significant difference between the two statistics remains even when the age structure used by the IBSS is imposed on LFS the data. 5 Specifically, the share of employed mothers in the IBSS is about 6% points above the LFS share in the Northern regions; 10 pp in the Centre and more than 12 pp in the South and Islands. 6 In the LFS, the transitions are only available at a one-year distance.
3 Balancing Work and Family
43
by part-time contracts: about 27% of full-time mothers shifted to part time, boosting the share of voluntary part-time from about 6% to about 30%. A part-time contract is probably the most attractive alternative to work exit. The reverse transition from part-time to full-time employment after childbirth only happens in less than 1% of cases. Childcare—Mothers still working after childbirth are mainly able to reconcile work and family responsibilities by relying on the help of grandparents. In 55% of households, the elderly provide vital support in caring for children. Many families prefer to trust their own relatives for reasons of reassurance. Moreover, the role played by relatives is of particular importance because of its low costs associated with high flexibility, advantages not always available from childcare services. About 20% of mothers use childcare services (both public and private) and only 10% use child-minders. Some 12% of new mothers are unable to use childcare, the main reasons cited (47% of cases) being a lack of available places and limited opening hours. This is especially the case of mothers in the South and Islands. The fairly limited use of private childcare facilities is principally due to the costs (20% of cases), which are particularly high for mothers resident in the North.
3.4 Selection into Motherhood and Endogeneity The main problem when studying new mothers’ employment patterns with the IBSS data was the potential selection into motherhood bias due to the correlation between fertility and employment decisions. In the IBSS, employment patterns are observed only for mothers. The unavailability of a control group of women without children could have given rise to biased estimates if a selection process underlying the fertility decision was at work and was ignored.7 The test for selection into motherhood was performed using the Heckman approach (Heckman 1979). For the sake of simplicity, we report only the results of the tests; a more detailed description of the procedure can be found in Casadio et al. (2008). Since the IBSS does not contain information about women without children, we were forced to use external data. We drew on the Survey on Household Income 7 In the IBSS data, the sample selection problem may heavily bias estimates of the probability of new mothers being in work 18–21 months after childbirth. Consider, for example, the extreme situation in which women’s working conditions are either fully protected (as in the public sector) or not protected at all (as in the case of a fixed-term contract with a small private firm). If the degree of protection were the only determinant of having children—total sample selection—only women benefiting from a high degree of job protection would have a significant probability of having a child. As a consequence, the sample would consist mainly of mothers working in protected sectors, and most of them would retain their jobs after childbirth. Ignoring this selection process would probably give rise to wrong conclusions (for example, according to the data, employment protection legislation would have little effect on new mothers’ employment patterns).
44
A. Neri et al.
and Wealth (conducted by the Bank of Italy), from which we selected a sample of women. A post-stratification of weights was then performed according to the distributions of women’s age, level of education, and geographical area of residence, using census data for the purpose. This technique made the selected sample of women representative of the whole population and, therefore, comparable with the IBSS survey. In the selection equation, the probability of having a child (0–3 years old) was estimated given a set of mothers’ characteristics: age, level of education, geographical area, size of municipality of residence, and distance (in hours) from relatives. The ‘excluded instrument’ was the number of mother’s siblings which was assumed to be a proxy for mother’s preference for children. After controlling for the mentioned covariates, the hypothesis of no selection could not be rejected with a significance level of 0.78.8 The correlation between fertility decisions and employment patterns may also have resulted in the endogeneity of the decision of whether or not to have more than one child. For instance, working mothers with one child may decide not to have another child so that they can more easily reconcile work and family (they may have an unobserved low propensity for large households). This hypothesis was tested, for the IBSS data, using a bivariate probit model. We jointly modelled the probability of having more than one child and the employment probability. The ‘excluded instrumental variable’ used in the first equation was the number of pregnancies undergone by the mother before having her last child (including abortions and miscarriages). This variable is a proxy for women’s difficulties in having a child, and it is strictly related to the decision to have more than one child. It can be assumed to be unrelated to employment decisions. More than one test was performed using different response variables (more/less than one child, more/less than two, more/less than three), and they showed that neither could the hypothesis of no endogeneity be rejected.
3.5 Mothers’ Probability of not Being in Work 18–21 Months After Delivery In this section we analyse the factors influencing the new mothers’ risk of not being in work 18–21 months after delivery. In the IBSS, the time span of 18–21 months after childbirth is the only period available to study mothers’ employment patterns. However, this period is also the most informative, given Italian legislation on
8
A similar result is reported by Bratti, Del Bono and Vuri (2005). In order to test for selection into motherhood, the authors estimated a probit model with sample selection, where the selection equation was represented by the decision to have a first child and the main equation was the employment equation. In none of the specifications of their model did they find a significant correlation between the error terms of the employment and fertility equations.
3 Balancing Work and Family
45
maternity leave—covering at most 12 months after delivery—and considering the results of previous studies on women’s career interruptions.9 The model used was the following bivariate probit: Y1i ¼ 1fX1i b1 þ e1i 0g Y2i ¼ 1fX2i b2 þ Z2i c þ e2i 0g where the error terms have a bivariate normal distribution with a correlation coefficient q = 0. In the first equation the dependent variable Y1i was a dummy variable indicating whether or not the woman is inactive/unemployed. To be noted is that, unlike in the LFS,10 in the IBSS survey no definition of unemployment is provided during the interview. Previous studies show that, in Italy, the work attachment of unemployed and inactive persons is very similar (Viviano 2002). We therefore merged the two categories. The explanatory variables were: mother’s individual characteristics, including age, nationality, and level of education; spouse/live-in partner’s characteristics (his employment position and level of education); composition of the household, including the presence of grandparents; mother’s working status and attainment before childbirth, such as job position and type of contract (part time/full time, permanent/fixed term); social services indicators (the share of available places in nursery schools for children aged under 3, on a regional basis).11 In the second equation the response variable was a dummy indicating whether or not the child attended a nursery school. The reason for including this equation was that it enabled us to take into account the simultaneity of mother’s decisions.12 The covariates were the same as those included in the first equation plus two variables intended to proxy the latent attitude towards nursery schools: i.e. dummy variables indicating that, according to the interviewee, socialisation and
9
As documented in Solera (2003), Italian women are unlikely to experience a career break more than once in their lives, and this usually occurs in correspondence with the birth of the first child. Moreover, Bratti et al. (2005), on analysing the new mothers’ employment decisions during the 3-year period following the birth of the first child, found that the probabilities of employment are very similar in each year of observation. 10 In the Labour Force Survey, the definition of an unemployed person given to the respondent was someone who had actively looked for a job in the 60 days previous to the interview and intended to start work immediately. This measure could be therefore quite different from that used by the IBSS. 11 We initially also included regional dummy variables and labour market indicators, but their effect was then captured by women’s employment status during pregnancy. These variables were therefore not included in the final specification of the model. 12 It is important to note that this variable could not be included among the covariates of the first equation because it was endogenous. Indeed, the probability of schooling a child depends, among other things, on whether the mother is working, on the household’s total income, and on nationality. We also tested for the absence of endogeneity of this variable, and the results showed that this hypothesis must be rejected.
46
A. Neri et al.
educational methods are of key importance in a child’s growth.13 The results show that there is a strong negative correlation between the use of a nursery school and the probability of not working (q = -0.55). The unconditional average probability of not working after delivery is about 49%. As expected, an increase in the mother’s age reduces this probability (see Table 3.1). By contrast, mothers aged under 24 seem to face greater difficulties: their average probability of not working is about 72% (23 points above average). Education level also plays an important role. The higher this level, the lower the probability of not working: on average, a mother with a university degree is about 43% points more likely to work than a mother with compulsory schooling. The risk of not working after childbirth is slightly greater for mothers with a previous child aged under 3. This result may reflect the fact that mothers whose youngest child is aged over 3 are likely to have organised a network that enables them to keep working. Therefore the arrival of a new baby has a lower impact on their employment patterns. The most important effect is exerted by the mother’s employment status before childbirth. An inherited non-working status is very persistent, and almost completely so for housewives and students with no previous working experience. Compared to those mothers, the chances of finding a new job significantly increase for women with some past working experience. At the opposite extreme, mothers who are managers or entrepreneurs, and mothers benefiting from the employment protection provided by the public sector, display a close attachment to their profession. A possible explanation for this is the high implicit costs that these mothers would incur if they had to leave their jobs. Moreover, in the public sector, mothers are likely to have more flexible working hours, which makes it easier for them to reconcile work and family life. The importance of employment protection is confirmed by the magnitude of the coefficients relating the type of contract. Working mothers with previous temporary contracts have a significantly higher probability of not working after childbirth compared to mothers with permanent jobs. The family network has a great influence as well. The presence of grandparents increases the probability of working by about 25%. Of course, the presence of elderly relatives may require women to devote time to caring for residing and/or non-residing elderly kin. Although this effect cannot be estimated, the results show that the positive effect is larger than the negative one due to the need to care for the elderly (this is consistent with previous research, see Marenzi and Pagani 2008). Finally, also the availability of childcare services is associated with the employment probability. In regions where the proportion of young children (0–2) using crèches is higher than 12%, the new mothers’ probability of being employed doubles with respect to those regions where less than 10% of 0–2 children use childcare.
13
Under the assumption that those variables are not correlated with the error terms in the first equation they represent two ‘‘excluded instruments’’. In case this assumption doesn’t hold, the results cannot be interpreted as casual effects.
3 Balancing Work and Family
47
Table 3.1 Probability of not being in work 18–21 months after childbirth Variable Marginal probability of Mother’s age at childbirth B24 25–29 30–34 35–39 40 ? Mother’s education Compulsory schooling High school diploma University degree Mother’s nationality (Foreign) Children before pregnancy None Youngest child aged 0–2 Youngest child aged 3–5 Youngest child aged over 6 Mother’s previous employment status Housewife without working experience Housewife with working experience Student without working experience Student with working experience Employed in private sector Manager, middle management Office worker, school teacher Factory worker Entrepreneur Member of arts or professions Sole proprietor and other self-employed Employed in public sector Mother’s type of contract Permanent status, working full time Permanent status, working part time Fixed-term contract working full time Fixed-term contract working part time Partner’s working status Not employed/single mother Office worker Manager, middle management Entrepreneur, self-employed Partner’s level of education High school diploma University degree Presence of grandparents
Not being in work
Using crèches
72.0 51.9 43.8 40.7 41.6
41.1 25.3 20.0 18.6 19.5
67.9 43.6 24.5 65.2
36.3 20.0 9.3 39.8
55.7 59.1 55.0 55.9
28.5 31.8 28.6 28.1
99.4 90.2 99.8 63.6
79.2 40.2 86.6 21.7
12.9 18.4 37.6 11.9 8.2 20.6 9.7
2.5 3.9 7.9 1.3 1.2 2.6 1.2
37.1 40.2 51.9 59.6
8.4 9.4 12.5 13.6
58.8 49.8 34.8 49.9
25.7 24.8 15.6 24.4
43.1 31.6 24.7
20.1 14.5 2.7 (continued)
48 Table 3.1 (continued) Variable Child-care systema Less than 10% 10–12% More than 12% Total a
A. Neri et al.
Marginal probability of Not being in work
Using crèches
59.4 35.5 29.1 49.2
30.3 15.2 13.9 24.1
Proportion of young children (0–2) using public child-care
3.6 The Determinants of Voluntary Transition to Non-employment After Childbirth In this section we focus on mothers who are employed at the beginning of pregnancy, and we analyse the determinants of voluntary withdrawal from the labour market after delivery. The sub-sample analysed consisted of 9,894 units. As in the previous section, we take into account the simultaneity of mother’s decision using a bivariate probit. In the first equation the dependent variable was a dummy indicating whether the mother had declared that she had left her job voluntarily (and was not working at the time of the interview). These women represented 14% of cases. The decision to leave was mainly motivated by ‘‘the need to spend more time with children’’ (60%) or by ‘‘difficulties in reconciling work and family life’’ (20%). This variable was therefore a good proxy for the mother’s decision to leave work to take care of her family if no reconciliation was possible. As expected, the availability of a nursery school was negatively related to the probability of voluntary exit (q = -0.65). Table 3.2 summarises the results of the analysis. As to employment status, employees, and in particular factory workers, show a higher propensity to leave their jobs than entrepreneurs and members of the arts and professions. The degree of job stability and protection confirms its importance in helping mothers into paid employment. For mothers working in the public sector, the average exit probability is 5% (about one-third of the overall mean). By contrast, in the trade and services sector this probability increases to 18%. A stronger effect is produced by fixed-term contracts, which increase the exit probability to 46% when the mother is also working part time.14
14
By contrast, fixed-term contracts do not seem to a have an appreciable impact. This is not surprising, given the dependent variable. In the analysis, if the contract expired and was not renewed, it was not classified as resignation.
3 Balancing Work and Family
49
Table 3.2 Probability of voluntary work exit after childbirth Variable Marginal probability of Mother’s age at childbirth B24 25–29 30–34 35–39 40 ? Mother’s level of education Compulsory or not schooling High school diploma University degree Mother’s nationality (Foreign) Mother expecting a new baby Children before pregnancy None Youngest child aged 0–2 Youngest child aged 3–5 Youngest child aged over 6 Employed in private sector Manager, middle management Office worker, school teacher Factory worker Entrepreneur Member of arts or professions Sole proprietor and other self-employed Employed in services/commerce sector Employed in public sector Mother’s type of contract Permanent status, working full time Permanent status, working part time Fixed-term contract, working full time Fixed-term contract working part time Partner’s working status Not employed/single mother Office worker Manager, middle management Entrepreneur, self-employed Partner’s level of education High school diploma University degree Presence of grandparents Childcare systema Less than 10% 10–12% More than 12% Total a
Voluntary exit
Using crèches
26.0 15.3 12.4 10.5 9.4
3.6 2.0 1.6 1.4 1.1
23.3 12.0 6.7 23.5 14.6
2.9 1.6 0.9 3.5 1.8
14.5 16.3 12.7 12.7
2.1 1.4 1.5 1.2
9.3 13.0 28.1 1.1 4.4 12.1 18.2 4.7
1.6 2.0 3.6 0.0 0.6 1.0 2.6 0.4
11.8 15.4 33.1 46.4
1.6 1.3 4.2 6.6
13.6 15.4 21.1 46.4
1.3 1.3 2.0 6.6
11.9 8.2 5.9
1.6 1.2 0.1
15.7 12.6 9.5 13.8
1.9 1.7 1.9 1.8
Proportion of young children (0–2) using public child-care
50
A. Neri et al.
Among demographic characteristics, age and level of education exert the strongest influence. For working mothers aged under 24, the probability of quitting work is 26% (twice the average). The level of education confirms its influence not only on labour-market entry but also on the decision not to leave it after childbirth. For mothers with a university degree, this risk decreases by about 16% compared to mothers with the lowest level of education. As in the previous analysis, the probability of withdrawing is—on average—higher for mothers with a previous child aged under 3. The partner’s working status has a significant influence on a mother’s decisions. When the partner is an entrepreneur or a self-employed worker, the probability of voluntarily quitting work increases to about 46%. The presence of grandparents significantly prevents withdrawal by mothers: the exit probability decreases by about a half. Moreover, the likelihood of leaving the labour market decreases in areas with a better supply of childcare services. In regions with a low availability of childcare places (less than 10%) the probability of voluntary exit is about 16% (two points above the overall average). This percentage drops to about 9% in areas with the higher supply of public childcare services.
3.7 Concluding Remarks Italy stands out among European countries in that it combines one of the lowest female employment rates with one of the highest long-term unemployment rates. Even the large increase in female participation since 1995 has been insufficient to close Italy’s gap with other European countries, and especially so in its Southern regions. In this paper we have focused on the employment patterns of Italian mothers during the childbirth period. In the 2 years surrounding childbirth, 20% of women in employment before pregnancy leave the labour market, while only 4% start to work after delivery. Most of the drop-out probability—about 70%—is due to voluntary work exit, while one-fourth is due to temporary employment or firm’s bankruptcy. Both the risk of not working and of voluntary leave after childbirth vary according to a similar set of characteristics. The presence of grandparents is probably one of the key factors in increasing the probability that a mother will remain in the labour market. A second important factor is the availability of public childcare facilities. Human capital variables have a significant influence on the continuation of mothers in paid employment. As education and job attainment levels increase, so the unemployment and voluntary exit probabilities fall, because of the high (implicit) costs incurred by highly-educated and well-paid mothers if they leave their jobs.
3 Balancing Work and Family
51
The mother’s age plays an important role as well. In particular, mothers aged under 24 seem to have the greatest difficulties in reconciling work and family. Their average probability of not working is about 70% (the average is 49%), and their probability of voluntarily leaving their jobs is about 25% (twice the average); the latter probability jumps to 53% for those women unable to obtain any help from family or public services. The probability of staying in work also grows with the degree of job protection. When mothers work in the public sector, the probability that they will not work after delivery decreases by about 25%. Conversely, the probability for mothers with fixed-term jobs increases by 23%. Overall, part-time contracts have a positive net effect on the employment rate. Nevertheless, some points should be noted. On the one hand, a part-time contract may be an important alternative to unemployment. About 17% of the women in the sample shifted from full-time to part-time work in order to reconcile family and work. On the other hand, the probabilities of unemployment and of voluntary exit increase significantly for mothers working part time during pregnancy. A possible explanation for this is that the previous choice of a part-time job may also be an indicator of lower labour-market attachment. Moreover, in most cases the low attractiveness of such jobs in terms of remuneration, working hours and selffulfilment may convince mothers that being employed is not worthwhile.
References Addabbo, T. (2003). Atypical work in Italy in a gender perspective. In D. Del Boca & M. Repetto-Alaia (Eds.), Women’s work, the family and social policy: Focus on Italy in a European perspective. New York: Peter Lang Publishing. Bardasi, E., & Gornick, J. C. (2008). Working for less? Women part-time wage penalties across countries. Feminist Economics, 14(1), 37–72. Beblo, M., & Wolf, E. (2002). The wage penalties of heterogenous employment biographies: an empirical analysis for Germany. Mannheim: Centre for European Economic Research. Boeri, T., Del Boca, D., & Pissarides, C. (Eds.). (2005). Women at work. An economic perspective. Oxford: Oxford University Press. Bratti, M., Del Bono, E., & Vuri, D. (2005). New mothers’ labour force participation in Italy: The role of job characteristics. Labour, 19(s1), 79–121. Casadio, P., Lo Conte, M., Neri, A. (2008). Balancing work and family in Italy: New mothers’ employment decisions after childbirth. Temi di discussione (Economic working papers) 684. Economic Research Department, Bank of Italy. CNEL (2003). Maternità e partecipazione delle donne al mercato del lavoro tra vincoli e strategie di conciliazione. Atti, Documenti, 49. Del Boca, D. (2003). Labour market participation and fertility of Italian women. In D. Del Boca & M. Repetto-Alaia (Eds.), Women’s work, the family and social policy: Focus on Italy in a European perspective. New York: Peter Lang Publishing. Dex, S., & Joshi, H. (1999). Careers and motherhood: Policies for compatibility. Cambridge Journal of Economics, 23, 641–659. Esveldt, I. (2003). Caring for children. The childcare dilemmas that working parents face [Special Issue]. Demos, 19. Eurostat (2005). Gender gaps in the reconciliation between work and family life. Statistics in focus, 4.
52
A. Neri et al.
Gornick, J. C., Meyers, M. K., & Ross, K. E. (1997). Supporting the employment of mothers: Policy variations across fourteen welfare states. Journal of European Social policy, 7, 45–70. Gutiérrez-Domènech, M. (2005). Employment after motherhood: A European comparison. Labour Economics, 12, 99–123. Heckman, J. (1979). Sample selection bias as a specification error. Econometrica, 47(1), 153–161. Istat. (2003). Annual report. Rome: Istat. Istat. (2004). Annual report. Rome: Istat. Istat (2006a). L’indagine campionaria sulle nascite. Obiettivi, metodologia e organizzazione. Metodi e Norme, 28. Istat (2006b). Avere un figlio in Italia—Approfondimenti tematici dall’Indagine campionaria sulle nascite (anno 2002). Informazioni, 32. Jaumotte, F. (2003). Female labour force participation: Past trends and main determinants in OECD countries. Economic department working papers, 376. Kamerman, S.B., Neuman, M., Waldfogel, J., & Brooks-Gunn, J. (2003). Social policies, family types and child outcomes in selected OECD countries. OECD social, employment and migrations working papers, 6. Marenzi, A., Pagani, L. (2008). The labour market participation of ‘‘sandwich generation’’ Italian women. Journal of Family and Economic Issues, 29(3), 427–444 OECD. (2003a). Babies and bosses: Reconciling work and family life—Austria, Ireland and Japan. Paris: OECD Publications. OECD. (2003b). Employment outlook 2003. Paris: OECD. Palomba, R. (2003). Reconciliation of working life and family life. Council of Europe European Population Papers Series, 10, Council of Europe. Sleebos, J. (2003). Low fertility rates in OECD countries: Facts and policy responses. OECD social, emplyoment and migration working papers, 13. Solera, C. (2003). Changes in women’s work histories across cohorts: to what extent are they due to a compositional effect ? A comparison of Italy and Great Britian. Working paper. European University Institute, Florence. Viviano, E. (2002). Un’analisi critica delle definizioni di disoccupazione e partecipazione in Italia. Temi di Discussione, 450.
Chapter 4
Part-Time and Temporary Employment: A Gender Perspective Tindara Addabbo and Donata Favaro
4.1 Introduction The focus in this chapter is on part-time and temporary work; we analyse these two different types of ‘non-standard’ work from a gender perspective. Our interest in these types of non-standard work is justified by their increasing weight with respect to total employment in Italy. At the same time the over-representation of women in jobs of this type—especially part-time work—and the risks of employment segregation to the disadvantage of women, as outlined by the literature (Sect. 4.2), require a gender perspective. We will try to disentangle not only the inequalities in the take-up of non-standard work by gender but also the gender differences in the determinants, and the factors conducive to the ‘choice’ of nonstandard employment (Sect. 4.4). The literature shows that a lower proportion of temporary workers than parttimers describe themselves as voluntary. Moreover, amongst voluntary part-timers, previous analyses have shown that for women there is a higher share than for men of those workers who state that they are voluntary part-timers for family reasons. Given the higher percentage of part-timers amongst employed women and the significant share of women indicating that they have chosen part-time work for family reasons, we have investigated more in depth in Sect. 4.5 the reasons for
T. Addabbo (&) Department of Economics, University of Modena and Reggio Emilia, Viale Berengario 51, 41121 Modena, Italy e-mail:
[email protected] D. Favaro (&) Department of Economics M. Fanno, University of Padova, Via del Santo, 33, 35123 Padova, Italy e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_4, Ó Springer-Verlag Berlin Heidelberg 2012
53
54
T. Addabbo and D. Favaro
working part-time in Italy, taking into account also the observable probability of people working part-time because there is no alternative. Another crucial issue in the analysis of non-standard employment relates to the costs of being in non-standard work. Previous research has uncovered the costs in terms of lower wages, lower career perspectives and lower training (Sect. 4.2). In this chapter we will analyse the costs in terms of hourly wages, and the probability of gaining a supervisory role. Further, we will try to assess the costs in terms of the development of other significant dimensions of well-being. For this purpose we use the Italian Survey IT SILC (Sect. 4.3) part of the wider European Statistics on Income and Living Conditions Survey, a set of data that allows us to recover not only the money costs of different employment positions but also the costs connected to other dimensions of well-being, such as access to health services, that we will take into account in Sect. 4.6.
4.2 Part-Time and Temporary Employment: Trends, Reasons and Outcomes Temporary employment has significantly increased in Italy: by 52% from 1993 to 2006, with an increase in the overall share of women in temporary employment (51% of temporary employees in 2006).1 Women are more often found in nonpermanent employment than men, as is true also in other European countries (Goudswaard and Andries 2002). The literature on temporary work conditions has underlined the wage penalty associated with temporary work (Booth et al. 2002; Kalleberg et al. 2000; Mertens and McGinnity 2004), with a higher gap at the bottom of the distribution (Comi and Grasseni 2009; Mertens et al. 2007). Temporary employees are also characterized by a lower access to training (Arulampalam and Booth 1998; Booth et al. 2002; OECD 2004; Connell and Burgess 2006), by a higher perception of job insecurity (De Witte and Naswall 2003) and by lower job satisfaction (Booth et al. 2002; Kaiser 2005; SiebernThomas 2005; Berton et al. 2009) with special regard to such job components as employment stability, wages and career prospects. The literature also shows that for most women temporary work is involuntary (Amuedo-Dorantes 2000; Morris and Vekker 2001; Polivka 1996). The costs associated with temporary work and the higher likelihood of this choice being involuntary, have led us to investigate more in depth the determinants of temporary labour supply (Sect. 4.4) and the implications in terms of hourly wages, possibility of supervisory status and access to health services and treatments (Sect. 4.6).
1
Altieri, Ferrucci and Doto (2008) elaboration based on Istat Labour Force Survey data.
4 Part-Time and Temporary Employment
55
Part-time work is a form of non-standard employment mostly concentrated on women. In 2008 the share of women working part-time ranged from 14.1% in Poland to 59.9% in the Netherlands, with an average of 31.5% at EU-15 level against 7.5% for men (OECD 2009). Though decreasing, compared to 1998 (when it was 78.7% at EU-15 level) the majority of people working part-time are female (77.2% in 2008 at EU-15 level, OECD 2009) and the gender gap in this type of employment is the highest (Burchell et al. 2007). Institutional factors have affected the late development of part-time work in Italy (Addabbo 1997, 2003a). Part-time work as a percentage of total employment in Italy has increased from 11.2% in 1998 to 16.3% in 2008, women’s share of total part-time work increasing from 71.9% in 1998 to 75.8% in 2008; the percentage of women employed in part-time jobs is now similar to the EU-15 average.2 Notwithstanding that the European directive on part-time work (Council Directive 97/81/EC of 15 December 1997) stresses the principle of non-discrimination, still there is evidence of employment segregation and wage differentials in European countries with respect to part-time work. The literature has shown that part-time work is more likely to be associated with menial jobs, with lower pay, lower upward mobility and lower satisfaction with regard to career perspectives (Berton et al. 2009; Blank 1990; Connolly and Gregory 2008a, b, 2009; Ferber and Waldfogel 1998; Francesconi, 1991; Manning and Petrongolo 2008; Mumford and Smith 2009; Tam 1997; Tilly 1996; Villosio 2001). However, after accounting for differences in accumulated skills, preferences and job characteristics there is a little evidence of a part-time wage penalty for women in the USA according to Hirsch (2005); and part-time wages are higher in the Australian labour market, according to the analysis made by Booth and Wood (2008). The different degree of labour market attachment shown by women working part-time has been found to be related to the disadvantageous wage gap differential; this was, for example, observed in the UK prior to the abolition of less favourable treatment rules for part-time workers (Bowlus and Grogan 2009). The higher concentration of part-time workers in low-paid jobs led Bardasi and Gornick (2008) to focus on occupational segregation as a cause of the wage differentials observed between part-timers and full-timers in their international (USA, Canada, Italy, UK, Germany and Sweden) analysis, which shows heterogeneity in respect of the weight of observed differences in explaining wage differentials between full-timers and part-timers, and a prevalence, amongst the observed differences, of the occupational component. Differences between the occupations of part-time and full-time women are found to be the most important cause of the wage penalty of women working part-time by Manning and Petrongolo (2008); analysis in Britain also shows that an increased differential can be explained by differences in occupational distribution. By extending analysis of the outcomes of part-time work to other dimensions of individual and family well-being, Booth and Van Ours (2009) found a positive aspect of part-time work in the job satisfaction felt by part-time women, in contrast
2
OECD (2009).
56
T. Addabbo and D. Favaro
to women working full-time, as shown by their analysis of the interaction between partners’ employment conditions in Australian couples. However, Booth and Van Ours (2008) had reached a different conclusion from a similar analysis of British women: part-time work raises job satisfaction but does not affect life satisfaction. Another key point in evaluating the costs of part-time employment is to assess the extent to which part-timers state that their part-time employment is voluntary. According to Barrett and Doiron’s (2001) findings on Canada, involuntary parttime workers are more likely to earn lower wages than other workers, and the wage penalty is mostly related to selection effects and differences in the returns, rather than in differences in characteristics. Involuntary part-time workers are also characterized by a lower degree of job satisfaction in Italy according to the analysis of Berton, Richiardi and Sacchi (2009) of Isfol Plus data. Involuntary part-time work turns out to be relevant also in terms of the likelihood of switching to full-time employment. Employees in involuntary part-time work in the USA have been found to be more likely to switch to full-time jobs afterwards (Stratton 1996). The higher heterogeneity in the individual motives stated with respect to part-time choice and the implications outlined by the literature have led us to investigate the extent and determinants of involuntary part-time employment in Italy (Sect. 4.5).
4.3 The Data With the aim of measuring the costs of being in part-time or temporary work positions not only in terms of income, but also in terms of other dimensions of well-being (e.g. access to health services and treatment), and of having a wider view on the interaction between non-standard employment and living conditions, we have looked for a set of microdata allowing a wider assessment of worker wellbeing. Another focus of this analysis is the question of whether employment conditions are an outcome of choice or are constrained by different factors. In addition, the regional differences observed in the Italian labour market require a data set that allows for regional disaggregation. In order to serve the different aims of analysis we have used microdata from the IT SILC 2006 survey in the wider European Statistics on Income and Living Conditions survey programme. The EU Survey on Income and Living Conditions (EU-SILC) is a European assessment of households’ living conditions, with special attention to different forms of deprivation and social inclusion; this provides the possibility of analysing the costs of being in non-standard work not only with reference to lower labour income, but also with respect to other relevant dimensions of well-being (such as access to health services). Moreover the survey collects information on individual education, health and employment conditions and allows us to collect data on the reasons for working less than 30 h that can be used to account for involuntary part-time work.
4 Part-Time and Temporary Employment
57
The Italian survey IT SILC has been carried out since 2004 on a sample that allows statistical significance at regional level.3 The data set used in this chapter refers to the IT SILC 2006 cross-sectional survey of 21,499 households and 54,512 individuals. The sample is made of 17,835 women and 17,384 men aged from 16 to 64. Amongst those who are employed, 21% of women are in part-time jobs, against 4% of men; and more employed women (21%) than men (15%) are in temporary jobs. Part-time work amongst women employees runs at 22% in the North, 20% in the Centre and 21% in the South, where more women and men are in temporary jobs (23% of men and 30% of women). Temporary work is more spread out amongst younger age groups in the phase of entry into the labour market. It involves from 44% of women in the youngest age group in the North up to 53% of women employees in the 16 to 24 age group in the South. However the incidence of temporary work is still high in the 35 to 44 age group for women (probably when they re-enter the labour market after an interruption connected to child birth) and the gender gap increases.
4.4 Being in Non-Standard Work: A Gender Analysis of Determinants The higher share of women in temporary and part-time positions compared with men, combined with the higher risk of experiencing costs in terms of wages, job satisfaction and career prospects, lead us to devote attention to the factors affecting the probability of being employed as temporary or part-time workers. We shall adopt techniques that allow us to disentangle gender differences in the determinants of employment conditions. For this purpose we have estimated separate equations for men and women on the probability of being in non-standard employment in order to compare the different effects of the same covariates by gender. Another key issue for the specification of the econometric model to be used for the estimation concerns the correlation between the employment choice and the probability of being employed in a non-standard position. In order to account for the correlation of these two probabilities we have estimated a bivariate Probit model that allows us to test the correlation of the residuals from the two models whose estimated coefficients are reported in Tables 4.1 and 4.2. As discussed in Sect. 4.2, according to the existing literature, being in a temporary work position leads to lower job security and lower job satisfaction, and is more likely to be a non-voluntary status. In the final section of this chapter we will try to estimate the costs of being temporary workers in terms of lower earnings, lower likelihood of being in a supervisory position, and lower access to health 3
For a wider description of the survey see Istat (2008).
58
T. Addabbo and D. Favaro
Table 4.1 Bivariate Probit estimation on the probability of being in temporary work Women Men Emp.
Temp.
Age
dy/dx
0.146** 0.052** 0.010 (20.33) (5.46) Age squared -0.002** -0.001** -0.000 (18.82) (7.06) Married or cohabiting -0.104* -0.107* -0.020 (2.48) (2.06) Separated or divorced -0.013 -0.034 -0.006 (0.22) (0.45) Widow -0.139 -0.059 -0.011 (1.74) (0.51) Chronic illness -0.010 0.008 0.001 (0.28) (0.16) Number of children -0.015 -0.004 -0.001 aged less than 15 (0.74) (0.17) High school Education 0.339** 0.047 0.009 (11.64) (1.23) University degree 0.461** 0.184** 0.037 or post-graduate (10.43) (3.41) Living in the South -0.599** -0.101** -0.018 (21.61) (2.71) Industrial, firm size and job position dummies included Constant -2.680** -1.799** (20.66) (10.59) Observations 17,835 17,835 Wald Test for rho = 0 Chisquare(1) 763.192 0.0000 Prob [ chi2
Emp.
Temp.
dy/dx
0.030* (2.45) 0.000 (1.07) 0.078 (1.63) -0.046 (0.66) 0.272 (1.36) 0.055 (1.10) -0.022 (0.95) -0.020 (0.58) 0.008 (0.16) -0.012 (0.32)
0.087** (4.72) 0.001** (4.85) -0.350** (4.39) -0.043 (0.37) -0.015 (0.05) 0.291** (3.82) 0.007 (0.16) -0.052 (0.70) 0.031 (0.26) 0.126* (1.99)
-0.003
1.500** (6.23) 11,799
0.224 (0.56) 11,799
-0.000 -0.031 -0.016 -0.011 0.002 0.004 -0.025 0.035 0.035
132.99 0.0000
Robust z statistics in parentheses *significant at 5%; **significant at 1%
services; while in this Section we try to detect the different factors affecting the probability of working in a temporary position and how their effect differs by gender. To account for the correlation between the probability of being employed and the probability of being in a temporary job we have chosen to estimate a bivariate Probit model that allows us to estimate the factors affecting each job while at the same time testing for the correlation of the residuals from the two models. As Table 4.1 shows, the Wald test for the null hypothesis leads to a rejection of the null hypothesis of absence of correlation of the residuals of the employment and temporary probability equations both for men and for women. The two equations appear to be strongly correlated. A different set of variables has been introduced as covariates in the two equations: together with factors such as demographics, civil status, education,
4 Part-Time and Temporary Employment
59
Table 4.2 Bivariate probit on the probability of working part-time and being employed Women Men Emp. Age
Part-time dy/dx
-0.001 -0.030 (0.09) (1.76) Age squared -0.000 0.000 (0.61) (1.32) Married or cohabiting -0.019 0.340** (0.32) (5.66) Separated or divorced 0.064 0.072 (0.77) (0.82) Widow -0.031 0.229 (0.23) (1.69) Chronic illness 0.081 0.173** (1.24) (2.78) Number of children aged less -0.043 0.260** than 15 (1.35) (8.94) High school education 0.319** -0.271** (6.55) (4.82) University degree or post0.137* -0.373** graduate (2.27) (5.04) Living in the South -0.070 -0.063 (1.32) (1.11) Constant 1.051** -0.009 (3.33) (0.03) Industrial, firm size and job position dummies included Observations 8,079 8,079 Wald Test for rho = 0 Chi2 (1) 4.19 0.04 Prob [ chi2
Emp.
-0.008 -0.030* (2.45) 0.000 0.000 (1.07) 0.092 0.078 (1.63) 0.020 -0.046 (0.66) 0.069 0.272 (1.36) 0.049 0.055 (1.10) 0.072 -0.022 (0.95) -0.076 -0.020 (0.58) -0.093 0.008 (0.16) -0.016 -0.012 (0.32) 1.500** (6.23) 11,799
Part-time dy/dx -0.087** (4.72) 0.001** (4.85) -0.350** (4.39) -0.043 (0.37) -0.015 (0.05) 0.291** (3.82) 0.007 (0.16) -0.052 (0.70) 0.031 (0.26) 0.126* (1.99) 0.224 (0.56)
-0.006 0.000 -0.027 -0.003 -0.001 0.026 0.001 -0.004 0.002 0.009
11,799
0.01 0.92
Robust z statistics in parentheses *significant at 5%; **significant at 1%
region, and number of children aged less than 15, that are common to the set of covariates in both equations, the covariates of the temporary employment equation include also a set of dummies to control for variables in firm size, type of industry, and job position. Turning to the results of our analysis we notice a different demographic involvement of men and women in the probability of being in a temporary work position, with temporary work being more likely to occur for younger men and for older women, in different phases of their life cycle and career. The effect of civil status is seen in relation to gender too, with married women being less likely both to work and to being found in a temporary work position, and married men to be more likely to be employed but less likely to be in temporary work. For men being married or cohabiting reduces by 3% their probability of working in a temporary job.
60
T. Addabbo and D. Favaro
The effect of chronic illness on the probability of being in a temporary job appears to operate only in respect of men: it increases their probability of being employed in a temporary job though by only 0.2%. Turning to the variables that measure the level of education, another interesting difference can be outlined: having a higher level of education has a positive and significant effect on women’s probability of being employed; however, having a university degree exposes more women to the risk of being in a temporary work position by 3.7%, a significant effect that does not operate in relation to men. This may be connected to a higher likelihood that graduate women enter employment through the channel of a spell of temporary employment. Living in the South decreases the employment probability of women and their probability of being employed in temporary job (by 1.8%), while significantly increasing the probability of men being in temporary work (by 3.5%). The industry-type, firm size and job position variables that have been included to control for labour demand factors and are not reported in the Table, confirm a lower probability for workers in firms with 50 or more employees to be found in temporary work, and a lower probability that employees in apical positions have a temporary contract. Industrial dummies also allow for the detection of a different effect as far as the temporary employment probability is concerned for men and women working in the same sector. In this regard we can see that, compared to being employed in the trade sector, being employed in the personal and protective services sectors is going to increase the probability of women being in temporary work, whereas men appear to be more exposed to the risk of being in a temporary positions in the hotel and restaurant sectors. To summarize: the results of this assessment confirm heterogeneity in the composition of temporary workers. For men, the probability of being in temporary work is significantly higher for younger men, men with chronic illness, and men who live in the South of Italy; whereas for women the probability of being in a temporary work position increases with age, and level of education. If they live in the South of Italy, the probability of being employed at all, even in non-standard work, diminishes. Moreover differences in the probability of being in temporary work positions, by gender, occur inside each industry with respect to being employed in the trade sector. On the whole these estimates suggest that it is crucial to distinguish the impact of different factors by gender and to take into account the simultaneity and joint distribution of the employment and temporary work probabilities equations. We then analysed the determinants of part-time probability by estimating, consistent with the need to allow for the simultaneity of the employment and type of contract probabilities and to detect different impacts of the covariates by gender, two different bivariate Probit models. However the test for the correlation of the residuals of the two equations, whose results are reported in Table 4.2, allows rejection of the null hypothesis of the absence of correlation only for the sample of women. The choices of being employed and being employed as part-timers seem not to be associated for men whereas they are correlated for women.
4 Part-Time and Temporary Employment
61
As the results of the estimation show, we can confirm the expected gender differences in the discouraging effect of being married or cohabiting on women’s employment probability, whereas the probability of working part-time increases for married or cohabiting women—by 9.2%—and decreases for men—by 2.7%. In a difference from what has been found with regard to the probability of being in a temporary work position (Table 4.1), married women are more likely to be in a part-time work position. This is in line with the higher share of women who accept part-time work for family reasons, thereby illustrating the role of part-time work in easing the work-life balance. The distribution of care work inside Italian families, the rationing of public childcare services and the low diffusion of full-time school for children aged less than 15 can affect the observed differences also in the coefficients connected to the number of children in the households in this age group: as the multivariate analysis shows, women are more likely to be working part-time the higher the number of children in the household aged less than 15, whereas men’s part-time labour supply appears not to be related to the family work-load. If on the one hand this result confirms a positive role for part-time work in easing the work-life balance, on the other hand it is in line with the literature which stresses the importance of the constraints affecting women’s labour supply and the effect of the presence of children in working hours decisions (amongst others: Addabbo 1999, 2003b; Addabbo et al. 2010; Del Boca 2002; Del Boca and Locatelli 2007; Del Boca et al. 2007; Paull 2008; Vosko and Clark, 2009; Weinkopf 2009). Turning to the effect of human capital investment on employment and part-time probability, the results confirm a higher involvement of more educated women in employment, consistent with previous evidence on female labour supply in Italy (Addabbo 1999; Bettio and Villa 1999), and a lower probability that more educated women are in part-time positions. The latter result differs from the effect of education on temporary work probability: we found a greater temporary employment probability for more educated women (Table 4.1). If a higher level of education can lead to entry into temporary work positions—and this is a widespread phenomenon in Italy for younger, more educated cohorts of women—then having a higher level of education leads women to opt for different choices as far as full-time and part-time work is concerned. On this topic we should point out that the term ‘‘part-time’’ includes also permanent positions that are also more difficult to revert to full-time in Italy. Previous analyses (Addabbo 1999) have shown that more educated women are also less likely to be affected in their working decision by the presence of children in the household. We can consider these different impacts of education on part-time and employment probability as a sign of segmentation within women in the Italian labour market with more educated women more likely to be found employed and in full-time work than less educated women. Turning to the effect of regional variables, consistent with what we have found in relation to temporary contracts, women living in the South are less likely both to be employed and to be employed part-time. These effects however do not appear to be statistically significant, whereas men appear to be more likely to work part-time
62
T. Addabbo and D. Favaro
if they live in the South. The latter can be more likely to be in involuntary parttime positions. The more likely occurrence of involuntary part-time work for men in the South of Italy will be investigated in more depth in the following Section. To account for labour demand factors we have controlled for industries, type of job and company size in the part-time probability equations. Like the probability of being in a temporary work position, the probability of being in part-time work significantly decreases for women and men in apical positions. This result is in line with the observed difficulties in career prospects for non-standard workers and will be further investigated in Sect. 4.6 of this Chapter. The estimates of the effect of industries and work positions variables, taking as a reference category white collar employed in the wholesale and retail trade sector, confirm the lower probability of blue collar men and women workers in manufacturing being employed in part-time positions. The estimates also show different effects by gender within sectors.
4.5 To What Extent Working Part-Time is an Outcome of Choice Descriptive statistics and multivariate analysis results, shown in the previous Sections, have outlined the gender differences on the probability of being in the two forms of non-standard employment on which we focus in this Chapter. If temporary work appears to be more likely to be involuntary, there is a higher degree of heterogeneity in the individual motivations for working part-time. This, together with the availability of data, lead us, having assessed the different factors affecting part-time employment probability, to try and measure to what extent this employment status is involuntary. The data set used allows us to measure the extent of involuntary part-time work by the use of information on the reasons for working less than 30 h a week. Involuntary part-time work is then defined as the status of those who are part-time and working less than 30 h a week for involuntary reasons. As the descriptive statistics in Table A1 show, on average involuntary part-time work is more spread amongst women aged 16–64 than amongst men in the same age group, in the northern and central regions, with a higher weight in the youngest age groups, whereas in the South of Italy for those aged 16–64 the percentage of involuntary part-time work is similar by gender. However the share of men’s involuntary parttime work is higher for younger workers while for women it is higher after they are 35 years old in the South of Italy. The latter can be related to the attempt to re-enter employment after interruptions connected with child-birth. Such attempts to find full-time employment can be unsuccessful and lead to spells of unemployment or underemployment. To capture the reasons leading to working part-time we have used the information provided by IT SILC on the reasons part-timers give for working less than 30 h a week. Motivations connected to family factors are more spread amongst
4 Part-Time and Temporary Employment
63
women and in the North of Italy, whereas part-time work for involuntary reasons is more spread amongst men, reaching the maximum in the South of Italy where 75% of men working part-time less than 30 h state that they do so because they cannot find full-time jobs. Voluntary part-time work is more spread amongst women living in the North of Italy (Table A2). We next focus on a particular group of part-timers—those who work part-time for less than 30 h—and we analyse how different factors affect the probability of people involuntarily working part-time for less than 30 h. This focus on involuntary part-time workers is justified because of the higher probability that these workers face a weaker situation in the labour market in terms of the different elements of their working conditions. In order to model the probability of people involuntarily working part-time for less than 30 h, we have used a Probit model with sample selection. The selection equation refers to the probability of being employed part-time and less than 30 h. The second equation is the Probit equation on the probability of involuntary parttime work of less than 30 h. The Wald test of the null hypothesis of absence of correlation between the residuals of the two equations can be rejected. The results of the estimation in Table 4.3 show that women are more likely to be found working involuntarily for less than 30 h, compared to the reference category (clerks with less than highschool education, employed in the trade sector in firms with less than 50 employees). Consistent with the findings of Barrett and Doiron (2001) on Canadian data, the probability of being involuntary part-timers is lower for the more educated employees. The negative coefficient of the variable on the number of children aged less than 15 mirrors the higher probability of their mothers (more likely to be found working part-time) stating that they work for less than 30 h, for family reasons. The estimation includes as covariates dummies for industry, job positions and size of the firms, in order to control for demand factors; their inclusion allows the detection of heterogeneity amongst sectors and job positions in involuntary part-time probability.
4.6 Evaluating the Costs of Part-Time and Temporary Employment We have seen in the previous sections how women appear to be over-represented, compared with men, in part-time work; similarly, their share is higher than men’s in temporary work. Analysis of the different factors affecting the probability of being in non-standard employment has brought out gender differences, the higher occurrence of involuntary part-time work amongst women, and the fact that voluntary part-time work for women appear to be strongly related to the presence of young children in the family, or to their marital status. The results of previous analyses bear out the need to account for gender differences in analysing the
64
T. Addabbo and D. Favaro
Table 4.3 Probability of involuntary part-time PT \ 30 h Female Age Age squared Married or cohabiting Separated or divorced Widow Chronic illness Number of children aged less than 15 High school education University degree or post-graduate Living in the South Firm with 50 or more employees Constant Observations Wald test of indep.eqns. Chi2(1) Prob [ chi2
0.847 (14.17)** 0.036 (2.41)* -0.001 (3.35)** 0.069 (1.18) -0.108 (1.21) 0.200 (1.50) 0.144 (2.64)** 0.114 (4.16)** -0.181 (3.36)** -0.365 (4.72)** -0.103 (2.09)* -0.153 (2.37)* -3.006 (9.04)** 22,141
Involuntary PT 0.653 (4.48)** 0.006 (0.27) -0.000 (1.01) -0.075 (0.71) 0.095 (0.62) 0.271 (1.25)
-0.111 (1.98)* -0.121 (1.60) -0.352 (3.03)** 0.202 (1.69) -0.135 (1.54) -2.685 (5.30)** 22,141
Inv.PT dy/dx 0.014 0.000 -0.000 -0.001 0.002 0.007
-0.002 -0.002 -0.005 0.004
3.19 0.074
Notes: Heckman Probit model where at the first stage, working part-time less than 30 h is modelled. At the second stage, involuntary part-time work for less than 30 h is estimated Industry and job position dummies included Robust z statistics in parentheses *significant at 5%; **significant at 1%
determinants of non-standard employment. In this section we turn to the consequences of this employment ‘choice’ by analysing the costs of part-time and temporary work by gender, in terms of the holding of supervisory positions, hourly wages and other dimensions of well-being. Descriptive statistics on supervisory responsibility by type of contract and gender are shown in Table A3. If we control for the type of contract computing the share of workers with supervisory roles, we can see that women are less likely to hold supervisory responsibilities than men in all types of contracts except part-time jobs. Moreover the share of employees with supervisory responsibility is significantly lower for temporary employees and part-timers.
4 Part-Time and Temporary Employment
65
In order to analyse the effect of being in non-standard work on the probability of having a supervisory role, we have estimated a two-step Heckman Probit model: the first step is on the probability of being employed, while the second step is on the probability of having a supervisory role.4 This two-step procedure has been followed due to the non-random selection of the employed sample; however, as the results of the Wald test on zero correlation between the residuals of the two equations show, the null hypothesis of absence of correlation between the two steps cannot be rejected for the male sample. The observed gap in access to positions of responsibility for part-timers and temporary workers is confirmed by multivariate analysis controlling also for personal, family and labour demand factors5 we can see that being part-time employed decreases the probability of supervisory role by 6% for women and 10% for men, and being in a temporary job decreases the probability of having a supervisory role by 15% for men and 10% for women (Table 4.4). Though both men and women are affected positively in their probability of having a supervisory role by the length of their past work experience and level of education, gender differences occur with regard to marital status: with respect to single married women and widows, a decrease is shown in the probability of their being employed and (though to a statistically insignificant extent) in being employed with a supervisory role; while at the same time men’s employment and supervisory role probabilities increase. We then turn to the evaluation of the wage penalty associated with holding parttime or temporary jobs. The observed gross wage gender gap is higher in part-time positions than in full-time positions, but this does not occur in temporary positions as compared to permanent jobs (Table A4). The descriptive analysis shows the higher penalty in terms of hourly wages connected to working in temporary jobs (the gap being 30% for men and 22% for women) and a lower but still significant penalty when one works in part-time positions (16% for men and 18% for women), the penalty being higher for men from a comparison of permanent and temporary jobs, and for women from a comparison of part-time and full-time positions. To measure the wage penalties associated with temporary and part-time jobs we have used multivariate analysis in order to control for individual, household, and labour market variables affecting wages. To account for the non-random selfselection of women into employment we have again estimated a two-step Heckman selection model.6 In the first step we have estimated the probability of being an employee, including as covariates individual variables on the level of education (high school, university or a higher level compared to a lower level of education), marital status (married or cohabiting, divorced or separated, widow or single),
4
See van de Ven and van Praag (1981) for the application of correction for sample selection to probit analysis. 5 In the second step of the model we controlled also sector and type of employment variables. 6 Heckman (1979).
66
T. Addabbo and D. Favaro
Table 4.4 Probit model on the probability of having supervisory responsibility Women Men Employee Supervisory Sup.dy/dx Employee Supervisory Sup.dy/dx Actual past work experience Part-time
0.179** (5.56) -0.259** (4.17) Temporary contract -0.418** (6.66) Married or cohabiting -0.175** -0.053 (5.64) (1.01) Sep. divorced 0.011 -0.052 (0.29) (0.70) Widow -0.214** -0.131 (3.83) (1.25) Number of children -0.009 0.042 aged less than 15 (0.64) (1.47) High school education 0.424** 0.220** (20.97) (3.79) University degree or 0.817** 0.542** post-graduate (28.79) (6.26) Living in the South -0.806** -0.019 (41.08) (0.25) Firm with more than 0.096 50 employees (1.87) Sectors and types of job dummies Age 0.203** (37.87) -0.002** Age2 (35.52) Chronic illness -0.055* (2.07) Constant -3.585** -1.342** (36.07) (8.60) Observations 16,282 16,282 Wald Test for rho = 0 Chi2 (1) 4.02 0.0450 Prob [ chi2 Robust z statistics in parentheses *significant at 5%; **significant at 1%
0.045 -0.058 -0.104 -0.013 -0.013 -0.031 0.010 0.056 0.160 -0.005 0.025
0.176** (5.89) -0.394** (3.24) -0.489** (7.43) 0.616** 0.359** (21.48) (6.68) 0.297** 0.210* (7.95) (2.53) 0.293** 0.612** (2.92) (3.20) -0.022 -0.010 (1.20) (0.38) 0.022 0.368** (0.96) (9.24) 0.042 0.893** (1.36) (13.65) -0.380** -0.247** (16.88) (5.84) 0.103* (2.46) 0.289** (61.55) -0.003** (58.86) -0.399** (14.38) -4.544** -1.175** (50.42) (10.44) 13,922 13,922 0.07 0.7858
0.053 -0.102 -0.148 0.109 0.068 0.219 -0.003 0.115 0.321 -0,073 0.032
4 Part-Time and Temporary Employment
67
regions (South compared to Centre-North), number of children aged less than 15, other households’ income, and health status.7 The covariates included in the wage equations are individual variables on past work history (measured by the declared number of years, since starting the first regular job, that the person has spent at work, as an employee or as self-employed), level of education (high school, university or a higher level compared to a lower level of education), marital status (married or cohabiting, divorced or separated, widow or single), industries (dummies on different sectors have been included compared to manufacturing), type of employment (different types of jobs have been included taking labourers as reference group), regions (South compared to Centre-North) and job contract (parttime and temporary compared to full-time permanent contract), and number of children aged less than 15. We have modelled men’s wages by estimating OLS regression and by introducing the same set of covariates as in the second step of the Heckman sample selection model estimated on the sample of women, to allow for comparability of the effect of the different factors analysed. Amongst the covariates the dummy variables on part-time work (taking the value as one if the individual is employed in a part-time position according to that individual’s self-defined employment status) and on temporary work (taking the value as one if the individual is employed in a temporary position in her main job) have been included with their estimated coefficients, thereby providing a first estimate of the existence of wage penalties associated with part-time and temporary employment.8 Being in a part-time position is found to significantly decrease women’s hourly wages by 9% on average, while a temporary position decreases the gross hourly wages of both men and women by 15% on average, taking into account family structure, human capital accumulation, and type of industry as the results of the multivariate analyses in Table A5 in the Appendix show. With the aim of enlarging the evaluation of the costs connected to being in nonstandard employment we have analysed how, amongst employed individuals, the unmet need for medical and dental care, related to its affordability, is affected by different types of contract. Amongst workers as a whole, 13% have unmet needs for medical or dental treatment and the share of those experiencing this problem because they consider it unaffordable is higher amongst temporary workers, as Table A6 in the Appendix shows. To account for the bivariate outcomes of whether there are unmet medical and dental needs, and whether they are unmet because unaffordable, we have estimated 7
The sample is made of employees not working in agriculture and working more than 9 h a week. The first step of Heckman’s selection model for women refers to the probability of being an employee working more than 9 h a week with respect to being unemployed or inactive. Selfemployed and employees working in agriculture sectors have been excluded due to higher difficulties in measuring without error their hourly wages. 8 However, though we have controlled for relevant observable variables affecting wages, the estimates provided can be still subjected to bias connected to unobserved heterogeneity.
68
T. Addabbo and D. Favaro
Table 4.5 Probability of unmet need for medical or dental examinations or treatments for reasons connected to cost Unmet dental and medical Unmet because Unmet because not examinations and treatments not affordable affordable dy/dx Female Age Age squared Married or cohabiting Separated or divorced Widow Chronic illness Number of children aged less than 15 High school education University degree or post-graduate Living in the South Employed in firms with 50 or more employees Part-time Temporary Log equiv.hous. Constant
0.009 (0.21) 0.031* (2.36) -0.000* (2.31) 0.004 (0.09) 0.225** (3.14) 0.157 (1.22) 0.467** (10.20) 0.027 (1.06) -0.058 (1.33) -0.117 (1.77) 0.159** (3.88) 0.064 (1.67) 0.077 (1.39) 0.118* (2.30) -0.241** (5.55) 0.182 (0.35)
Industry, firm size and job positions dummies included Observations 14,844 Wald test of rho = 0 198.583 Chi2 Prob [ chi2 0.0000
-0.022 (0.42) 0.030 (1.74) -0.000 (1.51) -0.010 (0.16) 0.286** (3.15) 0.229 (1.47) 0.465** (8.17) 0.029 (0.92) -0.100 (1.81) -0.330** (3.45) 0.093 (1.84) 0.060 (1.23) 0.025 (0.38) 0.164** (2.69) -0.351** (6.85) 0.818 (1.30)
-0.002 0.003 -0.000 -0.001 0.037 0.029 0.066 0.003 -0.011 -0.029 0.010 0.006 0.003 0.019 -0.037
14,844
Robust z statistics in parentheses *significant at 5%; **significant at 1%
a bivariate Probit model controlling for personal, household and type of job variables, where the error terms are assumed to be correlated with a cumulative bivariate normal distribution.
4 Part-Time and Temporary Employment
69
As the results in Table 4.5 show, being in a temporary position increases the probability of facing unmet needs when controlling for equivalised household income, health status, type of employment and industry and other individual and family structure variables. This can be connected to the higher uncertainty in the job income of workers in temporary employment, and brings to light another weak element in the sustainability of this type of contract in terms of employees’ wellbeing.
4.7 Conclusions The diffusion of non-standard work in Italy and the observed gender differences in their distribution suggested that attention should be devoted to the factors affecting labour supply in these positions, with special attention to the probability of people being employed in non-standard work involuntarily. Our analysis has also confirmed, in the Italian case, the higher probability for workers in non-standard posts to be female. Among employed people 16–64 years old, 21% of women are in part-time jobs against 4% of men; moreover, the overall percentage of women in temporary jobs (21%) is higher than the proportion of men in the same type of job (15%). Our findings highlight the way Italian women and men differ in their reasons for accepting non-standard work. Voluntary choice and family factors are the most common reasons among women living in the North. However, in both the North and the South, the highest incidence of working part-time for family reasons is encountered among females between 25 and 44 years old; this contrasts with the maximum incidence of part-time work for family reasons in Central Italy, which is with females 35–44 years old. Individual, family, institutional and labour demand factors are regarded as accounting for this take-up of part-time work. Controlling for individual, demand-side and family variables, and estimating different models by gender, allow us to detect different factors affecting the employment probability of men and women in temporary and part-time employment. With reference to part-time employment the estimates confirm the high weight of family structure variables affecting women in relation to part-time work (the high inequality in the distribution of unpaid work by gender is also mirrored in this result). Meanwhile, involuntary part-time work turns out to be more likely to occur for women, for less well-educated individuals, and for people living in the South. We then turned to the evaluation of costs connected to the observed nonstandard jobs. On average an hourly pay penalty is attached to temporary work both for men and for women, while for part-time work this penalty operates for women only. Part-time and temporary work are negatively related to the probability of holding supervisory positions, after the analysis has controlled for individual characteristics, family composition and industry. After these financial costs an attempt to extend the analysis to other dimensions of well-being, namely access
70
T. Addabbo and D. Favaro
to medical and dental assessment and treatments, shows a higher probability of temporary workers having unmet needs, related to affordability. The latter is probably connected to higher income uncertainty for temporary employees, and can magnify the cost of holding a temporary job by affecting health outcomes. In terms of policy proposals, we suggest that more attention should be paid to monitoring and improving the quality of non-standard jobs—in which women are over-represented—in a manner consistent with the achievement of gender equality goals (Burchell et al. 2007). This in turn could lead (as found in Del Boca et al. 2009, with reference to good quality part-time work) to an increase of female employment.
Appendix See Tables A1, A2, A3, A4, A5 and A6 Table A1 Involuntary part-time work by age groups, sex and area Age M (%) F (%) F–M (%) M (%) F (%) F–M (%) M (%)
F (%)
F–M (%)
16–24 25–34 35–44 45–54 55–64 16–64
28 27 28 47 50 32
-15 -18 4 28 26 -1
9 16 7 11 10 12
25 19 12 17 15 16
17 3 5 6 5 4
0 40 10 32 19 22
55 29 24 29 26 30
55 -11 13 -3 7 8
43 45 24 19 24 33
Table A2 Part-timers who work less than 30 h by reason, sex and area North Centre South Education Illness Underemployment Voluntary Considered fit Housework or care Other reasons
M
F
Gap
M
F
Gap
M
F
Gap
2.31 13.85 24.88 9.73 18.33 3.02 27.89
1.59 2.65 21.83 19.92 7.88 28.73 17.39
-0.72 -11.2 -3.05 10.19 -10.45 25.71 -10.5
3.59 6.6 45.9 8.6 17.96 0.88 16.46
0.69 1.22 39.8 13.33 9.79 19.48 15.69
-2.9 -5.38 -6.1 4.73 -8.17 18.6 -0.77
2.23 2.07 75.01 5.48 5.73 0 9.47
1.94 0.87 44.71 9.81 13.9 16.69 12.08
-0.29 -1.2 30.3 4.33 8.17 16.69 2.61
Table A3 Supervisory role by type of contract (PT-FT, permanent-temporary)
Full-time Part-time PT-FT Permanent Temporary Temp-Perm
M (%)
F (%)
Gap (%)
28 11 -17 30 8 -22
19 11 -8 18 6 -12
-9 0 -12 -2
4 Part-Time and Temporary Employment
71
Table A4 Gross wage differentials M
F
FT PT Gross wage gap Permanent Temporary Gross wage gap
10.83 8.89 18% 10.76 8.38 22%
11.04 9.29 16% 11.33 7.94 30%
Gross Gender Wage Gap9 (%) 2 4 5 -6
Table A5 Log gross hourly wage equations by gender Women
Men
Wage (logarithm) Probability of empl. Actual past work-experience Part-time Temporary Married cohabiting Separated divorced Widow Number of children aged less than 15 High school education University degree or post-graduate South Firms with more than 49 employees
0.112** (18.34) -0.094** (6.80) -0.150** (12.16) 0.080** (7.80) 0.047** (2.72) 0.036 (1.04) 0.030** (5.32) 0.101** (10.96) 0.357** (22.27) -0.087** (7.90) 0.106** (12.86)
-0.320** (12.70) -0.416** (10.51) -0.853** (10.85) -0.036** (2.65) 0.462** (22.77) 0.950** (33.51) -0.892** (43.77)
Wage (logarithm) 0.114** (15.54) -0.044 (1.30) -0.148** (9.24) 0.112** (8.74) 0.112** (5.34) 0.158* (2.26) 0.025** (3.61) 0.108** (10.83) 0.414** (19.32) -0.132** (11.46) 0.095** (8.82) (continued)
9
Gender Gap in wages is defined as [1- (woman’s wage/man’s wage)] specific to each type of job. The table shows also gross wage gap by type of jobs (part-time versus full-time and permanent versus temporary). The sample is made of employees not working in agriculture and working more than 9 h a week.
72
T. Addabbo and D. Favaro
Table A5 (continued)
Dummies on industry and type of employment Age
Women
Men
Wage (logarithm) Probability of empl.
Wage (logarithm)
Included
Included 0.226** (42.35) -0.003** (35.76) -0.298** (11.10) -0.357** (40.02)
Age squared Chronic illness Other household income Constant Observations Lambda Wald test of indep. eqns. Chi2(1) Prob [ chi2
1.784** (73.89) 19,144 19,144 -0.0562273
1.782** (62.71) 8,211
9.65 0.0019
Robust z statistics in parentheses *significant at 5%; **significant at 1%
Table A6 Unmet needs for medical or dental examinations or treatments and reasons connected to their costs: means and standard dev Type of Men Women contract Unmet needs for Of which Unmet needs for Of which medical or dental unmet because medical or dental unmet because examinations not affordable examinations not affordable Non temporary Temporary Full-time Part-time Employees
0.12 0.33 0.17 0.38 0.12 0.33 0.19 0.40 0.13 0.33
0.52 0.50 0.69 0.46 0.53 0.50 0.45 0.50 0.52 0.50
0.13 0.34 0.18 0.39 0.12 0.33 0.15 0.36 0.13 0.34
0.51 0.50 0.58 0.49 0.47 0.50 0.54 0.50 0.48 0.50
4 Part-Time and Temporary Employment
73
References Addabbo, T. (1997). Part-time work in Italy. In H.P. Blossfeld & C. Hakim (Eds.), Between equalization and marginalization. Women working part-time in Europe and the United States of America. Oxford: Oxford University Press. Addabbo, T. (1999). Labour supply and employment probabilities in Italy: A gender analysis in a regional perspective. Economia & Lavoro, XXXIII(3–4), 189–207. Addabbo, T. (2003a). ‘Atypical’ work in Italy in a gender perspective: The case of parasubordinate and part-timers. In D. Del Boca & M. Repetto (Eds.), Women’s work, the family, and social policy: Focus on Italy in a European perspective. New York: Peter Lang Editor. Addabbo, T. (2003b). Unpaid work by gender in Italy. In A. Picchio (Ed.), Unpaid work and the economy. London: Routledge. Addabbo, T., Caiumi, A., & Maccagnan, A. (2010). Unpaid work, well-being and the allocation of time in contemporary Italy. In T. Addabbo, M.P. Arrizabalaga, C. Borderias, & A. Owens (Eds.), Gender inequalities, households and the production of well-being in modern Europe. Aldershot: Ashgate. Altieri, G., Ferrucci, G. & Dota, F. (2008). Donne e lavoro atipico: Un incontro molto contradditorio. III Rapporto Osservatorio permanente sul lavoro atipico in Italia. Ires. http:// www.ires.it. Amuedo-Dorantes, C. (2000). Work transitions into and out of involuntary temporary employment in a segmented market: Evidence from Spain. Industrial and Labor Relations Review, 53(2), 309–325. Arulampalam, W., & Booth, A. (1998). Training and labour market flexibility: Is there a tradeoff? The British Journal of Industrial Relations, 36(4), 521–536. Bardasi, E., & Gornick, J. (2008). Working for less? Women’s part-time wage penalties across countries. Feminist Economics, 14, 37–72. Barrett, G. F., & Doiron, D. J. (2001). Working part-time: By choice or by constraint. Canadian Journal of Economics, 34(4), 1042–1065. Berton, F., Richiardi, M., & Sacchi, S. (2009). Flex-insecurity. Perchè in Italia la flessibilità diventa precarietà. Bologna: Il Mulino. Bettio, F. & Villa, P. (1999). To What Extent Does it Pay to Be Better educated? Education and Market Work for Women in Italy. South European Society and Politics, 2, Special Issue on Gender Inequalities in Southern Europe: Women, Work and Welfare in the 1990s. Blank, R. M. (1990). Are part-time jobs bad jobs? In G. Burtless (Ed.), A future of lousy jobs? The changing structure of U.S. wages (pp. 123–155). Washington, DC: Brooking Institution. Booth, A., Francesconi, M., & Frank, J. (2002). Temporary jobs: Stepping stones or dead ends? The Economic Journal, 112(480), 189–213. Booth, A. L., & van Ours, J. C. (2008). Job satisfaction and family happiness: The part-time work puzzle. Economic Journal, Royal Economic Society, 118(526), F77–F99. Booth, A. L., & Van Ours, J. C. (2009). Hours of work and gender identity: Does part-time work make the family happier? Economica, 76(301), 176–196. Booth, A. L., & Wood, M. (2008). Back-to-front down-under? Part-time/full-time wage differentials in Australia. Industrial Relations, 47(1), 114–135. Bowlus, A. J., & Grogan, L. (2009). Gender wage differentials, job search, and part-time employment in the UK. Oxford Economics Papers, 61, 275–303. Burchell, B., Fagan, C., O’Brien, C., & Smith, M. (2007). Working conditions in the European Union: The gender perspective. Dublin: European Foundation for the Improvement of Living and Working Conditions. Comi, S. & Grasseni, M. (2009, September, 24–25). Temporary jobs wage differential in Europe. Paper presented at the XXIV National Conference of Labour Economics University of Sassari. Connell, J., & Burgess, J. (2006). The influence of precarious employment of career development: The current situation in Australia. Education and training, 48, 493–507.
74
T. Addabbo and D. Favaro
Connolly, S., & Gregory, M. (2008a). Feature: The price of reconciliation: Part-time work, families and women’s satisfaction. Economic Journal, 118(526), F1–F7. Connolly, S., & Gregory, M. (2008b). Moving down? Women’s part-time work and occupational change in britain, 1991–2001. Economic Journal, 118(526), F52–F76. Connolly, S., & Gregory, M. (2009). The part-time pay penalty: Earnings trajectories of british women. Oxford Economic Papers, 61(S1), i76–i97. De Witte, H., & Naswall, K. (2003). Objective’ vs ‘Subjective’ job insecurity: consequences of temporary work for job satisfaction and organizational commitment in four European countries. Economic and Industrial Democracy, 24(2), 149–188. Del Boca, D. (2002). The effect of childcare and part-time work on participation and fertility of Italian women. Journal of Population Economics, 14, 549–573. Del Boca, D., & Locatelli, M. (2007). Motherhood and participation. In D. Del Boca & C. Wetzels (Eds.), Social policies, labour markets and motherhood. A comparative analysis of European countries. Cambridge: Cambridge University Press. Del Boca, D., Pasqua, S., & Pronzato, C. (2009). Motherhood and market work decisions in institutional context: A European perspective. Oxford Economic Papers, 61, i147–i171. Del Boca, D., Pasqua, S., Pronzato, C., & Wetzels, C. (2007). An empirical analysis of the effects of social policies on fertility, labour market participation and hourly wages of European women. In D. Del Boca & C. Wetzels (Eds.), Social policies, labour markets and motherhood. A comparative analysis of European Countries. Cambridge: Cambridge University Press. Ferber, M. & Waldfogel, J. (1998, May). The long-term consequences of non-traditional employment. Monthly Labor Review, 3–12. Francesconi, M. (1991). Determinants and consequences of promotions in Britain. Oxford Bulletin of Economics and Statistics, 63(3), 279–310. Goudswaard, A., & Andries, F. (2002). Employment status and working conditions. European foundation for the improvement of living and working conditions. Luxembourg: Office for Official Publications in the European Communities. Heckman, J. (1979). Sample selection bias as a specification error. Econometrica, 47, 153–161. Hirsch, B. T. (2005). Why do part-time workers earn less? The role of worker and job skills. Industrial and Labor Relations Review, 58(4), 525–551. Istat. (2008). L’indagine europea sui redditi e le condizioni di vita delle famiglie (EU-SILC). Metodi e Norme, no. 37. Kaiser, L. (2005). Gender-job satisfaction differences across Europe: An indicator for labor market modernization. IZA, Discussion paper no.1876. Kalleberg, A. L., Reskin, B. F., & Hudson, K. (2000). Bad jobs in America: Standard and nonstandard employment relations and job quality in the United States. American Sociological Review, 65, 256–278. Manning, A., & Petrongolo, B. (2008). The part-time pay penalty for women in Britain. Economic Journal, 118(526), F28–F51. Mertens, A., Gash, V., & McGinnity, F. (2007). The cost of flexibility at the margin. Comparing the wage penalty for fixed-term contracts in Germany and Spain using quantile regression. Labour, 21(4–5), 637–666. Mertens, A., & McGinnity, F. (2004). Wages and wage growth of fixed-term workers in East and West Germany. Applied Economics Quarterly, 50(2), 139–163. Morris, M. D. S., & Vekker, A. (2001). An alternative look at temporary workers, their choices, and the growth in temporary employment. Journal of Labor Research, 22(2), 373–390. Mumford, K., & Smith, P. N. (2009). What determines the part-time and gender earnings gap in Britain: Evidence from the workplace. Oxford Economic Papers, 61(S1), i56–i75. OECD (2004). Employment outlook. Paris: OECD. OECD (2009). OECD in figures 2009, OECD Observer, 2009, Supplement 1. Paull, G. (2008). Children and women’s hours of work. Economic Journal, 118(526), F8–F27. Polivka, A. E. (1996). Into contingent and alternative employment: By choice?’. Monthly Labor Review, 119(10), 55–74.
4 Part-Time and Temporary Employment
75
Siebern-Thomas, F. (2005). Job quality in European labour markets. In S. Bazen, C. Lucifora, & W. Salverda (Eds.), Job quality and employer behaviour. Houndmills, Basingstoke: Palgrave Macmillan. Stratton, L. S. (1996). Are ‘Involuntary’ part-time workers indeed involuntary? Industrial and Labor Relations Review, 49(3), 522–536. Tam, M. (1997). Part-time employment: A bridge or a trap?. Aldershot: Avebury. Tilly, C. (1996). Half a job: Bad and good part-time jobs in a changing labor market. Philadelphia, PA: Temple University Press. van de Ven, W. P. M. M., & van Praag, B. M. S. (1981). The demand for deductibles in private health insurance: A probit model with sample selection. Journal of Econometrics, 17, 229–252. Villosio, C. (2001). Un’analisi longitudinale dei percorsi lavorativi e dei differenziali retributive degli occupati a tempo parziale in Lombardia sulla base di dati Inps. In M. Samek Lodovici & R. Semenza (Eds.), Le forme del lavoro (p. 4). L’occupazione nonstandard: Italia e Lombardia nel contesto Europeo, Angeli, Milano, Approfondimento. Vosko, L. F., & Clark, L. F. (2009). Canada: Gendered precariousness and social reproduction. In L. F. Vosko, M. Macdonald, & I. Campbell (Eds.), Gender and the contours of precarious employment. London: Routledge IAFFE Advances in Feminist Economics. Weinkopf, C. (2009). Germany: Precarious employment and the rise of mini-jobs. In L. F. Vosko, M. Macdonald, & I. Campbell (Eds.), Gender and the contours of precarious employment. London: Routledge IAFFE Advances in Feminist Economics.
Part II
Non-Standard Employment, Workers’ Mobility and Labour Market Structure
Chapter 5
The Quality of Temporary Work Gianna Barbieri and Paolo Sestito
5.1 Introduction Temporary work has been an important component of employment growth in Italy since the early 1990s.1 This is not surprising as Italy has substantially removed the obstacles against the use of temporary workers (henceforth ‘temps’ for brevity), while maintaining the rather strict firing regulations applying to permanent ones.2 Whatever its origins, this dual structure has become increasingly controversial. The nature of temporary work, either as a stepping stone or as a dead end, remains unclear, and the ‘quality’ of temporary jobs is increasingly scrutinised by worried observers (see OECD 2002). Besides equity considerations, efficiency issues have also been raised, because temps may become trapped in a low-productivity market segment. To a large extent, such an examination requires analysis of the transitions into and out of temporary employment. We have conducted such analysis in a related paper by evaluating the impact of escaping unemployment through a temporary work experience on the subsequent work chances (Barbieri and Sestito 2008). That 1 The opinions expressed here are those of the authors alone and do not necessarily reflect those of the institutions to which they belong. 2 While such a route has been quite commonly pursued, Italy appears to be an extreme case which warrants specific examination. According to the OECD (see OECD 2004), Italy is the country recording the largest drop in the Employment Protection Legislation (EPL) subindex for temporary employment since the early 1990s.
G. Barbieri (&) Ministry of Education, Statistical Unit, Via Michele Carcani, 61, 00153 Rome, Italy e-mail:
[email protected] P. Sestito (&) Deputy Director Department for Structural Economic Analysis, Bank of Italy, Banca d’Italia, Via Nazionale, 91, 00184 Rome, Italy e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_5, Ó Springer-Verlag Berlin Heidelberg 2012
79
80
G. Barbieri and P. Sestito
paper focused on the decade-long rise in temporary employment since the early 1990s. In this paper, we restrict our effort to assembling a few indicators for the ‘quality’ of temporary employment. But we focus on the 2004–2006 period. In so doing we are able to draw on the latest waves of the Labour Force Survey (LFS), which are much richer and more precise in identifying temporary employment and the reasons for it. Although limited, this period is also important because of the further steps in ‘flexibilization’ of the Italian labour market associated with the hotly debated Biagi Law.3 Section 5.2 discusses both the Italian institutional set-up and the features of the LFS data, comparing the newest data with those furnished by the previous LFS editions and with the available administrative sources. Its aim is to clarify the counting of temporary employment in Italy, which is an area of much controversy. Section 5.3 presents a simple econometric accounting of the determinants of being a temp, comparing the results for 2004 and 2006. Section 5.4 summarizes the evidence on several indicators of job ‘quality’. The approach followed is highly empirical and aims at presenting broad evidence with which to disentangle whether temporary status in itself is related to certain qualitative features and whether the link has changed over time. The results are summarised in Sect. 5.5.
5.2 Contractual Arrangements and Measurement Issues Besides being controversial in itself, temporary employment has given rise to numerous measurement controversies. The shorter duration of temporary contracts implies that the incidence of temporary employment differs according to the use of either flow or stock measures. Moreover, until recently the statistical use of administrative data was underdeveloped and there was no precise mapping between legal concepts—the fact that a work contract has a termination clause, formally an exception possible only in certain circumstances in the Italian juridical framework—and the available statistical measures. These were mostly based upon the classification of employees provided by the LFS, in which the worker was asked about the presence of a termination date (or clause) in his/her work contract. Building upon the respondent’s perception may be an advantage because one cuts across the formal details of the many functional substitutes provided for by the legislation. On the other hand, there is no precise mapping from the LFS to formal contractual arrangements. Moreover, the replies by the respondent (who may be the worker or a proxy respondent) may be rather imprecise because s/he may be uninformed about many legal aspects of the work contract.4
3
For more details see Sestito (2004) and Pirrone and Sestito (2006). These ambiguities are reflected in the study by Tronti et al. (2003). A more precise comparison between LFS and administrative sources is provided by Anastasia et al. (2004) and by the Ministry of Labour Monitoring Reports, particularly the July 2006 issue, upon which much of this section is based Ministry of Labour (2006).
4
5 The Quality of Temporary Work
81
1,800,000 1,600,000 1,400,000 1,200,000 1,000,000 800,000 600,000 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1998
1999 CLAP
2000 LFS
2001
2002
new LFS
2003
2004
2005
reconstruction of LFS
Fig. 5.1 Temporary employment in the non-farm private sector—Survey and administrative data comparison. Source: our calculations on Istat and Ministry of Labour-Inps (National Institute for Social Security), CLAP data
Particularly until 2003, the breakdown provided by the LFS was rather imprecise.5 It also appears that the LFS was underestimating the total number of people formally employed with temporary arrangements according to administrative sources (Fig. 5.1).6 Since 2004, the adoption of a new frame in the LFS— the frame hereafter used in this paper—appears to have eliminated the problem, even if there is still no precise mapping between the breakdown of temporary employment in the LFS and the contractual arrangements allowed by the law. 5 It was based upon a single question investigating whether the temporary status was due to the fact that ‘‘the contract was covering a period of training (apprenticeship, research assistant, youth training)’’, ‘‘the person could not find a permanent job’’, ‘‘the person did not want a permanent job’’ or ‘‘the contract was covering a probationary period’’ (with a last item covering the cases in which no specific reason was provided by the respondent). Since 2004, involuntariness—defined as the fact that the temporary work was taken because of the lack of alternatives—has been investigated separately from the contractual arrangements breakdown. 6 The comparison was made using the CLAP (Campione Longitudinale Attivi e Pensionati) archives, which were available only until 2002. As these archives focus on the non-farm private sector, we excluded agriculture and the public sector from the LFS count as well. Because the new LFS series for the years before 2004 have been reconstructed only for the overall economy total, that aggregate reconstruction has been scaled back in order to purge from these two broad sectors. Besides the differences in the average level, to be noted is that the administrative count shows wider seasonal fluctuations. A possible reason for this is that the LFS survey considers the persons employed in a reference week (actually, until 2003 just 1 week per quarter was considered, while since 2004 the quarterly amount has been the average of all the quarter’s weeks), while the CLAP concept measures the number of people working at least once in a month.
82
G. Barbieri and P. Sestito
Table 5.1 Temporary employment—the contractual arrangements breakdown in the private non-farm sector 1998 1999 2000 2001 2002 A. People in work at least once during the month—(average of monthly stocks) Apprenticeships 336,796 407,042 468,826 491,412 404,848 Training and work contracts 350,924 331,929 267,095 251,273 183,808 Temporary agency work 2,313 29,670 68,182 105,846 90,978 Other temporary contracts 349,832 387,520 499,129 523,646 467,049 Total 1,039,866 1,156,162 1,303,232 1,372,177 1,146,683 Temporary as share of total employees 11.7 12.5 13.1 13.3 13.1 (%) B. Total yearly inflows Apprenticeships 402,843 419,748 457,325 436,738 371,717 Training and work contracts 282,249 269,097 210,478 215,414 135,408 Temporary agency work 10,683 108,119 217,791 312,196 277,312 Other temporary contracts 668,018 834,668 1,014,563 1,035,671 997,572 Total 1,363,792 1,631,632 1,900,156 2,000,018 1,782,008 Temporary as share of total employees 31.8 34.5 35.9 37.9 37.8 (%) Source: our calculations on Ministry of Labour-Inps, CLAP data
Focusing on the administrative data, Table 5.1 shows the breakdown among the different contractual arrangements which make up temporary employment. To be noted is the large difference that arises in the counting of temporary employment when one moves from a stock concept (the number of persons with temporary contracts at a given date) to a flow measure. On turning to more recent data, we must revert to the LFS count, which, as said, is more precise because of the new frame introduced in 2004. While in the next sections we will exploit the information on the individual worker and the job position in order to determine who temporary workers are and the ‘quality’ of their jobs, here we give a breakdown of temporary employment according to the duration and the nature of the work contract. Starting with the former, Table 5.2 shows that only one out of five ongoing contracts is expected to last more than 1 year (longer durations are for youths, males, and residents in the Northern regions). Note that the incidence of temporary workers who are involuntarily such—defined as those who take a temporary position because of the lack of alternatives, an indicator further analysed in Sect. 5.4—is not closely related to the expected duration. Indeed, involuntariness is minimum (but still above 75%) for those contracts whose duration is unknown, a situation which in principle can be considered one of high uncertainty. The same Table also shows that about 30% of temporary workers declare that their engagement with the current employer has already lasted more than 3 years. The length of the elapsed tenure should not be taken at its face value; and more specifically, it is likely that this group also includes multiple job spells (with the same employer) intervalled by episodes of joblessness. Quite strikingly,
5 The Quality of Temporary Work
83
Table 5.2 Expected further duration of ongoing temporary work experiences—2006 average Expected further duration Share of temps with tenure longer than 3 years Age
\1 month 2–6 months 7–12 months [12 months Unknown (000) % of total
15–24 5.8 25–34 5.9 35–54 6.3 55–64 4.9 65 e + 7.7 Centre- Males 5.5 North Females 5.8 Total 5.7 South Males 7.4 Females 5.5 Total 6.5 Italy Males 6.3 Females 5.7 Total 6.0 80.7 % workers involuntarily with a temporary status
25.1 33.0 34.8 40.0 19.0 31.3
18.0 36.1 42.3 32.9 38.0 30.8
42.3 18.4 10.0 12.2 22.7 27.5
8.7 6.7 6.5 10.0 12.6 4.9
69 170 331 49 5 129
12.8 22.5 40.0 54.0 57.9 20.6
26.8 28.8 39.0 34.7 37.0 34.5 29.5 32.0 81.5
41.4 36.5 23.3 37.3 29.7 27.6 40.0 33.9 83.7
21.9 24.5 16.9 12.8 15.0 23.0 18.8 20.8 82.8
4.2 4.5 13.5 9.6 11.7 8.5 6.1 7.3 79.1
203 332 150 145 294 279 348 627 78.9
27.4 24.3 32.4 36.8 34.4 25.6 30.7 28.2
the involuntariness indicator is slightly lower than average (78.9% vis-à-vis 82.3%) in this group, which is apparently the one closest to a situation of abuse of the temporary clauses by the employer. Tables 5.3 and 5.4 set out the types of work and contractual arrangements. As already said, the identification of contractual arrangements is very imprecise even in the LFS frame adopted since 2004. The incidence of apprenticeship contracts, in particular, appears to be somewhat underestimated. More generally, there are many cases in which the respondent (either the worker or a proxy respondent) is either unable to specify the contractual arrangement or declares that there is only a verbal understanding on the employment relationship (and its temporary nature), something which should not be possible according to the law. As for the types of work, seasonal jobs appear to be the most common, particularly among adults and in the South. These are the jobs with the lowest incidence of involuntariness, while training-related spells—more widespread among youths and in the North—are characterised by an higher incidence of involuntariness.
84
G. Barbieri and P. Sestito
Table 5.3 Typology of temporary contracts (% of total temporary employment)—2006 average Age
Training Probationary period
Seasonal work
Occasional job
Project related
Temporarily vacant job position
Others
15–24 25–34 35–54 55–64 65 e + CentreMales North Females Total South Males Females Total Italy Males Females Total % workers involuntarily with a temporary status
47.4 14.3 1.8 0.4 0.0 26.6
9.6 10.4 6.7 3.5 0.0 13.6
11.1 15.7 26.6 39.1 26.4 11.3
12.4 19.3 20.5 21.1 15.1 14.5
5.0 12.7 11.2 12.1 27.9 10.5
3.9 11.9 15.7 8.0 4.6 6.9
10.5 15.5 17.5 15.7 26.0 16.6
17.1 21.5 11.3 8.2 9.9 20.1 14.0 17.0 85.0
8.8 11.0 5.5 3.3 4.5 10.2 6.9 8.5 88.1
13.1 12.3 31.7 31.3 31.5 19.9 19.4 19.7 72.4
19.7 17.3 18.6 20.6 19.5 16.2 20.0 18.1 85.0
9.6 10.0 11.4 10.2 10.9 10.9 9.8 10.3 81.9
16.9 12.3 5.2 14.3 9.4 6.2 16.0 11.2 89.4
14.8 15.6 16.1 12.1 14.3 16.4 13.9 15.1 80.7
5.3 Who are Temporary Employees? Using the LFS count—more precisely, using the reconstruction of temporary (and total) employment coherent with the new LFS frame provided by Istat—Fig. 5.2 depicts the evolution of temporary employment since the early 1990s. It is apparent that the growth of temporary employment in the 2004–06 period, the one under scrutiny here, has deeper roots. Moreover, the most recent acceleration in the growth of temporary employment may be related to cyclical factors, because temporary employment usually tends to anticipate cyclical upswings in the economy (Figs. 5.3 and 5.4). Bearing this longer-run evolution in mind, in what follows we focus on the 2004–2006 period and analyse who temporary workers are. More specifically, we report estimates of a probit model for the probability of being a temp (conditionally upon being an employee), replicating the same specification for the 2004 and 2006 quarter surveys.7 Our specification included demographical variables:
7 We selected the II quarter surveys because they are the least affected by seasonal factors. To be noted is that temporary employment is strongly seasonal, more so than total employment, with a peak in the summer. We focused on the probability of being a temporary worker (conditional upon being an employee), but a similar broad pattern would emerge on treating temporary workers as a share of the working age population.
5 The Quality of Temporary Work
85
Table 5.4 Typology of contractual arrangements (% of total temporary employment)—2006 average Age Training Apprenticeship Temporary Contract Other and Unknown and work according through a verbal contracts to a job agreement collective placement contract agency 15–24 25–34 35–54 55–64 65 e + Total Centre- Males North Females Total South Males Females Total Italy Males Females Total % workers involuntarily with a temporary status
9.4 5.0 1.5 0.5 0.0 4.6 26.6
34.2 6.5 0.2 0.0 0.0 10.6 13.6
34.9 65.7 76.8 73.5 59.5 62.7 11.3
2.6 3.3 2.4 1.7 1.1 2.7 14.5
5.8 9.4 8.7 10.2 20.7 8.4 10.5
13.1 10.0 10.3 14.0 18.8 11.0 6.9
17.1 12.2 11.3 8.2 5.7 18.0 12.4 9.7 87.1
8.8 6.2 5.5 3.3 2.6 9.1 6.1 4.8 83.5
13.1 7.0 31.7 31.3 18.1 28.4 28.8 11.2 82.5
19.7 9.8 18.6 20.6 11.2 14.5 17.6 10.3 90.1
9.6 5.7 11.4 10.2 6.2 9.7 8.7 5.9 78.9
16.9 40.9 5.2 14.3 43.7 5.5 14.1 42.0 78.7
Source: our calculations on Istat data
Age (a set of dummies covering four age groups: \25, 25–34, 35–54, [54), Marital status (a set of dummies if an individual is single, married, widowed or divorced), Gender, Education (a set of dummies distinguishing among four education levels: primary, lower-secondary, upper-secondary and tertiary). We also included a dummy for the status of Student and a dummy variable marking the school-to-work transition phase (equal to one for those people who had exited from the education system during the last 5 years). We also controlled for the local labour market context by using a set of geographical dummies (AREA, distinguishing among North–West, North–East, Centre and South) and the provincial Unemployment rate.8 Conditioning upon being an employee, we also made use of some job-related covariates: Occupation (a set of dummies distinguishing among
8
We dropped a quadratic in the rate of employment growth at provincial level, because this variable—used in Barbieri and Sestito (2008) in order to take account of the complex cyclical behaviour of temporary employment—is not very reliable in year 2004 owing to the break in the LFS.
86
G. Barbieri and P. Sestito 10.5
2400
10.0
2200
9.5 2000 9.0 8.5
1800
8.0
1600
7.5 1400 7.0 1200
6.5
1000
O
ct Ap -92 r O -93 ct Ap -93 r O -94 ct Ap -94 r O -95 ct Ap -95 r O -96 ct Ap -96 r O -97 ct Ap -97 r O -98 ct Ap -98 r O -99 c Ap t-99 r O -10 ct 0 Ap -10 r 0 O -10 ct 1 Ap -10 r 1 O -10 ct 2 Ap -10 r 2 O -10 ct 3 Ap -10 r 3 O -10 ct 4 Ap -10 r 4 O -10 ct 5 Ap -10 r 5 O -10 ct 6 -1 06
6.0
% of total employment
absolute value (right axis)
Fig. 5.2 Temporary employment—data coherent with the new LFS frame. Source: our elaborations on Istat data
8000
80 60
6000
40 4000
20
2000
ct Ap 92 r-9 O 3 ct Ap 93 r-9 O 4 ct Ap 94 r-9 O 5 ct Ap 95 r-9 O 6 ct Ap 96 r-9 O 7 ct Ap 97 rO 98 ct Ap 98 rO 99 ct Ap 99 rO 00 ct Ap 00 rO 01 ct Ap 01 rO 02 ct Ap 02 rO 03 ct Ap 03 rO 04 ct Ap 04 r-0 O 5 ct Ap 05 r-0 O 6 ct -0 6
0 O
-20 -40
0 -2000
-60 -4000
-80 -100
-6000 Temporary employment
GDP
Fig. 5.3 Cyclical components of temporary employment and GDP. Source: our calculations on Istat data
legislators and managers, professionals, technicians and clerical workers, services and sales workers, craft and related workers, plant/machine operators, elementary occupations, armed forces) and Industry (a set of dummies for industry excluding construction, trade, private services, public administration, agriculture).
5 The Quality of Temporary Work
87
0.8
0.6
0.4
0.2
0
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
-0.2
-0.4
Fig. 5.4 Co-movements of detrended temporary employment (quarter t) and GDP (at various leads and lags). Source: our calculations on Istat data
The results do not change greatly between the 2 years. According to a standard Blinder–Oaxaca decomposition,9 the 1% point rise in the temps’ share between 2004 and 2006 is almost exclusively explained by the differences in coefficients. However, the only sizable difference is that in the constant, capturing an acrossthe-board rise in the probability of being a temp. Note that a similar across-the-board rise also emerges if one compares the provincial dummies across the 2 years in a specification where these dummies (interacted with a year dummy) substitute for the area dummies and the provincial unemployment rate. The differences across provinces (as estimated by the province dummies and thus controlling for the covariates shown in Table 5.5) are rather stable over time (Fig. 5.5). If anything, the provinces exhibiting the largest increase in the incidence of temporary employment are those with the worse unemployment performances: the effect is, however, not precisely estimated. Both the area dummies and the more detailed provincial dummies show that an employee’s probability of being temporary is greatest in the high unemployment 9
The Blinder–Oaxaca decomposition identifies an explained and an unexplained component, the former being defined in terms of differences in the values of the covariates, while the latter is associated with differences in the coefficients associated with those covariates (including the constant). In order to handle a non-linear model, we employed the methodology proposed by Yun (2004). The difference in probabilitiesi (Y)h between 2006 (A) andi 2004 (B) can be decomposed as h
: Y A Y B ¼ /ðXA bA Þ /ðXB bA Þ þ /ðXB bA Þ /ðXB bB Þ : In order to obtain proper weights, two approximations are used: the first, an approximation of the value of the average of the function, /ðXbÞ; with that of the function evaluated at the average value of the exogenous variables /ðXbÞ; the second, a first order Taylor expansion to linearize the characteristics and coefficients effects around X A bA and X B bB ; respectively.
Occupation
School-to-work transition period Sector
Education
Student Age
Gender Marital status
0.035 0.040 0.040 0.036 0.035 0.080 0.078 0.079 0.079 0.079 0.119
-0.564 -0.530 -0.793 -0.474 -0.275 0.142 0.117 0.276 0.218 0.731 -0.237
25–34 35–54 [54 Lower secondary Upper secondary Tertiary
Industry excluding construction Construction Trade Private services Public administration Professionals Technicians and clerical workers Services and sales workers Craft and related workers, plant/machine operators Elementary occupations Armed forces
Std. Err. 0.017 0.021 0.036 0.038 0.026 0.030 0.042 0.030 0.032 0.044 0.038
-0.270 -0.184 -0.153 0.298 -0.550 -0.902 -1.135 -0.136 -0.153 -0.074 0.507
Male Married Separated, divorced, widowed
Coef.
Table 5.5 Probit estimates of the determinants for an employee being temporary 2004 P[z
0.000 0.046
0.000 0.000 0.000 0.000 0.000 0.076 0.131 0.001 0.005
0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.091 0.000
0.843 0.043
-0.620 -0.584 -0.807 -0.558 -0.375 0.133 0.175 0.366 0.289
-0.273 -0.167 -0.103 0.236 -0.528 -0.868 -1.182 -0.171 -0.192 -0.098 0.482
Coef.
2006
0.076 0.107
0.034 0.040 0.038 0.035 0.034 0.078 0.075 0.076 0.075
0.017 0.020 0.034 0.040 0.028 0.032 0.042 0.030 0.032 0.043 0.445
Std. Err.
(continued)
0.000 0.690
0.000 0.000 0.000 0.000 0.000 0.088 0.019 0.000 0.000
0.000 0.000 0.002 0.000 0.000 0.000 0.000 0.000 0.000 0.022 0.028
P[z
88 G. Barbieri and P. Sestito
0.044
0.000
0.0112
0.0713
0.000 0.000 0.000 0.000 0.002
0.023 0.025 0.035 0.221 0.092
0.085 0.153 0.158 1.105 -0.286 0.138 61,254 39.46
0.097 0.186 0.276 0.521 -0.219 0.153 61,155
Coef.
P[z
Coef.
Std. Err.
2006
2004
0.023 0.025 0.035 0.480 0.090
Std. Err.
0.000 0.000 0.000 0.290 0.016
P[z
Baseline category: gender: female; marital status: single; age: \25; educational level: primary school; sector: agriculture; occupation: legislators and managers; area: North–West
North–East Centre South Provincial unemployment rate Constant Pseudo R2 N. obs. Chi Test (26) for the equality of all coefficientsapart from a year dummy on the constant Pooled model Dummy year (in a model with all other coefficients constant over time)
Area
Table 5.5 (continued)
5 The Quality of Temporary Work 89
90 Fig. 5.5 Probability of being temporarily employed: differences across provinces (net of the impact of the other covariates in Table. 5.5)
G. Barbieri and P. Sestito
2004
provincial dummies (-0.31,0.09) (0.09,0.20) (0.20,0.36) (0.36,0.87)
2006
provincial dummies (-0.40,0.04) (0.04,0.22) (0.22,0.37) (0.37,0.73)
Mezzogiorno area (and the provincial unemployment rate, when considered instead of a set of province dummies, has a positive impact on the incidence of temporariness).
5 The Quality of Temporary Work
91
As for the other effects, according to the estimates reported in Table 5.5, the probability of an employee being temporary diminishes (although not monotonically) with age and educational attainment, being highest for females, students, and during the school-to-work transition period. Temporary employment is most important in agriculture and in low-skilled occupations. We can summarise our results by stressing that temporary employment is overrepresented among weak groups of workers and in more menial industries and occupations. Over time, there has been a sizable rise in the incidence of temporary employment, and it has mostly occurred across the board. Although cyclical factors are likely to have played a role as well, we are not able to disentangle their contribution to the 2004–2006 change.
5.4 The ‘Quality’ of Temporary Jobs The concerns induced by the rise in temporary employment stem from the idea that temporary jobs may be bad jobs. The quality of a given job is intrinsically difficult to define. It is difficult to say above what threshold a job is a ‘good’ one. Even using the simplest metric—the wage—and fixing a given threshold, it is difficult to determine whether a given worker in a given job is in a bad position. Doubtless a better wage would make that person better off. However, given his/her actual capabilities and market chances, that job may be better than remaining unemployed. While recognizing these ambiguities, here we adopt the standard practice of considering whether temporary positions are associated with certain qualitative features (see for instance the special chapter in OECD 2002). The estimates are not intended to identify the impact of being temporary on those job features. Besides controlling for as many covariates as possible, in order to take into account that temps may differ from permanent ones and that well-known factors may explain those features, we do not attempt any structural interpretation. The qualitative features considered are those which can be captured within the LFS (2004, 2006), which comprises neither wage data nor summary job satisfaction measures.10 Before looking at the association between temporariness and those quality features, we consider the origin of the temporary status as evaluated from the worker’s perspective. More specifically, we focus on the involuntariness of the temporary status as identified by the fact that the worker states that s/he accepted that job because of the lack of suitable alternatives. As such, this is a first summary qualitative judgement on the temporary position. Note that there is some ambiguity 10
The latter have been examined in Sestito (2002), Chapt. 7 using the SHIW data produced by the Bank of Italy. The results show that jobs which are ‘‘discontinuous over the year’’, a concept related to but different from the temporariness here considered, are associated with an higher perceived risk of unemployment, but also, albeit less intensively, with more ‘‘consideration by other people’’. The summary measure of job satisfaction is entirely unrelated to the discontinuous features of the job itself. The results were obtained in a specification with controls for earnings.
92
G. Barbieri and P. Sestito 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0
O
ct Ap -92 r O -93 ct Ap -93 r O -94 ct Ap 94 r O -95 ct Ap -95 r O -96 ct Ap -96 r O -97 ct Ap -97 r O -98 ct Ap -98 r O -99 ct Ap -99 r O -00 ct Ap -00 r O -01 ct Ap -01 r O -02 ct Ap -02 r O -03 ct Ap -03 r O -04 ct Ap -04 r O -05 ct Ap -05 r O -06 ct -0 6
0.0
% of involuntary temporary empl.
Fig. 5.6 Share of involuntarily temporary workers. Source: our calculations on Istat data
in the question, because ceteris paribus the worker should gain nothing from the temporary nature of the contract; indeed, permanent status does not restrict the worker’s chances of leaving the firm, so that the lack of a permanent status, i.e. a temporary status, only creates additional uncertainties about work prospects beyond the contract’s expiry date (for the firm, by contrast, there is a costless opportunity to terminate the contract at the expiry date unilaterally, while terminating it before the expiry date may be quite costly). Hence it is unclear why temporary status per se should be voluntarily accepted by the worker. Our interpretation is that those respondents who declared that they did not want a permanent contract (the complement to one of those termed ‘involuntarily temps’) appreciated other specific features of their ongoing contract: for instance, the fact that they were employed only during a given season. The ambiguity is, however, that there may have been some respondents who appreciated the overall package of their ongoing temporary work but, quite understandably they did not like the temporary clause as such, and consequently stated that they accepted it for the lack of alternatives. So the involuntariness indicator may not be taken at face value, representing only a proxy of the underlying unsatisfactoriness of the ongoing work. Whatever the interpretation problems related to the current formulation of the LFS may be, it should be pointed out that the formulation adopted until 2003 was even more problematic because it mixed this feature with (some of) the details on the contractual temporary arrangements (see Sect. 5.2). Hence, whilst the current data may overestimate the incidence of temporary workers unsatisfied by their ongoing work experience, the previous data clearly underestimated it. Indeed, Fig. 5.6 documents the time break in the series before and after 2004.
Male Married Separated, divorced, widowed
25–34 35–54 [54 Education Lower secondary Upper secondary Tertiary School-to-work transtion period Sector Industry excluding construction Construction Trade Private services Public administration Occupation Professionals Technicians and clerical workers Services and sales workers Craft and related workers, plant/machine operators
Student Age
Gender Marital status
0.074 0.075 0.091 0.124 0.086 0.094 0.131 -0.037 0.097 0.114 0.111 0.093 0.091 0.223 0.221 0.227 0.228
0.310 0.333 0.170 0.439 0.729 1.014 0.769 1.014
0.049 0.066 0.112
Std. Err.
-1.027 0.031 -0.145 -0.611 -0.046 -0.019 0.197 -0.095 0.392
-0.030 -0.159 -0.047
Coef.
Table 5.6 Probit estimates of the determinants of being involuntarily a temporary employee 2004
0.001 0.000
0.007 0.003 0.066 0.000 0.001 0.000
0.000 0.682 0.110 0.000 0.595 0.839 0.132 0.074 0.000
0.548 0.016 0.675
P[z
0.955 1.149
0.409 0.363 0.207 0.299 1.099 1.086
-0.856 0.296 0.118 -0.535 0.119 0.161 0.290 0.613 0.472
0.006 -0.054 -0.040
Coef.
2006
0.227 0.229
0.117 0.104 0.092 0.088 0.230 0.222
0.075 0.073 0.087 0.117 0.083 0.091 0.125 0.013 0.098
0.048 0.066 0.104
Std. Err.
(continued)
0.000 0.000
0.000 0.001 0.024 0.001 0.000 0.000
0.000 0.000 0.173 0.000 0.155 0.077 0.020 0.071 0.000
0.900 0.414 0.702
P[z
5 The Quality of Temporary Work 93
P[z
0.0482
0.005
0.0187
0.0360
0.000 0.196 0.000 0.266 0.045 0.653 0.963
0.228 0.333 0.064 0.074 0.104 0.984 0.268
1.176 0.602 -0.255 0.011 -0.244 0.132 -0.562 0.098 6,225
Coef.
Std. Err.
Coef. 1.072 0.431 -0.253 0.082 0.209 0.984 0.012 0.098 5,594 39.05
2006
2004
0.229 0.286 0.064 0.076 0.104 4.920 0.266
Std. Err.
0.000 0.036 0.000 0.882 0.019 0.903 0.035
P[z
Baseline category: gender: female; marital status: single; age: \25; educational level: primary school; sector: agriculture; occupation: legislators and managers; area: North–West
Elementary occupations Armed forces Area North–East Centre South Provincial unemployment rate Constant Pseudo R2 N. obs. Chi Test (26) for the equality of all coefficients apart from a year dummy on the constant Pooled model Dummy year2006 (in a model with all other coefficients constant over time)
Table 5.6 (continued)
94 G. Barbieri and P. Sestito
5 The Quality of Temporary Work
95
Besides the difference in level, the pre-2004 data show a gently declining incidence of involuntariness—indeed, involuntary temps were a minor component of the overall positive trend for temporary contracts (see Barbieri and Sestito 2008)—while the data for the 2004–2006 period show an upward trend (the incidence of involuntariness rises from 75% in 2004 to 82% in 2006). Given the ambiguity of the question, it is rather difficult to understand the reasons for this rise, which may simply reflect increased social alarm about job precariousness (see Accornero 2006). Whatever the case may be, we again resort to a regression frame in order to examine the behaviour of the involuntariness indicator, controlling for compositional and contextual factors. The impact of the latter is not so obvious. For instance, in a buoyant economy one might expect the abundance of job offers to imply a decline in the involuntariness indicator because people do not have to accept the first offer received; on the other hand, with people having greater chances but also greater expectations, those ending up in a temporary position may be people feeling particularly unlucky and disappointed, while in bad times people may feel relatively happy when (at least) they find a temporary position. More specifically, we estimated a probit model for the chance of being involuntarily temporary (conditional upon being a temporary worker: Table 5.6). The specification largely resembled that already reported in the previous section, and we again examined the II quarter cross sections of 2004 and 2006. Broadly speaking, industry and occupation dummies are the most relevant regressors: involuntariness is highest in the public sector and lowest in agriculture and private services; involuntariness is lowest among managers and professionals, but also among service workers. With the exception of the dummy for the oldest age groups—people over 54 are least likely to be involuntary temps—the pattern of the other covariates (for instance gender or education) is quite imprecisely estimated. This is why the estimates are often unstable across the 2 years. Hence, in order to examine the changes experienced over time we resorted to a pooled model. The basic exercise, reported in the lowest part of Table 5.6, confirmed the across-the-board increase in the incidence of involuntariness (while the changes in the other coefficients are not significant). Using a pooled model with a year dummy on the constant in order to take account of this across-the-board rise, we further examined (see the results in Table 5.7) the link between certain features of the ongoing work experience— those already described in Sect. 5.2, more precisely in Tables 5.2, 5.3, 5.4—and involuntariness. It appears that the incidence of involuntariness is highest for jobs with some training content, those related to employment in temporarily vacant positions, and those which involve probationary periods, while occasional and seasonal jobs are the least likely to be characterised by involuntariness. By contrast, the contractual arrangements and the further expected duration of the ongoing work have less weight. Using the enriched formulation of Table 5.7 we experimented further by introducing a set of provincial dummies (interacted with a year dummy). Clearly, these absorbed the effect previously attributed to the unemployment rate and
96
G. Barbieri and P. Sestito
Table 5.7 Probit estimates of the determinants of being involuntarily a temporary employee (coefficients of expected further duration, typology of temporary contracts and contractual arrangements) Coef. Std. Err P[z Expected further duration of temporary job 2–6 months 7–12 months [12 months Unknown
-0.028 -0.073 -0.079 -0.034
0.084 0.084 0.092 0.096
0.735 0.381 0.394 0.723
Type of temporary contract Probationary period Seasonal work Occasional work Project related Temporarily vacant job position Others
0.109 -0.627 -0.294 -0.307 0.108 -0.219
0.096 0.084 0.086 0.088 0.097 0.081
0.254 0.000 0.001 0.000 0.268 0.007
Type of contractual arrangement Apprenticeship Temporary according to a collective agreement Contract through a job placement agency Other and verbal agreement Unknown
-0.168 -0.049 -0.047 -0.117 -0.144
0.085 0.083 0.090 0.110 0.108
0.049 0.556 0.606 0.288 0.184
Baseline category: expected duration: \1 month; type of temp contract: Training; type of contractual arrangement: Training and work contracts. The model was estimated over the pooled sample and included all the covariates of Table 5.6 (with coefficients assumed to be constant over time) and a year dummy
showed an higher incidence of involuntariness in the South. Two features should be stressed. Firstly, it appears that the ranking of the provinces is quite stable across the 2 years. While this stability resembles the stability already depicted for the incidence of temporary employment as such (see previous section), there appears to be no link between the ranking of the provinces for the two phenomena. Neither the levels nor the changes over time in the incidence of involuntariness and of temporariness as such are correlated to each other. As said, we do not consider the involuntary indicator to be very informative. Consequently, we examine two further indicators, which are somehow more ‘objective’ because they consider what is actually happening in the currently held job. The first indicator refers to worker’s search-behaviour. It considers, independently from the judgement on the currently held job, whether the worker is looking for an alternative job.11 Its rationale is simply that looking for alternatives 11
More precisely, the indicator identifies who is looking for an alternative job, as opposed to individuals not engaged in a search, or in a search motivated by the intention to find a second job (on top of the one currently held).
Occupation
School-to-work transition period Sector
Education
Student Age
Temporarily employed Gender Marital status
Industry excluding construction Construction Trade Private services Public administration Professionals Technicians and clerical workers Services and sales workers Craft and related workers, plant/ machine operators Elementary occupations Armed forces
25–34 35–54 [54 Lower secondary Upper secondary Tertiary
Male Married Separated, divorced, widowed
Table 5.8 Probit estimates of being an employee looking for another job
0.024 0.019 0.022 0.036 0.046 0.031 0.035 0.053 0.034 0.036 0.049 0.044 0.045 0.049 0.048 0.045
P[z
0.045 0.103 0.100 0.101 0.101 0.102 0.264
0.057 0.100 0.351 0.512 0.485 0.805 -0.812
0.000 0.002
0.203 0.336 0.000 0.000 0.000
0.000 0.000 0.000 0.000 0.848 0.000 0.503 0.000 0.986 0.082 0.000 0.000 0.004 0.000 0.003 0.000
0.752 -0.306
0.115 0.056 0.304 0.371 0.452
0.401 -0.126 -0.220 0.096 0.094 0.202 0.038 -0.430 -0.053 -0.024 0.114 0.112 0.189 0.186 0.210 0.279
Coef.
Std. Err.
Coef. 0.538 -0.074 -0.183 0.139 -0.009 0.141 -0.023 -0.542 0.001 0.062 0.236 0.156 0.130 0.211 0.143 0.213
2006
2004
0.102 0.179
0.048 0.105 0.100 0.101 0.100
0.024 0.020 0.023 0.037 0.049 0.034 0.039 0.056 0.037 0.039 0.051 0.037 0.048 0.053 0.050 0.048
Std. Err.
(continued)
0.000 0.088
0.016 0.592 0.002 0.000 0.000
0.000 0.000 0.000 0.010 0.052 0.000 0.331 0.000 0.152 0.532 0.027 0.002 0.000 0.000 0.000 0.000
P[z
5 The Quality of Temporary Work 97
Std. Err.
0.523 -0.107 0.666
-0.099
Year2006*temp empl
0.033
0.023 0.031 0.023
0.027 0.028 0.038 0.234 0.020 0.118
P[z
0.002
0.000 0.001 0.000
0.0461
0.007 0.000 0.064 0.000 0.000 0.000
-0.040 0.102 0.042 3.244 0.413 -2.502 0.127 61,155
Coef.
-0.072 0.130 0.070 2.605 0.337 -2.471 0.128 61,254 40.49
Coef.
Pseudo R2 N. obs. Chi Test (27) for the equality of all coefficients apart from a year dummy on the constant and its interaction with the temporary dummy Temporary employment Year2006*temp empl Temporary employment
North–East Centre South
2006
2004
0.028 0.030 0.041 0.330 0.021 0.121
Std. Err.
0.156 0.001 0.313 0.000 0.000 0.000
P[z
Baseline category: gender: female; marital status: single; age: \25; educational level: primary school; sector: agriculture; occupation: legislators and managers; area: North–West
Pooled model with province dummies and all covariates interacted with a year dummy
Pooled model
Provincial unemployment rate Tenure \ 2 years Constant
Area
Table 5.8 (continued)
98 G. Barbieri and P. Sestito
5 The Quality of Temporary Work
99
may be indicative of dissatisfaction with the current job. The other indicator considers participation in education and training activities. Its rationale is simply that life-long learning activities, often deemed to be essential in a knowledgebased economy, may capture longer-run quality features of a job. Also for the two ‘objective’ indicators we consider a probit model which captured compositional and context factors possibly impinging upon them. Besides the same factors already included in the previous models, this one also included a dummy for elapsed tenure, because tenure (which by definition is on average shorter for temporary positions than for permanent ones) may have an impact upon both phenomena. Shorter tenure jobs are those which have not yet passed through the very initial trial period during which the worker decides whether s/he likes the ongoing experience, while training and human capital investments may be systematically related to tenure (as newly-hired people may need some training, while the expected duration of the job may provide an incentive to invest in training).12 Here the estimates were run over the entire employee population and the focus was on the impact of a temporary dummy. The results for the search-behaviour indicator are shown in Table 5.8, which in regard to the temp dummy shows a positive and sizable coefficient. However, over time, the impact of temporariness on the probability of looking for another job has slightly (but significantly from a statistical point of view) declined. Also when a set of province dummies13 were used (in a model in which these dummies as well as the other covariates were fully interacted with a year dummy), the temporary employment dummy was positively related to the search for an alternative job, although its interaction with the year dummy showed a decline in its impact. The results for training activities (Table 5.9) are interesting in themselves. Traditional human capital theory would predict that the worker finances general human capital investments while the firm pays for the specific human capital. ‘Modern’ human capital theory has shown that also the firm may be ready to invest in general human capital provided some labour market imperfections allow it to reap the benefits of those investments.14 In such a situation, the firm may be less inclined to invest in training temps because they are more likely to quit the firm. More broadly, an increase in turnover rates due to a shift to the use of temporary workers would enable firms to poach trained workers instead of training them. On the other hand, a temporary worker, whose reference is the external labour market and not the internal labour market of a given firm, may become more inclined to pay for training as
12
Unfortunately, we were not able to construct a provincial turnover rate like the one which we had employed in Barbieri and Sestito (2008) as an additional control variable. 13 The province dummies absorbed the effect of the provincial unemployment rate. They also captured the possible effect of the turnover rate neglected in the previous exercise. The same applies to the similar exercise conducted for the training indicator in Table 5.9. 14 One of the first models of the ‘modern’ approach is Acemoglu and Pischke (1999). A summing-up of the issues is in Bassanini et al. (2005).
Occupation
School-to-work transition period Sector
Education
Age
Temporary employed Gender Marital status
Industry excluding construction Construction Trade Private services Public administration Professionals Technicians and clerical workers Services and sales workers Craft and related workers, plant/machine operators Elementary occupations Armed forces
Male Married Separated, divorced, widowed 25–34 35–54 [54 Lower secondary Upper secondary Tertiary
Table 5.9 Probit estimates of training participation
0.116 0.030 -0.123 -0.033 -0.434 -0.567 -0.792 0.382 0.727 0.927 0.112 -0.095 -0.050 -0.009 0.149 0.432 0.036 -0.141 -0.258 -0.530 -0.513 -0.462
0.028 0.018 0.022 0.038 0.033 0.038 0.049 0.065 0.065 0.069 0.100 0.062 0.070 0.065 0.062 0.062 0.058 0.057 0.060 0.061 0.067 0.094
0.000 0.092 0.000 0.379 0.000 0.000 0.000 0.000 0.000 0.000 0.031 0.127 0.471 0.887 0.017 0.000 0.531 0.013 0.000 0.000 0.000 0.000
0.052 -0.002 -0.099 -0.032 -0.429 -0.533 -0.806 0.209 0.569 0.699 0.001 -0.224 -0.234 -0.167 -0.012 0.294 0.013 -0.182 -0.336 -0.556 -0.546 -0.570
Coef.
P[z
Coef.
Std. Err.
2006
2004
0.072 0.917 0.000 0.406 0.000 0.000 0.000 0.002 0.000 0.000 0.030 0.000 0.001 0.008 0.845 0.000 0.822 0.001 0.000 0.000 0.000 0.000
P[z
(continued)
0.029 0.019 0.023 0.038 0.035 0.040 0.052 0.066 0.066 0.070 0.127 0.061 0.070 0.063 0.060 0.060 0.057 0.055 0.059 0.060 0.067 0.098
Std. Err.
100 G. Barbieri and P. Sestito
Std. Err.
P[z
Year2006*temp empl
-0.058
0.051 0.259
0.146 0.023 0.000 0.091 0.024 0.000 0.084 0.026 0.001 0.056 0.028 0.041 -0.055 0.040 0.169 -0.033 0.042 0.437 -0.752 -0.359 0.266 0.177 -1.041 0.366 0.002 0.023 0.930 0.039 0.024 0.103 -1.503 0.108 0.000 -1.211 0.108 0.000 Pseudo R2 0.128 0.129 N. obs. 61,254 61,155 27.39 0.3891 Chi Test (26) for the equality of all coefficient apart from a year dummy on the constant and its interaction with the temporary dummy Temporary employment 0.033 0.035 0.335 Year2006*temp empl -0.074 0.048 0.123 Temporary employment -0.001 0.035 0.986
North–East Centre South
Coef.
P[z
Coef.
Std. Err.
2006
2004
Baseline category: gender: female; marital status: single; age: \25; educational level: primary school; sector: agriculture; occupation: legislators and managers; area: North–West
Pooled model with province dummies and all covariates interacted with a year dummy
Pooled model
Provincial unemployment rate Tenure \2 years Constant
Area
Table 5.9 (continued)
5 The Quality of Temporary Work 101
102
G. Barbieri and P. Sestito
human capital becomes more marketable.15 The results show a strongly negative link between training and age, and a strongly positive relationship with previous educational attainment. The effect of tenure is not precisely estimated, contrary to our previous results over a different time span (see Barbieri and Sestito 2008): it is possible that either some training activities are financed by the worker or that some training is deemed necessary to integrate recently-hired workers into the firm’s workforce. With respect to the temporary dummy, we observe a positive, albeit slightly diminishing over time, coefficient. When experimenting with a model with provincial dummies (and full interactions with a year dummy), the temporary work dummy turned out to be statistically insignificant. This lack of precision was probably related to our inability to provide for a structural interpretation. It is likely that the results mix up different effects, with the standard poaching story being compensated (even more than compensated in some specifications and time periods) by the fact that firms may strategically combine training opportunities and their right to unilaterally terminate the contract at a given date in order to select the ‘best’ workers. Hence, even if temps are more likely to quit, firms may be willing to provide them with training opportunities (and the workers may be willing to invest in training because it signals their abilities).
5.5 Conclusions The paper has examined the incidence of temporary employment in Italy. It appears that temporary employment is more widespread among the groups of workers weakest in terms of skills and local labour market context. A temporary status is also more frequent among youths, students, and people still in the schoolto-work transition period. In the 2004–2006 period, there was a further rise in the incidence of temporary work, which mostly occurred across the board, with a difficult-to-examine contribution of business cycle factors, to which temporary work is usually strongly related. Temporary jobs are mostly accepted because of the lack of alternatives. Although we constructed an indicator of the involuntariness of temporary positions on this basis, it was stressed that this indicator inevitably suffers from ambiguities. Interpreted as an indicator of dissatisfaction with the ongoing work, the incidence of involuntariness is mostly affected by industry and occupation. However, there is no precise link between the incidence of temporary jobs and their degree of ‘involuntariness’ in different subgroups and local labour markets within Italy. Moreover, the lack of alternatives, as a cause of temporary status, is not necessarily related to the content and the contractual 15
Empirically the LFS data allow categorization of people who, within the last four weeks, were in regular education or attended courses, seminars, conferences or received private lessons or instruction outside the regular educational system. We considered all these taught learning activities to be training, both relative to the job and to personal/social training opportunities.
5 The Quality of Temporary Work
103
details of the temporary position: actually, repeated spells with the same employer (the cases closest to a situation of abuse by the employer of the temporary clauses) and seasonal activities appear to be the least likely to be taken solely because of the lack of alternatives. Whatever its origins, the temporary status appears positively associated with the search for job alternatives. We maintain that this may be a more reliable indication of the lower quality of temporary jobs. However, if there exists an association between temporary status and the search for alternatives (controlling for a wide array of compositional and contextual effects possibly affecting job search activities), it has nevertheless weakened over time. Neither does it appear that being employed on a temporary contract is negatively associated (again controlling for a host of covariates) with training activities
References Accornero, A. (2006). San precario lavora per noi. Milano: Rizzoli. Acemoglu, D., & Pischke, J. (1999). Beyond becker: Training in imperfect labour market. Economic Journal, 109, 112–142. February. Anastasia, B., Disarò, M. & Maurizio, D. (2004). Occupati stabili, mobili, temporanei in Veneto: misure di consistenza e di ‘‘lock in’’. Veneto Lavoro, I tartufi, 16. Barbieri, G., & Sestito, P. (2008). Temporary workers in Italy: Who are they and where they end up? Labour, 22(1), 127–166. Bassanini, A., Booth, A., Brunello, G., De Paola, M., & Leuven E. (2005 june 11). Workplace training in Europe. Report prepared for the seventh European conference of the Fondazione Rodolfo DeBenedetti, Venice. Ministry of Labour. (2006). Rapporto di monitoraggio sulle politiche occupazionali e del lavoro, http://www.lavoro.gov.it. OECD. (2002). Employment outlook. Paris: OECD. OECD. (2004). Employment outlook. Paris: OECD. Pirrone, S., & Sestito, P. (2006). Disoccupati in Italia. Bologna: il Mulino. Sestito, P. (2002). Il mercato del lavoro in Italia. Com’è. Come sta cambiando. Roma-Bari: Laterza Sestito, P. (2004). Compatibilità finanziarie ed effetti economici della legge Biagi. M. Tiraboschi (Ed.), La riforma Biagi del mercato del lavoro—Prime interpretazioni e proposte di lettura del d.lgs. 10 settembre 2003, n. 276. Il diritto transitorio e i tempi della riforma, Milano: Giuffré. Tronti, L., Ceccato, F., & Cimino, E. (2003). Measuring atypical jobs: Levels and changes. Rivista di statistica ufficiale, 1. Yun, M. S. (2004). Decomposing differences in the first moment. Economics Letters, 82(2), 275–328.
Chapter 6
Temporary Help Workers in Italy. Where Do They Come From and Where Do They Go? Federica Origo and Manuela Samek Lodovici
6.1 Introduction Temporary help work is a quite recent phenomenon in the Italian economy, but its share of total employment has been continuously increasing since its formal introduction in 1997 (with law 196/1997, also known as the ‘Treu Package’, after the name of the Ministry of Labour who proposed it). According to recent estimates, at the end of 2006 around 270 thousand workers were employed on temporary help contracts, corresponding to almost 1% of total employment (more than 1.5% of dependent workers). Temporary help work is still increasing, although at a slower pace than at the end of the 1990s (Ministry of Labour 2007). The main purpose of this type of contract is to help firms to cope with demand peaks without incurring the cost of selecting and recruiting new workers. According to the law, temporary help work should be always allowed, except for replacing workers on strike, in firms which have made collective lay-offs in the previous 12 months and for jobs that require medical supervision. Subsequent collective agreements have narrowed the scope for using this specific type of fixedterm contract to production peaks, temporary substitution of absent employees, and recruitment of skills not available within the firm. They also stipulated the maximum proportion of the workforce that can be hired on a temporary help contract (which, depending on the specific economic sector, is between 8 and 15% F. Origo (&) Department of Economics ‘‘Hyman P. Minsky’’, University of Bergamo, via dei Caniana 2, 24127 Bergamo, Italy e-mail:
[email protected] M. S. Lodovici Istituto per la Ricerca Sociale IRS, Via XX Settembre 24, 20123 Milan, Italy e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_6, Ó Springer-Verlag Berlin Heidelberg 2012
105
106
F. Origo and M. S. Lodovici
of permanent employees) and the maximum duration of each individual contract (no more than four extensions, for a maximum cumulated duration of 24 months). From a labour market policy perspective, temporary help work has also been seen as a good way to increase individual employability and to ease the school-towork transition. Temporary help contracts should then increase job opportunities for the unemployed (mainly when standard contracts are closely regulated and costly in terms of firing restrictions) and reduce the probability of experiencing long spells of unemployment. In light of this second aim, a broad strand of empirical literature has studied the transitions of temporary help workers in the labour market. The main purpose of these studies has been to evaluate whether temporary help work can enhance the probability of obtaining a stable job, rather than becoming a ‘trap’ whereby the longer a worker is hired on a temporary help contract, the lower his/her probability of exiting from this state. Although it may be the case that some individuals want to remain temporary help workers for a long time (Houseman et al. 2003), this type of contract is usually accepted while workers look for better and stable job opportunities. In this sense, temporary help work should be a sort of ‘springboard’ to permanent employment, allowing the most disadvantaged people in the labour market (such as the long-term unemployed, young first job seekers, prime-age women looking for a job after an inactivity spell, immigrants, etc.) to increase their human capital and to acquire new skills. Furthermore, temporary help work experiences should enable these workers to show their productivity and quality to employers (in spite of the prejudices that the latter often have against these categories), thus increasing their future likelihood of obtaining a stable job. From this perspective, temporary help work should increase labour market efficiency as long as it favours individual transitions to stable employment, mainly for involuntary temporary help workers (i.e., those who have accepted temporary help contracts because they have been unable to find a permanent job). By contrast, policy–makers may have to deal with new problems if temporary help contracts do not significantly influence the probability of finding a stable job (even if they increase the probability of exiting unemployment), eventually generating a vicious circle between unemployment and temporary work. Although it is reasonable to assume that individuals and society both prefer temporary jobs to (long-term) unemployment, from a policy point of view the problem may merely shift from the traditional ‘unemployment trap’ to a new ‘temporary help work trap’ (or to the above-mentioned ‘unemployment-temporary help work–unemployment’ vicious circle), raising new questions concerning the (mal)functioning of the labour market and the effectiveness of the measures in place to correct market failures. This problem may be even more serious if temporary help work is a ‘trap’ only for relatively disadvantaged groups in the population, so that the ‘stigma’ is shifted from unemployment to temporary help work. More in general, in order to improve the focus of public policy intervention, it is important to determine whether the probability of moving from temporary help work to different labour market states
6 Temporary Help Workers in Italy
107
(such as stable employment, other temporary contracts, unemployment, education) is influenced by unobserved individual characteristics (such as ability, motivation, psychological attitude towards work and life) or whether, regardless of these features, it depends on the duration of temporary help employment, so that the longer an individual stays in this state, the lower his/her probability of exiting it (state dependence). It is clear that, in the case of individual heterogeneity, labour market policies should be more closely targeted than in the case of state dependence: in the former case, specific measures should be addressed to workers whose characteristics negatively influence their probability of obtaining a stable job; in the latter case, public intervention should be more generally aimed at minimizing the duration of temporary employment for all the workers concerned. In light of these considerations, this study examines the transitions of temporary help workers in Italy, controlling for individual, job and firm characteristics. Taking into account of the complexity of individual employment histories, in the empirical analysis we shall model the probability of transition from temporary help work to different labour market states (stable employment, other forms of temporary employment, unemployment, or education) and the role played by state dependence. The rest of this study is organized as follows. The next section presents the theoretical background and the main results of previous empirical studies on the transitions of temporary help workers, paying special attention to evidence on Italy. Section 6.3 describes the main features of the dataset used in the empirical analysis, whose main results are reported and discussed in Sect. 6.4. The last section makes some concluding remarks and draws some policy implications.
6.2 Literature Background A strand of the socio-economic literature on temporary help work has studied the impact of this form of employment on the probability that individuals will transit to other states in the labour market, in particular to stable employment. Economic theory predicts two positive effects of temporary help work on the probability of finding a stable job: a direct (signalling) effect and an indirect (via human capital) effect. Firms may in fact use temporary help contracts to ascertain the real productivity and quality of workers before hiring them on a permanent basis. The probability of finding a stable job may also be increased because, during the temporary help work experience, individuals acquire new skills and increase their human capital. Furthermore, in markets with imperfect and incomplete information (as is usually the case of the labour market), temporary help work should favour the exchange of information between firms and workers, enabling the latter to enlarge their social networks and to acquire more information on vacancies (with permanent contracts).
108
F. Origo and M. S. Lodovici
On the other hand, signalling theory predicts that temporary help work may reduce the probability of obtaining a stable job if acceptance of a temporary help contract is interpreted as a negative signal by the firm, since this behaviour may hide difficulties in entering the primary labour market (characterized by stable jobs). Furthermore, according to human capital theory, firms may be less prone to pay for the specific human capital accumulated by these workers due to their high level of turnover. Given the lack of unambiguous theoretical predictions on the impact of temporary help work on the probability of finding a stable job, the issue must be addressed empirically. Many empirical studies have evaluated whether temporary help work experience facilitates entry into the labour market, leading the individual to stable employment. According to an international study by the OECD (2002), the probability of transition to stable employment differs significantly across countries, ranging from 21% in France to 56% in Austria. Despite these differences, the average characteristics of temporary help workers who obtain stable jobs are quite similar across countries: such workers are usually young (aged 25–34) and highly skilled/educated. Furthermore, almost everywhere, temporary help workers are much more likely to become unemployed than are workers on permanent contracts (7–24%, against 1–5%). International evidence on specific countries points up the different roles that temporary help work may play in the labour market. For example, in the case of Spain, Guell and Petrongolo (2007) found that this type of contract is used by firms for two main reasons: to increase labour flexibility (in the presence of high firing costs associated with permanent contracts) and to screen workers’ quality. Studying the probability of transition to stable employment in relation to the duration of temporary help contracts, Guell and Petrongolo found that this probability is relatively higher both 1 and 3 years after the contract has been signed. The former peak may be associated with the use of temporary help contracts as screening devices, while the latter is evidence of the need for flexibility within firms (3 years is in fact the maximum duration of this type of contract in Spain, after which the firm must either let the worker go or hire him/her on a permanent contract). Further evidence on Spain highlights the important role played by temporary help agencies in favouring the creation of stable employment, mainly in the case of continuing and long temporary help work experiences (Malo and Munoz-Bullon 2002). Empirical evidence on the Netherlands shows that the (positive) probability of finding a permanent job is closely influenced by individual characteristics, and that it is relatively higher for young men (mainly with medium levels of education) with a working partner, and for women without children (de Graaf-Zijl et al. 2011). Nonetheless, other evidence from the Netherlands indicates that temporary agency work does not contribute to creating a dual labour market: for instance, women and immigrants, who have greater probabilities of obtaining temporary help work, are as likely as other workers to subsequently achieve a stable job (Russo et al. 1997).
6 Temporary Help Workers in Italy
109
Evidence from Great Britain shows that temporary agency workers are less satisfied with their working conditions, receive less training, and earn lower wages than stable employees (Booth et al. 2002). Overall, the international evidence suggests that temporary help work may assist people in escaping unemployment (Garcia-Perez and Munoz-Bullon 2005), but this effect may be much weaker for specific groups in the population, mainly women with small children and older workers, who are more likely to become unemployed again once the temporary help contract has expired (D’Addio and Rosholm 2004). Empirical evidence for Italy is relatively recent, and it mostly confirms, on the one hand, the role of temporary help work as a springboard to stable employment, and on the other one, its twofold use as a screening device and as a means to increase flexibility. With reference to non-permanent workers aged between 18 and 40 and resident in Tuscany and Sicily, Ichino et al. (2003) estimated that temporary help work significantly increases the probability of obtaining a stable job one and half years later (28, against 14% without temporary work experience). However, this improvement is evident with other forms of temporary work as well. Another piece of evidence shows that the probability of transiting from a temporary help contract to a stable job is higher for contracts with a medium duration (around 6 months), a length of time long enough for the company to ascertain the worker’s productivity, but not so long that it becomes excessively costly with respect to other kinds of employment contract (Montanino and Sestito 2003). Furthermore, a more general study on temporary work (which considered also fixed-term contracts different from temporary help ones) shows that the probability of transition to stable employment is lower for individuals whose work careers are characterized by continuous transitions between employment and unemployment, or by a large number of short temporary contracts (Gagliarducci 2005). The current empirical evidence then seems to highlight the existence of quite different types of temporary help workers and different reasons for the use of this type of contract. Considering the number and the average duration of temporary help contracts, Incagli and Rustichelli (2002) define four different types of temporary help workers: occasional, characterized by a low number of relatively short temporary help contracts; fixed-term, who have experienced few but relatively long temporary help work experiences; marginal, who have cumulated a large number of short temporary help contracts; and permanent, whose number of days worked in a year equals that of similar permanent workers but is the result of subsequent temporary help work experiences (Fig. 6.1). The above classification also suggests that the probability of a certain worker ending up in one of these categories depends closely on both his/her individual characteristics and those of the employing firm(s). Furthermore, the quality of the match between workers and firms (and hence the characteristics of temporary help jobs) may also influence the probability of transition to stable employment. In the empirical analysis we shall take into account of all these aspects.
110
F. Origo and M. S. Lodovici Average length of temporary help work contract: N. contracts:
Short
Long
Few
Occasional worker
Fixed-term worker
Many
Marginal worker
Permanent worker
Fig. 6.1 Types of temporary help workers
6.3 The Data The dataset used in the empirical analysis is the result of an ad hoc (CATI) survey conducted on a representative sample of around 2,300 temporary help workers in Italy (for more details see IRS et al. 2005). More specifically, the sample contains individuals who had at least one temporary help work experience during the 12 months prior to the survey.1 The dataset contains detailed information on individual socio-demographic characteristics (such as gender, age, education, family background, socioeconomic conditions), temporary help work experiences (firms and job characteristics), employment histories (employment situation both before and after the temporary help work experience). The survey was aimed, in fact, at collecting information on different aspects of workers’ life, focusing on their condition in the labour market before the first temporary help job in the 12 months prior to the survey; the temporary help work experience during those 12 months (including its duration and subjective evaluation of this experience, mainly in terms of perceived pros and cons); condition in the labour market at the moment of the survey; training activities before, during and after the temporary help work experience (distinguishing between training received as a temporary help worker and other forms of training)2; the worker’s skills at the moment of the survey (and eventual demand for further training in the medium and long run). The sample contains 2,336 individuals. In light of the aim of our empirical analysis, we selected those individuals for whom information on the overall duration of the temporary help work experience was available. We then excluded 1 The survey was carried out between May and November 2004. Hence, the workers surveyed had at least one temporary help work experience between April 2003 and October 2004. 2 In Italy there is a specific public fund to finance training for temporary agency workers (Forma.Temp). Temporary help agencies contribute to the fund an amount equal to 4% of the gross wage of temporary workers. The fund became operative in 2002 through the issue of a ‘‘Training for temporary help workers guide’’ indicating the type of training that could be financed: basic training, vocational training, on-the-job training, and continuing training. Since 2002 further agreements between the social partners have specified that training activities should be mainly addressed to the long-term unemployed, the disabled, immigrants, and women reentering the labour market.
6 Temporary Help Workers in Italy
111
Table 6.1 Sample characteristics: a comparison with official data (%) Ad hoc survey 2004 Istat Labour Force Survey (LFS)
Female over 30 years old By education: College High school Vocational training Compulsory school or lower By economic activity Metalworking Other manufacturing Trade Public administration Other services
Temporary help workers
Temporary help workers
Other temporary workers
Permanent workers
51.4 46.4
47.9 61.6
51.9 56.5
41.1 81.9
16.3 46.8 9.4 27.5
14.2 41.4 6.5 37.9
14.3 36.1 7 42.6
13.3 37.2 8.5 41
26.9 28.4 11.9 7.1
17.7 27.3 10.4 5.3
7.6 19.3 11.7 6.3
12.5 22.3 11.4 9.5
25.7
39.3
55.1
44.3
those workers for whom this information was missing (93 individuals) and those who were already on a temporary help contract before the time span considered by the survey (390 workers, for whom we did not know the duration of temporary help jobs before the 12 months considered in the interview). To simplify the estimation model, we also excluded few temporary help workers moving to labour market states that could not be grouped with other states (76 individuals).3 After applying the above criteria, we ended up with a final sample of 1,777 individuals who had at least one temporary help work experience in 2003–2004. Comparison with official data (taken from the 2004 national Labour Force Survey) as reported in Table 6.1 shows that our sample is quite representative of the population of temporary help workers, with the partial exception of workers aged over 30 and workers employed in other services (both of which categories are under-represented). The larger share of young individuals is reflected in average educational levels (slightly higher in our sample than in the official data), while the lower share of workers in services is associated with a much higher share of workers in the metalworking industry (with no substantial differences for the other sectors).4 3 They were essentially self-employed, workers without formal contracts (neither permanent nor temporary), and inactive people who were not students. 4 Note that when the survey was carried out, no official statistics on temporary help employment in Italy were available. A specific question on temporary help work was introduced for the first time in the 2004 Labour Force Survey, but the data reported here were released a few months after the end of our survey. The sample was selected by drawing on the archives of two large temporary help agencies (Randstad and Obiettivo Lavoro) and five regional public employment services (two in the Centre-North of Italy, three in the South).
112
F. Origo and M. S. Lodovici
Table 6.2 Previous and following labour market status of temporary help workers Whole sample Restricted sample
Permanent employment Temporary help work Fixed-term employment a Self employed Other employed Unemployed Student Inactive Total (N observations) a
Before the temporary help work experience in 2003–2004 (%)
At the moment of the survey (%)
Before the temporary help work experience in 2003–2004 (%)
At the moment of the survey (%)
14.3
12.5
17.3
13.1
16.7
40.7
–
40.3
19.4
21.0
23.7
22.9
2.1 2.2 33.2 10.0 2.1 2.336
1.2 1.0 19.3 2.8 1.5
2.3 2.1 39.3 12.5 2.8 1.777
– – 20.2 3.5 –
Different from temporary help employment
The table also compares the average features of temporary help workers with those of workers hired on both other temporary and permanent contracts. It highlights that temporary help work, like other forms of fixed-term employment, is characterized by a larger share of women and young people. Nonetheless, it seems more widespread than other temporary contracts in manufacturing, particularly in the metalworking sector, while it is much less used in services. This type of specialization is distinctive of Italy; in other European countries (such as the United Kingdom, Sweden and Spain) manufacturing firms employ only around one-third of total temporary help workers (EIRO 2002).
6.4 Descriptive Statistics Table 6.2 reports the labour market condition before and after the temporary help work experience in 2003–2004. In order to take into account of the effect of the sample selection procedure described in the previous section, we present the results for both the whole sample and the selected sample. Both samples show that, before the temporary agency work experience, the majority of the subjects were unemployed or employed on temporary contracts. Comparison of labour market status before and after the temporary work experience reveals a sharp decline in the incidence of both the unemployed and students, accompanied by a marked increase in temporary help contracts. The other forms of
6 Temporary Help Workers in Italy
113
Fig. 6.2 Survival function in temporary help employment Kaplan-Meier estimates
employment are relatively stable, except for a slight decline of permanent workers in the selected sample. Overall, the figures in the table suggest that temporary help work is an effective way to exit unemployment, and it also seems to ease the transition from school to work. It is unclear, however, whether this type of contract is a real springboard to stable employment or whether, instead, it may become a ‘trap’. To provide some first (descriptive) insights into the influence of state dependence on the probability of exiting temporary help employment, Fig. 6.2 depicts the survival function (based on the non-parametric Kaplan-Meier estimator) for the restricted sample. The survival function shows the fraction of the sample still employed on temporary help contracts after a certain number of months have elapsed since the first temporary help contract was signed. The median duration of temporary help employment is around 6–8 months, but the slope of the function suggests that the survival rate tends to decline faster in the first 12 months. This means that it is relatively more difficult to exit from temporary help employment for workers who have been in this state for more than a year. Analysis of the survival function by individual characteristics does not show marked differences by gender, educational level and ethnic group (Fig. 6.3). Some differences seem to emerge in regard to age group, in that the probability of exiting temporary help employment is relatively lower for young people than it is for people aged over 30. Education seems to display some effects in the medium run: the probability of exiting after 1 year actually decreases with education, suggesting that the existence of state dependence may be more relevant for mid-high educated workers. Larger differences emerge when we consider firm and job characteristics, mainly the number of temporary help contracts and the number of employing firms (Fig. 6.4). More specifically, persistence in temporary help employment is higher for workers who have had more than one temporary help contract and/or been employed in more than one firm. Also the economic sector exerts some influence, given that the probability of exit is lower for workers in public sector jobs, mainly for longer durations.
114
Fig. 6.3 Survival function by individual characteristics
Fig. 6.4 Survival function by firms characteristics
F. Origo and M. S. Lodovici
6 Temporary Help Workers in Italy
115
Fig. 6.5 Survival function by personal and firm’s reasons
The survival function is also sensitive to the reasons that induce individuals and firms to resort to temporary help employment. (Fig. 6.5). Workers accepting temporary help contracts because they need work flexibility are in fact more likely than others to exit earlier from temporary help employment, mainly with respect to workers who have accepted a temporary help job because they could not find a permanent one. Regarding firms’ reasons, workers hired to substitute for employees who are temporarily absent (for instance, because of illness or maternity leave) exit temporary help employment earlier than the others in the first year. But for longer durations, the probability of exit is lower for workers whose temporary help contract has been used by the employer as a sort of probation period (worker screening).
6.5 Econometric Results Estimates of the transition probabilities of temporary help workers, with specific attention paid to the effect of the overall length of temporary help work experience, were based on duration models. Given the existence of different exit states, we specified a competing risk hazard model in which temporary help workers can eventually leave the initial state to enter: (1) permanent employment; (2) other types of fixed-term contracts; (3) unemployment; (4) any type of educational path leading to a formal degree. Furthermore, since we observed individuals at discrete points in time, we specified a discrete time duration model in which the probability of transition to a certain state depended on both observable (individual, firm and job) characteristics and the baseline hazard function k(t). Econometric results were then based on a multinomial logit model applied to a properly re-organized unbalanced panel dataset in which, for each person, there were as many rows as the
116
F. Origo and M. S. Lodovici
number of time intervals at risk of exiting temporary help employment for that individual.5 The model specification is as follows: Pr obðY ¼ jÞ ¼
expðbj xi þ kj ðtÞÞ PJ 1 þ k¼1 expðbj xi þ kj ðtÞÞ
j ¼ 1; :::; J
ð6:1Þ
where the dependent variable measures the probability of transition to a certain labour market state j (among the Jth available), k(t) is the baseline hazard rate (i.e., a function of the overall length of the temporary help work experience), and X a set of observable characteristics. The main advantage of this estimator is that its results are easy to read (in terms of Relative Risks Ratios). Furthermore, it allows one to model the hazard function very flexibly. In our case, the baseline hazard function was specified both parametrically (as the natural logarithm of t and as a polynomial function) and non-parametrically (using a piecewise constant duration dependence function defined by a set of dummy variables, one for each t).6 On the other hand, the main drawback of this estimator stems from its identification assumption (the so-called ‘Independence from Irrelevant Alternatives’ IIA), requiring that estimates are unbiased only if the individual choice of each exit state is independent from the choice of other states not considered in the model. The main estimates are reported in Table 6.3. As mentioned above, the model considered four different exit states (other than staying in temporary help employment): permanent employment, fixed-term employment different from temporary help work, unemployment, and education.7 As a robustness check, we estimated three different specifications which differed in the number and types of controls used: only individual characteristics (model a), only firm
5
It has been shown that a discrete time duration model can be estimated using either an extended multinomial logit or a complementary log–log model (Allison, 1982). Similar results were actually obtained using a complementary log–log (cloglog) duration model. The results are available from the authors upon request. 6 Note that a log specification of k(t) in a discrete time (logistic) model is the discrete time analogue to the continuous time Weibull model (Lancaster 1990). On the other end, the use of a full non-parametric baseline hazard function may impose less structure/constraints on the data. 7 In the table we report the Relative Risk Ratios (RRR). A RRR greater than one means a positive effect of that factor on the probability of transition to a certain state (with respect to the base category, that is, temporary help employment). By contrast, a RRR less than one means a negative effect of that variable on the same relative probability. We also performed a Hausman test, whose result allowed us to accept the IIA assumption.
b
c
a
b
Hazard rate (logt) 0.976 1.070 1.083 1.048 1.103* Individual characteristic Woman 0.792* 0.771* 1.182 30 and over 1.046 1.244 0.834 Italian 0.910 1.145 1.320 Married 0.909 0.760 0.629** Children 0.698* 0.696* 0.815 Living with parents 0.944 0.839 0.656** Centre-North 1.898** 1.664** 1.376* Education (ref group: compulsory school) College degree 0.910 l.083 0.906 High school diploma 0.945 1.036 1.050 Vocational training certificate 1.102 1.220 0.921 At least one high educated parent 1.076 0.993 1.002 Employment condition before temporary Permanent contract 1.398** 1.210 0.811 Part time 0.906 1.044 0.909 Reasons for accepting temporary help contract (ref: other reasons) First job opportunity 0.818 0.780 1.302* Hard to find permanent contract 0.658** 0.706* 1.050 Need for work flexibility 0.729 0.865 1.057
a
a
b
c
0.657** 1.342** 0.890 0.786* 0.841
0.745** 0.701** 1.005 1.318** 1.253 1.025 1.030
0.9394 0.8236 0.8834
0.683* 0.591** 0.558** 0.823
0.6334** 0.5939** 0.5723** 0.8852
0.984 1.107 0.899 0.931
1.411** 1.001 0.975 1.451** 1.043 1.444** 0.374**
1.131**
1.4495** 0.9375 0.9007 1.4017** 1.1186 1.3926** 0.4218**
1.101*
1.133 0.911 1.540* 0.604** 0.821 0.667** 1.201
1.120** 1.076
c
0.3882** 0.200** 2.890**
0.342* 1.781*
0.2764* 1.2587 0.8414 1.8919**
0.9149 0.4771* 1.5056 0.3472 0.8354 1.976* 0.7083
0.635**
a
c
(continued)
0.3779** 0.2162** 2.490**
0.305** 1.782*
0.280* 1.3051 0.8586 1.711*
0.7614 0.465* 1.4483 0.3089 1.073 2.2579** 0.5123**
0.616** 0.712**
b
Table 6.3 Probability of transition from temporary help employment to different labour market states, competing risks discrete time duration model Labour market status at the (A) Employed on (B) Employed on fixed(C) Unemployed (D) Student moment of the survey: permanent contract term contract
6 Temporary Help Workers in Italy 117
a
b
c
a
1.198 0.927 1.118 0.940 0.939 1.029
1.179 0.947 1.123 0.911 0.947 0.979
0.737 0.817 0.949 reasons) 0.815 0.828 1.586**
0.7867
a 0.772
c
0.834 0.817 0 858
1.1392 0.9208 0.9632 0.6857
1.220 1.4767**
1.641** 1.6466** 1.700** 1.5039** 0.766 0.7631
0.836 0.794 0.836
1.129 0.894 0.856 0.618**
1.116 1.219
0.672** 0.6964** 0.567** 0.5313** 0.454** 0.3694**
b
(C) Unemployed
0.624
a
0.6276
c
1124 0.908 2.294**
0.5123 1.3519 1.3559 0.9101
1.095 1.278
2.638** 2.0398** 2.119** 2.029* 0.609 0.5854
1.042 0.777 1.949*
0.403** 0.715 1.058 0.450
1.260 1.391
0.664 0.7009 0.350** 0.4033* 0.332** 0.2944**
b
(D) Student
Notes: The number in the table are Relative Risk Ratios (RRR). Reference state: temporary help employment. N observations: 13907; Pseudo R2 full model (column c): 0.0575 * Statistically significant at 10%, ** Statistically significant at 5%
0.970 1.342** 1.199 0.661*
0.924 1.286* 1.165 0.640*
1.448* 1.703** 2.609** 0.299**
1.108
c
1.402** 1.332** 1.173 1.063
b
1.037 0.541**
1.154
0.734** 0.730** 0.521** 0.526** 0.357** 0.358**
0.867
(B) Employed on fixedterm contract
0.662** 0.484** 0.292**
(A) Employed on permanent contract
Work experience while awaiting a 0.929 better job Firm and job characteristics No. temporary help contracts (ref group: one) Two 0.704* Three 0.498** Four or more 0.293** No. using firms (ref group: one) Two 1.035 Three or more 0.553* Economic activity (ref group: other services) Metalworking 1.459 Other manufacturing 1.710* Trade 2.627** Public administration 0.344** Firm size (ref group: small) Medium 0.733* Large 0.801 Quite different work experiences 0.959 Reasons for using temporary help contract (ref: other Production peaks 0.810 Substituting absent employees 0.745 Probation period 1.586**
Table 6.3 (continued) Labour market status at the moment of the survey:
118 F. Origo and M. S. Lodovici
6 Temporary Help Workers in Italy
119
and job characteristics (model b), the full set of observed characteristics (model c).8 We present the results of the model with the logarithmic specification of the hazard rate (first row of the table).9 The econometric estimates allow us to give better specification to the preliminary results of the descriptive analysis: firm and job characteristics are more relevant than individual ones in explaining transition probabilities, but this holds only for some labour market states. More specifically, the former are important in explaining the probability of moving to other employment contracts, either permanent or fixed-term, while individual characteristics are more significant in explaining the probability of moving (back) to unemployment or education. The duration of temporary help work experience does not significantly influence transitions to permanent employment, although it has a major role in shaping transitions to other labour market states. More specifically, it increases the probability of obtaining other fixed-term contracts or of becoming unemployed, while it reduces the probability of becoming a student. These results suggest that state dependence is not a relevant factor in determining the transitions from temporary help work to permanent employment, while positive state dependence (i.e., a probability of exiting temporary help employment which decreases with the length of temporary help work experience) is relevant in the case of education. Also past employment history significantly affects the probability of exiting temporary help work. Workers with permanent contracts before the temporary help work experience are in fact less likely to obtain other temporary contracts or to become unemployed, while there are no significant effects on the transition to stable employment. Furthermore, workers who have had part-time jobs are more likely to become either unemployed or students. Our results suggest that temporary help work forms a sort of ‘bridge’ between part-time employment and unemployment (for women) and education (for the young), which raises doubts concerning the quality of part-time jobs. The probability of transition from temporary help work to a stable job is instead strongly influenced by the number of temporary help contracts, the number and economic activity of employing firms, and firms’ reasons for hiring on this type of contract. The probability of obtaining a permanent contract in fact declines with 8
The vector of individual characteristics includes: gender, age, education, nationality, marital status, family composition (presence of children, a dummy for individuals still living with parents), family background, residence area, employment status before temporary help work and the reasons for accepting this type of contract. Firm and job characteristics include: the number of temporary help contracts, the number of employing firms, economic sector and size of the firm related to the longest contract, the existence of a link (in terms of job contents) between different contracts, firms’ reasons for using temporary help work. Summary statistics by gender for the whole set of controls are reported in the Appendix. Similar results were obtained using different functional forms (such as a parabolic function of age as a continuous variable) or a larger set of explanatory variables (for example, including training on and off the job). The results are available from the authors upon request. 9 Estimates with alternative specifications of the hazard rate were consistent with those presented and are available upon request.
120
F. Origo and M. S. Lodovici
the number of temporary help work experiences and the number of employing firms, while it increases when the firm hires on temporary help contracts in order to screen worker productivity. It is worth pointing out that the number of temporary help work experiences reduces the probability of moving to any labour market state, which suggests that this variable may be a good indicator to use in profiling those individuals that are more likely to become ‘permanent’ temporary help workers. The probability of finding a stable job is also higher for workers employed in manufacturing and trade than in services and public administrations. Employment in the latter also reduces the probability of moving to other types of fixed-term contracts, suggesting that the public sector is unlikely to create new jobs characterized by contracts different from temporary help ones. Temporary help contracts, in fact, have been the most common means used in the public sector to overcome the turnover limits imposed by law in order to reduce public spending. The hiring freeze may also explain the very low probability of getting a stable job in this sector. Furthermore, the larger share of women in the public sector (other than in small and medium firms; see Table A1 in the Appendix) may partly explain their difficulties in finding stable jobs after temporary help work experiences.10 Among personal characteristics, only gender has a significant influence on the probability of transition to permanent employment, since it is lower for women than for men. Of course, local labour market conditions matter as well: the probability of transition is much higher for workers in the Centre-North of Italy, where the labour market is tighter and unemployment rates lower than in the South. Analysis of personal reasons for using this type of contract shows that the probability of obtaining a stable job is relatively lower mainly for those who have accepted a temporary help contract because they have been unable to find a stable job. Individual characteristics, specifically gender and education, are more important in explaining transitions to unemployment: after a temporary help work experience, women and the low-educated are in fact more likely to become unemployed than, respectively, men and the high-educated. Furthermore, the probability of becoming unemployed is higher in the South, which confirms the role of local labour market conditions in shaping the transitions of temporary help workers. Among firms’ characteristics, the probability of becoming unemployed is influenced both by the number of temporary help contracts and by the consistency (in terms of job contents) among them: workers with more than one contract but characterized by similar tasks are less likely to become unemployed than are workers with the same number of contracts (and the same number of employing firms) but with different job contents. 10
Estimates of a model with interactions between gender and economic sector (available from the authors upon request) show that employment in the public sector significantly increases the probability that females will move either to other forms of temporary contracts or to unemployment.
6 Temporary Help Workers in Italy
121
Also the reasons that induce firms to use temporary help labour may influence the transition to unemployment, since the latter increases for temporary help workers hired to cope with production peaks or to replace absent employees. Finally, individual and family characteristics are important determinants of the probability of moving (back) to education. As expected, the oldest and the high-educated individuals are less likely to return to education, while this type of transition is more common for those still living with their parents, or with at least one highly-educated parent. Transitions to education are also more likely for individuals who accept temporary help jobs because of personal flexibility needs, or with inconsistent (in terms of job contents) temporary help work experiences. This suggests that students often prefer temporary help jobs so that they can gain financial support while studying, rather than doing so for longrun reasons (such as acquiring skills not provided by formal education, obtaining information on the actual jobs available, or establishing contacts with potential future employers). Estimates for local labour market conditions show that, as in the case of unemployment, the probability of moving (back) to education is higher in the South than in the Centre-North. Overall, our results clearly suggest that local labour demand is a crucial factor in favouring transitions from temporary help work to stable employment, thereby preventing the generation of continuous transitions between unemployment and temporary help work.
6.6 Concluding Remarks This paper was aimed at empirically studying the probability of exiting temporary help work in Italy by modelling both the transition to different labour market states (permanent employment, other types of fixed-term contracts, unemployment, and education) and the effect of state dependence (i.e., the effect of the overall duration of temporary help work experience). Confirming previous empirical findings, our empirical analysis suggests in general that, in Italy, temporary help work is not per se a ‘trap’. However, its effect as a ‘stepping-stone’ towards stable employment depends strictly not only on workers’ characteristics but also on those of firms and jobs. In particular, too many short temporary help work experiences and too many employing firms have a negative impact on transitions to stable jobs. We have also found evidence that these contracts are used as screening devices: some firms hire workers on a temporary help contracts in order to assess their productivity before hiring them permanently. Local labour market conditions are also crucial, since transition to permanent employment is more likely in low unemployment regions, while transitions (back) to either unemployment or education are more likely in high unemployment ones.
122
F. Origo and M. S. Lodovici
The analysis of the influence of individual characteristics on the probability of transitions to different labour market states has highlighted that temporary help workers are highly heterogeneous (also in relation to their labour market conditions before and after the interim work experience), and this calls for diversified policy interventions. The probability of moving from temporary help work to a stable job is higher for workers with previous stable work experiences, mainly in manual jobs in manufacturing. The temporary help work experience for this group of workers usually consists of few contracts with medium-long durations (around 6–8 months). For these individuals, temporary help work is an efficient way to gain access to new stable jobs, and temporary help agencies act efficiently as intermediaries in the job matching process. The young seem to be those using more temporary help work for personal flexibility needs, mainly to gain some financial support while studying. It is not clear whether they also have long-run reasons for doing so (such as increasing their future likelihood of finding stable employment by already acquiring work experience and skills not provided by formal education), because they usually accept short and fragmented temporary help contracts, often in different employing firms and characterized by different job contents. The probability of moving from temporary jobs to unemployment is higher for women and for individuals with low levels of education. Women are also more likely to remain in temporary help employment than to move to a stable job. They usually have many short and fragmented temporary help work experiences in different firms, but the overall duration of temporary help work experience is relatively long, especially if they are ‘involuntary’ temporary help workers—i.e., they have chosen this type of contract because they have been unable to find a stable job.11 This is the category of workers that should be targeted with preventive measures in order to reduce the risk that temporary work becomes a ‘trap’ (or a ‘bridge’ between two unemployment spells), a risk that is relatively higher for low-educated women.
Appendix See Table A1
11
Overall length is more than 8 months for women unable to find permanent jobs, 7 and 5.5 months for, respectively, women and men who have chosen a temporary help contract for personal flexibility reasons.
Woman 30 and over Italian Married Children Living with parents Centre-North Education (ref group: compulsory school) College degree High school diploma Vocational training certificate At least one high educated parent Employed on permanent contract before temporary help work Part-time employment before temporary help work Reasons for accepting temporary help contract (ref: other reasons) First job opportunity Hard to find permanent contract Need of work flexibility Work experience while awaiting a better job No. temporary help contracts (ref group: one) Two Three
Table A1 Summary Statistics
0.370 0.499 0.292 0.472 0.380 0.359
0.452 0.476 0.261 0.314 0.398 0.365
0.163 0.468 0.094 0.335 0.175 0.152
0.286 0.347 0.074 0.111 0.198 0.158
0.217 0.152
0.292 0.334 0.074 0.111
0.091
0.119 0.480 0.096 0.349 0.169
– 0.417 0.943 0.218 0.173 0.569 0.791
Mean
Std. dev. 0.499 0.499 0.240 0.463 0.426 0.499 0.353
Mean 0.514 0.464 0.939 0.311 0.237 0.472 0.854
Men
All
0.412 0.359
0.455 0.472 0.262 0.315
0.288
0.324 0.500 0.295 0.477 0.375
0.493 0.232 0.413 0.378 0.496 0.407
Std. dev.
0.179 0.164
0.280 0.360 0.073 0.111
0.209
0.205 0.456 0.092 0.322 0.181
– 0.509 0.934 0.398 0.299 0.381 0.914
Mean
Women
(continued)
0.384 0.371
0.449 0.480 0.261 0.314
0.407
0.404 0.498 0.289 0.467 0.385
0.500 0.248 0.490 0.458 0.486 0.281
Std. dev.
6 Temporary Help Workers in Italy 123
Four or more No. using firms (ref group: one) Two Three or more Economic activity (ref group: other services) Metalworking Other manufacturing Trade Public administration Firm size (ref group: small) Medium Large Quite different work experiences Reasons for using temporary help contract (ref: other reasons) Production peaks Replacement of absent employees Probation period
Table A1 (continued)
0.364 0.303 0.444 0.451 0.324 0.257 0.466 0.498 0.351
0.464 0.376 0.335
0.157 0.102 0.270 0.284 0.119 0.071 0.318 0.450 0.144
0.312 0.170 0.128
0.306 0.130 0.161
0.299 0.469 0.156
0.338 0.289 0.090 0.057
0.161 0.105
0.292
Mean
Std. dev. 0.448
Mean 0.279
Men
All Std. dev.
0.461 0.336 0.368
0.458 0.499 0.363
0.473 0.453 0.287 0.232
0.368 0.307
0.455
0.318 0.208 0.097
0.336 0.432 0.132
0.205 0.280 0.146 0.084
0.153 0.100
0.266
Mean
Women Std. dev.
0.466 0.406 0.297
0.473 0.496 0.339
0.404 0.449 0.353 0.278
0.360 0.300
0.442
124 F. Origo and M. S. Lodovici
6 Temporary Help Workers in Italy
125
References Allison, P. (1982). Discrete Time Methods for the Analysis of Event Histories. In S. Leinhardt (Ed.), Sociological Methodology (pp. 61–97). San Francisco, CA: Jossey-Bass. Booth, A., Francesconi, M., & Frank, J. (2002). Temporary jobs: stepping stones or dead ends? Economic Journal, 112, 189–213. D’Addio, A., & Rosholm, M. (2004). Exits From Temporary Jobs in Europe: A Competing Risks Analysis. Study for the European Commission, DG Employment and Social Affairs. de Graaf-Zijl, M., van der Berg, G., & Heyma, A. (2011). Stepping stones for the unemployed: The effect of temporary jobs on the duration until (regular) work. Journal of Population Economics, 24(1), 107–139. EIRO. (2002). Temporary Agency Work in the European Union. Dublin: EIRO. Gagliarducci, S. (2005). The dynamics of repeated temporary jobs. Labour Economics, 12(4), 429–448. Garcia-Perez, J. I., & Munoz-Bullon, F. (2005). Are temporary help agencies changing mobility patterns in the Spanish labour market? Spanish Economic Review, 7(1), 43–65. Guell, M., & Petrongolo, B. (2007). How binding are legal limits? Transitions from temporary to permanent work in Spain. Labour Economics, 14(2), 153–183. Houseman, S., Kalleberg, A., & Erickcek, G. (2003). The role of temporary agency employment in tight labor markets. Industrial and Labor Relations Review, 57(1), 105–127. Ichino, A., Mealli, F., & Nannicini, T. (2003). Il lavoro interinale in Italia. Rome: Ministero del Lavoro e delle Politiche Sociali (Ministry of Labour). Incagli, L., & Rustichelli, E. (2002). Il tanto e il poco del lavoro flessibile. Milan: Angeli. IRS, REF, Studio APS, Enaip Lombardia, & IAL Piemonte. (2005). Lavoro interinale e formazione. Caratteristiche del comparto e sperimentazione di nuovi percorsi formativi. Rome: Ministero del Lavoro e delle Politiche Sociali (Ministry of Labour). Lancaster, T. (1990). The Econometric Analysis of Transition Data. Cambridge: Cambridge University Press. Malo, M. A., & Munoz-Bullon, F. (2002). Temporary help agencies and the labour market biography: A sequence oriented approach. EEE, 132, FEDEA Ministry of Labour. (2007). Il lavoro interinale: un quadro aggiornato a novembre 2006. In Ministry of Labour (Ed.), Monitoraggio delle politiche occupazionali e del lavoro. Rome: Ministry of Labour. Montanino, A., & Sestito, P. (2003). Le molte funzioni del lavoro interinale in Italia: da strumento di flessibilità a contratto di prova. Rivista di Politica Economica, 93(3–4), 115–148. OECD. (2002). Taking the Measure of Temporary Employment. In OECD (Ed.), Employment Outlook. Paris: OECD. Russo, G., Gorter, C., & Moolenaar, D. (1997). Temporary jobs and temporary workers: The Dutch experience. Tinbergen Institute discussion papers, 97-133/3.
Chapter 7
The Dynamics of Unemployment, Temporary and Permanent Employment in Italy Matteo Picchio
7.1 Introduction In recent years, temporary employment has risen in almost all European countries, and atypical contracts have been used to obtain labour market flexibility and to respond to the high level of European unemployment. The standard arrangement of Italian employees has traditionally been full-time, permanent, and characterized by a high degree of employment protection. However, recent labour market reforms1 have changed the institutional set-up, and atypical employment forms, among them temporary contracts, have been growing in importance. Researchers have extensively debated the advantages and disadvantages of temporary contracts. Temporary jobs may increase labour market flexibility, provide firms with means to cope with demand uncertainty, and be ‘stepping-stones’ to longer employment relationships (Booth et al. 2002; Hagen 2003; de Graaf-Zijl and Berkhout 2007; de Graaf-Zijl et al. 2011; Ichino et al. 2005). On the other hand, it is has been pointed out that temporary workers face higher turnover and a greater probability of unemployment (Dolado et al. 2002; Farber 1999) and suffer wage penalties (Blanchard and Landier 2002; Booth et al. 2002; Brown and Session 2003; Jimeno and Toharia 1993; Hagen 2002; Picchio 2006). The purpose of this paper is to provide empirical evidence for the branch of the literature that assesses whether temporary jobs may be springboards to regular employment or to dead-end positions. Indeed, temporary job experiences may provide
As part of my PhD thesis, this study has been financed by the Marche Polytechnic University doctoral fellowship. M. Picchio (&) Department of Economics, CentER, ReflecT, Tilburg University, P. O. Box 90153, 5000 LE Tilburg, The Netherlands e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_7, Ó Springer-Verlag Berlin Heidelberg 2012
127
128
M. Picchio
the unemployed with opportunities to acquire skills and knowledge, and to avoid the deterioration of human capital and the negative signals associated with unemployment. But, if firms exploit this tool of labour market flexibility systematically to cope with demand uncertainty without investing in temporary employees, temporary work may become a trap. Temporary workers may not gain in terms of human capital, and they may return to the unemployment pool without any advantages with respect to the unemployed. It may be crucial from a welfare evaluation viewpoint to assess whether, in Italy, temporary contracts have provided stepping-stones to permanent positions. Therefore, the aim of this paper is to determine whether having a temporary job today, rather than being unemployed, increases or decreases the probability of having a permanent job within a few years, once individual observables and unobservable characteristics have been controlled for. The econometric analysis reported by this paper was performed using the 2000, 2002, and 2004 waves of the Survey of Italian Households’ Income and Wealth (SHIW), a representative survey conducted by the Bank of Italy every 2 years since 1989. Raw probabilities of permanent employment conditional on the past working status highlight that a temporary position, rather than unemployment, increases the probability of having a regular job 2 years’ later by about 28.4% points. It seems that temporary contracts provide stepping-stones to permanent jobs. Yet such raw evidence may be spurious. There may be certain individual characteristics that determine both the current working status and the future transition into a permanent job. For example, more able individuals may be more willing to accept a temporary job as an alternative to unemployment, and, at the same time, they may be those with a higher hazard rate towards a permanent job. Hence, the stepping-stone effect from raw data may entirely reflect the greater ability of the stock of temporary workers. In order to remove the spurious component from the stepping-stone effect, I estimated dynamic unobserved effects probit models for the probability of permanent employment; the model was dynamic because I introduced the lagged working status among the covariates. Unobserved heterogeneity was linearly approximated following Chamberlain’s (1980) approach, whilst the problem of initial conditions was addressed by using both Heckman’s (1981) method and Wooldridge’s (2005) conditional maximum likelihood estimator. The main finding is that, ceteris paribus, having a temporary contract today, rather than being unemployed, increases the probability of having a permanent position 2 years’ later by about 13.5–16% points. This finding is interpreted as a stepping-stone effect, since it derives from the estimate of a counter-factual outcome probability. Finally, I assessed the robustness of the stepping-stone result by relaxing the parametric assumptions on individual heterogeneity. I estimated: (i) dynamic linear probability models able to approximate the average partial effects, by fully controlling for unobserved heterogeneity and avoiding the problem of initial conditions; (ii) dynamic non-linear probability models where unobserved heterogeneity was randomly drawn from a discrete distribution with a finite number of support points. This paper is organized as follows. Section. 7.2 describes the data and reports descriptive statistics on the sample used in the econometric analysis. Section. 7.3
7 The Dynamics of Unemployment
129
presents the econometric specification of dynamic nonlinear models for the probability of permanent employment and sets out the estimation results. Section. 7.4 checks the robustness of the results relaxing the parametric assumptions on individual heterogeneity. Section. 7.5 concludes.
7.2 Data and Sample The empirical investigation was performed using the 2000, 2002, and 2004 waves of the Survey of Italian Households’ Income and Wealth (SHIW).1 The SHIW is a nationally representative survey conducted by the Bank of Italy every 2 years since 1989. Because the question on the type of contract was introduced in 2000, the longitudinal dimension of the sample was restricted to the period 2000–2004. I took all individuals belonging to the panel dimension of the SHIW from 2000 until 2004, and who were either unemployed or employees. I excluded individuals outside the 15–64 age-range in 2000 and observations with missing values for some of the variables used in the specification of the econometric models. The result was a balanced panel of 1,677 individuals, observed over three time periods. Since I estimated dynamic nonlinear models of order one, i lost the first time period, which was exploited only for the initial values. The dependent variable was a dummy indicator, y1it ; equal to 1 if individual i was a permanent worker at time t and 0 otherwise. The dynamic would be captured by the lag of this indicator and by the lag of the unemployment indicatory21it ; which is equal to 1 if individual i is unemployed at time t. Over 2000–2004, the average composition of the labour force was as follows: 9.8% unemployed, 6.3% temporary workers, 83.9 permanent workers.2 Table 7.1 displays the relative frequencies of permanent jobs, conditional on past labour market state, by some individual characteristics. Inspection of the first row of the table shows that there is a raw positive effect of temporary jobs. A temporary worker is more than twice as likely to find a stable job position 2 years later as someone who is currently unemployed. Having a temporary job today, rather than being unemployed, increases the raw probability of having a permanent job in the future by about 28.4% points. This shows that, in the raw data, a substantial stepping-stone effect is exerted by temporary contracts. The relative frequencies of permanent employment are lower for men, young people, and the less educated. People living in the South of Italy or in regions with higher unemployment rates are less likely to have permanent jobs. 1 The SHIW and further details on the dataset are available on the web-server of the Bank of Italy: http://www.bancaditalia.it/statistiche/indcamp/bilfait. 2 2 An individual is unemployed if s/he declares him/herself to be either a first-job seeker or unemployed in section B (Employment and Incomes) question ‘Apqual’ of the SHIW questionnaire. The information about the employees’ contract type comes from the answer to question ‘Contratt’ of annex B1 (information about the employees’ job).
130
M. Picchio
Table 7.1 Unconditional and conditional relative frequencies of permanent employment by individual characteristics Unconditional Permanent Unemployed Temporary Stepping at t-1 at t-1 at t-1 stone effect All Male Female Age C40 Age \40 Potential experience C20 Potential experience \20 High school or above Up to vocational school Centre-North South Regional unemployment rate C median Regional unemployment rate \ median Married Not-married Head of household Other household position Spouse not working Spouse working/no spouse Children \6 years No children \6 years Observations
0.850 0.839 0.866 0.923 0.737 0.923
0.951 0.947 0.958 0.966 0.922 0.965
0.232 0.224 0.250 0.260 0.225 0.250
0.516 0.515 0.517 0.588 0.467 0.554
0.284 0.291 0.267 0.328 0.242 0.304
0.748
0.926
0.228
0.490
0.262
0.901 0.797 0.920 0.702 0.782
0.968 0.932 0.968 0.902 0.933
0.310 0.188 0.296 0.203 0.213
0.573 0.481 0.574 0.462 0.481
0.263 0.293 0.278 0.259 0.268
0.926
0.969
0.328
0.574
0.246
0.718 0.921 0.924 0.791 0.870 0.844 0.872 0.848 3,354
0.918 0.965 0.968 0.935 0.940 0.954 0.932 0.953 2,783
0.207 0.321 0.297 0.218 0.321 0.216 0.294 0.223 349
0.455 0.573 0.559 0.500 0.534 0.510 0.632 0.506 252
0.248 0.252 0.262 0.282 0.213 0.294 0.338 0.283 –
Table 7.2 reports the summary statistics of the explanatory variables used in the econometric analysis. Table 7.3 displays the observed transitions between labour market positions: as expected, most of the individuals show a strong persistence in permanent jobs. Identification of the effects of interest here comes from observations out of the diagonal of this transition matrix.
7.3 Econometric Modelling and Estimation Results This section discusses and presents the econometric models for permanent employment, whose estimation results are set out in Table 7.4. The starting point (Sect. 7.3.1) is a bivariate dynamic unobserved effects probit model for permanent
7 The Dynamics of Unemployment
131
Table 7.2 Sample characteristics Unemployed Permanent employed Temporary employed Age Potential experience Female Education None or elementary Middle school Vocational school High school University degree or above Geographical area North-East North-West Centre South Islands Regional unemployment rate Permanent incomea Transitory incomeb Married Head of household Spouse not working Children \6 years Observations
Mean
Standard deviation
Minimum
Maximum
0.098 0.839 0.062 40.730 20.249 0.393
0.298 0.368 0.242 10.422 11.464 0.488
0 0 0 15 0 0
1 1 1 67 58 1
0.082 0.324 0.083 0.392 0.118
0.275 0.468 0.277 0.488 0.323
0 0 0 0 0
1 1 1 1 1
0.221 0.228 0.230 0.199 0.123 0.130 1119.5 0.000 0.650 0.440 0.217 0.104
0.415 0.419 0.421 0.399 0.328 0.106 5691.2 8145.2 0.477 0.496 0.412 0.306
0 0 0 0 0 0 -23251.8 -75513.2 0 0 0 0
1 1 1 1 1 0.384 45622.4 59038.3 1 1 1 1 5,031
Notes: Pooled data for SHIW waves (2000–2004) Permanent income is within-individual means across the period 2000–2004 of the individual non-labour income b Transitory income is defined as the deviation from the individual’s permanent income a
Table 7.3 Observed transitions between labour market positions Origin Destination Unemployment Temporary job Permanent job Total
Total
Unemployment
Temporary job
Permanent job
226 6.7% 25 0.7% 54 1.6% 305 9.1%
42 1.3% 75 2.2% 82 2.4% 199 5.9%
81 2.4% 123 3.7% 2646 78.9% 2850 85.0%
349 10.4% 223 6.6% 2782 82.9% 3354 100.0%
Coeff.
0.968 Permanent jobt1 Unemployedt1 -0.503 Experience 0.035 -0.064 Experience2 /100 Female -0.080 Education reference: none or elementary Middle school 0.338 Vocational school 0.356 High school 0.768 University degree or + 0.648 Area—Reference: North–East North–West 0.085 Centre 0.003 South -0.337 Islands -0.318 Head of household 0.391 Unemployment rate -0.818 Permanent income -0.036 Transitory income -0.023 Spouse not working 0.277 Married 1.149 Children \6 -0.207 0.336 0.356 0.767 0.647 0.084 0.002 -0.337 -0.320 0.392 -0.817 -0.036 -0.023 0.272 1.150 -0.208
0.120*** 0.157** 0.133*** 0.171*** 0.129 0.115 0.190* 0.175* 0.098*** 0.726 0.008*** 0.012* 0.314 0.564** 0.285
Coeff. 0.967 -0.504 0.035 -0.064 -0.078
S.E.
Wooldridge’s model
0.143*** 0.173*** 0.012*** 0.026** 0.079
a
Table 7.4 Dynamic probit models for permanent employment Variable Bivariate model S.E.
0.129 0.115 0.190* 0.174* 0.098*** 0.726 0.008*** 0.012* 0.312 0.564** 0.286
0.120*** 0.157** 0.133*** 0.171***
0.143*** 0.173*** 0.012*** 0.026** 0.078
a
0.077 -0.019 -0.463 -0.497 0.432 -1.247 -0.046 -0.026 0.302 1.271 -0.206
0.431 0.486 0.999 0.851
0.780 -0.547 0.070 -0.124 -0.130
Coeff.
Heckman’s model S.E.
(continued)
0.148 0.139 0.230** 0.218** 0.116*** 0.850 0.011*** 0.012** 0.241 0.575** 0.294
0.153*** 0.207** 0.207*** 0.229***
0.203*** 0.137*** 0.021*** 0.041*** 0.098
132 M. Picchio
Coeff.
Bivariate model S.E.a Coeff.
Wooldridge’s model S.E.a
D2004 0.033 0.073 0.034 0.073 Constant -0.580 0.200*** -0.580 0.200*** Random Effect: Initial condition and time-variant variables in all time periods 0.419 0.146*** 0.420 0.146*** Permanent employed0 Unemployed0 0.003 0.173 0.005 0.173 0.015 0.009* 0.015 0.009* Transitory income0 Transitory income1 0.036 0.010*** 0.036 0.010*** Spouse not working0 0.015 0.151 0.017 0.150 -0.252 0.238 -0.251 0.238 Spouse not working1 Spouse not working2 -0.283 0.193 -0.280 0.193 0.122 0.342 0.121 0.343 Married1 Married1 -1.000 0.472** -0.999 0.472** Married2 0.135 0.487 0.134 0.487 0.151 0.184 0.150 0.184 Children \60 Children \61 -0.341 0.259 -0.343 0.260 0.236 0.271 0.240 0.270 Children \62 H0 : ½d11 ; d12 ¼ 0 v211 ¼ 27:8; p-val ¼ 0:004 v211 ¼ 27:1; p-val ¼ 0:004 Observations 3,354 3,354 0.436 0.453 Pseudo R2 955.7 957.3 Wald v2 Log-pseudolikelihood -990.4 -776.5
Table 7.4 (continued) Variable
0.080 0.234
S.E.
(continued)
0.012 0.010 0.040 0.014*** -0.006 0.151 -0.266 0.202 -0.313 0.198 0.176 0.466 -1.058 0.619* 0.094 0.413 0.159 0.228 -0.405 0.312 0.272 0.280 v211 ¼ 18:4; p-val ¼ 0:072 5,031 0.245 418.6 -1284.8
0.028 -0.369
Coeff.
Heckman’s model
7 The Dynamics of Unemployment 133
v21
S.E.a
0.161 0.348 0.186
¼ 0:05; p-val ¼ 0:830
Coeff.
Bivariate model Coeff.
Wooldridge’s model
0.160 0.340 0.181
S.E.a Coeff.
Heckman’s model
Notes: Number of individuals: N = 1,677 *Significant at 10%; ** significant at 5%; *** significant at 1% a Standard errors robust to serial correlation b APETU is the marginal effect of a past Temporary job (the hyperscript T) instead of Unemployment (the hyperscript U) c APEPU is the marginal effect of a past Permanent job (the hyperscript P) instead of unemployment d APEPT is the marginal effect of a past permanent position instead of a past temporary contract
LR test of indep. equat. Average partial effects APETU b APEPU c APEPT d
Table 7.4 (continued) Variable
0.137 0.267 0.130
S.E.
134 M. Picchio
7 The Dynamics of Unemployment
135
Table 7.5 Correct predicted status and sequences Bivariate model Wooldridge’s model Heckman’s model (%) (%) (%) Correct predicted permanent employed status Permanent employed 96.42 96.31 97.40 Temporary employed or unemployed 59.72 60.20 50.20 Overall 90.91 90.82 90.31 Correct predicted permanent-temporary employment or unemployment sequences Always PC 97.02 96.87 97.91 From TC or U to PC 10.10 7.07 9.09 From PC to TC or U 0.00 0.00 0.00 Always TC or U 69.05 69.05 56.55 Overall 85.09 84.79 84.50 Note The acronyms PC, TC, and U respectively refer to permanent contract, temporary contract, and unemployment
employment, which is the most general specification presented and takes into account the endogeneity of the previous labour market state. Then, in Sect. 7.3.2, we move to a univariate framework, whilst Sect. 7.3.3 and Table 7.5 deal with the goodness of fit of the estimated dynamic nonlinear models.
7.3.1 Bivariate Unobserved Effects Probit Model Let y1it and y2it be the scalar indicator variables denoting the occurrence at time t of a permanent job and unemployment, respectively. The dynamic probability model for permanent employment was empirically specified using a bivariate unobserved effects probit model: ð7:1Þ y1it ¼ 1 y1it1 q11 þ y2it1 q12 þ x01it b1 þ c1i þ u1it [ 0 y2it ¼ 1½y1it1 q21 þ y2it1 q22 þ x02it b2 þ c2i þ u2it [ 0 if y1it ¼ 0
ð7:2Þ
where 1½ is the indicator function, x1it and x2it are vectors of skill, family, and individual structure variables that may explain working status, c1i and c2i are timeinvariant individual heterogeneities. Finally, ðu1it ; u2it Þ is the idiosyncratic error term, which is assumed to be bivariate standard normal with covariance qu : This model is a modified version of that developed by Alessie et al. (2004) and follows Stewart’s (2007). The model is bivariate so that y2it can be endogenous in Eq. 7.1. The coefficient of primary interest is q12 : it captures the effect of past unemployment, rather than temporary employment (the reference category), on the current probability of having a permanent job. It therefore conveys whether having a temporary job today, rather than being unemployed, reduces or increases the future probability of having a permanent position.
136
M. Picchio
The regressors vector x1it contains time-variant and time-invariant variables. The latter category comprises the constant, education (four dummies), geographical area of residence (four dummies), and gender. The time-variant variables are potential experience, its quadratic form, household position, marital status, spouse’s working status, presence of pre-school children, time intercept, and non-labour income. Following Hyslop (1999), non-labour income is decomposed into a permanent component hpi ; which is estimated by the within-individual mean across the sample period 2000–2004, and a transitory component hm it estimated by the deviation from the permanent component. The idea is that permanent income captures the total effect of income expectations on the likelihood of having a permanent job, whereas the transitory component has a direct income effect on this likelihood. The regressors vector x2it contains x1it and age and squared age. Correlation between unobserved heterogeneity cji with j 2 f1; 2g and observed characteristics is allowed by adopting a correlated random-effects specification (Chamberlain 1980): cji ¼
T X s¼0
dj1s wjis þ
T 1 X
dj2s hm ijs þ aji
ð7:3Þ
s¼0
where wjis is a three-dimensional vector comprising marital status, spouse’s working status, and presence of pre-school children and aji iidNð0; r2aj Þ and independent of xjit : In order to distinguish between spurious and true stepping-stone effects, I also had to make assumptions about the relationship between the initial observations of the dependent variables and individual heterogeneity. At this stage, I applied Wooldridge’s (2005) conditional maximum likelihood approach to the initial condition problem. I therefore modelled the density conditional on initial values, which entered the linear approximation of unobserved heterogeneity. The likelihood function to be maximized was given by L¼
N Y T Y
y1it Uðy1it Þ þ ð1 y1it ÞU2 ½y1it ; ð2y2it 1Þy2it ; ð2y2it 1Þqu ;
i¼1 t¼1
ð7:4Þ where U2 is the cumulative bivariate normal distribution function and yjit ¼ y1it1 qj1 þ y2it1 qj2 þ x0jit bj þ cji for j = 1, 2 and t ¼ 1; . . .; T: The estimation results of the bivariate model are reported in the first three columns of Table 7.4. The coefficient of the lagged permanent position is positive and highly significant. Since the lagged temporary working status is the reference group, having a permanent job today, rather than a temporary contract, significantly increases the probability of having a permanent job in the future. This result is as expected, whereas what I wanted to understand was whether a temporary position as an alternative to unemployment is able to increase the chances of obtaining a permanent job in the future. This is conveyed by the
7 The Dynamics of Unemployment
137
coefficient of the lagged unemployment status, which is negative and highly significant: having been unemployed at time t 1; rather than temporary employed, decreases the likelihood of having a permanent position at time t. In other words, an individual who accepts a temporary job today, rather than unemployment, has a significantly higher probability of being a permanent worker in 2 years’ time: this is the stepping-stone effect, since, ceteris paribus, a temporary job provides an opportunity to enter regular employment. On looking at the estimated coefficients of the other explanatory variables, one notes that the greater the worker’s potential experience, the higher his/her probability of having a permanent job. This finding is consistent with human capital theory and job-search explanations. Women and lower-educated individuals are less likely to have permanent jobs. The higher the permanent and transitory incomes, the lower the probability of having a permanent position. An explanation for this may be provided by job-search theory: when non-labour income increases or is constantly high, the individual has less incentives to look for and/or accept a stable job. Finally, individuals who are heads of household or married are more likely to have permanent jobs. It seems that the greater the employee’s household responsibility, the higher the probability that she/he will have a stable position. Note that a Wald test for significance of the coefficients of the linear approximation of the unobserved heterogeneity rejected the null hypothesis. Performing the analysis without introducing time-variant variables in all the time periods would have generated biased results due to their correlation with unobserved heterogeneity. Furthermore, the coefficient of the initial employment status was highly significant, and this finding definitively rejects a simpler bivariate probit that does not account for unobserved heterogeneity and initial conditions problem. Instead, the log-likelihood ratio (LR) test for independent equations did not reject the null hypothesis, meaning that this was a special case of the bivariate model: since the error terms are not correlated, Eq. 7.1 could be estimated in a univariate framework.
7.3.2 Univariate Models for the Probability of Permanent Employment Given the result of the LR test of independent equations, I replicated the analysis in a univariate framework by solving the initial conditions problem following Wooldridge’s (2005) approach and Heckman’s (1981) methodology. As said, Wooldridge (2005) suggested that the initial conditions problem can be dealt with by modelling the density of ðy1i1 ; . . .; y1iT Þconditional on initial conditions and explanatory variables. Under the linear approximation assumption of individual heterogeneity, we therefore have
138
M. Picchio
c1i ¼ w þ y1i0 e10 þ y2i0 e20 þ
T X
d11s w1is þ
s¼0
a1i Nð0; r2a1 Þ;
T1 X
d12s hm 1is þ a1i ;
s¼0
ð7:5Þ
a1i ?ðy1i0 ; y2i0 ; x1i Þ:
Model (1) can thus be rewritten as y1it ¼ 1 y1it1 q11 þ y2it1 q12 þ x01it b1 þ w þ y1i0 e10 þ y2i0 e20 þ
T X
d11s w1is þ
s¼0
T 1 X
d12s hm þ a þ u [ 0 ; 1i 1it 1is
ð7:6Þ
s¼0
and estimated using the standard random effects probit program integrating out a1i : In order to relax the implicit assumption of zero serial correlation of the score, I used a simple pooled probit estimator with standard errors robust to arbitrary serial correlation. The price that I had to pay was a loss in terms of efficiency, in that I obtained a scaled version of my parameters where the scaling factor was given by ð1 þ r2a1 Þ1=2 :3 The approach proposed by Heckman (1981) instead consists of specifying a latent variable model for the initial realization of the dependent variable: y1i0 ¼ z0i0 c þ hc1i þ ui0 :
ð7:7Þ
This equation is a linearized approximation to the reduced form equation for the initial value of the latent variable, where zi0 is a vector of exogenous variables (including x1i0 and a set of parental dummies)4 and ui0 is independent on c1i : This reduced form equation is jointly estimated with the dynamic probability r2c 1 2 c1 þru Þ
model (1). Let k ðr2
be the autocorrelation of the composite error term
between two time periods. Hence, the likelihood to be maximized is given by 9 8 0 U zi0 c þ hrc1 c ð2y1i0 1Þ > > Z N = < Y T Y L¼ ; > U y1it1 q11 þ y2it1 q12 þ x01it b þ rc1 c ð2y1it 1Þ > ; i¼1 : c
t¼1
ð7:8Þ where c ¼ c1 =rc and, following from the implicit normalization r2u ¼ 1; pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rc1 ¼ k=ð1 kÞ: Since c1i is unobservable, in Eq. 7.8 we integrate it out under
3 The scaled estimates of the parameters are required to estimate the average partial effects. See e.g. Wooldridge (2002), pp. 495, for further details. 4 The parental dummies are: four dummies indicating the parents’ maximum attained educational levels and one dummy indicating whether one of the parents is a public employee.
7 The Dynamics of Unemployment
139
the assumption that c1i is normally distributed. The integral over c1i is evaluated using Gaussian-Hermite quadrature with 20 points. The estimation results of these univariate models are reported in the last columns of Table 7.4. Consistently with the outcome of the independent equations LR test, the estimation results of the single equation models are in line with those of the bivariate model. There is still a highly significant stepping-stone effect of temporary jobs into permanent positions. In order to provide a quantitative evaluation of the stepping-stone effect, I estimated, following Stewart (2007), average partial effects (APEs hereafter) at the sample means, x1 :5 The intention was to predict the probabilities of having a permanent job conditional on different past labour market positions, so that I could compare counter-factual outcome probabilities. The estimated APEs are reported at the bottom of Table 7.4. Using the Wooldridge’s estimator, there is a stepping stone effect of about 16.2% points: if an individual accepts today a temporary job as an alternative to unemployment, his/her probability of having a permanent position in 2 years’ time increases by about 16% points. The stepping-stone effect from Heckman’s estimator is lower, being equal to 13.7% points. Given these estimation results, let us summarize the main findings. We have seen that estimating dynamic unobserved effects probit models for the probability of permanent employment depicts temporary jobs as a channel out of unemployment and springboards to stable employment. Indeed, on inspecting the estimated counter-factual probabilities evaluated at the sample means and the corresponding average partial effects, we can state that, given observable and unobservable characteristics, an individual accepting a temporary job today, rather than unemployment, increases his/her probability of having a permanent job in 2 years’ time by about 13.7–16% points. Finally, I sought to determine the direction of the biases if I did not take account of the presence of the unobservable and unobservable individual heterogeneity. We saw in Table 7.1 that the raw stepping-stone effect is about 28% points. If I estimated the APEs of a dynamic probit model with no unobserved heterogeneity, I found a stepping-stone effect of temporary jobs amounting to about 20% points.6 Therefore, one fourth of the initial stepping-stone effect is spurious because of observable heterogeneity x1 : When I moved to unobserved effects models and explicitly took account of the possible presence of the unobservable heterogeneity, I obtained an even lower estimated APE. These findings suggest that: (i) about one half of the raw stepping-stone effect is spurious and due to the presence of observable and unobservable individual characteristics; (ii) the estimated marginal effect of a temporary job is upward biased if one does not consider the presence of the individual heterogeneity. If we assume that more able workers are more likely to transit from a temporary
5
See Appendix A-1 for more details on the definition and estimation of the APEs. The estimation results of a simple dynamic pooled probit model are available from the author but not reported for the sake of brevity. 6
140
M. Picchio
to a permanent contract, this implies that more able unemployed workers are more likely to accept a temporary position instead of undertaking a further period of job-seeking. This self-selection of more able workers into temporary contracts is predicted by Loh’s (1994) theoretical model: temporary contracts can be viewed as probationary periods and as sorting mechanisms for firms. Hence, firms can attract more able workers by paying a low wage during the probationary period but promising higher wages in the future.
7.3.3 Goodness of Fit In order to provide a descriptive evaluation of the goodness of the fit of the dynamic nonlinear models, I report in Table 7.5 the percentages of correctly predicted employment statuses and of correctly predicted sequences. I follow the usual rule according to which one predicts, for each i and t, that y1it will be unity ^ it 0:5: If U ^ it \0:5; y1it is predicted when the estimated probability is C 0.5, i.e. U to be zero. The percentages reported in Table 7.5 are the percentages of the times when the predicted y1it matches the actual y1it : By ‘predicted sequences’ is meant the percentages of times when the predicted sequence matches the actual sequence fy1it ; y1it1 g: The first column reports the correct predictions of the bivariate dynamic model, whereas the last 2 columns set out the correct predictions when I estimated single equation dynamic probit models using Wooldridge’s (2005) and Heckman’s (1981) estimators, respectively. The three different estimation techniques are very close to each other in terms of goodness of fit. Permanent employment is very well predicted more than 96% of the time. As regards the correct predicted unemployment or temporary employment realizations, the models are correct more than 50% of the time. I now move to the correct predicted permanent employment sequences. In light of the overall results, we may argue that the dynamic models were indeed able to predict the transitions, since they correctly predicted sequences more than 85% of the time. But, on looking at each possible sequence, it is apparent that my models could not predict permanent/unemployment-temporary job sequences, and that they performed poorly in predicting unemployment-temporary job/permanent sequences (around 10% of the time). Conversely, the dynamic models performed well in predicting time-invariant sequences.
7.4 Robustness Analysis In this section I report assessment of the robustness of the stepping-stone effect by relaxing the parametric assumptions on the individual heterogeneity. Therefore, I replicated the Heckman’s procedure for initial conditions (Sect. 4.1), but the
7 The Dynamics of Unemployment
141
residual a1i of the linear approximation of the unobserved heterogeneity c1i was assumed to have a discrete mass point distribution. Section 4.2 instead focuses on dynamic linear probability models for permanent employment, since these provide satisfactory estimates of average partial effects near the centre of the distribution of the covariates: first-differencing is a straightforward way to eliminate the time-persistent individual heterogeneity and to avoid any problem of initial conditions. As pointed out by Stewart (2007), since no particular parametric assumption about unobserved heterogeneity is required, this approach may be considered a semi-parametric method, compared with the dynamic nonlinear models discussed thus far.
7.4.1 Discrete Distribution of the Individual Heterogeneity I used a specification of unobserved heterogeneity alternative to the ones that characterize the Heckman’s and Wooldridge’s estimators. Instead of imposing normality, the distribution of a1i was assumed to be discrete with mass point, ad1 ; d 2 f1; . . .; Dg; and corresponding probability pd : I specified probabilities p1 to pD using a multinomial logit model: exp kd pd ¼ PD ; j¼1 exp kj
kD ¼ 0; d ¼ 1; . . .; D:
ð7:9Þ
Therefore, I maximized the following discrete mixture likelihood function: 39 2 0 > U zi0 c þ had1 ð2y1i0 1Þ > ! " # 7= 6 T T T 1 7 Y X X L¼ pd 6 5 4 d > > d¼1 U y1it1 q11 þ y2it1 q12 þ x01it b1 þ d11s w1is þ d12s hm > i¼1 > 1is þ a1 ð2y1it 1Þ ; : 8 > <X N > D Y
t¼1
s¼0
s¼0
ð7:10Þ where D is chosen according to the Akaike Information Criterion (AIC). The estimation results of the discrete mixture probability model are reported in Table 7.6. According to the AIC, unobserved heterogeneity seems to be important, but only two points of support are detected: the first one has a 16.8% probability mass, whereas that of the second is 83.2%. The estimated coefficient of the lagged unemployment indicator is significantly negative and close to those presented above. Once again, a temporary job today, rather than unemployment, significantly increases the probability of having a permanent position in 2 years’ time. The marginal effect of a temporary contract instead of unemployment is in line with those obtained previously, and it lies between the estimates of the APEs obtained using Heckman’s and Wooldridge’s approaches. This finding assesses the robustness of the results obtained under the parametric assumption on the individual heterogeneity.
142
M. Picchio
Table 7.6 Discrete Mixture Dynamic Model for Permanent Employment Variable Coeff. S.E. Permanent jobt1 0.654 Unemployedt1 -0.643 Experience 0.080 -0.147 Experience2 /100 Female -0.102 Education—Reference: None or Elementary Middle school 0.492 Vocational school 0.492 High school 1.036 University degree or + 0.809 Area—Reference: North–East North-West 0.033 Centre -0.082 South -0.535 Islands -0.548 Head of household 0.407 Unemployment rate -1.360 Permanent income -0.044 Transitory income -0.018 Spouse not working -0.060 Married 0.938 Children \ 6 -0.158 D2004 -0.076 Unobserved heterogeneitya h 1.705 -1.347 a11 -0.040 a21 k1 -1.598 p1 0.168 p2 0.832 Average Partial Effects 0.153 APETU 0.240 APEPU 0.086 APEPT AIC Observations Log-likelihood
0.170*** 0.133*** 0.020*** 0.041*** 0.103 0.161*** 0.233** 0.203*** 0.222*** 0.177 0.166 0.251** 0.232** 0.126*** 0.865 0.012*** 0.010* 0.196 0.457** 0.325 0.118 0.524*** 0.315*** 0.206 0.298***
1.607 5,031 -1,283.3
a The estimated coefficients of the time-variant variables in all time periods are not reported for sake of brevity
7 The Dynamics of Unemployment
143
7.4.2 Dynamic Linear Probability Models I adopted the following dynamic linear probability model specification for Eq. 7.1: y1it ¼ y1it1 q11 þ y2it1 q12 þ x01it b1 þ c1i þ u1it ;
ði ¼ 1; . . .; N; t ¼ 0; 1; 2Þ ð7:11Þ
First differencing is a simple way to get rid of individual heterogeneity, yielding Dy1it ¼ Dy1it1 q11 þ Dy2it1 q12 þ Dx01it b1 þ Du1it ;
ði ¼ 1; . . .; N; t ¼ 1; 2Þ;
Since Dy1it1 and Dy2it1 are possibly correlated to Du1it ; this model can be consistently estimated by using y1i0 ; y2i0 ; and noncontemporaneous realizations of the explanatory variables as valid excluded instruments. The estimation results of dynamic linear probability model are displayed in Table 7.7. The upper panel reports ordinary least squares (OLS), instrumental variables (IV), and efficient generalized method of moments (GMM) estimates of the first-differenced dynamic model, whereas in the lower panel the dynamic model is in level. The coefficients of primary interest are those associated with the lagged unemployment status, and since their reference is lagged temporary employment, they directly provide an approximation of the average partial effect. The OLS estimates from the model in first-differences and the model in levels are -0.089 and -0.258, respectively. The former is biased upward due to positive correlation between Dy2it1 and Du1it ; while the latter is biased downward due to negative correlation between y2it1 and the unobservable heterogeneity. The central columns report IV estimation results using: y1i0 and y2i0 as instruments for Dy1it1 and Dy2it1 in the first-differences specification; Dy1it1 and Dy2it1 as instruments for y1it1 and y2it1 in the level specification. The estimated lagged unemployment status coefficients are now -0.105 and -0.138: hence they converge on each other. The F-tests for excluded instruments, as suggested by Staiger and Stock (1997), show no sign of weakness in the instruments. Finally, the last three columns report the efficient GMM estimation results, introducing as further instruments the initial values of time-varying explanatory variables. In this way, we gain in terms of efficiency and test the validity of the instruments with a standard over-identification test. The over-identification tests do not reject the null hypothesis, so that the instruments seem to be valid. The stepping-stone effect is between 13 and 15.8 percentage points. Therefore, fully controlling for unobserved heterogeneity and avoiding the initial conditions problem indicate that having a temporary job, rather than being unemployed, increases by about 13 and 15.8% points the future probability of having a permanent position. This is a further finding which gives robustness to the conclusions derived from dynamic nonlinear probability models.
144
M. Picchio
Table 7.7 Dynamic linear probability models estimation results Variable OLS IV Coeff. First-difference specification -0.392 DPermanent jobt1 D Unemployedt1 -0.089 -0.076 D Experience2 /100 D Head of household -0.066 D Unemployment rate -0.141 D Transitory income -0.002 D Spouse not working 0.054 D Married 0.063 D Children \6 0.003 Constant 0.090 Observations F-test exc. Instruments: – – D Permanent jobt1 F-test exc. Instruments: – D Unemployedt1 – Hansen J statistics – – Level specification 0.346 Permanent jobt1 Unemployedt1 -0.258 Experience 0.008 -0.014 Experience2 /100 Head of household 0.020 Unemployment rate -0.189 Transitory income -0.001 Spouse not working 0.003 Married 0.042 Children \6 -0.021 Constant 0.445 Observations F-test exc. Instruments: – Permanent jobt1 – F-test exc. Instruments: – – Unemployedt1 Hansen J statistics – – a
S.E.
a
Coeff.
S.E.
GMM a
Coeff.
S.E.
0.042*** 0.060 0.019*** 0.025*** 0.290 0.001** 0.055 0.039 0.026 0.020*** 1,677
0.254 -0.105 -0.038 -0.063 -0.176 -0.002 0.059 0.108 -0.025 0.040
0.086*** 0.109 0.023* 0.032* 0.356 0.001** 0.063 0.068 0.038 0.024* 1,677 F(2, 1667) = 115.4 p-value = 0.000 F(2, 1667) = 67.9 p-value = 0.000 – –
0.083** 0.106 0.022* 0.030** 0.343 0.001** 0.061 0.067 0.038 0.023* 1,677 F(11, 1658) = 23.3 p-value = 0.000 F(11, 1658) = 14.8 p-value = 0.000 v29 ¼ 7:577 p-value = 0.577
0.033*** 0.041*** 0.002*** 0.005*** 0.010* 0.119 0.001 0.013 0.012*** 0.019 0.044*** 1,677
0.077** 0.109 0.004*** 0.009*** 0.015 0.197* 0.001 0.019 0.018*** 0.034 0.081*** 1,677 F(2, 1657) = 229.8 p-value = 0.000 F(2, 1657) = 150.0 p-value = 0.000 – –
0.075*** 0.104 0.004*** 0.008*** 0.014 0.188* 0.001 0.016 0.018*** 0.033 0.080*** 1,677 F(9, 1650) = 52.8 p-value = 0.000 F(9, 1650) = 34.9 p-value = 0.000 v27 ¼ 5:566 p-value = 0.591
0.180 -0.138 0.016 -0.029 0.019 -0.361 0.000 0.011 0.069 -0.014 0.472
White (1980) robust standard errors have been computed
0.215 -0.130 -0.040 -0.069 -0.094 -0.003 0.042 0.100 -0.023 0.041
0.197 -0.158 0.015 -0.027 0.017 -0.333 0.000 0.006 0.065 -0.006 0.473
7 The Dynamics of Unemployment
145
7.5 Concluding Remarks The paper has assessed whether and to what extent temporary jobs have been springboards to regular employment or to dead-end positions in Italy, using for the purpose the 2000, 2002, and 2004 waves of the SHIW. The sample was made up of individuals who were unemployed, permanent employed, and temporary employed in 2000, 2002, and 2004. I estimated a bivariate dynamic unobserved effects probit model to predict the probability of having a permanent position given the lagged labour market state. The main finding was that, ceteris paribus, having a temporary contract today, rather than being unemployed, increases the probability of having a permanent job 2 years later by about 13.7–16.2% points. This evidence suggests that, given observable and unobservable characteristics, temporary contracts in Italy are stepping-stones to permanent jobs. They enable individuals to leave unemployment by giving them the opportunity to acquire general (and possibly) specific skills so that they are permanent employable afterwards. Finally, I estimated dynamic discrete mixture and linear probability models to assess the robustness of the stepping-stone effect to the parametric assumptions about the individual heterogeneity. The average partial effects estimated using these two approaches, which were nonparametric in the specification of the unobserved heterogeneity, are in line with those obtained with dynamic nonlinear unobserved effects probit models.
Appendix A.1
Estimation of the Average Partial Effects
In the analysis reported, the APEs, or marginal effects, were estimated following Stewart (2007): I estimated the counter-factual outcome probabilities evaluated at the sample means, x1 : Let us call pU ; pP ; and pT the probability of permanent employment given, respectively, unemployment, permanent employment, and temporary employment at time t 1: Hence, on performing the Wooldridge’s analysis, I had that ^pP ¼
N h i 1X ^ þq ^11 þ ^c1i ; U x1 b 1 N i¼1
^ pU ¼
N h i 1X ^ þq ^12 þ ^c1i ; U x1 b 1 N i¼1
and ^ pT ¼
N 1X ^ þ ^c1i ; U½x1 b 1 N i¼1
146
M. Picchio
where ^c1i is the estimated linear approximation (7.5) of unobserved heterogeneity. When I adopted the Heckman’s estimator, the outcome probabilities had to be rescaled because of a different normalisation, so that the arguments of the standard normal c.d.f were multiplied by ð1 ^ kÞ1=2 : Moreover ^c1i is, in this case, the estimated correlated random effect according to specification (7.3). ^ J ¼ ^pJ ^pH with J,H = Hence, the estimated APEs were defined as APE H ^ UT was the effect of a temporary job at time t 1; rather U,P,T. For example, APE than unemployment, on the probability of being a permanent worker at time t.
References Alessie, R., Hochguertel, S., & van Soest, A. (2004). Ownership of stocks and mutual funds: A panel data analysis. The Review of Economics and Statistics, 86(3), 783–796. Blanchard, O., & Landier, A. (2002). The perverse effect of partial labour market reform: Fixedterm contracts in France. The Economic Journal, 112(480), F214–F244. Booth, A. L., Francesconi, M., & Frank, J. (2002). Temporary jobs: Stepping stones or dead ends? The Economic Journal, 112(480), F189–F213. Brown, S., & Session, J. C. (2003). Earnings, education, and fixed-term contracts. Scottish Journal of Political Economy, 50(4), 492–506. Chamberlain, G. (1980). Analysis of covariance with qualitative data. Review of Economic Studies, 47(1), 225–238. de Graaf-Zijl, M., & Berkhout, E. (2007). Temporary agency work and the business cycle. International Journel of Manpower, 28(7), 539–556 de Graaf-Zijl, M., van der Berg, G., & Heyma, A. (2011). Stepping stones for the unemployed: The effect of temporary jobs on the duration until (regular) work. Journal of Population Economics, 24(1), 107–139. Dolado, J. J., García-Serrano, C., & Jimeno, J. F. (2002). Drawing lessons from the boom of temporary jobs in Spain. The Economic Journal, 112(480), F270–F295. Farber, H. S. (1999). Alternative and part-time employment arrangements as a response to job loss. Journal of Labor Economics, 17(4), S142–S169. Hagen, T. (2002). Do temporary workers receive risk premiums? assessing the wage effects of fixed-term contracts in west germany by a matching estimator compared with parametric approaches. Labour, 16(4), 667–705. Hagen, T. (2003). Do fixed-term contracts increase the long-term employment opportunities of the unemployed? ZEW discussion paper, 03–49. Heckman, J. J. (1981). The incidental parameters problem and the problem of initial conditions in estimating a discrete time-discrete data stochastic process. In C. F. Manski & D. McFadden (Eds.), Structural analysis of discrete data with econometric applications (pp. 179–195). Cambridge, MA: The MIT Press. Hyslop, D. R. (1999). State dependence, serial correlation and heterogeneity in intertemporal labour force participation of married women. Econometrica, 67(6), 1255–1294. Ichino, A., Mealli, F., & Nannicini, T. (2005). Temporary work agencies in italy: A springboard toward Permanent employment? Giornale degli Economisti, 64(1), 1–27. Jimeno, J. F., & Toharia, L. (1993). The effects of fixed-term employment on wages: theory and evidence from Spain. Investigaciones Económicas, 17(3), 475–494. Loh, E. S. (1994). Employment probation as a sorting mechanism. Industrial and Labor Relations Review, 47(3), 471–484.
7 The Dynamics of Unemployment
147
Picchio, M. (2006). Wage differentials and temporary jobs in Italy. Discussion paper 2006–33, Département des Sciences Économiques de l’Université catholique de Louvain. Staiger, D., & Stock, J. H. (1997). Instrumental variable regression with weak instruments. Econometrica, 65(3), 557–586. Stewart, M. B. (2007). The interrelated dynamics of unemployment and low-wage employment. Journal of Applied Econometrics, 22(3), 511–531. White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and direct test for heteroskedasticity. Econometrica, 48(4), 817–838. Wooldridge, J. M. (2002). Econometric analysis of cross section and panel data. Cambridge, MA: The MIT Press. Wooldridge, J. M. (2005). Simple solutions to the initial conditions problem in dynamic nonlinear panel data models with unobserved heterogeneity. Journal of Applied Econometrics, 20(1), 39–54.
Chapter 8
Labour Market Segmentation, Flexibility and Precariousness in the Italian North East Giuseppe Tattara and Marco Valentini
8.1 Introduction Since the late 1970s, inequality has been on the rise in a number of OECD countries. Many factors enter into the determination of income inequality: lowpaid jobs, new flexible ‘non-standard’ patterns of employment, and unemployment. These factors play an important role in determining household poverty because they are only partially counteracted by the various government transfer programmes and social policies. In Italy as in many other European countries, one of the main causes of economic inequality is segmentation of the labour market. The Italian labour market is currently described as deeply segmented between an insider market, with wellpaid and stable jobs protected by government laws and powerful labour unions, and a large outsider market made up of the unemployed and by people discouraged from entering the labour force. Outsiders are mainly young, unemployed, and geographically clustered in the centre-south of the country.1 The OECD’s views This research is part of the Miur project 2001–2002, n. 2001134473 and 2003–2004, n. 2003139334. A previous version of this paper was discussed at IWPLMS, International Working Party on Labour Market Segmentation, Rome, September 2003. 1
See Kerr (1954), Doeringer and Piore (1971), Wilkinson (1981). In regard to Italy see Garibaldi and Young (2003); in comparative terms Nickell (1997) and OECD (1994). G. Tattara (&) Department of Economics, University Cà Foscari Venice, Cannaregio 873, 30121 Venice, Italy e-mail:
[email protected] M. Valentini (&) Tolomeo Studi e Ricerche srl, Via S. Bona Vecchia 62, 31100 Treviso, Italy e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_8, Springer-Verlag Berlin Heidelberg 2012
149
150
G. Tattara and M. Valentini
have recently been disputed on the ground that they reflect the plethora of protective regulations imposed by Italian labour-market legislation rather than being an assessment of their actual effectiveness. For example, the protection against firing granted by the Statuto dei lavoratori to workers in large firms (=[ 15 employees) in the absence of misbehaviour applies only to half of the stock of private employees in Veneto. For a criticism of the OECD thesis, see Del Conte et al. (2004). In the insider market, entrepreneurs pay good wages and hire on the contratto di lavoro tipico, that is, a full-time open-ended work contract.2 Opportunities to move between these two labour markets are severely limited, and several attempts to spread new forms of labour contracts in order to render the market more flexible have been made by the various Italian governments since the 1980s. Most observers consider the wider diffusion of new forms of temporary labour contracts to be a way to remedy the rigidity of the Italian labour market and to induce Italian entrepreneurs to hire workers without fear of locking themselves into a permanent commitment. Hence more flexible contracts are frequently advocated as the main means to reduce Italy’s high rate of unemployment. How the spread of temporary contracts has changed the nature of the Italian labour market is difficult to assess. Official Italian statistics grossly underestimate the number of people working on short-term contracts, and a more thorough account of various forms of temporary contracts more than doubles the official figures, ranking Italy among the countries with the largest amounts of temporary work. The (supposedly) rigid Italian labour market is less rigid than appears at first sight. In the Italian manufacturing sector, which has a traditionally high quota of stable workers, a double shift has taken place in recent decades. First, the quota of stable workers has declined over time. Second, the number of unstable workers, low-qualified and low-paid, has increased, and represents a non-marginal proportion of total employment. Frequently, a young worker will move through a succession of temporary contracts at the beginning of his/her career. This initial precariousness can be regarded positively as developing specific human capital which prepares the worker for a more permanent position with the same firm or with other firms, and possibly with higher wages, so that the worse initial conditions are off-set by better conditions in the future. But the outcomes for temporary workers are frequently different: the precariousness extends across the entire career, with the consequence that these workers should be considered extreme cases of outsiders with bad working conditions and receiving low wages compared to those of workers hired on open-ended contracts. Several authors have stressed that workers at the bottom of the Italian wage distribution are’trapped’ in their jobs and seldom move to better ones (Capellari 2002; Lucifora 1998). Whilst temporary contracts avoid
2
Italian labour contracts are split into two broad classes, tipico and a-tipico. Atipico is defined residually and refers to all contracts which do not fall under the heading of the tipico contract, which is full-time and open-ended, or tenure. On the historical origin of the terms, which date back to the populism of the fascist regime, see Accornero (2000, pp. 191–192).
8 Labour Market Segmentation, Flexibility and Precariousness in the Italian North East
151
certain labour-market inflexibilities imposed by protective employment legislation, they also entail higher potential costs.3 These issues are important, for in June 2003 the Italian government introduced further forms of semi-subordinate work (midway between dependent employment and self-employment), and the desirability of this policy depends upon whether the number of these contracts is assumed to be at present insufficient to give the labour market the necessary flexibility, and whether such contracts are to be considered ‘dead ends’ or ‘stepping stones’, necessary transitions to more permanent careers (Booth et al. 2002). The first part of the chapter discusses the relation between flexibility and precariousness in the Veneto labour market. It furnishes quantitative evidence on the increase in ‘non-standard’ forms of employment during the late 1990s and draws some conclusions concerning the downward bias in the measures adopted in official Italian statistics and in international comparisons. The second part of the chapter restricts its treatment to private employees in manufacturing, dividing them between ‘movers’ and ‘stayers’. Both categories show signs of instability. The quota of tenured workers over total workers decreases, and movers significantly increase over time. Among these latter are permanent movers, whose fragmented and chaotic work histories are identified and compared with those of workers with more stable careers.
8.2 Underestimation of Temporary Work A flexible contract envisages possible changes in working hours, wages and functions. But flexibility also means that a firm can straightforwardly stipulate a new labour contract and lay the worker off without incurring redundancy payments or restrictions imposed by labour legislation. Self-employment is the first labour contract that comes to mind where flexibility is concerned. Self-employment has no stated working time, and no stated conditions or remuneration. It is noteworthy that, after a long period of decline, since the mid-1970s the self-employed fraction of the labour force has increased in several Western countries (Blau 1987; Evans and Leighton 1989; Magnac and Robin 1994). As in other countries, in Italy too the self-employment quota in terms of total employment has grown over time. From a minimum level of 14% in total manufacturing in 1974 it peaked in the late 1970s. It rose again thereafter, returning in the early 1990s to 20%, a high value which had been common 40 years previously (Rapiti 1997, pp. 176–180; Chelli and Rosti 2002).
3
The main reason why Italy ranks so high in comparison with other industrialized countries is that since approval of the ‘‘Statuto dei Lavoratori’’ in 1970, employers that have fired workers in the absence of misbehaviour have been forced to reinstate them, and to pay the entire wages lost during litigation plus social insurance contributions.
152
G. Tattara and M. Valentini
The net flow of self-employed is largely the result of a net flow away from dependent employment. The positive variation recorded by self-employment over the past 20 years is largely explained by the net outflow from the category of employees: people looking for work move directly into self-employment to only a very limited extent (OECD 1992, table 4.8; Rapiti 1997, p. 181; Chelli and Rosti 1998, p. 459): A large part of workers merely substitute a stable form of employment with a more precarious one. Worker-membership of small cooperatives, members of partnerships are mostly disguised forms of labour precariousness. In addition, the various kinds of labour contracts introduced in recent years—collaborazioni coordinate e continuative (employer-coordinated freelance work), lavoro interinale (Temporary agency work) and other forms of lesser importance—and considered in other chapters of this book are highly flexible forms of employment. This plurality of occupational categories makes flexibility difficult to measure. In order to make the problem more tractable, the discussion is often restricted to the employee labour market, but flexibility is still difficult to identify. Statistical measures for overtime, changes in functions, and other forms of flexibility are lacking within the more structured employee segment as well. In everyday discourse, ‘flexible’ is often identified with ‘atypical’, and a widelyused indicator of flexibility is the quota of atypical workers over total employees. Atypical contracts are the sum of those of part-time workers and temporary workers. Part-time work with no time limits (although sometimes involuntary) is an employment relationship profoundly different from temporary work because of its permanent character, and it is not to be considered atypical. Temporary workers are basically young people hired on training/work (contratti di formazione e lavoro, henceforth CFL), apprentices, and workers hired on fixed-term contracts (among these numerous seasonal contracts). This definition comprises a notion of flexibility that centres on the limited duration of the contract and the connected risk of instability (Anastasia and Maurizio 2002; Anastasia et al. 2004).4 Our study deals with Veneto, a region in the North-East of Italy which offers a significant point of observation. It is a rich and dynamic region, with an important manufacturing sector and a labour market close to full employment. According to the Italian Central Statistical Office (Istat), in 2001 the quota of temporary workers in the Veneto labour market was around 8% of the total employee stock, which in absolute value was 102,000 employees in a total of 1,300,000 (Istat, Rilevazione trimestrale delle forze di lavoro—Labour Force Survey, henceforth LFS). A similar ratio is reported for Italy as a whole. The figure rapidly increased in the 1990s, both in absolute and relative values, and doubled after 1993 (from 4 to 8% of total employment), with a yearly average rate of 4 Open-ended contracts may hide forms of precarious employment: as in the case of employers who ask female employees to sign undated resignation letters so that they can be sure that the women will leave with the onset of their first pregnancies. A survey by the Ministry of Labour cited in Saraceno (2002) reports that 20% of female voluntary quits in the Northern regions of Italy happen during pregnancy.
8 Labour Market Segmentation, Flexibility and Precariousness in the Italian North East
153
increase amounting to 5%, i.e. 5,000 employees per year. In the same years, OECD countries showed no tendency towards an increase in temporary work (Anastasia and Maurizio 2002; Anastasia et al. 2004). Turning to flow values, temporary work has been absolutely dominant among yearly hirings since the early 1990s, and this is the first reason for regarding temporary work as signalling the prospective erosion of the stock of stable employees, although stock values, which change at a slow rate, are still dominated by open-ended contracts. The official Italian figures on temporary work are rather low in comparison with those for Spain and Portugal, two countries marked by the wide diffusion of temporary labour, and they are below the levels of the large continental countries, Germany and France, and of the Nordic countries, the Netherlands and Sweden. Temporary work is more widespread in Italy than in the United Kingdom and Ireland, but these are countries where the absence of protective legislation makes the use of temporary labour much less interesting. Italy recorded the average OECD number of female temporary workers (12%) in 2000, but a much lower figure for males. The official Italian data on temporary work are taken from the LFS. As noted by de Angelini and Giraldo (2002, p. 105, 2003) and Anastasia et al. (2004) regarding the nature of the labour contracts of family members, the replies to the survey reflected the interviewee’s (generally the head of household) self-perception of the nature of the labour contract, rather than its legal nature. And the way in which the interview was structured gave rise to significant bias in the assessment of temporary work. For example, work/training contracts (CFL) and apprenticeships— two widely used types of temporary contract—are often perceived as stable by the interviewee because they are frequently converted into open-ended positions at the end of the training period; all the more so if the interviewee is not the worker him/ herself, but a member of the family, who is not normally aware of the legal nature of the relative’s labour contract.5 LFS estimates provide the basis for the Italian official data used in international comparisons, and overall, Italian temporary work figures appear to be heavily downwardly biased. The bias becomes very apparent if LFS data are compared with administrative records for the same period of time and for the same geographical area. Administrative data in Veneto are provided by the Ministry of Labour (Veneto Lavoro) and by the Veneto Worker History (VWH) panel compiled by the Department of Economics at the University of Venice (see Appendix).
5 Baretta, Leombruni, Trivellato, Rosati (Baretta et al. 2004) explain in detail the main problems that arise when comparing administrative data and LFS data. According to De Angelini Giraldo (2002; 2003), at least 40% of the apprentices interviewed reported that they had permanent jobs, and 30% of temporary labour contracts were due to be converted into open-ended contracts by the same firm. See Ministry of Labour (2003). The author’s computations on the Social Security longitudinal panel yield a larger quota of conversions for both CFL and apprenticeships.
154
G. Tattara and M. Valentini
According to these two sources, the number of temporary workers is more than twice that counted by the Istat-LFS. Table 8.1 reports the figures for Treviso and Vicenza (which have been kindly made available to us by Veneto Lavoro) and the VWH figures: for comparable sectors and data, the two sources are substantially consistent, taking into account that VWH has only gradually included fixed-term contracts, and that the procedure used to measure CFL has probably produced a forward temporal shift in the Veneto Lavoro dataset (Veneto Lavoro 2005). Both sources report a temporary Several activities in the private service sector are well known to be characterized by unstable employment; and restricting the analysis to manufacturing— construction excluded—appears to be a conservative measure According to the VWH, temporary work contracts amounted in 2000 to more than 13.0%, which was more than twice the proportion of temporary workers estimated by LFS for the comparable 2000 population (5.0–6.0%). Veneto Lavoro data for Treviso and Vicenza are quite closely in line with the figures based on the VWH panel (Table 8.2). If the same correction suggested for the Veneto were applied at the national level, Italy would rank just below Spain, and close to Portugal, among the countries with the highest numbers of temporary workers, leaving Germany and France well behind. With the significant difference that in Portugal and Spain the number of temporary contracts is decreasing, while in Italy it is increasing: a feature that makes the Italian situation all the more critical (Anastasia and Maurizio 2002; Anastasia et al. 2004).
8.3 From Temporary Labour Contracts to Labour Market Precariousness Temporary labour contracts and short employment spells are not immediately connected with each other: temporary contracts are often renewed and they pave the way to more stable forms of employment. Yet temporary contracts characterize the majority of workers that work for short spells. According to the VWH data, CFL, apprenticeships and temporary contracts account for 67% of the employment spells that conclude within 6 months; the remaining 33% concerns workers hired on open-ended contracts which end de facto after 6 months (quits, layoffs etc.). With the increasing number of self-employed workers and new forms of shortterm contracts, the Italian labour market has undoubtedly developed towards a larger number of flexible worker categories. As more forms of precarious work become available and spread among workers, labour contracts considered unacceptable by the trade unions a couple of decades ago, and of uncertain legality, are now commonplace.
21709 19821 16675 11946 8860
1996 1997 1998 20849 1999 15619 2000 11716 2001 8657 2002 6467 2003 4605
30704 32729 33305 30486 27443 24834
Veneto Lavoro
26293 25989 29375 31331 33543
VWH
23666 25470 26844 22530 22362 23301
Veneto Lavoro
13104 17396 21649
VWH
450 1336 2866 3267 4261
Veneto Lavoro
VWH
252 228 207 164 96 47
Veneto Lavoro
VWH 465447 463248 463111 467259 479025
VWH
16.24 15.80 15.00 12.85 11.72 10.99
Veneto Lavoro
10.31 9.89 12.77 12.98 13.37
VWH
Temporary/stockb
Veneto Lavoro: employment refers to the private manufacturing sector and to CFL, apprenticeships, fixed-term contracts, temporary agency work (introduced in 1998) and home workers. Both the Veneto Lavoro and the VWH data are yearly stock data computed from flow histories at 31.12 each year. Data have been kindly made available to us by Bruno Anastasia and Danilo Maurizio, Veneto Lavoro. www.venetolavoro.it Source and method a Ateco 1981, 3 and 4 (Istat 2007) b The stock values are from VWH 1996–2000 and from LFS from 2001-2003
VWH
Veneto Lavoro
Table 8.1 Number of employees in manufacturinga hired on temporary labour contracts in Veneto. December each year Year CFL Apprenticeships Fixed-term contracts Temporary agency work Others Total worker stock
8 Labour Market Segmentation, Flexibility and Precariousness in the Italian North East 155
156
G. Tattara and M. Valentini
Table 8.2 Temporary worker quota in total employment Manufacturing Total private sector Year 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
Treviso and Vicenza
Veneto
LFSa
VWH§
LFSa
VWH§
3.64 4.68 3.97 4.42 4.15 4.51 5.16 5.17
17.92 14.40 11.22 9.73 10.30 10.38 9.77 9.29 12.04 12.46 12.79
3.34 3.12 3.37 3.79 4.74 4.29 4.91 5.30 5.19 5.47 6.23
18.55 15.17 12.15 10.42 10.98 11.06 10.31 9.89 12.77 12.98 13.37
Treviso and Vicenza
Veneto
LFSb
VWHc
LFSd
VWHb
4.01 5.37 4.28 4.69 4.58 5.19 5.59 5.85
17.30 14.00 11.10 9.65 9.95 10.00 9.65 9.24 12.40 13.10 13.50
5.00 4.77 4.90 5.46 6.67 6.92 6.81 7.26 7.47 8.06 8.87
16.35 13.40 10.91 9.35 9.44 9.59 9.23 9.02 12.80 13.36 13.83
Method § 1990–1997: temporary workers are apprentices and CFL; since 1998 workers on fixedterm contracts have been added. Manufacturing: §Ateco 1981, 3 and 4. a Ateco 1981, 1 to 5. LFS manufacturing includes construction, where there is a large amount of temporary work, so that our conclusions concerning the alleged LFS undervaluation of temporary work are possibly reinforced. Total private sector b Agriculture excluded c Agriculture and public sector excluded d Agriculture and public sector included Source LFS data are not published by Istat at the provincial level and have been kindly supplied to us by Anna de Angelini, Veneto Lavoro
Short employment spells have increased over time. Spells lasting less than 12 months increased from 6% in 1982 to 13% in 1996, and spells lasting between 12 and 24 months rose from 12% in 1982 to 21% in 1995.6 The number of short employment spells, and of seasonal spells, exhibits a definite cyclical pattern, peaking when GDP per capita is high, as in 1989 and 1995, and declining in the low years, as in 1993. Short spells are not confined to new labour-market entrants and have gradually affected more and more workers. The increase in the number of short-term employment spells has several causes: the increased number of CFL in the late 1980s; the doubling of apprenticeships in the late 1990s as CFL declined; and the rapid increase in the number of seasonal workers, who have very short work spells by definition (seasonal contracts trebled
6
Short spells include voluntary quits (by workers possibly hired on open-ended contracts) and exclude renewals and conversions of temporary contracts into permanent ones: the conversion of CFL into open-ended contracts is encouraged by the law and apprenticeships can be renewed up to a maximum duration of five years.
8 Labour Market Segmentation, Flexibility and Precariousness in the Italian North East
157
Table 8.3 Treviso and Vicenza, employment spells (VWH) Year Number of short employment spells/total employees Manufacturinga 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
Total private
0–12 months
13–24 months
0–12 months
13–24 months
6.80 6.46 7.26 9.10 9.56 10.99 12.13 12.95 12.73 11.03 9.99 8.52 12.06 15.05 13.66
5.38 4.84 5.02 6.01 7.17 7.97 8.48 8.71 8.50 7.77 6.75 6.19 6.63 7.66
11.73 10.98 11.34 12.71 13.06 14.67 15.97 16.48 16.27 15.20 13.96 12.30 14.82 18.21 17.26
6.88 6.07 6.00 6.77 7.92 8.54 9.18 9.46 9.47 8.94 7.99 7.36 7.62 8.40
Short spells are spells at the beginning of the year plus new spells. Total employees are initial stock + yearly per capita hirings Method a Ateco 1981, 3 and 4
in 15 years, from 2,500 in 1982 to 7,650 in 1997).7 While most of the labour force is employed in long-term jobs, there are numerous short-term jobs, and their number has increased in recent years. In the remaining part of the article, workers’ careers will be evaluated ex-post according to the succession of employment spells, independently of the legal form taken by the labour contracts of the workers concerned (Table 8.3). Worker careers will be compared in relation to mobility among jobs and length of time in the same job. Here ‘job’ is synonymous with ‘firm’.8 The analysis is restricted to employees in Treviso and Vicenza working in private manufacturing from 1982 to 1997: a population of 194,000 employees in 1982 which had increased to 233,000 by the end of the period. For our present purposes, employees are termed ‘movers’ or ‘stayers’ (Blumen et al. 1955). Union of the two sets—movers and stayers—exhausts the population 7 Compared with other countries, in Veneto only around 15% of employees work less than twelve months. The figures reported for Italy in international comparisons are likely to underestimate the real Italian situation because they are derived from LFS-Istat. According to a comparison conducted by Burgess (1998), 20% of employees (26-45 age class) in the United States were employed for less than twelve months, 20% in Holland, 11% in France, 16% in Germany, 8% in Italy, and 4% in Japan. 8 A reasonable assumption given that the average firm size in Veneto is very small: 13 employees.
158
G. Tattara and M. Valentini
Fig. 8.1 The sliding tenure window (±T) centred in t
t-T
t
t+T
Fig. 8.2 The observations interval (1975–1997) and the sliding tenure window (±7 years) in the bounded interval 1982–1990
set: stayers may be constantly employed in the same job from the outset, or they may become permanent after some other employment episode, usually a sequence of short spells representing the search process both by the worker and by the employer (short spells preceding tenure). Alternatively, employees may move from one short spell to another and never stabilize: short spells are ‘dead ends’ in relation to the perspective-working careers, and workers that display this pattern are labelled ‘movers’. The entire working life of a mover is made up of short spells and is referred to as a ‘chaotic’ or ‘chequered’ working life. In order to identify stayers and movers, tenure must be defined and measured. The extended coverage of the VWH longitudinal panel enabled us to compute the tenure directly for a significant number of years. The computation was performed for manufacturing without construction. The complete tenure of length T in t was defined if the employee in t was employed with the same employer in both (t-i) and (t + T – i), i = 1,…, T. The VWH dataset comprises information on working lives from 1975 to 1997 and the tenure length, T, is assumed equal to 7 years. This enabled us to measure the number of stable jobs for the years 1982–1990 without encountering censorship problems,9 1982 B t C 1990 (Tattara and Valentini 2002). The population set was bounded to 27–54 years of age (the age range in which a worker is more likely to develop a tenure). Figs. 8.1 and 8.2 show the tenure computation strategy.
9
A ten-year tenure would have shortened our window to a couple of years, 1986 and 1987. The seven-year tenure has no particular meaning but is a compromise.
8 Labour Market Segmentation, Flexibility and Precariousness in the Italian North East
159
According to the definition of tenure, the worker population was split into the following components: Stayers: at least 7 years’ tenure in the same job • from their labour–market entry • after a succession of short work spells Movers: less than 7 years’ tenure in the same job
8.4 Stable Employment Comparative studies report that the USA is the country in which tenure is shortest, 6.6 years, while longer tenures are recorded in Greece, Italy and Sweden, followed by Belgium, Japan and Portugal (Auer and Cazes 2000, p. 381; Dell’Aringa and Piccirilli 2000). Less than one-year tenures are present in all countries, but they represent an especially large quota ([20%) in the United States, Spain and Denmark. Long-term commitment to a single job is traditionally considered to be a specificity of the Italian labour market and as resulting from protective labour legislation. In Veneto around 70–80% of employees aged between 27 and 54 have been in permanent jobs for 7 or more years. At the end of the seventh year the probability of a job break is rather low, and almost all jobs continue until the 10 year.10 The share of stable jobs in the employee population has declined over time, while shorter employment relationships have increased. Jobs lasting less than 60 months have increased from 15 to 26% despite the aging of the population: a trend that suggesting that the quota of stable workers in the employee stock has increased.11 Table 8.4 shows interesting tenure specifications in relation to the population aged 27–54. Gender does not seem to be relevant: both males and females are stable to a similar extent. More interesting is the distinction according to education. Education is not an entry in the VWH panel but can be identified indirectly. The age of the employee in the year of entry into the VWH panel (entry not compatible with the continuation of education12) is used to infer a divide between primary and secondary (or higher) education. The label ‘primary education’ is 10
The probability of working three additional years with the same firm, after seven years of tenure, is 75% for males. The probability of working three years with the same firm as that of first employment is much lower: 44%. 11 The number of years, given the age boundaries, has possibly been more favourable to employment in the more recent period, so that the conclusions about a decline in tenure are further borne out. 12 Basically, no seasonal work or other short-term summer employment. For details see Canu and Tattara (2005).
160
G. Tattara and M. Valentini
Table 8.4 Stayers (Manufacturing, Treviso and Vicenza, age 27–54) Year Employees in thousandsa Stayers (% quota of the respective stocka) Total Males Females White collars Blue collars entering VWH entering VWH [=19 years of age \19 years of age 1982 1983 1984 1985 1986 1987 1988 1989 1990
115681 114560 114414 116461 119389 122840 127489 134376 142076
84.2 85.0 85.0 84.0 83.2 81.8 80.4 78.2 75.9
84.7 85.3 85.5 84.7 83.8 82.4 80.8 78.8 76.5
83.1 84.2 83.9 82.7 81.9 80.6 79.7 77.2 74.8
Males
Females
Males
Females
78.1 78.7 79.1 79.2 78.7 77.5 76.4 74.6 73.3
84.5 84.1 83.4 82.9 82.2 81.8 80.4 79.1 77.3
78.0 78.1 78.0 76.8 75.9 75.1 74.8
83.6 81.4 80.2 78.0 77.2 74.3 72.3
Method Ateco 1981, 3 and 4. The four last columns refer to an employee population (1990) of 6,342 F, 13,601 M, 8,002 F and 19,450 M a Initial stock plus yearly hirings
Table 8.5 Yearly quota of stayers (Manufacturing, Treviso and Vicenza, age 27–54) Year Average firm size (Number of employees) at the beginning of the tenure 1982 1983 1984 1985 1986 1987 1988 1989 1990
\50
C 50 \100
C 100 \200
C200
70.4 67.4 68.4 66.3 65.8 64.7 64.2 61.9 60.0
83.7 84.7 85.8 86.1 86.3 85.7 84.7 80.8 78.3
86.8 88.5 86.6 85.6 86.0 85.4 85.1 83.7 81.5
92.1 94.3 94.0 94.4 94.2 93.9 93.8 91.9 91.4
Method see Table 8.4
attached to all blue collars entering the archive before 19 years of age, while white collars entering after 19 years of age are attributed a secondary education. This assumption is grounded on the idea that the Veneto labour market has been in full employment since the early 1990s, so that the interval between ending formal education and starting work as an employee is extremely short (Table 8.5). Nothing is said about blue collars entering the employee archive at the age of 19 years or over, or about white collars entering the labour market before 19 years of age, so that the result is a subset of the total population. Females with lower formal educations have tenures which are much shorter than both the average and that of females with higher formal educations.
8 Labour Market Segmentation, Flexibility and Precariousness in the Italian North East
161
Large firms have a larger quota of stayers by definition because larger units internalise many job changes which appear as tenure breakdowns in small firms. Moreover, small firms have high birth and high mortality rates, and this makes stability less likely. Nonetheless, tenure is very significant in small firms as well, and this is of specific importance in our two provinces, where firms with fewer than 100 employees account for 73% of the entire employee population. The stayer quota displays an overall declining pattern over time. The general trend is independent of job grade (white or blue collar), gender, and firm size: tenure decline characterises small and medium-sized firms (50 \ employees B 100) more than it does larger ones ([200). Stability is generally preceded by a series of short-term spells, on average two or three, which take place in a period of approximately 2 years.13 The bi-modal distribution of working life between many short spells and a subsequent long spell (tenure) is typical of males.14 Women have relatively uniform and longer spells.15 With the passage of time, the difference between the genders tends to disappear, and females exhibit the short and repeated spells of males.
8.5 Movers The rest of the labour market consists of unstable labour. Movers are workers that have never significantly stabilized during their working lives.16 Movers are defined as employees aged 35–54, i.e. of an age more favourable to the constitution of a stable employment relationship, who in the period 1975–1997 had no continuous job lasting 7 years or more (considering all the work spells available to them, jobs in firms outside Veneto included). We assume it extremely unlikely that a 35–54 year-old worker with no tenure during the 22-year 1975–1997 window would stabilize thereafter, and his working life has thus been marked by protracted
13
Pre-tenure months are truncated at 120. Few cases refer to workers that work continuously for 120 months with the same firm, and the tenure starting year, 1982 or 1990, in fact represents a transition from stable employment to stable employment. The pre-tenure period is assessed on the entire population, including employment spells undertaken outside the territory of Treviso and Vicenza and in all private sectors. 14 Pre-tenure spells are computed on a population aged 25–30. Males have pre-tenure periods shorter than those of females (the difference is around ten months) with a larger number of contracts with different firms and a definitely shorter average spell length. The average waiting period in pre-tenure status was 2 years in 1990, independently of gender. In 1990 males had on average 3.5 pre-tenure employment spells per capita, while females had 3. In 1982 the average pre-tenure spell was longer: 30 months for males and 44 for females. Because females had lengthy spells, the number of spells was smaller (2.1 females and 2.9 males). As time elapses females reduce their pre-tenure average duration from 44 months to 39. 15 This does not imply that the bulk of pre-tenure working time is evenly distributed among all workers: individual differences are marked. 16 The definition is independent of the labour contract at the reference date.
162
G. Tattara and M. Valentini
Fig. 8.3 Number of working spells (Y axis) in relation to their average duration (X-axis). Female movers (left) and male movers (right). Number of months (average working spell) in total private employment
instability. The analysis is restricted to manufacturing (Ateco 3 and 4) in Treviso and Vicenza. Movers are strictly defined so as not to confuse them with workers who work short spells which afterwards lead to a tenure. Are movers really workers with chaotic careers, or are they just workers on the borderline of the stayer’s definition? Do these workers embody a potentially stable relation or do they form a quite different category? In the following analysis we exclude employees close to the stayer definition (i.e. workers with two working episodes with an average duration greater than or equal to 5 years) from the candidate movers. And we exclude very short movers as well, although for a different reason, namely that very short episodes are liable to be confused with temporary commitments—commissions, service on examination boards, bonus committees, etc.—which do not substitute for a open-ended labour contract but are mostly additional to it (1/3 of the total is discarded). The circled area in Fig. 8.3 represents the ‘bulk of authentic movers’: people with work episodes ranging between 3 and 14 in number and with an average duration of between 6 and 30 months. The average number of working spells lies between 7 and 10, and the single spell average duration between 17 and 20 working months. These are employees with chequered careers. The spell distribution is rather even, and the average spell value and the average number of spells represents the behaviour of the majority of employees. Movers account for a significant quota of employees in manufacturing for the relevant age cohort. Movers are counted by summing work spells, and they are related to the employee population in the same period in Tables 8.6 and 8.7. Their number varies between 14% and 22% of the total. Movers have a definite gender and professional specification because the larger quota is made up of females and by lower-educated persons. The divide is clear and stable over the entire period. It highlights a structural segmentation of the employee labour market in manufacturing which is not usually taken into due account and is worth considering. Education is significant for both males and females: among workers with lower educations, movers are double the number of those with secondary educations, and the divide seems to be increasing markedly over time.
8 Labour Market Segmentation, Flexibility and Precariousness in the Italian North East
163
Table 8.6 Movers in manufacturing in Treviso and Vicenza, age 35–54 Year Work spells Movers (% on yearly flowsa) Total Males Females White collars entering VWH C19 years of age 1982 1983 1984 1985 1986 1987 1988 1989 1990
116759 115310 115141 117168 120164 123822 128728 136474 144418
13.8 12.4 12.1 12.2 13.0 14.2 15.6 18.6 21.8
12.1 10.9 10.9 11.0 11.9 13.3 15.1 18.2 21.5
17.1 15.2 14.5 14.5 15.1 15.9 16.6 19.3 22.4
Blue collars entering VWH \19 years of ageb
Males
Females
Males
Females
14.18 12.86 12.06 11.01 10.52 10.38 10.62 11.69 12.21
10.68 11.12 9.60 8.71 8.00 7.30 7.79 8.06 8.62
13.81 15.24 16.51 17.52 19.06 22.11 23.42
9.96 11.23 11.77 13.32 14.11 16.91 18.97
a
Column 2 Age 27–54 Method Ateco 1981, 3 and 4. Yearly movers are measured by the number of work spells pertaining to people labelled ‘movers’. The number of employees and the number of work spells differ because employees who move are often the same in different years, and, on average, movers have several work spells in the same year. Work spells pertaining to movers in 1990 are 1531 M, 405 F, 3832 M, 2458 F for the last four columns b
Movers are more common in the apparel, leather, furniture and food industries. Apparel has a large female quota and the tanning industry (part of the leather sector) employs numerous immigrants, which form a highly unstable component of the labour force. The rate of increase of movers is high and positive for most sectors as time elapses, and movers overflow into sectors previously excluded, like mechanics and plastics (moulding). Movers are not important in machine production, where the demand for high specialization and the required learning time are possibly greater than in most traditional sectors. As to be expected, movers are much more significant in small firms, although their quota stabilizes at just under 50 employees. Movers are mainly low-wage, low-educated, precarious workers. Sample restrictions have prevented the inclusion of highly-qualified workers and managers that jump from job to job seeking better salaries. The working lives of movers can be described with the notion of the chaotic career, that is, a working life with a chequered trajectory, where the term ‘chaotic’ has a distinctly negative meaning. While ordered careers are accompanied by strong integrating elements, chaotic careers, owing to high worker mobility among different jobs—particularly when movers occupy the bottom level of the market—prevent professional learning processes and induce workers’ isolation. The creation of a solid relational network is prevented and, as time passes, the impoverishment of human capital and the debasement of social relations reduce the chances of moving to a better job (Bianco 2004). We constructed different careers for movers and stayers by taking the entire working lives of movers into account. In what follows, we shall refer to the ‘representative’ permanent mover as the average mover: that is, someone who
11.16 9.09 7.83 6.43 5.15 4.87 5.27 6.04 6.47
12.99 12.81 12.67 12.94 12.00 11.26 12.41 13.85 15.97
6.77 6.05 5.96 5.84 6.11 5.73 5.64 6.67 7.80
17.96 16.67 14.74 14.18 15.17 15.71 17.20 20.88 22.80
17.92 16.09 14.74 13.96 13.73 13.92 13.78 15.25 16.57
13.10 11.02 9.51 8.38 8.20 8.29 8.87 10.50 12.55
15.35 11.79 12.37 11.38 10.93 10.97 11.07 12.59 15.20
19.66 17.27 15.63 15.20 14.99 15.18 16.10 18.08 20.09
C 50\100 11.14 9.62 8.71 8.28 8.07 7.94 8.02 9.08 10.28
C100 \200 9.50 7.78 6.80 6.53 6.24 5.94 5.97 6.98 7.83
C200 5.40 4.57 4.11 3.44 3.06 2.83 2.58 3.14 3.66
Method see Table 8.6. Firm size and sector are measured at the beginning of the labour spell. The population numbers (first eight columns) in 1990 are: 6,853 (31), 1,934 (32), 906 (41), 4,035 (43), 2,455 (44), 5,600 (45), 3,220 (46), 1,504 (48)
1982 10.35 1983 8.50 1984 7.59 1985 7.96 1986 7.82 1987 8.29 1988 9.55 1989 11.44 1990 13.37
31 Metal 32 Mechanics 41 Food-bev 43 Textiles 44 Leather 45 Apparel 46 Furniture 48 Plastic \50
Table 8.7 Mover work spells in manufacturing in Treviso and Vicenza, age 35–54 (% on yearly flows) Year Ateco 1981 Average firm size (number of employees) at the beginning of tenure
164 G. Tattara and M. Valentini
8 Labour Market Segmentation, Flexibility and Precariousness in the Italian North East Fig. 8.4 The working careers of stayers and movers. Males
Working episodes duration. Months
Males 202,176
Mover Stayer Interval between successive employment spells
192 168
165
144 120
134,114
140,114
96 118,65
72 75,63 48
131,65
83,63 75,45 88,45
34,22 43,27 57,27 22,22 12,12 28,12
24
0
24
48
72
96
120
144
168
number of months
works five periods if male, four periods if female, with an average duration of 18 months divided by long periods of non-employment. Female non-employment intervals are twice as long as those of males.17 The relative working spells for a mover were computed by taking account of all possible employment spells in every sector and in every firm located in and outside Treviso and Vicenza. Mover self-employment spells have not been considered because few workers had experienced such episodes. Figures 8.4 and 8.5 show the number of months worked (vertical axis) as time passes (time on the horizontal axis: worked and non-worked episodes in sequence). The 45 line represents the working life of a permanently employed person. The segments parallel to the 45 line represent the series of employment spells; the horizontal shifts away from the 45 line are the number of non-employment
17
The representative mover is an ideal type. A set of representative movers was identified in the entire VWH population: 40,000 workers were selected and their average numbers of spells were computed: 5 for males and 4 for females. In order to compute the duration of each single episode in the work spells sequence, the birth cohort was restricted to the interval 1955–1960 (around 10,000 individuals) so as to avoid a possible left censorship. The average of the durations of the individual episodes, according to their place in the sequence, was rounded up to the nearest integer. The same procedure was used for stayers, who amounted to around 12,000 individuals.Mover and stayer wages were the average wages for the respective sets. They are expressed in real terms through the cost of living index.
166 Fig. 8.5 The working careers of stayers and movers. Females
G. Tattara and M. Valentini Females working episodes. dutarìtion. Months
Mover Stayer interval between successive employment spells
employees
187,158
187,247
144 120 110,97
96 72
96,58 48 24
42,42
141,79 120,58
55,42
54,34
15,15
126,97
76,34
35,15 0
24
48
72
96
120
144
168
number of months
months, put in chronological order. The figures illustrate the series of employment/ non-employment spells for stayers and movers: time is limited to approximately 15 years, a period sufficient for efficacious description of large part of the career of the average permanent mover, either female or male.18 Permanent movers have long waiting periods between one employment spell and the next; the waiting periods are of approximately the same length for males, but they increase as time passes for females: in other words, the sequencing of periods of employment by movers does not increase their employment prospects but possibly damages them, particularly in the case of females Conversely, stayers have very short waiting periods between two successive employment episodes: the duration of the employment period is on average four times longer than that of the preceding non-employment spell for females, and seven times longer for males. Over 15 years, male stayers on average total 16 months of inactivity relative to three or four job changes that can be considered frictional periods of nonemployment. The inactivity periods between successive jobs for permanent movers amount to 72 months, i.e. 6 years in a total working career of 11 years: non-employment is a basic element in the permanent mover’s working career. The time spent in inactivity by stayers, in relation to worked time, declines positively, in particular for males, as new working episodes accumulate (from 12 to 8 to 6 months in Fig. 8.3), while movers exhibit a less clearcut pattern. Different career perspectives can be evaluated in terms of the incomes generated. Although social security data on wages are a matter of considerable debate, they represent a unique source of information: they are recorded in VWH and are carefully exploited here. VWH wages are probably contractual wages, not the
18
12,000 stayers and 8,200 movers, born in 1955–1960, 30–55 years of age.
8 Labour Market Segmentation, Flexibility and Precariousness in the Italian North East
167
300 stayer mover
250 200 150 100 50 0 1
15
29
43
57
71
85
99 113 127 141 155 169 183 197
number of months
Fig. 8.6 Labour incomes according to the number of months worked in manufacturing. Males
actual remuneration paid to workers: the number of declared worked days may be less than the number of actual days worked (to appear coherent with the contractual wage paid), and overtime is not (usually) declared.19 The first element can be dealt with by considering only spell episodes greater than or equal to 12 months for stayers (with a range of working days between 290 and 320) and monthly intervals for movers (the monthly wage is divided by the number of days worked). The comparison between stayer and mover wages is restricted to work episodes in manufacturing (Ateco 3 and 4): although this drastically reduces the number of episodes, it makes the comparison more significant because stayer wages in manufacturing are meaningfully compared with mover wages computed in relation to their work spells in manufacturing alone.20 Wages are expressed at constant 1995 prices (Table 8.8). Male wages are generally higher than female wages; stayer wages—for both males and females—are higher than those of movers, with a wage gap of around 15%. The wage difference is larger for females. The main reason for the difference between stayer and mover wages is that movers never progress in their careers. Average wages for movers and stayers are multiplied by the time spent as employees so as to provide a rough visual measure of the difference in earned income. The result of the two elements—lower wages and fewer periods worked— is rather low work incomes for movers compared with those of stayers. This is illustrated in Fig. 8.6 and 8.7 for both males and females. The horizontal axis 19
On the appropriateness of using social security data on wages to represent the actual remuneration paid to workers, see Ginzburg et al. (1998; 1999), Gavosto and Rossi (1999), Contini et al. (2001, 2002). The Italian social security records provide data for total wages paid but do not detail the hours worked. Hence it is difficult to determine whether overtime, and how much of it, has been included. 20 We have excluded managers, whose salaries are so different and so variable that the inclusion would distort the average.
168
G. Tattara and M. Valentini 200 stayer mover
150
100
50
0 1
16
31
46
61
76
91 106 121 136 151 166 181 196
number of months
Fig. 8.7 Labour incomes according to the number of months worked in manufacturing. Females Table 8.8 Average weekly gross wages of movers and stayers in manufacturing Employment successive Wages at 1995 prices, in Euros spells Stayers Movers Malea
Femaleb
Malec
Femaled
Mean Std.dev. Mean Std.dev. Mean Std.dev. Mean Std.dev. 1 2 3 4 5
293 294 313 325
63 75 114 121
253 260 258 0
52 64 69 0
276 287 306 320 313
87 110 126 150 134
228 225 233 222
64 65 81 68
Method Ateco 1981, 3 and 4. Average number of workers: 4704a , 2725b , 1585c , 1215d
measures the number of months (worked and non-worked episodes in sequence), the vertical axis cumulates the monthly real wages earned for each month worked so as to provide the total earned income referred to the respective month measured on the horizontal axis (Table 8.8).
8.6 Towards More Precariousness? The increase in inequality in Veneto has been discussed in light of the distinction between the new flexible ‘non-standard’ patterns of employment and the tracditional Italian full-time open-ended labour contract. Non-standard patterns of employment or atypical labour contracts have been on the government agenda, and they have given rise to reforms which have substantially increased the number of atypical contracts (Tursi 2004; Accornero 2006).
8 Labour Market Segmentation, Flexibility and Precariousness in the Italian North East
169
Our research has provided conclusive evidence that the official Italian statistics grossly underestimate the number of people employed on atypical labour contracts: a more accurate account of the various forms of short-term contracts should more than double the official figures, ranking Italy among the European countries with the largest amounts of temporary work. The number of atypical labour contracts has increased rapidly over time, and although the increase seems to have halted in recent years, the erosion of permanent employment]over time is very apparent. More thorough examination of the dynamics of the Italian labour market would have provided evidence that temporary labour contracts were widespread in Italy, and that there was no need for labour-market reforms to foster their further diffusion. Other aspects of the labour market, primarily the Italian anomaly of the absence of universal unemployment benefit, should have been addressed directly. Do repeated spells of temporary employment lead to more permanent jobs and better working conditions in the future, or should they be considered as dead ends? Training contracts represent the majority of short employment spells. They enable young workers to ‘shop around’ for better matches, and they often develop into permanent careers. But there is a substantial and increasing quota of mature workers—movers—for whom short working episodes are a permanent situation of precariousness and low income that lasts for a considerable number of years. Movers are workers at the bottom level of the market, blue collars with low educations, and mainly females. They repeatedly enter and exit employment; they never stabilize and their situation worsens as they grow older. The recent spread of new forms of temporary employment among mature workers has exacerbated the recourse to short employment spells—chequered careers—and the cost of this new pattern of employment, in terms of both income and human capital loss, is possibly much larger than the benefits that can be expected from greater labour-market flexibility—if flexibility can be considered an issue in the contemporary Italian labour market.
Appendix: The VWH Database The ideal dataset for analysing the divergence between worker and job flows is provided by the universe of employers matched by the universe of workers, because job flows are defined on the employer’s behaviour over time. We have been able to exploit a long panel of such data. The longitudinal panel used in this study has been constructed from the administrative records of the Italian Social Security System (Inps) and is compiled at the Department of Economics, Venice University. It concerns the entire population of employees and workers in two provinces, Treviso and Vicenza, of the Italian region of Veneto. The database covers every firm and every individual employed in the private sector (not, with few exceptions, the state and local government) except for self-employed workers, farm workers, and people not earning wages. For almost a decade, the Veneto labour market has been characterized by almost full employment and by a positive rate of job creation in manufacturing, in
170
G. Tattara and M. Valentini
contrast to a negative national rate. It is a dynamic manucentric area, with a large population of small firms; the average firm size is thirteen employees. The stock of manufacturing workers in the two provinces of Treviso and Vicenza has varied between 194,000 employees in the early 1980s and 233,000 employees in 1996, with a yearly positive average rate of variation of 1.4%. The average rate of growth in employment is the result of a marked increase in white collars and female workers (Occari et al. 1997). The Veneto longitudinal panel has records on establishment and worker flows from 1975 to 1997, a rather long period of time compared with other studies of the same kind; employers are classified with the three-digit Ateco 1981 standard classification.21 The VWH data include register-based information on all firms, and on all employees hired by those firms for at least one day during the period of observation, independently of the worker’s place of residence.22 The unit of observation is the employer-day. These items of information are used to build a monthly history of the working life of each employee. Employers are identified by their identification number, which changes if ownership, in the strict sense, changes. This has been amended, and whenever more than 50% of all employees are taken over by the new legal employer, the employment spell is considered to be continuing. Similarly, if there are short breaks in the employment spell, as long as the worker continues at the old employer, his/her spell is considered to be uninterrupted.23 Data include all individual employment spells with an employer, of whatever duration, and this has probably produced a large number of very short spells. Although short spells characterize the average job, they are concentrated at a young age, whilst long spells characterize the mature worker’s current experience. All plant sizes have been considered, because Veneto is characterized by a multitude of very small production units (firms with B5 employees account for almost 12% of total manufacturing employment).24
21
Revelli (1995) and Rapiti (1998). Strictly speaking, the term ‘establishments’ should be used, and not firms, because social security contributions can be paid at the establishment, although the payments for different establishments can be united on demand (and the database seeks to unify scattered payments through detailed study of the most relevant situations). 22 The entire working lives of all employees that have worked at least one day in Treviso and Vicenza has been reconstructed by considering employment spells outside Treviso and Vicenza as well. 23 A ‘cleaned’ social security archive has been used. Engagements/separations and creations/ destructions due to a change in the unit paying the social security contributions and not matched by a corresponding change in the working population assessed at the firm level have been defined as ‘spurious’ and have been deleted. The complex matching procedure – which is common practice when working with social security data – is explained in Occari and Pitingaro (1997). For a similar procedure, see Bingley and Westergård-Nielsen (2002). 24 The absolute importance of small firms impedes comparison with other countries: for example, in Veneto the percentage of employment in firms with C100 employees is 27%, while in Denmark it is more than 40%, and even larger in the United States. On the uncertain meaning of mobility measures for small firms see Tattara and Valentini (2003).
8 Labour Market Segmentation, Flexibility and Precariousness in the Italian North East
171
References Accornero, A. (2000). Contratti a tempo determinato e prospettive di flessibilità di lavoro. In Agenzia per l’impiego (Ed.), Solo una grande giostra? La diffusione del lavoro a tempo determinato. Milano: Angeli Accornero, A. (2006). Primo bilancio della ‘‘riforma Biagi’’: presupposti e risultati. Economia e società regionale, 92, 58–71. Anastasia, B., Disarò, M., & Maurizio, D. (2004). Occupati stabili, mobili, temporanei in Veneto: Misure di consistenza e di ‘‘lock in’’. I tartufi, 16, novembre. Anastasia, B., & Maurizio, D. (2002). Misure dell’occupazione temporanea: consistenza, dinamica e caratteristiche di uno stock eterogeneo. I tartufi, 11, abridged in Ministero del Lavoro e delle Politiche Sociali (Ministry of Labour) (Ed.) (2003), Monitoraggio delle politiche occupazionali e del lavoro. La misurazione statistica del lavoro a tempo determinato. Box 1, p. 84. Auer, P., & Cazes, S. (2000). The resilience of the long-term employment relationship: Evidence from the industrialized Countries. International Labour Review, 139(4), 378–408. Baretta, P., Leombruni, R., Trivellato, U., & Rosati, S. (2004). Worker mobility from social security registers and household surveys: A comparative assessment and updates analyses. Final workshop for the Miur Research Project Dynamics and inertia in the Italian labour market, San Servolo-Venezia, 15–17 Aprile. Bianco, M. L. (2004). I risvolti di genere della flessibilità. In G. Mari (Ed.), Libertà, sviluppo e lavoro. Milano: Bruno Mondadori. Bingley, P., Eriksson, T., Werwatz, A., & Westergård Nielsen, N. (2000). Beyond ‘‘Manucentrism’’. Some fresh facts about job and workers flows. Centre for Labour Market Social Research Blau, D. (1987). A time-series analysis of self-employment in the United States. Journal of Political Economy, 95, 445–467. Blumen, I., Kogan, M., & McCarthy, P. J. (1955). The industrial mobility of labour as a probability process. Ithaca,NY: Cornell University. Booth, A. L., Francesconi, M., & Frank, J. (2002, June). Temporary jobs: Stepping stones or dead ends? Economic Journal, 112, 189–213. Canu, R., & Tattara, G. (2005). Quando le farfalle mettono le ali. Riflessioni sull’ingresso delle donne nel lavoro dipendente. Economia & Lavoro, 2. Capellari, L. (2002). Do the ‘working poor’ stay poor? An analysis of low pay transitions in Italy. Oxford Bullettin of Economics and Statistics, 64(2), 87–110. Chelli, F., & Rosti, L. (1998). Un’analisi sui flussi nel mercato del lavoro: l’occupazione indipendente. Politica Economica, 8(3), 447–476. Chelli, F., & Rosti, L. (2002). Age and gender’s differences in Italian workers mobility. International Journal of Manpower, 24(4), 313–325. Contini B., Filippi M., & Malpede C. (2001). Differenziali retributive Nord-Sud: distorsioni attribuibili alla normativa previdenziale. Labor WP, 13, Centre for Employment Studies. Contini, B., Filippi, M., & Malpede, C. (2002). Safari nella giungla dei salari. Labor WP, 3, Centre for Employment Studies. de Angelini, A., & Giraldo, A., 2002. Mobilità e percorsi di stabilizzazione nel mercato del lavoro veneto. Confronti fra evidenze statistiche e evidenze amministrative. In Veneto Lavoro (Ed.), Il mercato del lavoro del Veneto. Tendenze e politiche. Rapporto 2002, Milano: Angeli de Angelini, A., Giraldo, A. (2003). La mobilità dei lavoratori nel Veneto. Confronto fra misure su dati LFS e su dati Netlabor, Miur Project Dinamiche e persistenze nel mercato del lavoro italiano ed effetti sulle politiche, Working paper, 61, Padova. Del Conte, M., Devillanova, C., & Morelli, S. (2004). L’indice OECD di rigidità nel mercato del lavoro: Una nota. Politica Economica, 20(3), 335–355. Dell’Aringa, C., & Piccirilli, G. (2000). La mobilità occupazionale nelle fasi iniziali della carriera. Lavoro e relazioni industriali., 1, 3–35.
172
G. Tattara and M. Valentini
Doeringer, P.B., & Piore, M.J. (1971). Internal labor markets and manpower analysis. Lexington, Mass: Heath. Evans, D., & Leighton, L. (1989). The determinants of changes in US Self-employment. Small Business Economics, 1, 111–120. Garibaldi, P., & Young, D. (2003). Employment protection legislation: Its Economic impact and he case for reform. European Economy-Economic papers, 186. Gavosto, A., & Rossi, F. (1999). Giornate retribuite e differenziali salariali nei dati Inps. Politica economica, 15(2), 253–257. Ginzburg, A., Scaltriti, M., Solinas, G., & Zoboli, R. (1998). Un nuovo autunno caldo nel mezzogiorno Note in margine al dibattito sui differenziali salariali territoriali. Politica Economica, 14(3), 377–410. Ginzburg, A., Scaltriti, M., Solinas, G., & Zoboli, R. (1999). Il mistero dei salari in Italia Una risposta a Gavosto e Rossi. Politica Economica, 15(2), 259–266. Istat (2007). Classificazione delle attività economiche. Roma. Kerr, C. (1954). The balkanization of labor markets. Bekeley: University of California Press. Lucifora, C. (1998). Working poors? An analysis of low wage employment in Italy. In R. Asplund, P.J. Sloane, & I. Theodossiou (Eds.), Low pay and earnings mobility in Europe. Chelthenham: Edward Elgar. Magnac, T., & Robin, J. M. (1994). An econometric analysis of labour market transitions using discrete and tenure data. Labour Economics, 1, 327–346. Ministero del lavoro e delle politiche sociali (Ministry of Labour). (2003). Monitoraggio delle politiche occupazionali e del lavoro. Nickell, S. (1997). Unemployment and labour market rigidities: Europe versus North America. Journal of Economic Perspectives, 11(3), 55–74. Occari, F., & Pitingaro, S. (1997). Demografia di impresa e mobilità del lavoro. WP CNR, 7. Occari, F., Tattara, G., & Volpe, M. (1997). Occupazione, mobilità e componente femminile nel mercato del lavoro: i lavoratori dipendenti a Treviso e Vicenza. In Regione del Veneto, & Veneto Lavoro (Eds.), Il mercato del lavoro nel Veneto (pp. 460–488). Milano: Angeli OECD (1994). The OECD jobs study, evidence and explanations (Vols. I and II). Paris: Author. OECD (various years, July). Employment Outlook. Rapiti, F. (1997). Lavoro autonomo, lavoro dipendente e mobilità: un quadro statistico sull’Italia. In S. Bologna & A. Fumagalli (Eds.), Il lavoro autonomo di seconda generazione Scenari di postfordismo in Italia (pp. 173–191). Milano: Feltrinelli. Regione del Veneto. (various years). Il mercato del lavoro nel Veneto. Rapporto. Milano: Franco Angeli. Revelli, R. (1995). Potenzialità degli archivi Inps ai fini della stima degli aggregati dei conti economici territoriali. Quaderni di ricerca Istat. nuova serie, 1 Saraceno, C. (2002). Paradossi della flessibilità: una prospettiva di genere e generazionale. In M. Magatti & G. Fullin (Eds.), Percorsi di lavoro flessibile. Roma: Carocci. Tattara, G., & Valentini, M. (2002, June 17). Flessibilità e attaccamento al posto di lavoro. Due tipi di carriera o due aspetti di una sola carriera lavorativa? Venice workshop: La mobilità del lavoro in un mercato di piena occupazione. Tattara, G., & Valentini, M. (2003). Un mercato del lavoro molto mobile? Si, ma con molti se e molti ma. Economia e Società Regionale, 1, 24–54. Tursi, A. (2004). Un anno di Legge Biagi. http://www.lavoce.info/articoli/pagina1249-351.html. Veneto Lavoro (2005). I lavoratori dipendenti in Veneto 1998-2003: profili e percorsi. Statistiche sistematiche da ‘‘Giove 2005’’, 20. Wilkinson, F., (Ed.), (1981). The dynamics of labour market segmentation. London/New York: Academic Press.
Part III
Part-Time Employment, Working Conditions and Job Attributes
Chapter 9
Quality of Work: The Case of Part-Time Work in Italy Brendan J. Burchell
9.1 Introduction While the rate of part-time work has been consistent in the EU15 in the period 2000–2005, Italy has seen a large rise in women’s part-time work over that period. The likely cause of this rise in women’s part-time work is presumed to be the Italian enactment of the EU Part-Time Work Directive (Faro 2004). Before 1984 Italian legislation had little provision for part-time work, and the Italian rate was well below the European average. But, the weak 1984 legislation was repealed by the then-conservative government to make way for the new directive adopted in 2000, and subsequently revised. As well as providing a clearer legislative framework on issues such as flexible working by part-timers, part-time work was claimed to offer important solutions to the Italian labour markets including reducing unemployment, increasing women’s employment and enhancing flexibility. Understanding the nature of part-time work is central to understanding current European labour markets, and planning for the labour markets deemed desirable for the future. Part-time work is particularly fascinating for a number of reasons that will hopefully become apparent in this paper, as it cuts across debates on employment, work-life balance gender and families. There is no universally agreed stance on part-time work. Some, perhaps coming from a traditional trade-union background, see part-time work as a form of nonstandard work which is a threat to standard employment rights. Other feminist researchers view part-time work as a ghetto which diverts women from competing with male employees and maintains their lower status in the labour market. B. J. Burchell (&) Department of Sociology, University of Cambridge, Free School Lane, Cambridge, CB2 3RQ, UK e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_9, Ó Springer-Verlag Berlin Heidelberg 2012
175
176
B. J. Burchell
By contrast, some groups have traditionally been strongly supportive of part-time work. ‘Conservative’ or ‘traditional’ views on gendered roles for men and women see part-time jobs (but not full-time jobs) as acceptable for women with child-care responsibilities. Recent concerns to increase work-life balance of families tend to see part-time jobs as having an important role that permits women (and perhaps men too) to maintain their career whilst combining employment with domestic work. And as the male-breadwinner model declines, part-time employment at both ends of the employment spectrum has become an increasingly important feature for students and for older workers making the transition to retirement. And EU, for instance through the Lisbon summit, tends to see part-time work as an important tool to increasing women’s participation in the labour market (Kok 2003). Rather than starting from a ideological position, this paper takes a pragmatic and empirical approach to assess the quality of part-time and full time employment using existing European datasets. But, first, the paper will examine the various definitions of part-time research employed by researchers, and then look at the prevalence of part-time work in the old EU countries.
9.2 Definitions of Part-Time Employment Because there are national differences in the normal weekly hours of work for fulltime jobs, and different institutional arrangements for setting such norms, there is no simple, universally agreed and unproblematic threshold to separate part-time workers from full-timers. Instead, setting such a definition in terms of hours is only one of three definitional frameworks commonly employed. They are: 1. Objective hours. If an objective threshold is to be used, a figure of 30 or 35 h per week is typically used. These are low enough to be below the normal full-time hours in a country like Sweden with norms of short hours, but long enough to classify, for instance, someone working 4 days a week in a high-hours country (such as the UK) as a part-timer. Even then there are problems. For instance, if a part-timer works overtime, is it appropriate that they then become classified as a full-timer, or should it be the contracted hours of work that count? 2. Subjective definition. As the defining features of part-time work vary between countries, hours of work alone may not be sufficient or accurate as a way to define employees. So another approach is to ask them directly ‘‘Are you a full-time or part-time employee?’’. Typically, if they ask for clarification from the interviewer, they may be given some guidance in answering the question, including an hours threshold if they explicitly request it. This is the approach taken in the Labour Force Surveys, but wording varies between countries to reflect local norms. 3. In some cases, it is the contract of employment that is most important to differentiate part-time and full-time employees. This is particularly important where those with part-time contracts are explicitly excluded from employment protection legislation.
9 Quality of Work: The Case of Part-Time Work in Italy
177
A further measurement issue arises if part-time workers are themselves further sub-divided into categories. For instance, some researchers have differentiated ‘‘substantial’’ part-time work from ‘‘marginal’’ or ‘‘mini-jobs’’. Again, a useful threshold for such a division will vary between countries, but 16 and 8 h are often used. Such thresholds can be important, as some benefits and legal protections are denied to employees working very short hours in some countries. Yet another problem arises in the analysis of part-time work: individuals who have several part-time jobs. This will not be dealt with in this paper, but it is a nontrivial issue in some economies. Italy has a relatively low rate of multiple job holders, unlike some of the central and eastern European post-transition countries.
9.3 The Prevalence of Part-Time Work A comparison of the second, third and fourth waves of the European Working Conditions Survey (Table 9.1) permits an analysis of the changing levels of parttime work across Western Europe. Interviewees were asked ‘‘How many hours do you usually work per week in your main paid job?’’ Here the number of hours was split into ‘‘30 and below’’ and ‘‘31 and above’’.1,2 There was a general rise in the level of part-time work in the EU between 1995 (17.4%) and 2000 (21.0%), then a period of stability until 2005 (21.0%). This pattern is similar for men and women separately: women’s rates of parttime work increased from 31.4 to 37% then remained similar on 36.4%, men’s rates in these 3 years were 7.3, 9.1 and 8.7%. But as Table 9.1 shows, the rates for Italian women have increased dramatically in between 2000 and 2005, from 30 to 43%, but the rate for Italian men remained stable at 11%. The same trend, though at different levels, is also evident from comparing part-time contracts in the Italian Labour force data from 1993 to 2005, which also shows a dramatic rise in women’s part-time employment compared to a small rise for male employees (Istat 2007). The other fact evident from Table 9.1 is that the country differences are large, and there is no evidence of convergence over time. The Netherlands and the UK had the highest rates in 1995 and continued to increase, but at the other extreme Portugal started low and decreased. It is beyond the scope of this paper to examine the cause of these Europeanwide changes in the rate of part-time work, but it has been the focus of other enquiries. For instance, Rubery (1998) concludes that where increases have occurred it is ‘more of the same’ rather than employers developing new models of 1
The pattern of results in Italy and the EU is similar if the threshold is changed to 34 h, the pattern is also similar using the European Labour Force Surveys variable that asks respondents directly whether they work part-time or full-time (Fernández 2007). 2 For all of the statistical analyses where we imply that there is a difference between categories, this difference is statistically significant at the 5% level or below.
178
B. J. Burchell
Table 9.1 The prevalence of part-time work (B30 h) Sex of respondent Percentage of female
Percentage of male
Country of interview
1995
2000
2005
1995
2000
2005
Belgium Denmark Germany Greece Spain France Ireland Italy Luxembourg Netherlands Austria Portugal Finland Sweden UK
30 27 27 25 26 25 28 28 26 53 25 16 15 34 48
43 30 40 30 28 31 35 30 44 61 35 21 19 28 47
42 26 30 27 28 27 47 43 34 59 33 12 22 31 54
8 9 4 10 9 9 6 9 3 8 2 7 9 7 9
11 8 7 24 9 8 8 11 3 14 6 5 7 11 8
11 10 6 10 7 7 9 11 7 13 5 5 11 7 13
Source: European Working Conditions Survey (EWCS) Fig. 9.1 Part-time employment in the EU, by gender and age (% of total employment), 2002. Source: Eurostat Labour Force Survey, 2002 (adapted from Corral and Isusi, 2003)
part-time work. But this dramatic shift in Italian women from significantly below to significantly above the EU-15 average seems to have been a direct result of the EU-imposed Part-Time Working Directive. Unintentionally the effect has increased the gender gap in part-time work, from a ratio of about 3:1 to about 4:1 female:male. It is also clear that the function of part-time work for men and women in the EU-15 is different. As can be seen from Fig. 9.1, part-time work for men is concentrated amongst young men and retired men, and still at very low levels for men between the ages of 25 and 64, but for women the rate of part-time work increases monotonically with age. Women’s part-time work is closely associated with childcare, but for men, part-time work occurs at those points in their lifecycle when they are least likely to have childcare responsibilities. As well as a clear association with age, some occupations are also working-time specific. For instance, Table 9.2 shows that women’s part-time employment is very under-represented in managerial jobs, but over-represented in less skilled blue-collar work.
9 Quality of Work: The Case of Part-Time Work in Italy
179
Table 9.2 Full- and part-time employment in the EU, by gender and occupation Occupational status group Men Women
All
Full-time Part-time All Full-time Part-time All White-collar managerial jobs White-collar professional jobs White-collar clerical and service jobs Blue-collar craft and related manual Blue-collar operative/labour manual jobs All
59 44 28 79 57
5 7 4 5 8
64 51 32 84 65
30 29 39 12 18
6 20 29 4 17
36 49 68 16 35
100 100 100 100 100
50
6
56 26
18
44 100
Source: Fagan and Burchell 2002, Table 15
9.4 The Quality of Part-Time Work 9.4.1 Part-Time Employment and Health The 2000 European Working Conditions Survey asks several questions of individuals about their self-reported health, and its relationship to work. Table 9.3 compares several health variables between full-time and part-time employees, separately for men and women. Although none of the differences are great, every one of the measures of health and health-related absenteeism, part-time employees fair better than full-time employees, and this is the case for both male and female employees. So, from this evidence, part-time employment would seem to be the ‘healthy option’. But a note of caution should be sounded before taking this conclusion as unequivocal evidence. When comparing part-time with full-time employees, we are not comparing like with like; there are many other variables such as age and occupation that are also different, as was shown in Fig. 9.1 and Table 9.2. With the exception of gender, in most cases this paper does not attempt to control for these potentially confounding variables, so in this case (as in several others in this exploratory paper) it may be that a multivariate or panel analysis would arrive at different conclusions.
9.4.2 Part-Time Work as Temporary or Unstable Work? In many literatures part-time work is discussed as a form of non-standard work, but figures from the EU15 in 2000 suggests that this is questionable. For instance, if one examines the types of employment contract held by part-timers, it is by no means true that part-time work is always insecure. Figure 9.2 shows that 16% of workers on indefinite contracts are part-time. The situation is very different in
180
B. J. Burchell
Table 9.3 Full- and part-time employment in the EU, by gender and health outcome Men Women Full- Part- All Full- Part- All time time time time Job affects their health in some way 62 Health or safety is at risk because of their job 32 At least 1 day due to health problems caused by work 10 At least 1 day due to other health problems 34 78 Compatibility of working hours with family and other commitments (Percentage of respondents that answer very or fairly well)
55 25 7 29 81
61 31 16 33 78
61 24 12 40 80
53 19 9 34 91
58 22 10 36 84
Source: Fagan and Burchell 2002
Fig. 9.2 Employees part time (spontaneous answer), by type of contract, 2000. Source: Paoli and Merllié, 2001. Adapted from figure 34
some non-EU contries (for instance, Australia) where ‘part-time’ is almost synonymous with ‘casual’, but EU legislation has increased dismissal protection for part-time employees on the grounds that to discriminate against part-timers was a form of gender discrimination. But employment contracts are often only weakly linked to actual labour-market behaviour; do turnover or tenure data suggest that labour markets are associated with higher levels of instability? Table 9.4 below shows again, part-time jobs are again associated with higher labour turnover, but the differences are again small. For instance, 36% of part-time men and 33% of part-time women have been with their current employer for more than 10 years (albeit not necessarily part-time for all of that duration). These figures are not far below the figures for full-timers of 45 and 40% for men and women respectively. Another characteristic of non-standard jobs is that employees often take such jobs involuntarily because of a lack of suitable standard jobs; for instance very few workers would choose a temporary or fixed-term contract if an equivalent job with an indefinite contract were available. Table 9.5 shows the reasons that employees give for working part-time. One should always be sceptical when interpreting employees’ reasons for making choices, as their choices are often heavily constrained by factors such as inadequate child-care or an unequal division of domestic labour. Nevertheless less than 13% of women gave the unavailability of full-time work as a reason for taking a part-time job; for part-time men, the figure was significantly higher at 19%. If one compares rates of involuntary part-time work across countries, the results are quite paradoxical. One might have expected that where the number of part-time
9 Quality of Work: The Case of Part-Time Work in Italy
181
Table 9.4 Employment tenure with company and in present job, by gender and full-/part-time status Men Women
Short tenure Employed in their current job for 1 year or less Employed by the company for 1 year or less Long tenure Employed in their current job for 10 ? years Employed by the company for 10 ? years Average tenure Average number of years in current job (median) Average number of years with the company (median)
Fulltime
Parttime
All Fulltime
Parttime
All
18 18
32 32
19 20 19 20
28 27
23 22
42 45
33 36
42 37 44 40
31 33
35 37
8
4
8
6
5
6
7
4
7
6
4
5
Source: Fagan and Burchell 2002, Table 11
Table 9.5 Main reasons for working on a part-time basis by gender, EU-15, 2002 Main reason Men Women
All
Impossible to find a full-time job Do not want to work full time Involvement in education or training activities Sickness or disability Child and adult care Other reasons No reasons Total
14.1 31.9 10.9 3.1 25.8 11.7 2,5 100
19 31 23.6 5.9 4.2 11.8 4.5 100
12.8 32.2 7.6 2.4 31.5 11.7 2 100
Source: Eurostat Labour Force Survey, 2002 (Corral and Isusi 2003, Table 4 )
jobs was low (for instance in Greece) these posts would be filled with employees who had a strong preference for part-time work, but where the proportion of employees in part-time work was large, then one might expect more women would be forced to take part-time jobs because of a lack of choice. In fact the very opposite seems to be true, that a higher prevalence of part-time work is associated with higher levels of voluntary part-time work. For instance, at the two extremes, involuntary part-time work is almost non-existent in the Dutch case (1.9%), but very prevalent in Greece (42%). And Fig. 9.3 shows that between these two extremes is a very strong linear relationship. There are many things that could give rise to such a relationship. For instance, it could be that in those countries where part-time rates are low, this is for a good reason because part-time jobs are, for legal or historical reasons, of very poor quality, so will tend to be taken only when alternatives are not available. But nevertheless, it does suggest that the greater the familiarity of a society with parttime employment, the more they are accepted as desirable jobs.
182 50.00
Involuntary part-time work
Fig. 9.3 Rate of involuntary part-time work, 2002 plotted against rates of women’s part-time work for EU15 countries. Source: Figures taken from Labour Force Surveys, Table 4 in Corral and Isusi 2003
B. J. Burchell
Greece
40.00 Finland
30.00
Italy
Sweden Portuga
20.00
France Denmark
Spain
Belgium
Germany Ireland
10.00
Austria
UK
Netherl
0.00 0.00
20.00
40.00
Rate 2002
60.00
80.00
9.4.3 Job Satisfaction The European Working Conditions survey has separate questions about satisfaction with working conditions and satisfaction with work-life balance. In Table 9.3, one can see that female part-timers are more satisfied with their ‘compatibility’ of working hours with family and other commitments, with 9% of part-time women expressing dissatisfaction compared to 20% of full-timers. For men the difference is in the same direction, but much weaker. And Table 9.6 shows that, in the more general question about job satisfaction, part-timers expressed more satisfaction with their work than full-timers, and the effect was again stronger for women than for men. The social psychological literature on job satisfaction advises us to be cautious about interpreting job satisfaction data too literally. In answering such questions, respondents are making an implicit comparison, and therefore employees with lower expectations about other jobs will express higher satisfaction with their current job. For instance, we know that on many ‘objective’ criteria, men tend to have better jobs than women, yet as is commonly found, and is replicated in Table 9.6, women are more likely to express satisfaction with their jobs than men. There are several ways of getting beyond this ‘superficial’ satisfaction (Agassi 1982). For instance, one way is to ask: ‘‘Would you be pleased if your son [daughter for females] were to have a job like yours?’’. This question then gives responses which resemble more closely the objective quality of jobs. Another approach would be to get those employees who have held both fulltime and part-time jobs to choose which of their jobs they liked best. This question asks them to explicitly compare real jobs, rather than to make a hypothetical judgement about a job.
9 Quality of Work: The Case of Part-Time Work in Italy
183
Table 9.6 Satisfaction with working conditions, by gender and full-/part-time status Men Women
All
Very satisfied Fairly satisfied Not very satisfied or not at all Satisfied Total
Full-time
Part-time
All
Full-time
Part-time
All
23 59 18
25 56 19
23 59 18
24 59 17
28 57 15
25 58 16
24 59 17
100
100
100
100
100
100
100
Source: EWCS, 2005
One survey that permits this comparison to be made is the 1986 Social Change and Economic Life Initiative dataset (Burchell et al. 1997). In this survey, of 1,022 UK adults aged between 20 and 60, they were asked about every job that they had held since leaving full-time education. After they had listed them all in chronological order (with the assistance of the interviewer) they were asked ‘‘[of] all the jobs you’ve ever had, which one was the job you liked best?’’ Just looking at the data for women, one-third selected, as their best job, one that had been part-time. This was the same proportion of all the jobs in the women’s work histories that were part-time. Again, this is not unequivocal data, but it does again suggest that, for women, part-time jobs are no less satisfying than full-time jobs. Yet another suggestion that individual preferences favour part-time more than full-time work comes from surveys of working time preferences (Bielenski et al. 2002; Fagan 2003). When employees are asked how many hours a week they would like to work (for the same hourly rate of pay) full-timers tend to say that they would prefer to work slightly less than full-time (part-timers tend to say that they would prefer to continue working part-time, but a few more hours a week).
9.4.4 Community-Friendly Work In order to consider the quality of different forms of employment, one can go beyond simply looking at the benefits for the individual employee. For instance, one topic that has generated much discussion is the effect of women’s employment on the educational attainments of their children (O’Brien and Jones 1999), with controversial claims that children of mothers working part-time had higher levels of qualifications at the age of 16 than children whose mothers worked full-time. An emerging topic is the impact of employment types on communities, using outcome variables such as rates of community activities or levels of social capital. Putnam (2000) hypothesised that women out of the labour market might be an important social-capital resource for societies, as they have more time for voluntary organisations. He therefore argues that the declining stocks of social capital over the twentieth century might have been caused, in part, by rising female participation in the labour market. But no empirical evidence for this hypothesis was found by Putnam for the US or by Hall (1999) for the UK.
184
B. J. Burchell
But if one examines the rates of voluntary and community activities in the European Working Conditions Survey (EWCS), there is evidence that employees in part-time jobs have slightly higher levels of these types of civic participation, and this effect is maintained in multivariate models when other variables, such as children and occupation are controlled for (Kamerade and Burchell 2004).
9.4.5 Part-Time Work and Career Development In considering the quality of a job, one needs to consider not only the present but also the future; for instance does the job have training and promotion prospects, or is it a dead-end job or a ‘‘trap’’. This is a complex question that can be considered from a number of perspectives. For instance, other research has suggested that part-timers are not considered as seriously for promotion as full-timers. For some managers there may be a perception that part-time work and the supervision of others are mutually incompatible. For other managers, it is assumed that parttimers have a low commitment to the organization, and are therefore unsuitable for human capital investment through training for more responsible positions. There is also another way to approach the relationship between employment and career development. Kohn and Schooner (1983) suggests that some jobs are challenging and lead to cognitive development or learning, which facilitates career progression and an upwards spiral. In contrast, mundane jobs do not present the opportunity for learning and lead to stagnation. Table 9.7 shows that this may well be the case with part-time jobs. They are less likely to be described as providing opportunities for problem-solving or learning than full-time jobs, and more likely to be described as monotonous or as not complex. Table 9.7 also shows that part-timers are slightly less likely than fulltimers to have received training through their employer in the past year. Parttimers are more likely to describe their job as being below the skill level that they are capable of, and they are only about half as likely to have supervisory responsibility compared to full-timers. It is perhaps puzzling that jobs described as mundane and below the skill level of the employee should also have high levels of job satisfaction. One possible explanation is that many part-timers have very challenging and intensified lives outside of work, with many other responsibilities including housework, childcare or part-time study. Perhaps one way of reducing the stress in the case of these dualrole or multiple-role lives is to take a job that is easily within one’s physical or mental capacity? Economists tend to use pejorative terms to describe such jobs as ‘‘dead-end’’ or ‘‘traps’’, but for the incumbents. they may experience such jobs as being a liberating, permitting them to gain some of the social and economic benefits of employment where the only realistic alternative for them, within the current institutional arrangements, would be non-employment and social exclusion.
9 Quality of Work: The Case of Part-Time Work in Italy
185
Table 9.7 Characteristics of job content and % of workers receiving training, by gender and full-/part-time status Men Women
Problem-solving No problem-solving or learning Some problem-solving or learning Both problem-solving and learning Task complexity Monotonous tasks, no complex tasks Both monotonous and complex tasks Neither monotonous nor complex tasks Complex tasks, no monotonous tasks Matching of skills with job demands The demands of my job are too high for my skills The demands of my job match my skills The demands of my job are too low for my skills % who have supervisory responsibilities % who received no training from their employer in the last 12 months
Fulltime
Parttime
All Fulltime
Parttime
All
9 24 67
11 33 56
9 11 26 26 65 63
15 31 54
12 29 59
16 25 21 38
25 18 30 27
17 24 22 37
18 23 26 33
25 14 35 26
21 19 30 30
9 85 6 25 69
5 81 14 12 74
85 8 7 24 69
9 84 6 17 66
5 85 10 8 72
85 8 8 13 69
Source: Fagan and Burchell, 2002 , Tables 16 and 18
Further considerations of the effects of part-time work on women’s longer-term careers have been considered by several authors (e.g. O’Reilly and Bothfeld 2002 and Connolly and Gregory 2005). Such analyses are complex, and will differ between different countries. Connolly and Gregory’s 20-year longitudinal analysis of the UK 1958 birth cohort suggests that it is useful to hypothesise that there are two different types of career trajectories for women that use part-time work. For some women, more attached to the labour market, they make the transition from full-time to part-time work as a way of coping with domestic work, but are then likely to return to full-time employment. But for other women, they enter part-time work from economic inactivity, and are then likely to leave the labour market again when they terminate their part-time job. O’Reilly and Bothfeld (2002) use a transitional labour market theoretical framework to analyze the conditions that would permit part-time employment to simultaneously benefit employees, reduce labour market inequalities and to promote flexibility for both employers and employees.
9.5 Conclusions If one starts from the premise that part-time work is a non-standard form of work forced upon employees, mainly female employees, as a way of achieving flexibility or depriving the workforce of their rights, then the evidence presented here
186
B. J. Burchell
makes surprising reading. Using many standard, and a few innovative, measures of quality of employment, one finds that part-time work, more often than not, seems to be superior to full-time employment (and this is generally true also for the small minority of men who work part-time). This paper concurs with the conclusions of other research that one of the main drawbacks of part-time work is that it is less oriented towards an upward career trajectory than full-time work. But even here, the differences in measures of training and stimulating job content between full-timers and part-timers are not great. Furthermore, these differences are certainly not intrinsic to differences in hours of employment, so there is scope for interventions to improve the career prospects of part-timers. The incentives on employers to do this will depend on their reasons for using part-time contracts. If it is simply to cut costs, then it might be an uphill struggle to improve personal development opportunities for parttimers. But increasingly employers are using part-time work to retain valuable staff and to keep their skills current through periods when they otherwise might not be employed, such as when women have caring responsibilities for pre-school children. There are two other drawbacks that are not discussed in greater detail in this paper, but that also need to be taken into account if one is to give a wider assessment of the quality of part-time work: wages and pensions. Some researchers have reported that, controlling for the type of work and levels of human capital, part-timers are paid a lower hourly wage than full-timers. But this wage-penalty is not consistent across European countries, being negligible in some countries, but large in others. Secondly, there may be longer term disadvantages of working part-time for a significant proportion of a career. In welfare systems where the pensions are set relative to contributions made over a lifetime, part-time employment may increase the risk of poverty in retirement. Again, it depends what one is comparing it to; if the alternative is unemployment or non-employment, then part-time employment will be an improvement, but compared to full-time employment, part-timers will again suffer a delayed cost. One needs to be very cautious in generalizing about part-time employment in the EU when the function and rates of part-time work vary so greatly between the member states. Taking the evidence in this paper together, a quality-of-work perspective finds much to commend part-time employment as a method of combining labour-market activity with domestic work or other non-labour market activities. But taking a lifecycle perspective, the picture is more mixed, with evidence that it impedes career progression and perpetuates gender inequalities. The increase in part-time employment for Italian women needs to be monitored carefully; a short-term increase in the quality of life for Italian women and Italian families might be followed by an increased gender gap as women’s career advancement slows down, so increasing vertical gender segregation. The relatively high rate of involuntary part-time employment in Italy should be of some concern, and suggests that the quality of part-time work in Italy may still be lagging behind other western European countries Or can regulatory mechanisms be found to
9 Quality of Work: The Case of Part-Time Work in Italy
187
minimize the negative impact of part-time work on women’s careers? Given the trend towards deregulation, and the fact that the increase in part-time work in Italy has only really affected women, perhaps there is a genuine cause for concern that part-time employment is a move away from gender equality in Italy. Acknowledgements I would like to thank Catherine O’Brien for assistance with the data preparation and Tindara Addabbo and Giovanni Solinas for valuable feedback on earlier drafts. I would also like to thank the Aiel for the conference invitation that lead to this paper.
References Agassi, J. B. (1982). Comparing the work attitudes of men and women. Aldershot: Gower. Bielenski, H., Bosch, G., & Wagner, A. (2002). Working time preferences in sixteen European Countries. Dublin: European Foundation for the Improvement of Living and Working Conditions. Burchell, B. J., Dale, A., & Joshi, H. (1997). Great Britain. In P. Blossfeld & C. Hakim (Eds.), Between equalization and marginalization: Women working part-time in Europe and the United States. Oxford: Oxford University Press. Connolly, S., & Gregory, M. (2005, September). Part-time work–a trap for women’s careers? an analysis of the roles of heterogeneity and state dependence. University of Oxford, Department of Economics, discussion paper, 245. Corral, A., & Isusi, I. (2003). Part-time work in Europe. Dublin: European Foundation for the Improvement of Living and Working Conditions. Fagan, C. (2003). Working-time preferences and work-life balance in the EU: Some policy considerations for enhancing the quality of life. Dublin: European Foundation for the Improvement of Living and Working Conditions. Fagan, C., & Burchell, B. J. (2002). Gender, jobs and working conditions in the European Union. Dublin: European Foundation for the Improvement of Living and Working Conditions. Faro, A. L. (2004). Italy: Adaptable employment and private autonomy in the Italian reform of Part-time work. In S. Sciarra, P. L. Davies, & M. R. Freeland (Eds.), Employment policy and the regulation of part-time work in the European Union: A comparative analysis. Cambridge: Cambridge University Press. Fernández, E. (2007). The plausibility of the results of the 4th EWCS a reality check against previous surveys and the LFS. Dublin: European Foundation for the Improvement of Living and Working Conditions (Mimeo). Hall, P. A. (1999). Social capital in Britain. British Journal of Political Science, 29(3), 417–461. Istat (2007). Le statistiche di genere, Istat, Approfondimenti, March, 7, 2007 from (http://istat.it/ salastampa/comunicati/non_calendario/2007307_00/). Kamerade, D., & Burchell, B. J. (2004). Teleworking and social capital: Is teleworking an isolating or a community-friendly form of work? European Sociological Review, 20(4), 345–361. Kohn, M. L., & Schooner, C. (1983). Work and personality: An inquiry into the impact of social stratification. New Jersey: Ablex Publishing Corporation Norwood. Kok, W. (2003). Jobs, jobs, jobs: Creating more employment in Europe. Report of the Employment Taskforce, European Communities. O’Brien, M., & Jones, D. (1999). Children, parental employment and educational attainment: An English case study. Cambridge Journal of Economics, 23(95), 599–621. O’Reilly, J., & Bothfeld, S. (2002). What happens after working part-time? integration, maintenance or exclusionary transitions in Britain and Western Germany. Cambridge Journal of Economics, 26(4), 409–439.
188
B. J. Burchell
O’Reilly, J., & Fagan, C. (Eds.). (1998). Part-time prospects: An international comparison of part-time work in Europe, North America and the Pacific rim. London: Routledge. Putnam, R. D. (2000). Bowling alone: The collapse and revival of American Community. New York: Simon and Schuster. Rubery, J. (1998). Part-time work: A threat to labour standards. In J. O’Reilly & C. Fagan (Eds.), Prospects: An international comparison of part-time work in Europe, North America and the Pacific Rim (pp. 137–155). London: Routledge.
Chapter 10
Does Part-Time Mean Part-Satisfaction? Sara Depedri
10.1 Introduction The increase in part-time jobs has characterized many European countries in recent decades. This trend is mainly explained by the evolution of the regulation of labour markets. Countries have promoted various forms of labour flexibility, both with specific legislation and through deregulation and increase in the parties’ autonomy. The principal purpose of rendering the labour market more flexible has been to respond to new organizational needs and competitive pressures. On the other hand, flexible employment has the advantage of developing new job opportunities and reducing the level of unemployment. Moreover, the tendency to use part-time contracts is increasingly explained by changes in the demand and supply of labour. On the one hand, organizations require job flexibility, and they use part-time contracts in order to satisfy specific production needs, increase production and productivity, and compete on the goods market. On the other, workers may want fewer working hours in order to meet family needs and other personal commitments. Furthermore, part-time contracts should cater to a specific area of the labour supply, where the participation constraint and the individual supply of labour are characterized by lower numbers of hours supplied. In all these cases, part-time contracts are efficient because they enable both parties to achieve their goals (i.e. needs and constraints) and to maximize their (also social) well-being. However, this positive view becomes controversial when part-time relationships match the needs of only one contractual party. Part-time may be wanted by workers but not be in the interest of the organization; conversely, the principal may impose a reduction in working hours to the detriment of workers’ expectations, S. Depedri (&) Department of Economics, University of Trento, via Inama 5, 38100 Trento, Italy e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_10, Ó Springer-Verlag Berlin Heidelberg 2012
189
190
S. Depedri
needs, and incomes. Furthermore, when the working hours of employment relationships are shortened, there may be negative consequences on both the perceptions of workers and their performances. Linked to this last aspect, part-time work may also reduce human capital accumulation, even when it is voluntary. The final effect of part-time contracts is therefore controversial and should be analysed in terms of its various possible impacts on the parties’ well-being. This chapter evaluates part-time workers’ well-being through analysis of parttime jobs in the social services sector. This sector of activity is of interest for two reasons. Firstly, workers’ performances depend not only upon the amount of hours worked, but also and mainly upon the effort devoted to job quality. Thus, workers’ effort should be judged in terms of relationships with clients, involvement in the activity and in the mission of the organization. In this context, the provision of a standard economic contract (or better, a transactional contract as defined by Rousseau 1995) is insufficient because most aspects of the job and of the employment relationship must be explained in psychological terms. This is clearly the case of the emergence of psychological contracts, which meet the needs of organizations and workers on intrinsic aspects, relationships, perceptions, and beliefs, and which commit the parties outside the pure contractual aspects of the job. The other salient feature of the sector analyzed here is the composition of its workforce. The sector is characterized by a large number of women and a high percentage of part-time workers. The latter aspect permits the development of analyses in which part-timers represent a significant portion of interviewees and yields understanding of whether problems in the treatment and in the satisfaction of part-timers compromise the performance of organizations and the sector itself. The prevalence of women instead allows one to judge whether part-time jobs satisfy their specific need for time to meet familial needs, or rather whether they respond to organizational needs and dissatisfy both females and males. The aim of the paper is to demonstrate by empirical means the main factors influencing part-time workers’ satisfactions, doing so by investigating their perceptions about their jobs, internal fairness, and the quality of the work. For these purposes data from a survey conducted in Italy in 1998 on a sample of organizations delivering social services will be used. Questionnaires were administered to 228 organizations, 266 managers, 2,066 paid workers and 724 volunteers (Borzaga 2000). Fifty-four of the organizations surveyed were public, while the others included both for-profit (17) and non-profit (157) organizations, mostly of small size. Most of the 2,066 employees were women, and belonged to the middle age group (more than two-thirds were aged between 30 and 49). More than one-third of the workers had upper secondary school leaving certificates, and 16.5% were graduates. Tenure in the organization was generally rather short (however, to be noted is that one-third of organizations had been in existence for less than 10 years). The questionnaire for employees contained questions on workers’ satisfaction, with regard not only to their work as a whole but also to some of its aspects which can be grouped into intrinsic, extrinsic and relational. Furthermore, specific questions were asked in order to investigate workers’ attitudes and motivation,
10
Does Part-Time Mean Part-Satisfaction?
191
perceptions of fairness, and perspectives on loyalty to the organization. The following sections concentrate on these aspects, besides personal and professional characteristics, in order to understand the differences between full-time and parttime workers and to support the hypothesis put forward in the literature concerning part-timers’ well-being and role within organizations.
10.2 The Controversial Evaluation of Part-Time Employment Part-time employment has been studied along two main dimensions. From a macroeconomic perspective, part-time employment is interpreted as a way to reduce unemployment and to increase labour-market flexibility. The analysis focuses on international comparisons and evaluates interrelations among the use of part-time contracts, labour markets, and goods markets. On the other hand, microeconomic analyses seek to identify the determinants of part-time work (Houseman 2001) and the consequences of flexible jobs (Barrett and Doiron 2001). Close to this view, part-time jobs are evaluated in their ability to meet the needs of organizations and workers. On the demand side, organizations estimate the convenience to them of hiring part-time workers by comparing costs of investments in the selection and training of workers with the benefits accruing to their productivity, and with the possibility of responding to demand peaks for their products or specific requirements of the production cycle (Booth and Wood 2008). Furthermore, organizations are attracted by part-time contracts according to the national tax regime, the general regulation of part-time work, and specific agreements with trade unions on wage levels and other rules and protections. The supply side is instead characterized by individuals with heterogeneous needs and preferences. Consequently, part-time jobs should respond to specific needs of workers and satisfy this (marginal) part of the supply of working time. At the same time, part-time jobs (as in general the excessive use of flexible contracts) may negatively impact on the expectations and perceptions of both employers and employees (Buddelmeyer et al. 2004). For workers, part-time contracts may be imposed by the organization and must consequently be treated as involuntary part-time. As a consequence, workers not only receive lower incomes but are also dissatisfied with their working conditions. Furthermore, part-time work is often associated with low career chances, accumulation of human capital, fringe benefits, and other non-wage rewards. Workers may therefore perceive their jobs as less secure, with negative consequences on their lives: the employment relationship restricts workers’ expectations and decreases their opportunities, employees are less prone to invest in their futures, to leave their families of origin, get married, or have children (Fullin 2004). This is due to an increase in the perception of risk linked with sociological investments, the overlap between the duration of work and life perspectives, and mainly the limitations on freedom to fulfil personal ‘capabilities’— i.e. ability to achieve the desired standard of life (Sen 1993).
192
S. Depedri
The main problems of part-time employment are thus related to involuntary part-time work, unfair treatment compared with full-time jobs, and unexplained penalizations with regard to career prospects and opportunities for vocational training. The consequences of these perceptions are both direct on workers’ wellbeing (i.e. level of satisfaction or utility achieved on the job) and indirect on their performances. With regard to this last point, workers may perceive themselves as less involved in the organization and not adequately rewarded and recognised. Consequently, they do not internalize the organization’s goals and mission in their decision-making processes; they reduce their effort in response to a perception of unjust behaviour by the principal; and finally they may negatively influence the quality or the quantity of organizational outcomes. For organizations, the costs thus consist not only, as said, in specific investments for recruiting and training workers but also in the inability to control and adequately incentivize part-time workers. In order to achieve increases in productivity and better organizational performance, organizations must devise human resources policies able to maximize their part-time workers’ satisfaction, involve them in the organizational process, and improve internal fairness. Policies of training rather than rewarding are a crucial aspect. Furthermore, organizations must also develop complete psychological contracts with all their workers. As defined by Rousseau (1995), a psychological contract consists in the perceptions of the two parties regarding their mutual obligations. These obligations arise from psychological and subjective perceptions, are mainly seen as ‘promises’ or as ‘expectations’, and are often informal and imprecise. This is because the psychological contract mainly depends upon the interpretations and perceptions of the actors, taking into account the other’s actions, the fairness of the treatment, and the transparency of information. When one or more of these elements are lacking, the psychological relationship is compromised and the behaviours of the parties may deteriorate. And this risk is higher when the duration of the employment relationship is shortened, because workers have fewer exchanges with employers, colleagues, and the working environment itself. The contraposition of positive and negative aspects of part-time employment is therefore crucial. It should be evaluated by considering parameters relative to both the effective treatment received by part-time workers and their perceptions and psychological considerations. Furthermore, analysis should evaluate the impact on both workers’ well-being and performances or the quality of the job. Finally, it should determine the ability of part-time contracts both to respond to workers’ needs for working time flexibility and to yield organizational comparative advantages.
10.3 General Characteristics of Part-Time Workers The majority of interviewees were full-time workers, but also the percentage of part-timers was significant (20.6%, equal to 422 cases), a little higher in private than in public organizations (respectively 24 and 13.4%), and certainly larger than
10
Does Part-Time Mean Part-Satisfaction?
193
in other sectors and in the Italian labour market in general. In order to understand the characteristics of the sample better, it is useful to compare part-timers with full-time workers (as in Table 10.1). This allows evaluation of whether workers employed on part-time contracts have different personal characteristics that explain the different needs of workers or the policies of organizations. Firstly, such comparison confirms that females are more frequently employed on part-time contracts (22.1 of females and 15.7% of males are employed part-time), but the percentage of part-time males is neither close to zero nor significantly lower than that of full-time workers (17.5 against 24.4%), given the characteristics of the sector. The number of part-time workers is significantly higher for workers with a degree (30.2%) and the distribution of part-time workers by level of education progressively increases in comparison with the distribution of full-time workers. The propensity of organizations to hire workers on part-time contracts and then convert them into full-time relationships is confirmed by the higher percentage of full-time workers employed in the organization for more than 10 years (25 against 11.9% of part-time workers) and the lower percentage employed for fewer than 3 years (respectively 45.9 and 28.9%). Secondly, in regard to workers’ age, part-time contracts are applied to youngest workers in 28.2% of cases, while only 23.4% of full-time workers are less than 30 years old. Moreover, even when one inspects the entire sample, no difference emerges among workers with different civil statuses. On distinguishing males from females, the data show that married males are employed full-time in 91.5% of cases, while part-time contracts are more frequently applied to singles (24.7%). Conversely, females are employed part-time at a higher percentage when they are married (24.7%) than when they are single (19.3%). These results confirm the hypothesis that part-time contracts are mainly chosen by females wanting to meet family needs, while among males only singles may accept this reduction of income (other than working hours). Another explanation for choosing part-time jobs is the decision to engage in other activities, specifically external voluntary work: the percentage of workers who are also volunteers is in fact somewhat higher among part-timers (given their greater free time) (22 against 16.2% for full-time workers), and this difference increases further among males. An analysis of the principal characteristics of workers’ positions in the organization and in the labour market (Table 10.1) shows that, firstly, workers employed part-time are also more frequently employed on temporary contracts: only 53.9% of part-timers are employed on open-ended contracts (81.1% of fulltime workers), while 29% have fixed-term contracts.1It thus seems that
1 Italian labour law has in recent years introduced a series of flexible forms of employment which in general can be called ‘contingent contracts’. In regard to the specific question used in the questionnaire, it is possible to distinguish between workers on fixed-term contracts and employer-coordinated freelancers. The former are employed as subordinate workers, while the latter have para-subordinate contracts, i.e., are similar to self-employed workers. Nevertheless, given that their roles in the organization and the temporary nature of the contracts are very similar, in what follows they will be considered jointly as temporary workers.
194
S. Depedri
Table 10.1 Principal characteristics of part-time workers (per cent) Variable Values Type of organization Gender Educational level
Age
Tenure in the organization
Civil status
Voluntary work
Contractual form
Area of activity
Received training
Previous employment status
Wage
Public Private Male Female Compulsory schooling Technical education High-school diploma University degree or higher Less than 25 years old 25–30 30–39 40–49 More than 50 years old Fewer than 3 years 3–5 years 5–10 years More than 10 years Single Married Divorced Widowed Not currently Yes, occasionally Yes, continuously Open-ended contract Job training Fixed-term contract Freelance Other contract Co-ordination Supply of the service Accounting Human resources Management Complementary activity Other No training Short-term training Long-term training Employed Unemployed Student Housewife Retired person Other Monthly Hourly
Full-time
Part-time
86.6 76.0 24.4 75.6 25.9 20.7 38.8 14.6 5.8 17.6 39.1 26.4 11.1 28.9 21.2 24.9 25.0 31.6 58.0 8.0 2.4 83.8 9.2 7.0 81.1 1.6 7.5 2.2 7.5 9.8 74.2 6.3 2.4 10.0 7.0 45.5 32.4 22.2 43.5 27.6 10.5 9.7 0.2 8.5 832.21 5.77
13.4 24.0 17.5 82.5 31.1 12.1 33.7 24.3 6.8 21.4 42.2 21.1 8.5 45.9 23.0 19.6 11.5 32.2 60.4 6.0 1.4 77.0 10.2 11.8 53.9 3.6 17.1 11.9 13.5 8.3 71.9 5.7 3.6 11.7 12.9 50.0 31.4 18.6 31.7 26.9 15.8 13.9 1.4 10.3 551.30 6.39
10
Does Part-Time Mean Part-Satisfaction?
195
organizations offer flexible contracts both and simultaneously in terms of duration and timing as forms of first-employment or first-hiring in the organization. Nevertheless, this may increase the risk that part-time jobs are perceived by workers, not as meeting their needs, but mainly as decreasing job security and as responding to the preferences of the principal. However, no discrimination between full-time and part-time workers emerges in the assignation of roles and the provision of training. Workers on part-time contracts are, in fact, employed in coordination and responsibility activities in percentages very close to those of full-time workers. Differences emerge only when the analysis distinguishes according to certain characteristics of workers and the job. The presence of part-time workers employed in the direct provision of services is higher (84.6%) in public organizations, which instead reserve roles of coordination and management for full-time workers; private organizations record similar percentages of full-time and part-time workers in different roles and activities. As to gender, while females are similarly employed in different roles when employed on part-time or full-time contracts, the percentage of males with coordination and management roles decreases when they are employed part-time (8.2% of the sample, against 17.9% of full-time workers). Consequently, it is possible to claim that the social services sector and specifically non-profit organizations differ from the average treatment of part-time workers in the labour market. They offer more career opportunities; or at least they do not specifically discriminate between part-timers and full-timers when offering job opportunities. The only differences in workers’ roles seem to be explained by the involvement of part-time (mainly male) workers in initial training, without ruling out future career advancement. Nor does training significantly discriminate against part-time workers, since organizations tend to invest only slightly more in preparing their full-time workers (45.5% of them had received no training in the 3 years preceding the interview, and this percentage was 50% for part-time employees). Nevertheless, all workers enjoy a similar level of short-term vocational training (close to 32%). Differences in training penalize part-time workers in private organizations (who receive no training in 51.7% of cases) to a greater extent, while public organizations provide specific training for most of their part-time workers (60.8%). Vocational training is also reduced for males employed on part-time contracts (50.7% without training in the 3 years considered) in comparison with their full-time colleagues (who received internal training in 61.4% of cases). But the most evident difference emerges in regard to graduates: 43% of graduate part-timers do not receive on-the-job training, while this percentage is only 23.5% for full-time workers with a degree, who instead received long-term training in 36.4% of cases. Finally, pure economic treatment seems on average not to discriminate between full-time and part-time workers: while monthly wages are obviously proportional to the number of hours worked, the average hourly wage of part-timers is slightly higher than that of full-timers. The explanation for this is not the better general treatment of part-time workers, but the presence of two classes of part-timers. The first consists of workers with lower educational levels (more than 50% with
196
S. Depedri
compulsory schooling), of the youngest or oldest age classes (aged under 25 or over 50), and just employed in the organization or with little experience (35% of workers with fewer than 3 years of tenure in the organization). This class is characterized by low wages (9.7% of part-time workers earn less than 4 Euros per hour, compared with 6.3% of full-timers). The second and opposite class consists of workers with degrees (30% earning more than 10 Euros per hour) and belonging to the middle-age class (from 30 to 40 years old), who earn salaries also significantly above the average (more than 20% above 7 Euros per hour, compared with 7% of full-timers). The wage structures are thus equal in terms of how they reward both specific education and training and tenure in the organization. No significant differences affect salaries by gender, even if the average data cannot exclude wage gender discrimination and must therefore be further analyzed. The linkage between organizational form and wages is clearer, because private (specifically non-profit) organizations offer lower salaries for part-time workers (with pay levels very close to those applied to full-time workers, in order to select intrinsically motivated people). Turning to previous employment conditions, part-time workers are in a higher percentage than full-timers, students or housewives (respectively 15.8 and 13.9%), while the percentage of part-time workers with previous job experiences is lower (31.7% against 43.5% of full-time workers). These few cases comprise a significant percentage of self-employed people (24.2%, while full-time workers with previous work experiences were subordinate employees in 84% of cases)with low percentages of part-time workers with experiences as blue or white collars. The explanations for job changes differ significantly between part-time and fulltime workers: the former have principally (and more than their full-time co-workers) accepted a new job because it offered an opportunity to meet family needs better (average score 4.42 on a scale from 1 to 7), while the wage was certainly not an explanation for the change (average score 2.51). The importance of having a more flexible job accounts for the change especially among married people (assigned relevance of 4.6), but also among the youngest workers (4.8 among workers aged under 25). However, part-time jobs are not chosen solely for their reduced working hours: part-time workers declare themselves greatly attracted by the work as well, and specifically by the opportunity to help other people and share the organization’s mission (respectively 4.63 and 4.54). These data demonstrate that part-timers are frequently influenced in their employment choices by time constraints. Nevertheless, they also conscientiously choose their job, and over time they tend to create a strong psychological contract and a positive sharing of organizational goals. Comparison between current and previous jobs shows that many workers currently employed on part-time contracts admit that it has significantly improved the organization and flexibility of working time (65.6% of cases, which decreases to 52.5% for full-time workers), even at the expense of the final wage (for 28.6% of part-time workers lower than in the previous job). Furthermore, the shortening of working time and of the employment relationship
10
Does Part-Time Mean Part-Satisfaction?
197
has worsened neither the level of relationships with colleagues and superiors nor moral satisfaction. Interestingly, when the interviewees were asked to judge their motivations and attitudes towards work, their general perceptions did not vary between workers employed part-time and full-time. They were similarly and mainly interested in work as an opportunity for self-fulfilment, but they also thought that was a necessity and a way to earn a living for themselves and for their families. Analysis of the specific aspects of the job that attracted workers shows instead that part-time workers had chosen their jobs mainly (and more than full-time workers) to meet other personal interests and family needs (average score of 4.09), whilst job security was less attractive (though on the contrary it was very important for fulltime workers, with average scores of respectively 2.57 and 3.34). Other aspects of the job were evaluated similarly by full-time and part-time workers. Some attitudes towards the work and the organization differ when the personal and professional characteristics of part-time workers are considered. Females are significantly more attracted than males by the consistency of the job with their educational level and professional training, the opportunity to meet other familial commitments, and by the stability of the job. These preferences and differences in the needs of males and females, which reflect different motivations for choosing the job and the organization, are also typical of the personal family situation. When the data are separated for married people and singles, it emerges that, in regard to some aspects of the job, differences between the two genders increase. Specifically, interest in the sector of activity is the most important aspect for singles (especially women, who assign it a score of 4.98); matching educational level and personal training mainly motivated married people (especially men, with an average score of 3.48 in comparison with 2.62 of singles); flexibility and meeting familial needs equally attract married men and women (4.31 with respect to 3.66 of singles); involvement in the organization is more interesting for singles than for married people, consistently with their need to dedicate more time to their families than to their jobs. Finally, a lack of other job opportunities partly explains the decision of several singles (and the youngest respondents) to accept the job.
10.4 Partial Satisfaction? What are the consequences of differences between part-time and full-time jobs on workers’ satisfaction? Assuming satisfaction with the job as a whole as the dependent variable, a first multinomial model evaluated the impact of the personal and professional characteristics of workers on their well-being. The differences found between full-time and part-time workers (Table 10.2) are very significant. The job satisfaction of the former significantly depends upon their civil status (since both married people and singles are less satisfied than are divorced and widowed persons), their level of education (where
198
S. Depedri
Table 10.2 Job satisfaction and workers’ characteristics (ordered probit model) Full-time Part-time (Constant) Gender Married Single Diploma Degree Specific training Age Tenure in organization Coordination Supply of the service Open-ended contract Collaboration Short-term training Overtime Log number of workers Public–private nature Hourly wage
Std. coeff. t
P(t)[|t| Sig. Std. coeff. t
1.2613 0.0600 20.1759 20.1839 0.3015 0.4490 0.2313 20.0092 20.0077
2.313 0.835 21.946 21.719 3.752 3.760 3.129 22.282 1.814
0.0207 0.4039 0.0517 0.0855 0.0002 0.0002 0.0018 0.0225 0.0697
0.4918 -0.0794 0.0541 0.0112 0.1021 0.1508 0.0651 -0.0025 20.0162
0.535 -0.524 0.262 0.051 0.688 0.838 0.468 -0.394 2.108
P(t)[|t| Sig. 0.5928 0.6002 0.7935 0.9597 0.4917 0.4018 0.6395 0.6933 0.0351
20.4919 -0.0800 0.0691 -0.2527 -0.0950 -0.0747 20.0011 0.3248 0.0000
23.182 –1.177 0.915 –1.273 –1.488 –1.113 23.334 5.107 1.110
0.0015 0.2392 0.3604 0.2030 0.1367 0.2656 0.0009 0.0000 0.2668
-0.3715 -0.1268 -0.1218 -0.2401 -0.0233 -0.1155 20.0016 0.1244 0.0000
-1.201 -1.054 -0.918 -1.358 -0.199 -0.784 22.878 0.79 -0.222
0.2298 0.2917 0.3585 0.1746 0.8427 0.4329 0.0040 0.4294 0.8245
Model FT: Observations: 1,503; Iterations completed: 25; Log likelihood function: -2295.13; Restricted log lik.-2338.1; Chi-squared: Degrees of freedom: 17; Sign. level: 0.000 Model PT: Observations: 402; Iterations completed: 24; Log likelihood function: -611.21; Restricted log like. -622.84; Chi-squared: Degrees of freedom: 17; Sign. level: 0.000
people with degrees or diplomas are less satisfied), and their age (positively correlated to job satisfaction). Furthermore, full-time workers are more satisfied when employed in coordination roles. Job satisfaction instead decreases with tenure in the organization, in large and in public organizations. On the other hand, the level of job satisfaction declared by part-time workers is correlated only to their tenure in the organization (where just-employed are more satisfied) and the size of the organization (negatively correlated with job satisfaction). An ordered probit model on workers’ motivations in the choice of the organization and attitudes toward work yielded understanding of whether part-timers are satisfied with their jobs only when they respond to their needs for time to spend outside the job, or whether other motivations are sources of satisfaction. Comparison between full-time and part-time workers (Table 10.3) first shows that the general satisfaction of the former depends upon more crucial aspects of the job and of their motivations. Instead, the level of satisfaction of part-timers is only related to few, although very important, variables. Specifically, both part-time and full-time workers are more satisfied if they have been attracted to the organization by the sector of activity and involvement in organizational decisions. Furthermore, they see work mainly as a necessary experience for self-fulfilment.
10
Does Part-Time Mean Part-Satisfaction?
199
Table 10.3 Job satisfaction and workers’ motivations and attitudes (ordered probit model) Full-time Part-time Std. coeff. (Constant) Interest in the sector Getting to know other workers Getting to know users Way in which the organization works with users Match with educational training Compatibility with other commitments Only job available Pay and career Job stability Workforce involvement An opportunity for self–fulfilment A necessity An opportunity for new relationships A hobby A contribution to society To earn a living To earn as much as possible To support the family A way to gain recognition
t
P(t)[|t| Std. Sig. coeff.
t
P(t)[|t| Sig.
-0.0641 -0.481 0.6308 -0.1895 -0.608 0.5429 0.0583 3.971 0.0001 0.1041 3.182 0.0015 0.0150 0.984 0.3250 -0.0112 -0.371 0.7110 20.0489 22.669 0.0076 -0.0510 –1.411 0.1581 0.0621 4.387 0.0000 0.0561 1.630 0.1030 0.0155 0.0318 -0.0194 0.0531 -0.0207 0.0749 0.1244 -0.0055 0.0417 0.0114 0.0203 -0.0154 -0.0206 0.0310 -0.0204
1.177 2.280 -1.500 2.605 -1.499 4.925 7.437 -0.304 2.595 0.59 1.332 -0.804 –1.263 1.782 -1.195
0.2391 0.0226 0.1335 0.0092 0.1340 0.0000 0.0000 0.7614 0.0095 0.5553 0.1828 0.4214 0.2067 0.0747 0.2322
0.0267 0.0300 -0.0132 0.0236 -0.0127 0.0864 0.1098 0.0705 0.0245 0.0300 0.0457 20.1068 0.0177 0.0308 20.0804
1.004 1.026 -0.467 0.583 -0.476 2.634 2.972 1.627 0.711 0.779 1.401 22.662 0.437 0.92 22.012
0.3152 0.3048 0.6406 0.5599 0.6344 .0084 0.0030 0.1038 0.4768 0.4360 0.1611 0.0078 0.6619 0.3574 0.0442
Model FT: Observations: 1,514; Iterations completed: 28; Log likelihood function: -2,201,830; Restricted log lik. -2,356,882; Chi-squared: Degrees of freedom: 19; Sign. level: 0.000 Model PT: Observations: 403; Iterations completed: 28; Log likelihood function: -5,757,962; Restricted log lik. -6,241,765; Chi-squared: Degrees of freedom: 19; Sign. level: 0.000
Surprisingly, no correlation emerges between job satisfaction and interest in hourly flexibility. It is therefore possible to claim that part-time workers feel very satisfied when they have intrinsic motivations in the sector of activity and find that the job is a way to achieve professional growth and satisfy their life-needs (Table 10.3). They differ from full-time workers principally because these latter are more sensitive to all types of motivation, both intrinsic and extrinsic. In fact, for this group of workers, job satisfaction increases when they are attracted to the organization, on the one hand, by involvement, the opportunity to begin new relationships, and the way in which the organization works with clients, but, on the other, also by hourly flexibility (expressed by the item on compatibility with other commitments), economic rewards, and career opportunities. Consequently, fulltime employees appear to be more complex and differentiated workers, whose satisfaction must be achieved with multiple incentives. The satisfaction of parttimers depends more upon their intrinsic motivation (Table 10.4). The perception of fairness, as shown by the binomial analyses, is quite high and homogeneous for part-time workers. This has consequences also on the correlation
200
S. Depedri
Table 10.4 Job satisfaction and workers’ satisfaction with different aspects of the job (ordered probit model) Full-time Part-time
(Constant) Professional development Decision-making autonomy Recognition of one’s own contribution Variety and creativity of the job The working environment The social usefulness of the job The salary Working hours Previous career advancement Future career advancement Job stability Relationships with superiors Relationships with colleagues
Std. coeff.
t
P(t)[|t| Sig.
Std. coeff.
t
P(t)[|t| Sig.
–1.3805 0.1489 0.0548 0.0749
–11.918 8.625 2.973 3.988
0.0000 0.0000 0.0030 0.0001
–1.5980 0.2119 0.0401 0.2034
–5.748 5.651 0.923 4.636
0.0000 0.0000 0.3560 0.0000
0.1204 0.0518 0.0878 0.0728 0.0738 –0.0178
6.658 2.903 4.878 4.154 4.070 –0.781
0.0000 0.0037 0.0000 0.0000 0.0000 0.4348
0.1607 0.0450 0.0745 0.0758 –0.0223 0.0076
4.347 1.244 1.751 2.034 –0.574 0.155
0.0000 0.2134 0.0800 0.0420 0.5658 0.8770
0.0119 0.0343 0.1152
0.51 0.6100 2.074 0.0381 6.533 0.0000
–0.1193 0.0074 0.1326
–2.308 0.0210 0.234 0.8147 2.773 0.0055
0.0393
1.899 0.0576
0.0254
0.605 0.5451
Model FT: Observations: 1,514; Iterations completed: 23; Log likelihood function: –1,918,160; Restricted log lik. –2,356,882; Chi-squared: Degrees of freedom: 13; Sign. level: 0.000 Model PT: Observations: 403; Iterations completed: 23; Log likelihood function: -5,016,396; Restricted log lik. - 6,241,765; Chi-squared: Degrees of freedom: 13; Sign. level: 0.000
with job satisfaction (Table 10.5), since items of distributive fairness are not significantly related to the general satisfaction of part-timers. The only factor that influences part-timers’ satisfaction is the fairness of the wage considering the economic resources of the organization. Part-time workers are thus more satisfied when the principal is fair, and they tend to reciprocate or at least to internalize vertical fairness (i.e., the comparison between pay-offs of employer and employees) in their utility function. There is a very different relationship between job satisfaction and distributive fairness for full-timers. These workers are significantly more satisfied with their jobs when they perceive the wage as very fair considering mainly their training and effort, but also responsibility on the job and their experience. The proxy that influences the satisfaction of both full-time and part-time workers is procedural fairness. All workers are similarly sensitive to communication, the attention of managers to workers’ ideas, and opportunities to improving abilities. Considering these results all together, part-time workers are again particularly appreciative of opportunities for training and growth. The elements of fairness that define their utility do not depend on economic aspects, but rather on being part of a
10
Does Part-Time Mean Part-Satisfaction?
201
Table 10.5 Job satisfaction and workers’ perceptions of fairness (ordered probit model) Full-time Part-time Std. coeff. (Constant) Procedural fairness Balance incentive/contribution Communication Professional growth and career Being listened to Growth of skills and capabilities Transparency of promotions Work better if more remunerated Not very good security Cause of stress and tension Distributive fairness Responsibility Training Experience Effort Quality of the job Stress and tension Economic resources of the organization
0.5070
t
P(t)[|t| Sig. 4.705 0.0000
Std. coeff.
t
P(t)[|t| Sig.
–0.0743
–0.303 0.7620
0.0162 0.1051 –0.0279 0.0948 0.1359 –0.0520 –0.0012 0.0114 –0.0693
0.866 5.953 –1.302 4.765 6.242 –2.467 –0.084 0.773 –4.529
0.3864 0.0000 0.1928 0.0000 0.0000 0.0136 0.9334 0.4397 0.0000
0.0118 0.1013 –0.0348 0.1456 0.1502 –0.0544 0.0399 0.0392 –0.0347
0.347 2.488 –0.787 3.579 3.703 –1.324 1.329 1.201 –1.100
0.7282 0.0128 0.4315 0.0003 0.0002 0.1855 0.1838 0.2299 0.2711
0.0421 0.1201 –0.0534 0.0759 –0.0506 –0.0087 –0.0049
1.825 4.723 –1.871 2.411 –1.628 –0.391 –0.282
0.0679 0.0000 0.0613 0.0159 0.1036 0.6961 0.7782
0.0220 0.1098 0.0460 –0.0138 0.0297 –0.0142 –0.0685
0.437 1.618 0.652 –0.2 0.411 –0.301 –1.904
0.6618 0.1057 0.5141 0.8417 0.6813 0.7636 0.0569
Model FT: Observations: 1,508; Iterations completed: 25; Log likelihood function: –2,089,278; Restricted log lik. –2,48,199; Chi-squared: Degrees of freedom: 16; Sign. level: 0.000 Model PT: Observations: 402; Iterations completed: 25; Log likelihood function: -5,518,030; Restricted log lik. -6,221,952; Chi-squared: Degrees of freedom: 16; Sign. level: 0.000
work environment which interrelates with them and considers them as an essential part of the productive process.
10.5 Conclusions Despite the increase in part-time employment, this contractual type is frequently criticized by economists, who argue that part-time undermines workers’ opportunities to improve their abilities and skills, to pursue careers, and to be fully satisfied. Analyses have generally concentrated on wage treatments and differentials with respect to full-time workers. Nevertheless, investigations and theory generally agree that part-timers are less secure in their jobs, receive lower recognition both in the present and in the future (given the difficulty of career advancement), and consequently achieve lower levels of well-being. This is especially true when part-time is involuntary, i.e. it does not respond to workers’ needs for time to deal with other commitments, but is rather an organizational imposition for financial and productive reasons.
202
S. Depedri
This chapter has investigated part-time workers in the social services sector, concentrating on a sector characterized by a high proportion of part-time contracts, a large presence of women, and the centrality of workers’ satisfaction in order to improve their effort and the quality of services. In fact, the data should be interpreted in light of the fact that the workers’ involvement in an organization’s mission and the development of good psychological contracts with employees are central factors in organizational performance. And both of them, together with personal well-being, may be significantly correlated to workers’ perceptions of their jobs. The results first show that part-time contracts are mainly stipulated with married females on the one hand, and single, young, and highly educated people on the other. Also the analysis of workers’ preferences and motivations has supported the idea that part-time is explained by two different factors. Firstly, it is a response to the needs of employees, who want jobs that match their familial needs and give them more time to spend on other activities, such as voluntary work. Secondly, part-time contracts are offered by organizations to just-employed people and are thus means to assess and integrate workers into the organization. This feature also reflects the intention of organizations to impose flexibility only on workers with less experience and to cover their organizational needs and production peaks with workers that may be considered more flexible because of their personal characteristics (age, civil status, work experience, needs of training, etc.). As a result, when the analysis focuses on average data on part-time workers, a general sense of well-being emerges: part-timers are not less satisfied than fulltimers with the different aspects of their work, they receive internal training and professional growth, transparent and complete information, and concrete opportunities for future careers (sustained by the presence of part-timers also in roles of coordination and managerial activities). Furthermore, most elements of psychological contracts—such as involvement and relationships—are satisfied, and non-economic benefits (or intrinsic incentives) are provided. As a consequence, part-time employees are just as satisfied as their full-time colleagues. Furthermore, they have similar prospects of staying with the organization, but their loyalty depends mainly upon their perceptions of distributive and procedural fairness. However, two specific features should be considered. Firstly, part-time and fulltime workers should be treated in different ways because their job satisfaction is correlated with different factors. In particular, dissatisfaction is mainly correlated with no prospects of careers and future stability, while part-timers are more satisfied when they perceive that they are being treated in a fair and nondiscriminating way, in general and primarily in regard to training and professional growth. In this context, a good policy for organizations is to select workers with high intrinsic motivations and altruism (i.e., ones who internalize others’ wellbeing in their utility functions). The data also show that part-time workers differ and can be classified according to different needs, expectations, and behaviours. As already reported in the literature, involuntary part-timers represent a first group of workers. The analysis developed in the paper has demonstrated that differences in levels of satisfaction
10
Does Part-Time Mean Part-Satisfaction?
203
with work in general, and with the wage, job stability, and autonomy in particular, are significantly lower among workers who are not motivated or satisfied with their working hours, and who can therefore be considered involuntary part-timers. In conclusion, the policies of part-time employment adopted by the organizations in the social services sector seem satisfactory and in line with the requirements of the activity, and they sustain a positive psychological contract. In fact, firstly, part-time contracts respond to the flexibility needs of women and other workers with family commitments. Secondly, hourly flexibility is a first approach to the labour market and to the organization for both marginalized workers (housewives or unemployed persons) and young people (mainly students). Thirdly, the offer of part-time contracts as a mechanism to assess just-employed people should be a good method for selecting and giving long-run job opportunities to workers really motivated and committed to organizational goals. The only aspect to which managers should pay closer attention is the possible dissatisfaction of even motivated workers with their working hours. The clear communication and transmission of workers’ preferences and organizational needs would be the best ways to prevent under-effort and the desire to quit the organization. The match between the demand for and supply of part-time jobs is thus possible, and may be a source of quality performance rather than partial worker involvement and satisfaction.
References Barrett, G. F., & Doiron, D. J. (2001). Working part-time: By choice or by constraint. Canadian Journal of Economics, 34(4), 1042–1065. Booth, A., & Wood, M. (2008). Back-to-front down under? Part-time/Full-time wage differentials in Australia. Industrial Relations, 47(1), 114–135. Borzaga, C. (2000). Capitale umano e qualità del lavoro nei servizi sociali. Un’analisi comparata tra modelli di gestione. Rome: Fondazione italiana per il volontariato. Buddelmeyer, H., Mourre, G., & Ward, M. (2004). The determinants of part-time work in EU countries: Empirical investigations with macro-panel data, IZA. Discussion paper, 1361. Fullin, G. (2004). Vivere l’instabilità del lavoro. Bologna: Il Mulino. Houseman, S. (2001). Why employers use flexible staffing arrangements: Evidence from an establishment survey. Industrial and Labor Relations Review, 55(1), 149–169. Rousseau, D. M. (1995). Psychological contracts in organizations. Thousand Oaks, CA: Sage. Sen, A. K. (1993). Capability and well-being. In M. Nussbaum & A. K. Sen (Eds.), The quality of life. Oxford: Clarendon Press.
Chapter 11
Workers’ Lifestyle Choices, Working Time and Job Attributes Giovanni Russo and Edwin van Hooft
11.1 Introduction The Netherlands has been dubbed ‘‘the only part-time economy’’ (Freeman 1998). This expression reflects the popularity of part-time jobs in the country, particularly among working women. The beginning of the boom in Dutch part-time work can be traced back to the tripartite agreement of 1982 (the Wassenaar agreement), which dealt with issues concerning working-time reduction. Legislation that converted the option to work part-time into what was essentially a workers’ right was enacted during the 1990s (Euwals 2005; Euwals and Hogerbrugge 2006). Since then, the incidence of part-time employment has been on the rise. Indeed, whilst part-time employment is still rather uncommon among men, it has become almost the norm for women. In 2004 around 62% of all jobs filled by women had part-time schedules (the share had steadily increased from 56% in 1995 to 62% in 2003 and 2004); the incidence of part-time employment among working women in 2005 fully 73% (including women working fewer than 12 h/week; it
The views expressed herein are the authors own and do not necessarily represent those of Cedefop. G. Russo (&) CEDEFOP, PO Box 22427, 55102 Finikas (Thessaloniki), Greece e-mail:
[email protected] E. van Hooft (&) Work and Organizational Psychology, University of Amsterdam, Roetersstraat 15, 1018 WB, Amsterdam, The Netherlands e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_11, Springer-Verlag Berlin Heidelberg 2012
205
206
G. Russo and E. van Hooft
dropped to 68% among women working more than1 This difference is usually ascribed to gender differences in labour-force participation—influenced by favourable marginal tax rates for dual earner couples when the second earner works only a few hours per week (Schettkat and Yocarini 2001)—and in human capital accumulation that can be related to the role of women within households (Goldin 1997).2 In fact, women with young children represent the majority of part-time workers (Riedmann et al. 2006). Three possible schedules of part-time work can be found in the Dutch labour market: a few hours every day, full-time work on fixed days of the week (or month), and other fixed patterns.3 The third part-time schedule, which entails a few hours of work on given days, is the modal offer among Dutch firms (Riedmann et al. 2006). Despite generally positive attitudes towards part-time work, part-time workers are found to fare worse in the labour market than their full-time colleagues (Fuchs Epstein et al. 1999).4 They appear to incur a (modest) part-time wage penalty— they earn less than an otherwise equal full-time worker (Manning and Petrongolo 2008; Russo and Hassink 2008)—and the incidence of training and promotions is lower among part-time workers than among full-time workers (Green and Ferber 2005a, b, c; Riedmann et al. 2006; Russo and Hassink 2008). However, doubts about the desirability of part-time work arrangements are countered by the finding that part-time workers are just as satisfied with their lives and their jobs as their full-time counterparts. Therefore, low wages and slow career
1
There has been a steady increase in the proportion of employed women working part-time. Among all women (including those working fewer than 12 h per week) the incidence of part-time employment rose from 66% in 1996 to 73% in 2005, while the incidence of part-time among women working more than 12 h per week rose from 58% in 1996 to 68% in 2005. 2 There is a gender aspect to the analysis that we will not pursue here. Nor will we deal with estimation of the part-time wage gap. Likewise, we will not discuss whether workers accept parttime jobs because full-time jobs are rationed. Aggregate evidence from the Dutch labour force survey shows that in 2004 about 176 (146) thousand workers working fewer than 12 h (between 12 and 35 h) per week were trying to increase their working time and 418, 000 full-time workers were trying to reduce their working time. This suggests that it is the supply of part-time jobs that is likely to be rationed. 3 For example, the retail sector has peaks in demand during Thursday evening openings and at the weekends (food retailers experience a demand peak on Sunday mornings). 4 Managers perceive the motivation of part-time workers as being no lower than that of their fulltime counterparts. However, about one quarter of managers tend to assess the career prospects of part-time workers as worse than those of their full-time colleagues. The perception of diminished career opportunities is stronger among workers than among managers (Riedmann et al. 2006).
11
Workers’ Lifestyle Choices, Working Time and Job Attributes
207
progression are understood to be in line with part-time workers’ preferences.5 It is argued that, because non labour-market activities—which range from leisure to important and non-leisure activities such as taking care of children, parents, and people in need—may be very (intrinsically)6 rewarding activities, part-time workers must be less interested than their full-time counterparts in a career, in training, and in the monetary aspect of their jobs.7 Nonetheless, career issues may be especially important for part-time workers. In fact, 60% (22%) of part-time jobs are offered by firms with more than 100 employees (firms with more than ten but fewer than 100 employees). About 82% of all part-time jobs are in firms big enough to offer substantial career possibilities. Overall, only 18% of part-time jobs can be found in small firms (with ten employees or fewer) where the concept of a career may be less meaningful. The purpose of the present study is to explore whether workers who prefer parttime work differ from their full-time counterparts. It also investigates whether the observed differences sustain a part-time wage gap, doing so by analysing data on self-reported preferences for desired job attributes collected in the Netherlands in 2001 (Van Hooft et al. 2005).
5 This conclusion runs counter to standard economic reasoning: assuming standard (quadratic) utility functions in income (consumption) and leisure, in that part-time workers derive higher utility from leisure than do full-time workers, other things being equal, part-time workers should receive a wage premium to reach the utility level of full-time workers. In other words, the quadratic utility functions may suffer from misspecification. The presence of a part-time wage gap and of equal utility levels among part-time and full-time workers can be reconciled by a more complex utility function that lets workers derive utility from compliance with a role standard: an identity. Membership of a group would be expressed in terms of compliance with a given norm (working time) shared by all members of the group. The members of a group may derive satisfaction by adhering to a norm consisting in an ideal number of hours worked by (and consumption level of) the group members (Akerlof and Kranton 2000; Grodner and Kniesner 2006; Hakim 2000). Similarly, workers’ utility may be affected by comparison of their incomes and wages with aspiration levels (Brown et al. 2008; Clark and Oswald 1996). 6 Moreover, roles may prompt people to undertake care activities in relation to life-course contingencies; as these subside, individuals may revert to their preferred activities. Of course, individuals may decide to purchase these services from the market. 7 Women’s attitudes toward labour-force participation are becoming more diversified over time. That is, human resources officers encounter very different types of women: those who have fully internalized feminine stereotypes (homemakers who would consider work less rewarding) and women who are relatively androgynous in their preferences (who would regard work as very important and rewarding). Research shows that women’s conditions have changed dramatically in the US and UK. Women’s labour-force participation (and men’s attitudes towards it) has improved, Equal opportunity practices have been introduced, the number of divorces has increased, and the number of children born to single mothers has increased as well. Moreover, gender roles have also evolved. Women in commercial jobs tend to show increasingly fewer feminine stereotypes; by contrast, men tend to exhibit an increasing number of male stereotypes (such as the breadwinner model). Hence, women tend to be more diverse than men in their workrelated attitudes. Some women retain the ‘homemaker’ stereotype while others have become more similar to men, so that the spectrum of women’s preferences has widened (Konrad et al. 2000).
208
G. Russo and E. van Hooft
We find that part-time workers differ from their full-time counterparts in one respect: they place more emphasis on working-time flexibility than do full-time workers. However, if firms’ incentive schemes (career and training) make additional demands on workers’ time, time-constrained part-time workers will respond to them less effectively than full-time workers. In so far as firms use this information to allocate their training and promotion opportunities among their workforce, part-time workers forgo career opportunities more often than their full-time counterparts. Consequently, part-time employment spells will have long-term effects on workers’ wages (Francesconi and Gosling 2005; Roman et al. 2004). The paper is organized as follows: Section 11.2 presents the theoretical background, Sect. 11.3 describes the data, Sect. 11.4 illustrates the results, and Sect. 11.5 concludes.
11.2 Identity, Short Hours, Flexibility, and Job Attributes The term identity denotes a person’s self-image. Group identity is important because it prescribes (behavioural) norms that affect (labour market) behaviour (Ajzen 2001; Godin et al. 2005; Hagger and Chazisarantis 2005). Identity, among other things, dictates the appropriate proportion of time to be allocated to labour market and non-labour market activities (usually familial) and how people belonging to a given group behave in certain circumstances (Akerlof and Kranton 2000, 2005). For example, identity prescribes appropriate labour-market behaviours around childbearing. Men (and women) may act differently according to the extent to which their self-image is shaped by labour-market success. Career women, whose self-images are centred on their jobs, may stop breastfeeding and hire child-minders sooner than housewives who centre their self-images around the family. In the same vein, men characterized by a work-oriented self-image may continue to work in the days immediately following the birth of a child, whilst men characterized by a more family-oriented self-image may take a period of leave (or holiday) to assist their spouses. In fact, it is no coincidence that the majority of women working on part-time schedules have young children, or are returning to the labour market after a spell of non-participation because of childbearing and childrearing (Hakim 2000; Riedmann et al. 2006). Nowadays, however, society expects everybody to engage in labour-market activities at least to some extent. Some individuals will therefore experience work and non-work activities as in conflict with each other, while others will not. Some individuals with greater demands on their time may feel under pressure, while others may feel busy and productive. This different perception may depend on whether these individuals like what they are doing, and on whether those close to them (their ‘significant others’) think that they are doing the right thing (Thompson and Bunderson 2001). It follows that individuals who adhere to identities that stress the importance of non-labour market activities may perceive a conflict between the behaviour prescribed by their identities and what society expects from
11
Workers’ Lifestyle Choices, Working Time and Job Attributes
209
them. This particular kind of conflict may take the form of a time constraint (Fuchs Epstein and Kalleberg 2004; Thompson and Bunderson 2001). To the extent that the time constraint (work and non-work activities laying conflicting claims to workers’ time) arises because of workers’ commitment to prescribed non-work activities, it will not affect their utility. Because of the presence of different identities, otherwise identical workers may hold very different views about the importance of work in their lives. In particular, because workers whose identities prescribe that they must engage in non-labour market activities derive satisfaction from performing the non labor-market activities prescribed, they may suffer when work instances intrude into their private sphere (work-to-family conflict). Adherence to a certain group identity may thus require a limit to be set on labour-market time, which may be achieved through the decision to work on a reduced time schedule: that is, to work part-time. Moreover, identity may influence individuals’ preferences in regard to job attributes in two ways. Firstly, individuals whose identities prescribe that they must engage in non-work activities (and who are committed to those activities) will assign particularly high value to those job characteristics—such as the opportunity to work at convenient times—that let them pursue their other interests. In addition, it is likely that adherence to an identity that places minor emphasis on labourmarket success may entail little interest in career opportunities or in high wages. Secondly, identity may indirectly influence workers’ preferences towards job attributes: for instance, when workers anticipate certain reactions from supervisors and colleagues and adjust their preferences accordingly. Therefore, the preference that part-time workers allegedly display for reduced career opportunities may in fact be spurious. We propose to use (subjective) data about the locus of individuals’ interests (work or non-work activities) to infer their identities, using this construct to infer how part-time workers evaluate job attributes. All in all, adherence to a given group identity may entail the decision to work part-time. But this decision has far-reaching consequences, for the decision to work part-time in its turn affects career and training opportunities, doing so in at least three related ways (which we will not be able to disentangle in the present article): 1. Part-time workers may believe that work has a limited importance for achievement of their life-goals, and they may not invest enough in their jobs. 2. Firms that accommodate part-time workers’ preferences for flexibility do so in order to please workers and to improve their morale (Riedmann et al. 2006). When wage bargaining takes place, these firms may be in stronger bargaining position and may be able to extract compensating differentials from part-time workers in the form of low career and training opportunities and low wages (Francesconi and Gosling 2005). 3. In a signalling setting, workers who prefer to work on a part-time basis (or prefer to be able to decide when to work and for how long) do not respond well to incentives schemes designed to induce workers to put in greater effort
210
G. Russo and E. van Hooft
through long(er) working hours (longer than contractually agreed upon). The lukewarm response of part-time workers to incentives may induce firms to withhold human capital investments from this segment of their workforce (Booth et al. 2003; Cowling 2007; Fuchs Epstein and Kalleberg 2004; Landers et al. 1996).8 In all these cases, part-time workers will experience a part-time wage gap that will not be closed once they decide to return to a full-time schedule.
11.3 The Data The data set used was a sample of 1,828 employed or job-searching individuals (47% of them were women) which can be considered as representative of the Dutch population. The data were collected with structured questionnaires administered electronically. The questionnaire consisted of items about the centrality of work, desired job attributes, and preference for part-time or full-time employment. The data collected, however, did not contain any information about job title, occupation, industry, and firm’s size.9 Here we describe the features of the data that are relevant to the analysis; the interested reader can find a more detailed data description in Van Hooft et al. (2005). About 77% of the respondents were employed at the moment of the data collection: 62% had permanent contracts, 9% were in temporary jobs, and 6% were self-employed. Of the employed workers, 61% were in full-time employment (more than 34 h per week), 30% had long part-time jobs (fewer than 35 h but working 12 h or more), and 9% had a small part-time jobs (fewer than 12 h per week).10 Only 5% of the respondents in full-time employment would have preferred a part-time job, while about 68% (62%) of those holding a long (small) part-time job would have preferred a part-time job. There was a striking gender difference: 81% of the individuals preferring parttime employment were women. Only 12% of the men (either employed or not) had a preference for part-time employment, while about 57% of the women would have liked a part-time job (the length of the desired part-time job was not 8 Agency theory posits that, in the presence of moral hazard, firms may use incentives to induce appropriate behaviour and to induce workers to supply the optimal amount of effort (Prendergast 1998, 1999; Salaniè 1997). Such incentives are typically wage contracts and promotion and training opportunities (Gibbons and Waldman 1999; Malcomson 1999). 9 The data are not suited to investigating the relation between part-time workers’ labour-market attitudes and the reasons inducing workers to opt for part-time work arrangements, or their labour-market experiences (Green and Ferber 2005a, b, c; OECD 2003). 10 The classification of part-time jobs adopted appears to be rather accurate: 6.6% of all jobs between 30 and 35 working hours per week are full-time jobs, while 2.7% of all jobs involving more than 35 working hours per week are, in fact, part-time jobs (Dutch central bureau of statistics, survey on employment and wages).
11
Workers’ Lifestyle Choices, Working Time and Job Attributes
211
Table 11.1 Life goals (scale 1–5) by preferred number of hours and gender Item Men Women
1 2 3 4 5 6 7 8
9 10 11 12 13
Having enough money Feeling that one is useful Social contacts Respect Sense of security Having order and regularity in life Variation Having tasks that match knowledge and skills Being able to grow and develop Status Work-life balance Having responsibilities Having enough leisure time Number of cases 1,828
Fulltime
Parttime
T-test score
Fulltime
Parttime
T-test score
3.46 3.86
3.25** 4.01
2.05 -1.45
3.32 3.90
3.29 3.90
0.45 0.12
3.83 3.79 3.75 3.31
4.03* 3.93 3.79 3.36
-1.85 -1.40 -0.48 -0.41
3.97 3.89 3.84 3.44
3.97 3.87 3.81 3.27**
0.00 0.34 0.45 2.59
3.70 3.83
3.81 4.07**
-1.03 -2.28
3.67 3.73
3.62 3.80
0.88 -1.08
3.84
3.97
-1.30
3.77
3.80
-0.43
2.51 3.81 3.57 3.91
2.37 4.07** 3.55 4.33**
1.19 -2.27 0.16 -3.76
2.37 3.72 3.42 3.90
2.21** 3.89** 3.41 4.02**
2.52 -2.56 0.29 -1.95
888
73
554
313
*significant at 10%; **significant at 5%
specified). The preference for part-time employment was stronger among individuals without paid employment (57%) than it was among the respondents with paid jobs (28%). However, 68% of the individuals preferring part-time jobs were, in fact, employed. There was also a gender difference in family roles: 62% of the respondents reported themselves as household heads, the incidence of household heads being 86% (31%) among men (women). Similarly, 32% of the respondents were spouses, with the incidence of spouses being as low (high) as 6% (61%) among men (women). Life-goals and preferences concerning work-related attributes differed between part-time and full-time workers. Given the importance of gender, Table 11.1 compares the importance attached to life goals by individuals preferring full-time and part-time jobs, broken down by gender. The Table shows that men preferring fulltime jobs attach more importance to the availability of sufficient income, and less importance to tasks matching their skills, leisure time, social contacts, and work/life balance than do men preferring part-time employment. Conversely, women preferring full-time jobs attach more importance to an orderly life and to status, and less importance to leisure time and work/life balance than do women wanting part-time employment. Furthermore, women appear to place more importance than men on the
212
G. Russo and E. van Hooft
Table 11.2 The importance of having a job to attain life goals (scale 1–5) by preferred number of hours and gender Item Men Women
14 Having enough money 15 Feeling that one is useful 16 Social contacts 17 Respect 18 Sense of security 19 Having order and regularity in life 20 Variation 21 Having tasks that match knowledge and skills 22 Being able to grow and develop 23 Status 24 Work-life balance 25 Having responsibilities 26 Having enough leisure time Number of cases 1,828
Fulltime
Parttime
T-test score
Fulltime
Parttime
T-test score
3.85 3.63
3.77 3.48
0.81 1.41
3.69 3.57
3.69 3.53
0.04 0.50
3.53 3.60 3.66 3.44
3.40 3.53 3.48 3.52
1.15 0.65 1.64 -0.68
3.52 3.60 3.62 3.46
3.52 3.61 3.55 3.29**
0.06 -0.20 1.10 2.56
3.34 3.69
3.30 3.73
0.30 -0.29
3.42 3.63
3.38 3.63
0.52 0.01
3.62
3.68
-0.55
3.59
3.58
0.23
2.83 2.95 3.49 2.70
2.74 2.71 3.58 2.49
0.66 1.82 -0.81 1.58
2.65 3.05 3.45 2.77
2.60 2.96 3.35 2.71
0.61 1.33 1.41 0.81
888
73
554
313
*significant at 10%; **significant at 5%
sense of security, whilst men appear to be more concerned than women with variation, the opportunity to grow and develop, status, and responsibility. The second aspect considered was the extent to which having a job is perceived to be instrumental in achieving specific life-goals. The average scores on these items, again broken down by desire to work part-time and gender, are shown in Table 11.2. There is hardly any difference between workers who want to work part-time and full-time as far as the perceived instrumentality of work in achieving life goals is concerned. However, individuals who want to work full-time perceive work as more instrumental to the achievement of a satisfactory income, status, order and regularity in life, responsibilities, security, and a sense of usefulness compared to individuals wanting to work part-time. We now turn to evaluation of a series of job attributes. Table 11.3 presents the average scores attached to various job attributes broken down by desired work schedule (part-time versus full-time) and by gender. Interestingly, job security, and career opportunities score significantly higher among individuals who prefer to work full-time than among individuals who prefer part-time work, regardless of their gender. On the other hand, the opportunity to
11
Workers’ Lifestyle Choices, Working Time and Job Attributes
213
Table 11.3 Importance of job attributes (scale 1–5) by preferred number of hours and gender Item Men Women
27 28 29 30 31 32 33 34 35 36 37 38 39 40
41 42 43 44
Pay level Benefits Career possibilities Job security Working conditions Favourable working time Firm’s location Holidays (and days off) Possibility to work with colleagues Firm’s reputation Training provision Stimulating and challenging job Selection process Demographic composition of the workforce Childcare provision Equal opportunities (for minorities) Opportunity to use own skills and abilities Good relationship with colleagues and managers Number of cases 1,828
Fulltime
Parttime
T-test score
Fulltime
Parttime
T-test score
4.05 4.01 3.78 3.88 4.04 3.66
3.78** 3.96 3.49** 3.58** 4.10 4.12**
2.57 0.50 2.67 2.71 -0.59 -4.10
3.83 3.93 3.58 3.82 4.06 3.92
3.91 3.91 3.34** 3.70** 4.08 4.18**
-1.33 0.38 3.88 2.07 -0.27 -4.47
3.60 3.74 3.56
3.81* 3.99** 3.63
-1.73 -2.27 -0.60
3.81 3.85 3.77
3.91* 4.00** 3.68
-1.66 -2.45 1.55
3.64 3.57 4.03
3.53 3.40 4.12
0.97 1.56 -0.96
3.71 3.58 3.98
3.59** 3.44** 3.85**
1.95 2.19 2.44
3.21 2.90
3.27 2.84
-0.54 0.51
3.32 3.01
3.23 3.00
1.46 0.15
2.24 2.59
2.15 2.66
0.64 -0.53
2.64 2.94
2.85** 2.84
-2.24 1.45
4.03
4.16
-1.45
3.99
3.99
-0.12
3.86
4.10**
-2.34
3.92
3.98
-1.21
888
73
554
313
*significant at 10%; **significant at 5%
take days off, the firm’s location (i.e. commuting time), and the opportunity to work favourable shifts are all items that score higher among workers who would like to work part-time than among those who want to work full-time, irrespective of gender. Among men, those who prefer to work full-time assign more (less) importance to pay (good relationship with colleagues and managers) than those who prefer to work part-time. A supportive work environment (managers and colleagues) is perceived as a necessary prerequisite by men who prefer to work part-time. Women preferring part-time work deem the firm’s reputation, training provision and the possibility of have interesting and stimulating work as less important job characteristics than do women who prefer full-time jobs. On the other hand, women who desire to work part-time place more emphasis than women who prefer to work full-time on the availability of childcare services.
214
G. Russo and E. van Hooft
Table 11.4 General level of life satisfaction (Panel a) and job satisfaction (Panel b, for employed individuals only) by preferred number of hours and gender Panel (a) Men Women
No problem in making ends meet Satisfied with one’s life Work attitude Item Work attitude 45 Work is an important part of life 46 Work is a source of satisfaction in one’s life 47 Work gives a meaning to one’s life 48 If I had enough money, I would work 49 I enjoy talking about work with others 50 Work means more to me than just money Number of 1,828 cases Panel (b) Job satisfaction Intention to quit Number of 1,390 cases
FullTime
Parttime
T-test Score
FullTime
Parttime
T-test score
3.81
3.78
0.24
3.77
3.60**
2.54
3.95
3.77*
1.86
4.03
3.89**
2.78
3.59
3.40*
1.89
3.53
3.52
0.16
3.74
3.48**
2.37
3.57
3.60
-0.43
3.62
3.36**
2.49
3.47
3.34**
1.97
3.34
3.10**
2.05
3.26
3.10**
2.40
3.18
3.22
-0.28
3.33
3.33
0.00
3.37
3.25
1.14
3.37
3.44
-1.22
3.68
3.66
0.23
3.71
3.78
-1.10
888
73
554
313
3.99 2.09 750
3.58** 2.89** 53
4.00 2.18 375
3.76** 2.42** 212
2.83 -4.51
2.69 -2.17
*significant at 10%; **significant at 5%
Overall, individuals (and women in particular) who prefer to work part-time may be less sensitive than their colleagues preferring full-time work to the means that firms generally use to provide incentives. This result matches the low incidence of promotions among part-time workers, but it may not be in line with part-time workers’ preferences. Cognitive dissonance theory suggests, in fact, that the experience of reduced career and training opportunities and workplace frictions (Blau and Ehrenberg 1997; Fuchs Epstein et al. 1999) may cause part-time to take a different stance on these issues when they explain to themselves why they have accepted such treatment (Akerlof and Kranton 2005). We now analyse the differences among individuals’ attitudes towards work and life satisfaction broken down by gender and preferred working-time schedule. The results are presented in Table 11.4.
11
Workers’ Lifestyle Choices, Working Time and Job Attributes
215
Panel (a) shows that individuals with a preference for part-time jobs are less satisfied with their lives than are individuals preferring to work full-time. Women preferring part-time work report a lower ability to make ends meet than women preferring a full-time work schedule. Men preferring part-time work display a lower score on the construct ‘work attitude’, which proxies intrinsic motivation (Benabou and Tirole 2003; Deci and Ryan 2002), than men preferring to work fulltime. Because the construct ‘work attitude’ was obtained by aggregating six items (the average scores obtained on the six items), we now investigate differences among the scores relative to these items by preferred work schedule and gender. Consistently with the view that preferences for working-time schedules are related to people’s identities, individuals who prefer to work part-time find that work is a less important aspect of their lives, and regard it as less important in giving meaning to their lives than do workers preferring full-time work. Moreover, men who prefer a part-time work schedule find that work is a less important aspect of life compared with men preferring to work full-time. This pattern of results has also been found to characterize the difference between part-time and full-time workers in the US (Thorsteinson 2003). However, on the three items measuring intrinsic motivation (items 48–50), workers who prefer part-time work score no differently from workers who prefer a full-time work schedule. In other words, part-time and full-time workers tend to show an equal level of interest in their work: personal interest in a given task is a key motivational element, especially as far as intrinsic motivation is concerned (Deci et al. 1999a). From this point of view, we suggest that voluntary part-time and full-time workers may display a similar level of intrinsic motivation (as far as work is concerned). Furthermore, panel (b) shows that individuals preferring a part-time work schedule are characterized by a lower level of job satisfaction and by a stronger intention to quit than are their counterparts who prefer full-time work. In fact, involuntary full-time workers may be keener to quit than voluntary full-time workers because adjustments in their working hours may involve a job change (Altonji and Paxson 1988, 1992; Manning and Petrongolo 2005). Why might work be less central for individuals who prefer part-time employment than for those who prefer to work full-time? Table 11.5 suggests that the answer may be forthcoming from the role of these individuals outside the labour market. Individuals who prefer to work part-time perceive a job as a source of rigidity more than do people who prefer to work full-time. Furthermore, women preferring part-time jobs tend to experience time-related problems, especially in relation to caretaking activities. Men who prefer part-time work, on the other hand, emphasise the loss of independence, having to obey orders, repetitiveness of tasks, and the amount of pressure and stress, while women emphasise the scant increase in income that part-time jobs provide. In addition, individuals who prefer part-time work emphasise the firm’s location (and commuting time). Moreover, the lack of childcare facilities is more often
216
G. Russo and E. van Hooft
Table 11.5 Importance of work related problems (scale 1–5) by preferred number of hours and gender Item Men Women
51 Too little time for care 52 Day too tightly organized 53 Not enough increase in income (relative to no job) 54 Too little time for friends 55 Too little time for household work 56 Loss of one’s independence 57 Too repetitive 58 Having to spend time with people you might not like 59 Too much pressure and stress 60 Having to obey the orders of others 61 Lack of Childcare 62 Transportation (commuting) 63 Dutch language 64 Discrimination 65 Having to do tasks beneath one’s level Number of cases 1,828
Fulltime
Parttime
T-test score
Fulltime
Parttime
T-test score
2.61 2.33
2.68 2.63**
-0.61 -2.30
2.89 2.55
3.24** 2.75**
-4.62 -2.81
2.61
2.44
1.10
2.52
2.69**
-1.98
2.68
2.75
-0.54
2.97
3.04
-0.96
2.54
2.68
-1.20
2.88
2.99
-1.60
2.62
2.93**
-2.22
2.73
2.79
-0.78
2.72 2.55
3.01** 2.60
-2.21 -0.40
2.58 2.55
2.59 2.53
-0.10 0.25
2.44
2.88**
-3.59
2.61
2.70
-1.35
2.13
2.49**
-3.10
2.04
2.06
-0.31
1.39 1.54
1.53 1.81**
-1.41 -2.22
1.63 1.62
2.08** 1.78**
-5.08 -2.17
1.08 1.29 2.23
1.05 1.22 2.52**
0.56 0.74 -1.97
1.06 1.22 2.00
1.04 1.25 2.03
0.80 -0.62 -0.30
888
73
554
313
*significant at 10%; **significant at 5%
an obstacle to work for women who prefer part-time work, while men who prefer a part-time work schedule report that having to do tasks beneath their job grade is an obstacle to work. Finally, we grouped the items to obtain three constructs (indexes) informative of individuals’ identity. The three indexes are set out in Eq. 11.1 P 1 WI ¼ 115 ITi ITiþ15 A : fIT1 . . .IT10 ; IT12 g i2A P B : fIT51 ; IT52 ; IT54 . . .IT56 ; IT24 ; IT26 g TC ¼ 17 ITi ð11:1Þ i2B P WN ¼ 15 ITi C : fIT53 ; IT57 . . .IT60 g i2C
11
Workers’ Lifestyle Choices, Working Time and Job Attributes
217
Table 11.6 Work attitude and work importance, work resistance, and time constraint indexes (scale 1–5) by preferred number of hours and gender Men Women Full-time Part-time T-test score Full-time Part-time T-test score WI: work importance WN: work resistance TC: time constraint Number of cases 1,828
2.62 2.43 2.73 888
2.60 2.72** 2.92** 73
0.21 -3.41 -2.52
2.59 2.46 2.88 554
2.52 2.53 3.02** 313
1.11 1.29 -3.08
*significant at 10%; ** significant at 5%
Where WI is the work importance index, TC is the time constraint index, and WN is the work resistance index, and IT denotes the item number (as appearing in the first column of Tables 11.1, 11.2, 11.3, 11.4, 11.5). The three indexes relate to identity in the following ways. WI captures the extent to which an individual’s identity centres on the labour market: it is a weighted average of life-goals, where the weights are the importance of having a job in order to attain particular lifegoals. TC captures the conflict between (possibly incompatible) roles (market and non-market): it is obtained by averaging all the items that signal the presence of competing claims on an individual’s time (work in conflict with other activities, friends, leisure, family, or care activities, or generating work-to-family conflict). WN captures resistance to work independently of the presence of alternative activities to which individuals may be committed: it measures the strength of the beliefs that individuals hold against having a job. The three indexes, broken down by preferred work schedule and gender, are summarized in Table 11.6. Individuals who prefer to work on a part-time schedule score higher on the time constraint index than their counterparts who prefer a full-time schedule. Moreover, men who prefer a part-time schedule score higher on the work resistance index than men who prefer a full-time schedule. The relationship between the preference to work part-time and preferences for job attributes is important. However, these are aggregated data, and the relationship found may be spurious. We therefore put our conjectures to the test of a multivariate analysis.
11.4 Empirical Analysis The empirical analysis focused on differences between individuals who preferred to work part-time and their counterparts who preferred full-time work, controlling for—insofar as this is actually possible—the effects of identity (i.e. the importance of work and non-work activities). We analysed workers’ evaluations of the importance of ten job attributes: pay, benefits, career, job security, the possibility to work at favourable times, holidays,
218
G. Russo and E. van Hooft
training opportunities, opportunity to do interesting tasks, the availability of childcare facilities, and the possibility to use skills.11 These items are interesting in their own right. Moreover, because they are of such importance we could plausibly assume that, over time, individuals have gathered considerable experience on these matters, and that this would reduce the influence of measurement errors (Bertrand and Mullainathan 2001). Because of the salience of gender in these matters we also estimated the equations for men and women separately. The results for the whole sample are presented in Table 11.7, while the results obtained on the sub-sample of men (women) are set out in Tables 11.8 and 11.9.12 Because of the large amount of information presented we will discuss the results of the three estimates jointly.13 Women find the opportunity to work at favourable times, the amount of holidays and days off, and the presence of childcare facilities to be more salient than men, who, for their part, find pay and career opportunities to be more salient than women do. Predictably, individuals experiencing difficulties due to a lack of childcare facilities positively evaluate opportunities to work at favourable times and the presence of childcare facilities. Individuals who display a preference for leisure also give importance to job attributes that can be considered inputs to the leisure production function: money (pay and benefits) and time (the opportunity to work at favourable times and the amount of holidays and days off). The positive relationship between preference for leisure and hourly wage, although surprising, is not new in the literature (Bertrand and Mullainathan 2001). The importance of work/family balance increases the salience of benefits (especially for men) of the possibility of doing interesting and stimulating tasks, and of deploying skills and abilities. Among women, the importance of workfamily balance increases the salience of the possibility to work at favourable times. 11
The nature of the variables, which measured the importance of job attributes based on a fiveitem Likert scale, required the use of ordered probit regression models. 12 The explanatory variables included in all models requires further description. The number of hours worked was actually an interaction term between being in employment (either permanent, temporary or self-employed) and working hours (full-time, the reference group, part-time or small part-time). Financial situation was the (self-reported) ability to make ends meet (based on a five-point Likert scale). We preferred financial situation to income because of the latter’s lack of accuracy (measurement errors). All regressions included: 11 regional dummies (the 12 Dutch provinces), four tenure dummies (detailed in the summary statistics), two level-of-education variables (primary and university), age and age squared, a dummy signalling the presence of children (never significant, apart from a positive effect on the salience of the availability of childcare facilities), a dummy marking the desire to live in an urban area (never significant), four labour-market-position dummies (non-participating, unemployed, self employed, temporary contract, and permanent contract, the reference group). The list of covariates also included household composition: a breadwinner dummy, four role dummies (household head, spouse, cohabiting partner, and child or other positions). 13 We report the coefficients relative to selected variables. The estimated models, however, included all controls.
–0.10 (0.05)* Work importance 0.23 (0.05)** Work resistance 0.04 (0.05) Time constraint 0.04 (0.05) Work/home 0.03 balance (0.04) Leisure 0.12 (0.04)** Lack of child –0.01 care (0.03) Working Hours Part–time –0.11 (0.08) Small part–time –0.15 (0.13)
Work attitude
0.06 (0.05) 0.40 (0.05)** –0.05 (0.05) 0.09 (0.05) 0.04 (0.04) –0.00 (0.04) –0.05 (0.03)
–0.01 (0.08) –0.15 (0.13)
–0.07 (0.05) 0.28 (0.05)** 0.01 (0.05) 0.06 (0.05) 0.06 (0.04) 0.15 (0.04)** 0.00 (0.03)
–0.01 (0.08) –0.16 (0.12)
0.10 (0.08) 0.07 (0.12)
–0.04 (0.05) 0.38 (0.05)** 0.03 (0.05) 0.04 (0.05) 0.03 (0.04) 0.06 (0.04) 0.00 (0.03) 0.26 (0.08)** 0.25 (0.13)**
–0.09 (0.05)* 0.05 (0.05) 0.03 (0.05) 0.20 (0.05)** 0.08 (0.04)** 0.30 (0.04)** 0.08 (0.03)** 0.05 (0.08) –0.07 (0.13)
–0.17 (0.05)** 0.11 (0.05)** –0.01 (0.05) 0.29 (0.05)** 0.01 (0.04) 0.41 (0.04)** –0.01 (0.03)
Table 11.7 Ordered probit models, the whole sample 1,828 cases, all controls included Pay Benefits Career Job Favourable Holidays and level security working time days off
0.01 (0.08) –0.09 (0.13)
0.14 (0.05)** 0.36 (0.05)** 0.00 (0.05) 0.07 (0.05) 0.07 (0.04)* –0.04 (0.04) –0.01 (0.03)
Training provision
0.11 (0.09) –0.12 (0.13)
0.19 (0.06)** 0.49 (0.05)** 0.04 (0.05) 0.05 (0.05) 0.08 (0.04)** 0.04 (0.04) –0.04 (0.03)
Interesting challenging tasks
0.08 (0.08) 0.00 (0.13)
0.12 (0.05)** 0.02 (0.04) –0.06 (0.05) 0.07 (0.05) 0.00 (0.04) –0.02 (0.04) 0.48 (0.03)**
Child care
0.06 (0.09) 0.13 (0.14) (continued)
0.25 (0.05)** 0.43 (0.05)** 0.01 (0.05) 0.05 (0.06) 0.13 (0.04)** 0.04 (0.04) –0.01 (0.03)
Use own abilities and skills
11 Workers’ Lifestyle Choices, Working Time and Job Attributes 219
0.01 (0.07) –0.13 (0.07)* –2124 –2023 200**
–0.08 (0.08) –0.09 (0.07) –1975 –1855 241**
–0.26 (0.07)** –0.25 (0.07)** –2345 –2178 334**
Benefits Career
*significant at 10%; **significant at 5%
Restricted Full model Chi-squared
Gender
Want part–time
Table 11.7 (continued) Pay level –0.24 (0.07)** 0.08 (0.07) –2260 –2101 318**
0.20 (0.08)** 0.23 (0.07)** –2300 –2108 383**
Job Favourable security working time 0.13 (0.08) 0.18 (0.07)** –2237 –2023 429**
Holidays and days off –0.14 (0.07)** 0.02 (0.07) –2310 –2182 255**
Training provision –0.11 (0.08) –0.09 (0.07) –2010 –1771 479**
Interesting challenging tasks –0.11 (0.07) 0.19 (0.07)** –2756 –2428 655**
Child care 0.04 (0.08) –0.05 (0.07) –1917 –1704 426**
Use own abilities and skills
220 G. Russo and E. van Hooft
0.03 (0.07) 0.39 (0.07)** –0.09 (0.06) 0.17 (0.07)** 0.06 (0.05) 0.03 (0.05) –0.03 (0.05)
–0.17 (0.13) –0.24 (0.24) –0.37 (0.15)** –1190 –1100 180**
–0.12 (0.07)* 0.30 (0.07)** 0.02 (0.06) 0.07 (0.07) 0.09* (0.05) 0.13 (0.06)** –0.02 (0.05)
–0.22 (0.14) –0.20 (0.20) –0.01 (0.16) –1043 –973 141**
*significant at 10%; **significant at 5%
–0.11 (0.07) Work importance 0.25 (0.07)** Work resistance 0.09 (0.06) Time constraint 0.03 (0.07) Work/home 0.06 balance (0.05) Leisure 0.12 (0.05)** Lack of child –0.04 care (0.05) Working hours Part–time –0.30 (0.14)** Small part–time –0.05 (0.23) Want part–time –0.29 (0.15)** Restricted –1101 Full model –1038 Chi–squared 126**
Work attitude
–0.09 (0.13) 0.13 (0.21) –0.33 (0.14)** –1213 –1131 166**
–0.10 (0.07) 0.36 (0.07)** –0.03 (0.06) 0.15 (0.08)** 0.04 (0.05) 0.07 (0.05) –0.04 (0.05) 0.08 (0.14) 0.06 (0.22) 0.42 (0.15)** –1264 –1175 179**
–0.17 (0.07)** 0.08 (0.07) 0.02 (0.07) 0.24 (0.07)** 0.03 (0.05) 0.27 (0.05)** 0.08 (0.05)
Table 11.8 Ordered probit models, men 961 cases, all controls included Pay Benefits Career Job Favourable level security working time
–0.12 (0.14) –0.17 (0.23) 0.19 (0.15) –1209 –1084 251**
–0.19 (0.07)** 0.09 (0.07) –0.04 (0.06) 0.30 (0.07)** 0.03 (0.05) 0.45 (0.06)** 0.02 (0.05)
Holidays and days off
0.08 (0.14) –0.07 (0.23) –0.28 (0.14)** –1212 –1152 120**
0.12 (0.07)* 0.30 (0.07)** –0.00 (0.06) 0.09 (0.07) 0.07 (0.05) –0.02 (0.05) –0.01 (0.04)
Training provision
0.06 (0.15) –0.23 (0.25) 0.09 (0.17) –1031 –913 236**
0.18 (0.08)** 0.49 (0.07)** 0.11 (0.07) 0.01 (0.07) 0.08 (0.05) 0.06 (0.06) –0.06 (0.05)
Interesting challenging tasks
–0.03 (0.13) –0.02 (0.23) –0.15 (0.16) –1363 –1239 247**
0.18 (0.08)** 0.01 (0.06) –0.10 (0.06)* 0.17 (0.07)** 0.01 (0.05) –0.04 (0.05) 0.45 (0.05)**
Child care
0.12 (0.16) 0.34 (0.24) 0.09 (0.16) –1007 –894 226**
0.21 (0.08)** 0.46 (0.08)** 0.05 (0.07) –0.01 (0.08) 0.17 (0.06)** –0.00 (0.06) –0.01 (0.05)
Use own abilities and skills
11 Workers’ Lifestyle Choices, Working Time and Job Attributes 221
0.10 (0.08) 0.44 (0.07)** –0.01 (0.07) –0.00 (0.08) 0.01 (0.05) –0.04 (0.06) –0.05 (0.03)
0.10 (0.12) –0.01 (0.16) –0.21 (0.08)** –1126 –1039 173**
0.01 (0.08) 0.26 (0.07)** –0.00 (0.07) 0.08 (0.08) 0.02 (0.05) 0.16 (0.06)** 0.02 (0.04)
0.16 (0.12) –0.09 (0.17) –0.11 (0.09) –926 –856 141**
*significant at 10%; **significant at 5%
–0.07 (0.08) Work importance 0.17 (0.07)** Work resistance –0.02 (0.07) Time constraint 0.10 (0.08) Work/home 0.01 balance (0.05) Leisure 0.14 (0.06)** Lack of child 0.00 care (0.04) Working hours Part–time 0.06 (0.12) Small part–time –0.09 (0.16) Want part–time 0.10 (0.08) Restricted –1007 Full model –956 Chi-squared 102**
Work attitude
0.22 (0.12) 0.08 (0.17) –0.21 (0.09)** –1038 –928 219**
0.03 (0.08) 0.45 (0.08)** 0.10 (0.07) –0.09 (0.08) 0.00 (0.06) 0.05 (0.06) 0.04 (0.03) 0.42 (0.13)** 0.41 (0.17)** 0.11 (0.09) –1003 –907 193**
0.01 (0.09) 0.01 (0.07) 0.03 (0.07) 0.15 (0.08)* 0.14 (0.05)** 0.36 (0.06)** 0.09 (0.04)** 0.22 (0.13)* 0.06 (0.18) 0.11 (0.08) –1020 –920 200**
–0.13 (0.08) 0.15 (0.07)** 0.04 (0.07) 0.29 (0.08)** –0.01 (0.05) 0.38 (0.06)** –0.03 (0.04)
Table 11.9 Ordered probit models, women 867 cases, all controls included Pay Benefits Career Job Favourable Holidays and level security working time days off
–0.08 (0.12) –0.11 (0.16) –0.08 (0.09) –1097 –1008 178**
0.17 (0.08)** 0.45 (0.07)** –0.01 (0.07) 0.03 (0.08) 0.06 (0.05) –0.09 (0.05)* –0.01 (0.04)
Training provision
0.10 (0.12) –0.11 (0.16) –0.19 (0.09)** –970 –837 267**
0.23 (0.09)** 0.50 (0.07)** –0.02 (0.08) 0.10 (0.08) 0.07 (0.06) 0.04 (0.06) –0.03 (0.04)
Interesting challenging tasks
0.14 (0.12) 0.04 (0.17) –0.10 (0.09) –1354 –1167 374**
0.04 (0.08) 0.03 (0.06) –0.03 (0.07) –0.01 (0.08) –0.02 (0.06) 0.02 (0.05) 0.52 (0.04)**
Child care
–0.05 (0.12) –0.09 (0.18) 0.04 (0.09) –907 –798 218**
0.29 (0.09)** 0.40 (0.07)** –0.01 (0.07) 0.11 (0.08) 0.08 (0.06) 0.07 (0.06) –0.01 (0.04)
Use own abilities and skills
222 G. Russo and E. van Hooft
11
Workers’ Lifestyle Choices, Working Time and Job Attributes
223
We now discuss the effects of the number of hours worked. Part-time workers appear to find the opportunity to work at favourable times more important than do full-time workers. This effect is especially strong among women, while men parttime workers pay less attention than their full-time counterparts to pay issues. Individuals who prefer to work part-time (regardless of whether or not they work part-time) are found to pay more attention to the opportunity to work at favourable times and less attention to job security, career and training opportunities than their colleagues preferring full-time work. Moreover, for men preferring a part-time work schedule pay is less salient than for men wanting to work full-time. Consistently with our interpretation that the work attitude index can be a proxy for ‘intrinsic motivation’ we find that, as it increases, the salience of pay (extrinsic motivator) declines, and so does the salience of the possibility to work at favourable times and of the number of holidays and days off. By contrast, a strong work attitude increases the salience of job attributes such as the opportunity to: receive training, to do interesting and challenging tasks, and to use one’s abilities.14 The work resistance index in no way affects evaluations of job attributes. By contrast, the work importance index and the time constraint index appear to be quite important. High values of the work importance index, which typify identities that emphasise the instrumentality of labour-market success to achievement of lifegoals, increase the evaluation of all job attributes except those related to the availability of childcare facilities and the opportunity to work at favourable times (for both men and women). On the other hand, high scores on the time constraint index, which signal adherence to identities that stress the instrumentality of success in non-work activities to achievement of life-goals, heighten the importance of attributes (such as the opportunity to work at favourable time and the amount of holidays and days off) that enable pursuit of the non-work activities prescribed.
11.4.1 Discussion We find that workers who prefer part-time jobs tend to place less importance on career and training opportunities. However, we do not want to stress this point excessively because of the effects of ‘cognitive dissonance’. If employers were to offer part-time workers worse working conditions (i.e. no fringe benefits, no training, low wages, no career opportunities etc.) than those of their full-time colleagues, the former, in order to protect their self-esteem and to deal with feelings of perceived unfairness, would reduce the importance attributed to such items. This behaviour would be the result of ‘cognitive dissonance’. In other words, people always seek to have consistency between their attitudes (what they 14
These items may influence an individual’s sense of control over his/her action (Benabou and Tirole 2003; Deci et al. 1999a, b; Eisenberger et al. 1999).
224
G. Russo and E. van Hooft
deem important) and their situation (what they experience). When the situation cannot be easily changed, people alter their attitudes or the psychological meanings that they attach to things in order to cope with the situation.15 The issue is a difficult one and we do not claim that we can solve it here. However, our interpretation finds some support from the observation that, all in all, workers wanting to work part-time appear to be time-constrained but otherwise intrinsically motivated. In fact, their evaluations of the opportunity to work on interesting tasks, and to deploy their knowledge and abilities, are no different from those of their fulltime counterparts. Workers preferring part-time jobs appear to attach more importance than those who prefer full-time jobs to the opportunity to work at favourable times (for them, practical times) and to the amount of holidays and days off. This should not come as a surprise, because part-time workers have other interests outside the labour market and they need time to attend to them. Most importantly, part-time workers appreciate the opportunity to use time flexibility in that they welcome the chance to squeeze work between their non-work activities. This preference may have long-run costs because managers report that they find it difficult to evaluate workers on flexible-time work arrangements (IFF 2005). This difficulty extends to the evaluation of part-time workers (a specific form of flexible time contract). In addition, part-time workers may not stand out in managers’ evaluations, and thus may not be short-listed for promotions, because of the time constraint to which they are subject. This may be the case when output is difficult to measure and managers must rely on workers’ (time) input to rank them. In the same vein, part-time workers may appear to be uninterested in training when it conflicts with their non-work activities. In so far as career decisions are taken on the basis of managers’ evaluations— which may be based on workers’ use of time—part-time workers may be expected to have reduced career and training opportunities. When this is the case, workers in part-time jobs will forgo training and career opportunities which they cannot regain once they have decided to return to a full-time schedule. In this situation, the decision to work part-time may have long-run effects, it would leave a scar.
11.5 Conclusions In the Netherlands part-time work is a widespread phenomenon with a long history. Despite the popularity of this work arrangement, part-time workers tend to receive lower pay and less career and training opportunities than do their full-time 15
This process takes place over time. Because women tend to remain in part-time employment for longer spells of time than their male colleagues and because the time span spent in temporary employment does not appreciably differ between sexes, ‘cognitive dissonance’ may help explain why women are more satisfied than men in part-time employment, while they are just as equally dissatisfied as men when in temporary employment (Petrongolo 2004).
11
Workers’ Lifestyle Choices, Working Time and Job Attributes
225
counterparts. Because part-time workers are less likely than full-time workers to progress in the labour market, the (negative) effect of forgone training and career opportunities will also be felt once workers return to a full-time schedule. In so far as part-time workers are time-constrained, and firms’ incentive schemes make additional demands on workers’ time (working long hours to obtain a promotion or additional pay, or attending training courses after the regular working day), part-time workers may respond poorly to such schemes. The low sensitivity of part-time workers to firms’ incentive schemes may thus induce managers to withhold investment from this segment of the workforce. However, part-time workers do not appear to be poorly motivated, in that they are concerned to have interesting and stimulating jobs and want to put their knowledge and skills to good use. To the extent that the career gap of part-time workers is related to their ability to receive a fair evaluation, the part-time career penalty should be lower in firms with performance evaluation systems that rely on multiple indicators. Because effective performance evaluation systems are a prerogative of large firms, the parttime career penalty (or wage penalty) could decrease with firm’s size; and it could increase with the proportion of small firms in the economy, so that the part-time career (or wage) gap could be large in countries, like Italy, characterized by a large proportion of small firms. Finally, part-time workers view the opportunity to work at (for them) favourable times as a highly important job amenity because it enables them to combine work and non-work activities. Hence, firms’ attempts to regulate part-time time slots (to accommodate firms’ needs) will reduce workers’ satisfaction (and possibly their productivity). In other words, the advantages accruing to firms (in terms of flexibility) from the ability to change part-time work shifts according to their needs may be annulled by the decrease in productivity that would likely characterize dissatisfied part-time workers who can no longer work in their preferred time slots.
Appendix: Variables
Wants part-time Dummy Labor market position Permanent Dummy Temporary Dummy Self employed Dummy Unemployed Dummy Dummy Out of the labor force (students…)
variable, 1: individual would like to work part-time variable, variable, variable, variable, variable,
1: 1: 1: 1: 1:
individual individual individual individual individual
has a permanent job has a temporary job is self employed is unemployed is out of the labour force
(continued)
226 (continued) Working hours Full-time Part-time Small part-time Jobless Education High Medium Low Gender Age Tenure Shorter than 1 year 1–2 years 2–5 years 5–10 years Longer than 10 years Unemployed l.t.
G. Russo and E. van Hooft
Dummy Dummy Dummy Dummy
variable, variable, variable, variable,
1: 1: 1: 1:
individual individual individual individual
works works works works
more than 35 h per week between 12 and 35 h per week fewer than 12 h per week 0 h per week
Dummy variable, 1: individual holds an university de- gree or comparable Dummy variable, 1: individual holds a secondary school diploma (general or vocational) Dummy variable, 1: individual holds a primary school diploma (general or vocational) Dummy variable, 1: woman Individual’s age (in years) Dummy Dummy Dummy Dummy Dummy
variable, variable, variable, variable, variable,
1: 1: 1: 1: 1:
tenure tenure tenure tenure tenure
in in in in in
present present present present present
job job job job job
shorter than 1 month between 1 and 2 years between 2 and 5 years between 5 and 10 years longer than 10 years
Dummy variable, 1: if life-long total unemployment spell longer than 1 year Households characteristics Head Dummy variable, 1: individual is head of the household Spouse Dummy variable, 1: individual is spouse of the head of the household Partner Dummy variable, 1: individual is (non-married) spouse of the head of the household Child Dummy variable, 1: individual is the child of the head of the household Partner in the Dummy variable, 1: partner present in the household household Breadwinner Dummy variable, 1: individual is the breadwinner in the household Children Dummy variable, 1: children are present in the household Urban type Individual reports him/herself to be an urban type, much or very much Table A1 Descriptive statistics Wants part-time Labor market position Permanent Temporary Self-employed Unemployed Out of the labour force Working hours Full-time Part-time
Mean
Standard deviation
0.21
0.41
0.62 0.09 0.06 0.05 0.19
0.49 0.28 0.23 0.22 0.39
0.23 0.47
0.42 0.50 (continued)
11
Workers’ Lifestyle Choices, Working Time and Job Attributes
227
Table A1 (continued) Small part-time Jobless Education High education Medium education Low education Women Age Tenure Less than 1 year 1–2 years 2–5 years 5–10 years More than 10 years Unemployed l.t. Household characteristics Household head Spouse Partner Child Partner in the household Breadwinner Children in the household Urban type Financial situation
Mean
Standard deviation
0.23 0.07
0.42 0.25
0.36 0.45 0.20 0.47 40.12
0.48 0.50 0.40 0.50 12.36
0.14 0.10 0.16 0.10 0.26 0.15
0.35 0.30 0.37 0.30 0.44 0.36
0.60 0.32 0.00 0.08 0.81 0.59 0.42 0.42 3.76
0.49 0.06 0.47 0.27 0.39 0.49 0.49 0.49 0.95
References Ajzen, I. (2001). Nature and operation of attitudes. Annual Review of Psychology, 52, 27–58. Akerlof, G. A., & Kranton, R. E. (2000). Economics and identity. The Quarterly Journal of Economics, 65, 715–753. Akerlof, G. A., & Kranton, R. E. (2005). Identity and the economics of organizations. Journal of Economic Perspectives, 19, 9–32. Altonji, J. G., & Paxson, C. H. (1988). Labor Supply preferences, hours constraints, and hourswage trade-offs. Journal of Labor Economics, 6, 254–276. Altonji, J. G., & Paxson, C. H. (1992). Labor Supply, hours constraints, and job mobility. Journal of Human Resources, 27, 256–278. Benabou, R., & Tirole, J. (2003). Intrinsic and extrinsic motivation. Review of Economic Studies, 70, 489–520. Bertrand, M., & Mullainathan, S. (2001). Do people mean what they say? Implications for subjective survey data. American Economic Review, 91, 67–72. Blau, F. D., & Ehrenberg, R. G. (1997). Gender and family issues in the workplace. New York: Russel Sage Foundation. Booth, A. L., Francesconi, M., & Frank, J. (2003). A sticky floors model of promotion, pay, and gender. European Economic Review, 47, 295–322.
228
G. Russo and E. van Hooft
Brown, G. D. A., Gardner, J., Oswald, A. J., & Qian, J. (2008). Does wage rank affect employees’ well-being? Industrial Relations, 47(3), 355–389. Clark, A. E., & Oswald, A. J. (1996). Satisfaction and comparison income. Journal of Public Economics, 61, 359–381. Cowling, M. (2007). Still at work? An empirical test of competing theories of long hours culture, MPRA Paper No. 1614. Deci, E. L., Koestner, R., & Ryan, R. M. (1999a). A Meta-analytical review of experiments examining the effects of extrinsic rewards on intrinsic motivation. Psychological Bulletin, 125, 627–668. Deci, E. L., Koestner, R., & Ryan, R. M. (1999b). The undermining effect is a reality after all— extrinsic eewards, task interest, and self determination. Reply to Esienberger, Pierce, & Cameron (1999) and Lepper, Henderlong, & Gingras (1999)’’ Psychological Bulletin, 125, 692–700. Deci, E. L., & Ryan, R. M. (2002). Handbook of self-determination research. Rochester, NY: The University of Rochester Press. Eisenberger, R., Pierce, D. W., & Cameron, J. (1999). Effects of rewards on intrinsic motivation—negative, neutral, and positive: a comment on Deci, Koestner and Ryan (1999). Psychological Bulletin, 125, 677–691. Euwals, R. (2005). Recent developments in part-time work in eu countries: trends and policy discussion. In R. Gómez-Salvador, A. Lamo, B. Petrongolo, M. Ward, & E. Wasmer (Eds.), Labour supply and incentives to work in Europe. Cheltenham: Edward Elgar. Euwals, R., & Hogerbrugge, M. (2006). Explaining the growth of part-time employment: Factors of supply and demand. Labour, 20(3), 533–557. Francesconi, M., & Gosling, A. (2005). Career paths of part-time workers. EOC working paper series, 19. Freeman, R. B. (1998). War of models: Which labor market institutions for the 21st century? Labour Economics, 5, 1–24. Fuchs Epstein, C., & Kalleberg, A. L. (2004). Fighting for time: shifting boundaries of work and social life. New York: Russell Sage Foundation. Fuchs Epstein, C., Seron, C., Oglensky, B., & Sauté, R. (1999). The part-time paradox: Time norms, professional life, family and gender. New York: Routledge. Gibbons, R., & Waldman, M. (1999). Careers in organizations: theory and evidence. In O. A. Ashenfelter, & D. Card (Eds.), Handbook of labor economics. Amsterdam: Elsevier Science. Godin, G., Conner, M., & Sheeran, P. (2005). Bridging the intention-behaviour ‘gap’: The role of moral norm. British Journal of Social Psychology, 44, 497–512. Goldin, C. D. (1997). Career and family: College women look to the past. In F. D. Blau, & R. G. Ehrenberg (Eds.), Gender and family issues in the workplace. New York: Russel Sage Foundation. Green, C. A., & Ferber, M. A. (2005a). Do detailed work histories help to explain gender and race/ethnic wage differentials? Review of Social Economy, 63, 56–85. Green, C. A., & Ferber, M. A. (2005b). The long-run effect of part-time work. Journal of Labor Research, 26, 56–85. Green, C. A., & Ferber, M. A. (2005c). The Impact of part-time work on women’s and men’s careers: Timing, length, and motivation. University of South Florida, COBA-ECON-05-01. Grodner, A., & Kniesner, T. J. (2006). Social interactions in labor supply. Journal of the European Economic Association, 4, 1226–1248. Hagger, M. S., & Chazisarantis, N. L. D. (2005). First- and higher-order models of attitudes, normative influence, and perceived behavioural control in the theory of planned behaviour. British Journal of Social Psychology, 44, 513–535. Hakim, C. (2000). Work-lyfestile choices in the 21st century: Preference theory. Oxford: Oxford University Press. Konrad, A. M., Ritchie, E. J. J., Lieb, P., & Corrigall, E. (2000). Sex differences and similarities in job attribute preferences: A meta analysis. Psychological Bulletin, 126, 593–641.
11
Workers’ Lifestyle Choices, Working Time and Job Attributes
229
Landers, R. M., Rebitzer, J. B., & Taylor, L. J. (1996). Rat race redux: adverse selection in the determination of work hours in law firms. The American Economic Review, 86, 329–348. Malcomson, J. M. (1999). Individual Employment Contracts. In O. A. Ashenfelter, & D. Card (Eds.), Handbook of Labor Economics. Amsterdam: Elsevier Science. Manning, A., & Petrongolo, B. (2008). The part-time pay penalty for women in britain. Economic Journal, 118(526), F28–F51. OECD. (2003). Employment outlook. Paris: OECD. Petrongolo, B. (2004). Gender segregation in employment contracts. Journal of the European Economic Association, 2, 331–345. Prendergast, C. (1998). What happens within firms? In J. Haltiwanger, M. E. Manser, & R. Topel (Eds.), Labor statistics measurement issues. Chicago, IL: The University of Chicago Press. Prendergast, C. (1999). The provision of incentives in firms. Journal of Economic Literature, 37, 7–63. Riedmann, A., Bielenski, H., Szczurowska, T., & Wagner, A. (2006). Working time and work-life balance in European companies. Dublin: European foundation for the improvement of the working conditions. Roman, A., Fouarge, D., & Luijkx, R. (2004). Career consequences of part-time work: Results from Dutch panel data 1990–2001, OSA Publicatie, A 206. Russo, G., & Hassink, W. H. J. (2008). The part-time wage gap: a career perspective. De Economist, 156(2), 145–174. Salaniè, B. (1997). The economics of contracts: A primer. Cambridge, MA: The MIT Press. Schettkat, R., & Yocarini, L. (2001). Education driving the rise in dutch female employment: Explanations for the increase in part-time work and female employment in the Netherlands, contrasted with Germany. IZA Discussion Paper, 407. Thompson, J. A., & Bunderson, S. J. (2001). Work-nonwork conflict and the phenomenology of time: Beyond the balance metaphor. Work and Occupations, 28, 17–39. Thorsteinson, T. J. (2003). Job attitudes of part-time vs full-time workers: A meta analytic review. Journal of Occupational and Organizational Psychology, 76, 151–177. Van Hooft, E. A. J., Born, M. P., Taris, T. W., & van der Flier, H. (2005). Predictors and outcomes of job search behavior: the moderating effect of gender and family situation. Journal of Vocational Behavior, 67, 133–152.
Part IV
Non-Standard Employment, Quality of Work and Firms’ Outcomes
Chapter 12
Non-Standard Employment and Quality of Work: Towards New Forms of Measurement Tindara Addabbo and Giovanni Solinas
12.1 Introduction In the wake of an intense season of research in the 1960 and 1970s, of which perhaps the key European point of reference was the Tavistock Institute of London (Baldamus 1961), in the following two decades, the studies by historians, sociologists and economists into the conditions and the quality of work underwent a substantial setback. Only at the end of the 1990s did we witness renewed interest in the theme, of which an emblematic example may be found in the surveys promoted by the Oecd and the Eurofound (European Foundation for the Improvement of Living and Working Conditions) of Dublin. This renewed interest on the part of the social sciences in the theme of work and the conditions in which it is supplied was due to a number of different factors. Among these, a role was certainly played by the tertiarisation of the main economies, the technological change and that of the organisational and production models, the change of welfare regime and the rise of flexible/precarious work, the changes in the representation mechanisms and the changing role of the trade union in the most advanced economies. These factors, interconnected in various ways and, at least in part, retraceable to the transformation/integration of the world economies, have led to a radical change in work and its traditional functions in terms of defining individual and collective identities, with evident consequences on the processes of labour market structuring, posing new questions to social and work policies on a national
T. Addabbo (&) G. Solinas (&) Department of Economics, University of Modena and Reggio Emilia, Viale Berengario 51, 41121 Modena, Italy e-mail:
[email protected] G. Solinas e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_12, Springer-Verlag Berlin Heidelberg 2012
233
234
T. Addabbo and G. Solinas
and international level. In this situation, the study on the conditions and quality of work has taken on new importance. In the present chapter, we shall start out from the taxonomy proposed in a groundbreaking study by Gallino (1993). In defining work and connoting it in terms of qualitative characteristics, Gallino identifies four key dimensions, which the author presents as follows: ‘‘The mass of data available makes it possible for us to state that the concept of the quality of work would not be adequate to the functions of interpretation, critical analysis and planning […] if it did not cover at least four different dimensions: a. the ergonomic dimension, which corresponds to the psycho-physical needs of the worker; b. the complexity dimension, which corresponds to the needs of commitment in terms of overcoming difficulties, of creativity, of professional training, and of the accumulation of work experience; c. the autonomy dimension, which corresponds to the need to self-determine the rules to follow to carry out the activities allocated to an individual (or group) in order to achieve given production goals; d. The control dimension, which corresponds to the need to control not only the ongoing conditions of the work process (like in c.), but also the general working conditions, such as the object of production, its application, the organisation and the activities to assign to oneself and to others.’’ Gallino concludes: ‘‘High quality work is thus that which in each of these dimensions presents properties able to satisfy the relative needs to a high degree’’ (Gallino 1993, p. 393).1
Re-examined 30 years on from its formulation, despite maintaining an extraordinary heuristic/explanatory power, Gallino’s taxonomy has at least two clear limitations. The first is its not considering adequately the relational/social dimension of work; in other words, that which may be defined as the recognition of a status within the community that the worker belongs to. The second, highlighted by a great number of gender studies, concerns the possibility, given the organisational model, of making compatible or ‘‘reconciling’’ living and working conditions. On the basis of these considerations, we have proposed an interpretational model which—drawing on Gallino’s taxonomy, introduces two new dimensions: the social dimension, meant as a need for social recognition, and the work-life balance dimension, aimed at satisfying the need for compatibility between working conditions and the management of everyday life. Gallino’s model, modified as suggested above, was thus broken down into more than 40 single items measurable quantitatively according to a suitable scale, and complete with a vast range of control variables which may be traced back to the various dimensions of the quality of work. The model was subjected to an initial verification in various research projects: it was proposed using a sample of some of the most representative companies to be found in the province of Modena; it was replicated on a sample of engineering firms in one of the most advanced industrial sectors (the packaging companies present within the province); it was also proposed in the sector of care for the elderly (including social cooperatives, public rest
1
This and the other citations from works originally in Italian have been translated by the Authors. Usual disclaimers apply.
12
Non-Standard Employment and Quality of Work
235
home structures and private homecare assistants). Over the course of three years, these studies provided the most quantitatively significant database on the issue of the quality of work of all those available on a national level. Although it does not constitute a statistically representative sample, it provides individual evaluations on the quality of work as perceived by a total of around 2,500 workers. The results make it possible to highlight significant differences within single groups and, on a common and homogeneous basis of comparison, between the different groups, and which outline the characteristics of workers, as well as prevalent organisational models, contractual conditions, the technology of productive processes and, more in general, the conditions of the request for work. The results obtained, in other words, constitute a promising basis on which to reason on the patterns of segmentation/structuring of the labour markets emerging in an advanced economy in rapid transformation. The interpretational scheme proposed by Addabbo et al. (2007) exploits fuzzy logic to provide the formulation of a fuzzy expert system, and will be applied in this chapter to the interpretation of the quality of work in non-standard employment. The research path chosen is not immune to criticism: many studies by work psychologists underline how there exists a systematic distortion between the subjective perception of the quality of work and the objective conditions in which work is performed. Also Sen convincingly argues that work which is ‘‘poor’’ in terms of capabilities will tend to be evaluated positively by the subjects that perform such work. The more the worker is ‘‘fragile’’, the lower his/her expectations of work are and the better his/her evaluation will be of the various dimensions that characterise his/her work. In Sect. 12.2, we shall examine the state of the art with regard to the definition of work and of the quality of work. In Sect. 12.3 the various dimensions of the quality of work are presented, including those used to define the quality of work following the taxonomy proposed by Gallino, and the empirical evidence is also presented. In Sect. 12.4 we shall examine the interpretational scheme followed in this chapter through the presentation of an expert fuzzy system. The expert system is thus applied to the measurement of the quality of work with special reference to non-standard employment (Sect. 12.5). A number of brief notes then follow in the conclusion.
12.2 The State of the Art and the Empirical Evidence 12.2.1 The Central Role of Work in Western Thought Work, working conditions and the quality of work have long been (and will presumably continue to be so in the foreseeable future) a fundamental element of Western thought. The most immediate way to understand this is perhaps to make reference not to the corpus of the social sciences, in which—albeit on radically
236
T. Addabbo and G. Solinas
differing suppositions—in one way or another work lies at the heart of the process of creating value, and neither to the political and social movements which have built the foundations of their collective identity on work, but rather on the thought of the great Western religions. It is well known that in the Protestant ethic, ‘‘work is meant as Beruf, i.e. as the task for which a person feels that he is particularly suited on the basis of his innermost qualities and by which he will be subjected to divine justice’’ (Lazzarini 2009, p. 94). In this sense, the valorisation of work comes across as a response to a divine calling. Also in the Catholic tradition, work plays a key role. In the social doctrine of the Church, for all intents and purposes the vision of work as a malediction has been overcome, and the expressive rather than the instrumental dimension of work prevails. Through work, man transforms nature and realises himself as a man and, ‘‘in a certain sense, becomes more of a man’’ (John-Paul II, Laborem Excercens, 1981). In the Compendium it is said that ‘‘Beyond being a decisive paradigm in social life, work has all the dignity of a field in which both the natural and supernatural vocation of the person must be realised’’ (Compendium of the Social Doctrine of the Catholic Church 2004, No. 101, p. 56). A notion of work emerges as a space of relationships, of the shared construction of the common good, and as a fundamental area of an individual’s life. Today there are a lot more people working than there have ever been in the entire history of humanity. And the underlying questions on work, as Manghi observes, are substantially the same as they were a century ago (‘‘what space does work take up in life in quantitative terms, how much emotional intensity surrounds it […], how and how much does it contribute to defining social status, or to producing identity, […], conflicts and sense of belonging’’, Manghi 2009, p. 99). And yet there is no doubt that a great number of onlookers and scholars—from at least partially differing perspectives—have supported the loss of the centrality of work in defining individual and collective identities (from Touraine, among the first and most original, to the Italians Accornero and De Masi, to the somewhat controversial contribution of Rifkin himself). Processes of stasis and decline in studies on work have arisen in various disciplines, from sociology to the anthropology of history, to the historiography of work (Causarano et al. 2008; Musso 2002; Van der Linden 1993). In the field of the social sciences, perhaps the most emblematic path is that of the progressive separation of industrial sociology and the theory of organisations, and the substantial merging of the latter into the neoclassical paradigm of business theory. The following section is dedicated to providing a brief outline of this path and to highlighting its limitations.
12
Non-Standard Employment and Quality of Work
237
12.2.2 The Evolution of the Theory of Organisations and the Disappearance of Work In the first half of the last century—from the works of Taylor (1911) up to those of Blau (1955)—the study of work (as the primary object of industrial sociology) and the forms of labour organisation were effectively one and the same thing (in particular, Chinoy 1955; Dalton 1950; Gouldner 1954; Walker and Guest 1952; and the above-cited Blau 1955). With the exception of the studies based on the socio-technical paradigm, which revolved largely around the Tavistock Institute in London (in particular, Trist and Bamforth 1951; Rice 1963; Mumford 1967; Herbst 1974) as well as the studies drawing on the participatory models in Germany and the Scandinavian countries throughout the 1960 and 1970s, the analysis of organisational forms slowly became detached from the analysis of work. As Barley and Kunda (2001) note, the concept of reference for organisational theories focused on the general organisational principle which allows for and promotes forms of collaboration within a group (Miller and Rice 1967). These processes had far-reaching consequences: they changed disciplinary parameters; organisationalists and work scholars founded new journals of reference as so on. The idea emerged according to which the actions of man even within production processes are not isolatable from the resulting organisational/institutional forms. The analysis of work became a specific (complementary) branch of sociology, and the scholars of bureaucratic forms (and thus of organisations) clearly started to move towards the schools of business and economic analysis tout court. Perhaps the clearest symbolic act is the configuration that the sections of reference were provided in the USA by the American Sociological Associations from the 1970s onwards. This process has not been a painless one from a cognitive point of view: in the studies of organisational forms, working conditions are largely overlooked in the analyses, and replaced with ever more stereotypical canons. The figure of the man at work is no longer. There is merely his petrified image (to cite the term coined by Barley and Kunda) of which the greatest testimony is perhaps constituted by the flood of essays on Fordism and post-Fordism a few years later. The ideal-type of the worker on the production line (and, conversely, of all that which differs from it), inspired thousands of pages of somewhat blurred and approximate analyses of working conditions and processes. Stereotypes on working conditions and quality were adopted yet which were generally inadequate to describe the transformation processes taking place. Of all these, the difficulty in classifying professions (which involves all the main statistics offices in industrialised countries) provides a clear example. Following a completely different path, the company theory independently evolved towards a theory of organisational practices. This path is outlined in great detail in the articles by Milgrom and Roberts (1992) and Roberts (2004). Company theory, breaking away from the study approaches which emerged in the 1930s, is basically configured as a theory of organisational practices. As Anna Grandori
238
T. Addabbo and G. Solinas
(an organisationalist in disciplinary terms) observes in the introduction to the Italian edition of Roberts’ volume, organisational practices ‘‘include mechanisms of production management, of human resource management, of planning and control, as well as the division and coordination of work, considered in their real applicative form in the company: flexible versus rigid machinery; specialisation or polyvalent skills; vertical or horizontal communication; the intensity of teamwork, of hierarchical interventions, of monetary incentives, of outsourcing, etc.’’ And nevertheless, from this perspective, work no longer exists. The conditions in which work is performed, just like the various analyses of working processes and production processes in the theory of industrial organisation, basically constitute the ‘black box’, not giving any account of the conditions in which work takes place, nor of the evaluation that workers themselves give of it: both remain external with regard to the object of the analysis. In other words, there are two main underlying points (and limitations) of convergence between the organisational theory and modern company theory. The first was reproposed recently in Italy by Butera (2009), re-examining the point of view of the socio-technical school (Touraine, Perrow, Gouldner). ‘‘It is not possible to understand social relationships unless the technological resources and production process on which social processes are built are studied.’’ He continues, ‘‘Certain aspects of the social relationships of the necessary cooperation between working people cannot be understood unless the product creation stages are studied […]: what goes in and what comes out of a factory, how prime materials are transformed, what kind of machinery is used, how long it takes, where the production line may be interrupted etc.’’ (Butera 2009, pp. 21–22; see also La Rosa 1993). Obviously, the same reasoning may also be applied to services. The second element, common to both organisational theory and contemporary company theory, is that the incentive schemes (and contracts) are entirely based on individual motivation, yet nothing is said of the weight of the subjective perception of work, which is the underlying basis of the motivation. That which Barnard highlighted as far back as the early twentieth century (1938) is completely overlooked: behaviour depends on latent structures of which the basis is inevitably made up of elements of sharing/estrangement, thus providing a positive/negative perception of one’s own working conditions.
12.2.3 Work and the Quality of Work in a Changing World: The Great Transformations For the above-mentioned reasons, reintegrating the studies of organisational forms and of working conditions should be one of the high-priority issues on the social sciences research agenda. Bringing back the attention to the characteristics of work, just as took place to a certain extent with product analysis, is a key issue. And even more so in a changing world.
12
Non-Standard Employment and Quality of Work
239
12.2.3.1 The Rise of Flexible Labour, Changes in the Welfare Regime and Trade Unions The first great transformation was the rise of flexible labour. Flexible labour coupled with a generally unfavourable demand for work, an inadequate and contradictory right to work, a training system falling apart, a fragile welfare system, and lastly, the behaviour of companies attempting to minimise the almost fixed costs of labour (according to the canonical articulation through which, in his day, Oi defined them) over a short time span, make workers precarious, with permanent consequences on their working careers and the income expected by a great number of newcomers to the job market (Berton et al. 2009). And not only in Italy. In several activities (training, research) and in several professions (journalists, professional consultants, translators, etc.) a stable contract is not necessary in order to guarantee good conditions of work and pay (Paci 2005; Saraceno 2005; Semenza 2000). However, in many other areas it is, and here most observers underline risks of social marginalisation and exclusion. Gallino (2001) observes that fragmented and uncertain work risks producing fragile life stories and identities, to the point of mining workers’ self-esteem. Sarchielli (2008) suggests that in this sphere, the changes in work ‘‘are so radical as to risk compromising the very social status of work itself, that which— according to Jahoda’s well-known distinction (1982)—may be defined as explicit positive functions (first and foremost that of guaranteed income), and latent positive functions (identity, relations, social integration, etc.). Beck (2008) speaks of ‘‘endangered life stories’’; Sennet (1999) of the ‘‘corrosion of character’’. Albeit with a different background, discipline and language, the conclusions that Boeri (2009) reaches are not all that different. Conversely, an analytical element shared by many is that ‘‘an efficient organisational framework’’ calls for ‘‘not so much the flexibilisation of the working relationship, but rather its stabilisation’’ (Leoni 2008). There is one further element which completes and renders this framework even more articulate: the changes in collective action and workers’ representation. For economists, this has traditionally constituted unchartered territory. In a wellknown passage, Ezio Tarantelli observed that: ‘‘economists’ approach to trade unions has often been akin to that of mediaeval monks when faced with a Greek word found in a manuscript. Unable to read it, they would note it down and carry on regardless: graecum est, legi non potest’’ (Tarantelli 1986, p. 23). The history of efficient negotiation models provides no grounds on which to modify Tarantelli’s judgement. And yet, if we look at the labour market from the supply side, then as now, the role of worker representation and the trade unions is fundamental. In fact what has changed the supply conditions is the growing difficulties in providing trade union protection. As Baglioni observes, ‘‘in order to explain the difficulties of trade union protection, the prevailing thesis—well-founded yet not exhaustive—identifies the main cause in the rise of globalised markets, starting from the labour market’’. In
240
T. Addabbo and G. Solinas
general terms, this is the thesis that Reich (2008) popularised. Yet it is clearly over-simplistic. ‘‘Other factors (not always separate from globalisation) also obstruct the presence and efficacy of trade union action. These factors have to do with the aging of the population and the increasing costs of welfare spending; the growing spread and importance of an imposed minimum wage; […] the difficulties of representing the vast range of professional, occupational and contractual profiles; the increased role of trade unions (be it real or presumed) as a ‘corporative’ and conservative player […]; the tendency to look upon workers as just another item on the list of company stakeholders’’ (Baglioni 2009, p. 152). This range of factors, in Italy in particular, has not led to a fall in the level of trade union membership (on the contrary, those enrolled in the main three trade union confederations in Italy over the last 20 years has grown by around 32%), but to a radical change in its make-up, with a sharp decline of around one million members in the private sector and, in particular, in manufacturing, against a rise among public sector workers and immigrants (especially in the services sector) and, above all, among pensioners. The underlying datum is that the growing share of young people with fixed-term contracts in thus structured trade union organisations, unsurprisingly, is not given suitable expression and representation (Tronti 2009).
12.2.3.2 Changes in Technology, Production Models and Organisational Models The second aspect that must be borne in mind is that of technical and organisational change. The bonds between the changes in technology, the forms of work organisation and the requirements demanded of the worker are somewhat complex, and to this day have been given no theoretical framework. Several scholars argue that the introduction of digital technologies increases the complexity of productive processes and requires a higher level of skills (Johnston and Packer 1987). Others have pointed out changes in the opposite direction (Aronowitz and De Fazio 1994). Yet others have highlighted contrasting tendencies in terms of the qualification of work (Diprete 1988; Spenner 1995; Zuboff 1988). And there are also those who claim that the change brought about by technology per se has far less radical effects on the qualification of work than one might think (Gallie 1994). Once again, whether and to what extent technological change influences the requirements demanded of the worker depends on the way in which the technology has been designed, implemented and used in a specific organisational context (Barley and Kunda 2001). While the link between technology and the contents of work/professional skills required is controversial, nobody, however, doubts that the advent of microelectronics, of electronics applied to processes, of robotics and information technologies in general have radically changed (and are still doing so) organisational and production models. Furthermore, the change of techniques and the crisis of Taylorism/Fordism (which has been growing with increasing speed since the end
12
Non-Standard Employment and Quality of Work
241
of the 1970s) have led to a proliferation of industrial models (Kern and Schumann 1991). Boyer and Freyssenet (2005) state that in the car industry alone, today at least six completely different production/organisational models (in terms of product policy, the organisation of production, wage management, tasks required of workers and general working conditions) coexist. The very emergence, over a relatively brief time span, of a flood of notions such as ‘‘virtual organisations’’ (Byrne), ‘‘shamrock organisations’’ (Handy), ‘‘network organisations’’ (Powell), ‘‘boundaryless organizations’’ (Arthur and Rousseau), lean structures (Woomack et al.) and many others is proof of it. There is perhaps also another phenomenon, which may be noted particularly with regard to services. For many working figures, on the intermediate or executive level, traditional professional profiles are blurred and we are faced with a complete redefinition of their roles. Alongside numerical flexibility, a functional flexibility emerges (and not always with positive connotations) with great force. The following example is typical of this trend: ‘‘Once in all banks, […], there were the desk clerk and the cashier with two distinct professional profiles; then came the desk clerk-cashier, and then the desk clerk–cashier–consultant, leading to the desk clerk–cashier–consultant–promoter and so on, merging, subtracting and redesigning the role on the basis of the various banks’ varying commercial and management policies.’’ (Ceri 2009, p. 147). In these fields, the task base in broadened, but not enriched. This is a far cry from that process of labour evolution that the theorists of job enlargement/job enrichment had hoped for (Davis and Taylor 1972). The underlying point is that on the production line, in a chemical or metalworking factory or a press shop, in automated workstations, in the engine room of a ship, in the control centre of a power plant, in banks, in hospitals, in rest homes, in the care of children and the elderly, whatever the reference parameters adopted are, work has changed radically. Suffice to look through the recent volume edited by Riccardo Leoni (2008) to understand how much this process is widespread and pervasive, also in Italy (see also Brugnoli 2008 and Leoni and Albertini 2009). For all of the reasons that have been outlined in brief in the previous pages, each of which are interconnected to a greater or lesser extent, today once again ‘‘there is an air […] of widespread anxiety with regard to the quality of work, its professional content, its stability, the adequacy of its remuneration, without forgetting the human climate in which it takes place’’ (Manghi 2009, p. 107). It is also for this reason that the redefinition of work in people’s existence and in society is once more one of the key topics examined by economics and the social sciences. It is in the very era of large-scale transformation, shown once more by seminal works of Weber, Durkheim and Marx, that work and the quality of work return to the spotlight. Among the many possible developments deriving from the considerations examined, the one that we have chosen to privilege here has to do with the conditions and the quality of work according to the subjective perception of the worker. The way in which work returns to centre stage is the way in which it is judged by those who carry out that work.
242
T. Addabbo and G. Solinas
12.3 The Quality of Work: Definitions and Empirical Evidence As we attempted to show in the previous section, in work—meant as the manufacturing of goods and the provision of services, in industrialised nations—over a quarter of a century everything has changed. The technology has changed, with the more or less widespread introduction of electronics and information technologies, and of information in process management and control. The organisation has changed, designed to guarantee quality and conformity based on pre-defined product standards, the customisation of goods and services, speed of delivery, in a system in which a potential competitor is no longer (only) one in the same geographical area, but could be anywhere in the world. These two transformations—in technology and organisation—in turn have radically modified the abilities and skills required by companies: even when there is large-scale customisation of goods, growing (although not exclusive) responsibility for this is placed on the control mechanisms of machinery and not the skills and versatility of ancient crafts, in other terms recreating the functional conditions of work traditionally associated with mass production. These transformations are no less radical if we look not at the company but the workers: the definition of professional roles has become more and more blurred; the bond with the company and the workplace has weakened; career expectations have become far more uncertain, and the identity mechanisms associated with work have also been altered. Lastly, the mechanism of collective action and representation has been transformed. As said before, everything has changed. The study presented in these pages looks at the transformation of work from a specific point of view: albeit in a circumscribed environment—and with a methodology part of which has yet to be decided on—we shall attempt to provide a measurement of the subjective perceptions of workers of the quality of the work they do. In particular, the research focuses on the theme of the quality of work, in its various dimensions, and investigates the connections that exist between workers’ perceptions and the quality of work (and hence their degree of satisfaction), comparing them to the contractual and remunerative characteristics of their employment agreements, with particular reference to non-standard employment (Sect. 12.5). In the first part of this section, while starting out from the taxonomy put forward by Gallino, we shall define the various dimensions of the quality of work before analysing the empirical evidence gathered.
12.3.1 The Dimensions of the Quality of Work The notion of the quality of work, as has been widely recognised by scholars from a diverse range of disciplines, draws on a number of different dimensions, each of which is measured by a different group of indicators/reference variables. From the
12
Non-Standard Employment and Quality of Work
243
socio-economic point of view (Gallino 1993; La Rosa 1983; Merllié and Paoli 2001; Sestito 2002) there is a broadly shared abandonment of the income dimension as the element which might best sum up and define a good job. ‘‘The concept of quality of work,’’ notes Gallino (1993, p. 393), ‘‘summarises the entire field of variation of the work phenomenon—of all the types and forms in which it may manifest itself.’’ High-quality work is a job able to satisfy the needs of the worker. From this point of view, by modifying the scheme originally proposed by Gallino, the six following dimensions are taken into consideration: 1. 2. 3. 4. 5. 6.
The The The The The The
social dimension; economic dimension; work-life balance dimension; complexity dimension; organisational dimension, of autonomy and control; ergonomic dimension.
12.3.1.1 The Social Dimension of Work The first dimension taken into consideration is the social one. There are four aspects for which a form of measurement is provided: 1. relationships within the workplace (with co-workers and superiors), attributing great significance to the possibility for workers to take part in the decisions concerning their own working context; 2. the social recognition of work (based on the social recognition of merit, of specific professional skills, and of the consideration—in terms of status—received by members of a given community); 3. the individual satisfaction of the worker, also expressed in relationship to the perception of being able to grow in professional terms and to gain promotion; and lastly, 4. the relationship between work and social life (including the possibility of managing one’s time on a weekly and daily basis, and also with reference to holidays). The social dimension, insofar as it is inherently relational, also sums up aspects connected to the economic and organisational dimensions, as well as that of the work-life balance, which are explored more thoroughly below.
12.3.1.2 The Economic Dimension of Work In the economic dimension, most attention was focused on the aspects connected with the present and future perception of workers’ income and their welfare coverage as a whole. In particular, on examining the economic dimension of work, attention was paid most of all to four main aspects: 1. the level and variability of retribution (overall retribution, possibility of making use of remuneration in kind, performance-linked bonuses, seniority-based bonuses); 2. the perception of job
244
T. Addabbo and G. Solinas
security and social safety nets; 3. family rights (perception of maternity rights and the possibility of making use of holidays and leave in order to maintain family relationships); 4. expected income (an approximate evaluation of future work training and perceived career opportunities).
12.3.1.3 The Work-Life Balance The work-life balance dimension specifically concerns those environments (and those means) that the company uses to make working time compatible with the other interests of the worker, first and foremost the worker’s relationship with his/ her own family nucleus. Once more, the notion is that a job is of good quality if it is able to respond to the need for the job to co-exist with ‘‘other’’ aspects. In this perspective, all of the aspects that may be implemented throughout one’s working career, the year and in the everyday management of work times are examined.
12.3.1.4 The Complexity of Work Dimension For a range of objectives, typical of the various company functions, the complexity of work reflects the ability of the worker to respond to the needs of the company, as expressed in the tasks assigned to him/her. In this perspective, a job is of good quality if it is able to develop such abilities over time. In rendering this notion operational, essentially two fields were taken into examination: 1. the difficulty/ variety of the job, the repetitivity of the tasks and the overall commitment required, and 2. the skills required (measured in terms of the work preparation necessary and the need for professional training).
12.3.1.5 The Organisational Dimension, of Autonomy and Control The organisational and control dimension concerns the worker’s potential for being informed of, affecting and interacting with the decision-making process of a given organisation. In the analysis of this dimension, prevalently three aspects were focused on: 1. the degree of autonomy with regard to the ways in which the work is organised; 2. the internal relationships with co-workers, direct superiors and top management; 3. participation in the decision-making process with regard to one’s own specific working context.
12.3.1.6 The Ergonomic Dimension of Work The ergonomics of work may be functional in terms of satisfying three of the worker’s needs: a safe working environment which does not put the worker’s physical health in danger; an organisation of the process which does not entail a
12
Non-Standard Employment and Quality of Work
245
waste of physical resources, and lastly, a working environment which maintains the psychological wellbeing of the worker. Gallino’s analytical scheme, later re-elaborated in Italy by La Rosa (1983), is substantially modified here by the writers, introducing two new dimensions: the social dimension and that of the work-life balance. The reasons behind this extension are easily understandable, and concern both the change in the role of work in contemporary societies as it is perceived by the main social actors, and at the same time, the way in which observers and social scientists take into account or, in some cases, anticipate the way in which society (and in particular, western society) look upon work. The social dimension concerns the network of relationships (within the single workplace and with the outside world) that a specific network of relationships is able to activate. The work-life dimension concerns the conditions of compatibility, and therefore the use of time as split between work and other activities. Both dimensions, in different ways, take account of the growing weight of that which is ‘‘non-work’’ in the definition of individual identity, or self-perception, of the worker. The second innovation proposed in this study is that of trying (albeit as an initial approximation) to activate and supply a measurement of the initial analytical scheme. Each of the dimensions, in fact, is measured by a range of variables, precisely defined for each dimension, and measured exclusively in relationship to the subjective perception expressed by the worker (Sect. 12.4).
12.3.2 The Quality of Work in Empirical Research both in Italy and Europe With the aim of monitoring one of the objectives of the European Strategy of Lisbon—that of creating ‘‘more and better jobs’’—indicators were proposed based on criteria partially different from those discussed in the previous pages. Both the European Commission and the International Labour Office (Ilo) identified various dimensions with regard to the quality of work, and proposed a system of indicators by which to measure them. Leschke et al. (2008) put together an indicator known as the Job Quality Indicator (JQI) which could be broken down into six sub-indicators aimed at measuring salaries, skills, training and career prospects, working conditions and job security, participation and representation, contractual conditions, work time and the work-life balance. The index is obtained by carrying out a weighted average of the results obtained in the various dimensions, and using different sources of data including not only the European Foundation for the Improvement of Living and Working Conditions (Eurofound) surveys, but also the European Union Statistics on Income and Living Conditions (EU SILC) and the European Countries Labour Force Surveys.
246
T. Addabbo and G. Solinas
The dimensions on which the measurement proposed in European Communities (European Communities 2008) focused also referred to the system of indicators defined by the European Council of Laeken, and are to do with wages and socioeconomic security, working conditions and intensity, skills and training, and the work-life balance. European Communities (2008) makes use of an analysis of the main components as well as a cluster analysis, also showing the accuracy of the measurements obtained through the various methodologies, and leading to the grouping of the European countries analysed on the basis of the results obtained in the various dimensions analysed. Other proposals may be found in the studies of Clark (2009), Tangian (2007) and Seifert and Tangian (2009) and in the works presented at the fourth international seminar on the measurement of the quality of work (Unece 2007). Using the International Social Survey Programme between 1989 and 1997, and the British Household Panel 1991–1999 (with regard to the answers on the questions on job values), Clark (2005) shows that if we focus only on the monetary dimensions or on working time, we risk underestimating the deterioration of other elements with regard to the quality of work in Oecd countries which came into play over the 1990s. On the basis of Clark’s analysis (2005, p. 386) Italy in particular shows a low level of job satisfaction (along with the UK and Hungary). A similar result by comparing the studies of the European Foundation may be seen in Isfol (2004). We may note a reduction in the level of satisfaction in work itself, while at the same time there is greater satisfaction with regard to wages and job security. Clark also shows that there was a rise in inequality as concerns job outcomes (the young and the most highly educated obtained the best results in terms of job quality, and the members of trade unions in terms of job security). Considering job satisfaction on the basis of the various characteristics of the job itself, when comparing genders, with various data sources for the UK, Rose (2005) shows that several different trends were present, with a fall in terms of job satisfaction among women and a stable or rising trend among men. On the basis of the analysis carried out by Green (2004) and by Green and Tsitsianis (2005) using survey samples (1992, 1997 and 2001), the decline of job satisfaction in the UK appears to be associated with the intensification of energies concentrated on work, and the lower degree of discretion granted to workers in the everyday performance of their tasks. The European Foundation for the Improvement of Living and Working Conditions in Dublin carries out a survey every five years into working conditions. The survey makes use of a random and statistically significant sample of the population in various European countries. There are currently five surveys available (from 1991 to 2010). On the basis of the analysis of the European Working Conditions Surveys, Burchell et al. (2007) identify the permanence of gender inequalities in many aspects of working conditions in European countries. The use of information on job satisfaction with regard to various items shows a higher average level of job satisfaction among women (Clark 1997; Sousa-Poza and Sousa-Poza 2000) albeit in decline (Sousa-Poza and Sousa-Poza 2003) and
12
Non-Standard Employment and Quality of Work
247
with differences between countries (Sousa-Poza and Sousa-Poza 2000; Kaiser 2005). The link between intensification and the various dimensions of work analysed by gender was examined by Burchell and Fagan (2004) using the European Foundation’s European Working Conditions Survey. Burchell and Fagan (2004) show the existence of a negative effect on the work-life balance linked to the presence of long and unsociable hours. However, this difference is lower among the younger and the most highly educated, those in professional and managerial positions, and among the female workers whose mothers were professionals or who had more male-dominated positions (Clark 1997). The difference does not seem referable to the sample-selection (Clark 1997; Sousa-Poza and Sousa-Poza 2007), but rather linked to the lower expectations of women (Clark 1997). According to Kaiser’s analysis (2005) of European Community Household Panel (ECHP) data, the differences in levels of job satisfaction are connected to the differences that may be observed with regard to job opportunities and the degree of job market modernisation. In fact, there appears to be a positive correlation between the degree of restriction in access to the job market among women and the emergence of the ‘gender-satisfaction paradox’ (meaning a greater level of satisfaction for women, despite their working conditions being worse on average). The literature has also highlighted differences in the degree of job satisfaction with regard to a range of job items, and the importance of different elements that contribute to the quality of work together with the heterogeneity of gender with regard to age, type of work and level of education (Clark 1997; Sloane and Williams 2000; Sousa-Poza and Sousa-Poza 2007; Bender et al. 2005), and these results lead us to bear gender differences in mind more than other individual characteristics when analysing the type of work and the different elements of working conditions. The difference in job satisfaction by gender in more female-dominated roles is found to be connected to the greater level of flexibility that these jobs are able to provide leading to a better life-work balance (given the role that women often have in terms of housework and caring duties): a dimension to which women attribute a greater value (Bender et al. 2005). A link lesser than that generally expected in the literature was found between the demand for a reduction in working hours, the work-life balance and the presence of children in the family, according to the British Social Attitudes Survey cited by MacInnes (2005). However, this result is attributable to the already relatively low number of hours worked by women in the UK and the spread of parttime work among those with children. Not only is a reduction in terms of job satisfaction to be found when there is a gap between actual working hours and expectations depending on the intrinsic value attributed to work, but also an effect on living conditions, considering the levels of stress to which married women with children are subjected to in Japan (Boyles and Shibata 2009). On the basis of this evidence, it is suitable not only to consider working hours but also the difference that exists between actual working hours and the desired number of hours as a decisive element in the level of job satisfaction and that of living conditions.
248
T. Addabbo and G. Solinas
According to the ISFOL (2004), on the basis of the analysis of micro-data gathered as part of a European survey on the quality of work with reference to Italy, the different weight of domestic workloads and caring leads to the greatest dissatisfaction among women on average with regard to working time and rhythms. With reference to atypical work models, the ISFOL (2004) shows that in Italy, given the same conditions, there is a lesser degree of job satisfaction among atypical workers towards the stability of their employment compared to other forms of contract. There also appear to be a range of different reasons behind the level of satisfaction with regard to working conditions on the basis of forms of contract. Among temporary workers, those who have a higher education qualification appear to be less satisfied with regard to their employment stability.
12.4 Measuring the Quality of Work Using Fuzzy Logic In the literature, the main analyses adopted to measure the quality of work are multivariate econometric and descriptive statistical analyses. In this chapter, however, we apply fuzzy logic2 to the measurement of the quality of work. The interpretational scheme proposed in Sect. 12.3.1 has been broken down into more than 40 single items, measurable quantitatively using a suitable scale, and backed up by a range of control variables, retraceable to the various dimensions of the quality of work.3 The model was subjected to an initial check in various research projects: it was proposed to a sample of representative companies in the province of Modena; it was repeated using a sample of light-engineering firms in one of the most advanced industrial sectors (the packaging companies present in the province); it was also proposed in the sector of care for the elderly (including social cooperatives, public rest homes and private homecare assistants). Over the course of three years, these studies provided the most quantitatively significant database on the issue of the quality of work of all those available on a national level. Although it does not constitute a statistically representative sample, it provides individual evaluations on the quality of work as perceived by a total of around 2,500 workers. With regard to the measurement of the quality of work, we believe that the ‘‘measure’’ of intangible quantities in a multidimensional context implies, on one hand, the use of quantitative instruments to convert the variables into an appropriate numerical scale, and on the other hand, the definition of suitable aggregation operators to synthesise the initial qualitative data into indexes which also envelop
2
For a description of the underlying characteristics of a fuzzy model, see Altrock (1995), Piegat (1999), Zadeh (1965), Facchinetti (2002), Facchinetti et al. (2002) and Facchinetti and Pacchiarotti (2006). 3 A more detailed description of the fuzzy expert system is to be found in Addabbo et al. (2007).
12
Non-Standard Employment and Quality of Work
249
the information provided by the data. The classical methodologies of the theory of multi-criteria decisions offer a wide range of mathematical–statistical tools designed to provide aggregate evaluations. Most of them are formulated via real value functions, linear in terms of single attributes. Suffice to think of averages or weighted averages. The limitations of linearity in the above-mentioned methodologies have been overcome in various ways by more recent techniques, arising from the field of artificial intelligence. Of all these techniques, which are commonly known as ‘‘soft computing’’, we have applied a fuzzy expert system. There are a great number of advantages to this system: from the mathematical point of view, the aggregating function is extremely nonlinear in terms of its variables, which then serve as the micro-indicators that will supply the aggregate evaluation of the quality of work. From the point of view of significance, the fact of being constructed not ‘‘starting from the data’’, but from the experience of those who create it provides a clear and perfectly intelligible picture of what the experts think: it is all transparent, criticisable, justifiable and modifiable. From an understanding point of view, its strength lies in the possibility to create aggregations with qualitative (and not quantitative) rules of the same micro-indicators, tools that may also be used by people who are not entirely at ease with refined mathematical tools. The methodology exposed allows for better interaction between experts from different sectors, providing a result allowing for a clearer interpretation of the choices made and the results obtained. In brief, the use of fuzzy logic makes it possible to obtain synthetic indicators of the quality of work without losing the complexity of its definition, and coming to the check of hypotheses regarding the existence of compensating differentials and/ or of job market segmentation. Thus one might ask in which working positions (in terms of qualifications and type of contract) and in which contexts (which sectors, company types, territorial areas) jobs of different quality may cluster together, and which individual and family characteristics place the worker at greatest risk of finding him/herself in jobs defined overall as being of low quality. In this field, one may even set out to check what relations exist between changes in company organisation structures and various other elements of the quality of work. In the application proposed in Sect. 12.5, the fuzzy system is used to compare the quality of work in its various dimensions, with special reference to the various forms of contract. Within the analytical framework described in the above sections, in the analysis of the quality of work six different dimensions are identified, as cited schematically in Table 12.1, and which, obviously, are an integral part of the survey tool used and constructed with the aim of exploiting them in the measurement of the various dimensions.4
4
The questionnaire was devised together with Vando Borghi as part of the Mario Del Monte Foundation and University of Modena and Reggio Emilia project on the quality of work.
– Needs of communication relative to the human working environment (in general terms)
– – – – –
– Needs of commitment, development and training
– – – Organisational, autonomy and – Needs of discretion – control dimension – Needs of control over production and participation in – the production choices –
Complexity dimension
Type of items examined
Job security, social protection Parenthood rights Family-friendly policies Working hours Content, difficulty and variety of work Commitment required Complexity of tasks etc Training and Experience Conditions of responsibility Possibility to propose changes Participation in decision-making process
– Needs of recognition within the working community – Vertical social relationships and that of reference (superiors, supervisors, bosses) – Horizontal social relationships (coworkers) – Recognition of status within the community – Needs of financial subsistence – Career and remuneration
– Welfare regime Work-life balance dimension – Needs of reconciling life and work time
Economic dimension
Social dimension
Table 12.1 The dimensions of the quality of work Dimensions Needs
– Autonomy – Control
(continued)
– Variety – Training required
– Career – Remuneration – Protection – Reconciliation of life and work time
– Internal relationships – Recognition of work – Relationship between work and social life – Overall satisfaction
Intermediate dimensions
250 T. Addabbo and G. Solinas
Needs
– Needs relative to the time spent at work and other connected aspects – Psychophysical needs relative to the (physical) working environment; safety needs
Source Partial elaboration of Gallino 1993 and La Rosa 1983, p. 96
Ergonomic dimension
Table 12.1 (continued) Dimensions Type of items examined
– Quality of technological instruments – Layout and organisation of workspaces – Safety in the workplace
– Cognitive fatigue and stress – Flexibility of working hours – Environmental health hazards
Intermediate dimensions – Physical fatigue – Mental fatigue – Environmental health hazards
12 Non-Standard Employment and Quality of Work 251
252
T. Addabbo and G. Solinas
Each indicator gives rise to aggregations, produced with fuzzy expert systems of micro-indicators (single items), re-aggregatable in turn in further synthetic indexes. The reader will have noted in the above table (and even more clearly in the description of the characteristics of work) that a set of items may be found in more than one dimension. This is not a typing mistake or, even worse, a badly thought out taxonomy. In fact, one of the characteristics of the fuzzy expert systems is that of being able to examine items—even with different weighting—which by their very nature are of a multidimensional value.
12.5 The Quality of Work in Non-Standard Employment The fuzzy expert system was applied to the micro-data gathered in the province of Modena, from 719 workers in a company sample ranging from the light-engineering sector to building, services, food, IT and clothing. The companies were not chosen at random and are of different sizes, sectors and structures. Despite the fact that through an analysis of the single case studies it is possible to reconstruct the specific determining factors of the quality of work compared to the companies analysed,5 in this section there is no intention of proceeding with the analysis of case studies, but rather we shall use the measurement of the various dimensions of the quality of work in order to put forward an initial reading of the differences that may be found between different jobs with different forms of contract, considering the observations gathered as a whole. By using the fuzzy methodology, one obtains the measurement of the overall indicator of the quality of work in the various dimensions of which it is made up, thus being able to detect interactions between the various dimensions which make up the quality of work. Thus, considering the values assumed by the various dimensions, the results are proposed of the analysis of the correlation between the various observable dimensions. A positive correlation emerges between the output of the economic dimension and that of the complexity and the social/organisational dimension, while there is a weaker correlation with the ergonomic dimension (Table 12.2). Furthermore, the analysis of the various dimensions showed lower values for the overall index of the quality of work and for the economic dimension with reference to women and manual workers employed in manufacturing, where growing values may be found when coupled with the rising level of qualification. For women doing voluntary part-time work, a significant improvement may be noted in the work-life balance dimension and a worsening in the economic
5 Case studies were carried out on the basis of the data gathered using traditional multivariate and descriptive statistic techniques. For these studies, see Addabbo et al. (2006), Bigarelli et al. (2009) and Bertolini et al. (2008).
12
Non-Standard Employment and Quality of Work
253
Table 12.2 Correlation between different quality-of-work dimensions Comp. w-l.b Econ. Erg. Org.
Soc.
q.w.
complexity Work-life balance Economic Ergonomic Organisation Social Quality of work
1.00 0.83
1.00
1.00 0.39 0.60 0.32 0.61 0.59 0.74
1.00 0.46 0.42 0.48 0.58 0.69
1.00 0.26 0.44 0.63 0.70
1.00 0.35 0.35 0.59
1.00 0.82 0.78
Table 12.3 Dimensions of the quality of work for women (total and voluntary part-time) Women Women PT voluntary Variable
Mean
Std. Dev.
Mean
Std. Dev.
Complexity Work-life balance Economic Ergonomic Organisational Social Quality of work
0.55 0.52 0.42 0.57 0.57 0.55 0.54
0.18 0.17 0.17 0.23 0.20 0.17 0.18
0.57 0.64 0.40 0.61 0.60 0.57 0.57
0.16 0.16 0.13 0.24 0.15 0.13 0.15
dimension (Table 12.3). This result is in line with the empirical studies carried out in other parts of the country, and it may be linked to the same motivations that push women (especially in the Centre and North) to offer part-time work in their unpaid care work which are more easily reconcilable with part-time paid working hours. The micro-data gathered make it possible to detect to what extent part-time or full-time contracts are voluntary, and show that more than the type of work contract itself, it is the involuntary nature of working hours that negatively influences the output of the various dimensions. Lastly, the evidence gathered in these analyses does not show the existence of compensating wage differentials, but rather seems to indicate the presence of a job market segmentation between standard and non-standard employment, and between different qualifications. The OLS estimate of the same multivariate model for the analysis of the factors which influence the value of the various dimensions allows us to verify its weight and, by checking personal and company characteristics, to verify to what extent jobs that differ in contractual, qualification or content terms may be low skilled/ routinary (defining as low skilled/routinary the jobs in which the interviewee states that s/he carries out exclusively operational activities and not to have ever—or only rarely—had the chance to intervene in problem-solving) and may reach a different level of development in those dimensions that (in the model which we propose) comprise the quality of work.
High respons.
White collar
Managerial
Low skilled/routinary
Degree
High school
Woman with children
Age
Woman working part time
Temporary
Woman
-0.023 (1.01) -0.022 (0.57) 0.051 (1.78) -0.002 (1.74) -0.023 (1.94) 0.015 (0.64) -0.015 (0.49) -0.031 (1.55) 0.082* (2.07) 0.022 (0.92) 0.051 (1.85)
-0.008 (0.36) -0.033 (1.09) -0.023 (0.81) -0.002* (2.18) -0.003 (0.20) -0.020 (0.88) -0.011 (0.37) -0.040* (2.08) 0.110* (2.32) -0.007 (0.30) 0.009 (0.33)
Table 12.4 OLS models on the quality-of-work dimensions W.L. Balance Economic -0.001 (0.04) -0.024 (0.56) -0.069* (2.18) -0.001 (0.77) 0.007 (0.47) -0.002 (0.10) -0.055 (1.76) -0.071** (3.68) 0.160** (2.82) 0.071** (3.09) 0.073** (2.81)
Complexity -0.011 (0.34) 0.037 (0.74) 0.018 (0.44) -0.000 (0.05) -0.014 (0.64) 0.079* (2.53) 0.037 (0.90) 0.001 (0.05) 0.139* (2.24) 0.045 (1.39) 0.083* (2.19)
Ergonomic -0.008 (0.36) 0.030 (1.03) -0.046 (1.55) -0.002* (2.45) -0.004 (0.28) 0.009 (0.37) -0.021 (0.64) -0.117** (5.96) 0.230** (6.42) 0.036 (1.39) 0.056 (1.88)
Organis. 0.010 (0.42) -0.013 (0.40) -0.025 (0.93) -0.003** (3.30) -0.012 (0.95) -0.025 (1.07) -0.037 (1.22) -0.077** (3.97) 0.186** (4.40) 0.011 (0.48) 0.039 (1.49)
Social
(continued)
-0.004 (0.16) 0.010 (0.28) -0.026 (0.83) -0.002* (2.27) -0.009 (0.59) 0.019 (0.70) -0.024 (0.71) -0.068** (2.97) 0.176** (4.27) 0.038 (1.43) 0.078** (2.60)
Quality
254 T. Addabbo and G. Solinas
0.119** (3.42) 0.399** (6.87) 411 0.27
W.L. Balance 0.180** (5.46) 0.374** (7.02) 412 0.22
Economic 0.179** (5.49) 0.353** (6.45) 411 0.23
Complexity
Robust t statistics in parentheses; * significant at 5%; ** significant at 1%
Observations R-squared
More than 100 Employees Constant
Table 12.4 (continued) 0.135** (2.94) 0.340** (4.45) 403 0.12
Ergonomic 0.113** (3.37) 0.511** (9.54) 409 0.26
Organis. 0.078* (2.56) 0.582** (11.41) 411 0.20
Social
0.161** (5.01) 0.407** (7.32) 414 0.22
Quality
12 Non-Standard Employment and Quality of Work 255
256
T. Addabbo and G. Solinas
The multivariate analysis highlights how while for certain types of contract (specifically for women working part-time), there seems to be a sort of compensation between an improvement in terms of reconciliation, yet accompanied by a worsening in the complexity dimension (in this regard, suffice to think of the parttime work carried out by women, most likely on their own behalf, given the current division of unpaid domestic and care work, in order to reconcile family time and working hours). For other types of work, such as merely low skilled work (Table 12.4) or employment contracts characterised by unchosen working times (be they full or part-time) we rather witness a lower value of fuzzy indicators in more than one dimension. It may be noted how in the case of merely routinary work, apart from the ergonomic dimension, the value assumed by all the quality of work dimensions worsens (Table 12.4). Considering the work-life balance dimension, the worst result for routinary jobs is in line with that to be found in the literature (Roberts 2007; Warren 2003). The same worst situation in terms of work-life balance experienced by women with children is in keeping with their greater exposition to reconciliation problems, felt more acutely by the ‘‘main carer’’ (Roberts 2007). These results are in line with the existence of segmentation and clustering of the worst working conditions in several employment positions.
12.6 Conclusions In this chapter we started out from the definition of the quality of work as put forward by Gallino, which includes six dimensions: the economic dimension, that of complexity, organisation and ergonomics, while introducing two new dimensions: the social dimension and that of the work-life balance. The empirical literature on the quality of work has made use of statistical and econometric techniques as well as drawing on the availability of surveys dedicated to the measurement of working conditions, or ones from which subjective judgements on the level of job satisfaction could be inferred. In this essay, we chose to experiment with fuzzy logic in order to maintain the complexity of the definition adopted of quality of work and—through a complex methodology—come to the measurement of the various dimensions, thus with the use of descriptive statistics and multivariate analysis making it possible to interpret their values with particular reference to the non-standard forms of employment observed. This reading, albeit carried out on a limited number of cases and not chosen at random from the entire population of reference, leads us to identify lowquality jobs. Part-time work produces an improvement for women in the work-life balance dimension at the expense of the economic dimension. Lastly, the evidence gathered in this analysis does not indicate the existence of compensating wage differentials, but rather seems to indicate the presence of a job market segmentation between standard and non-standard employment and between different types of qualification.
12
Non-Standard Employment and Quality of Work
257
References Addabbo, T., Facchinetti, G., Mastroleo, G., & Solinas, G. (2007). A fuzzy way to measure quality of work in a multidimensional perspective. In J. Pejas & K. Saeed (Eds.), Advances in information processing and protection. Heidelberg: Springer. Addabbo, T., Lugli, L., & Maiani, B. (2006). Il caso Tetra Pak Carton Ambient. Quaderni della Fondazione Mario del Monte, 3, from http://www.mariodelmonte.it/. Altrock, C. (1995). Fuzzy logic and neurofuzzy applications in business and finance. New Jersey: Prentice Hall PTR. Aronowitz, S., & De Fazio, W. (1994). The jobless future: Sci-tech and the Dogma of work. Minneapolis, MN: University of Minnesota Press. Baglioni, G. (2009). Per una tutela sindacale meno tradizionale. Formazione e lavoro, 1, 151–158. Baldamus, W. (1961). Efficiency and effort. London: Tavistock Institute. Barley, S. R., & Kunda, G. (2001). Bringing work back. Organization Science, 12(1), 76–95. Barnard, C. (1938). The functions of the executive. Cambridge, MA: Harvard University Press. Beck, U. (2008). Costruire la propria vita. Bologna: Il Mulino. Bender, K. A., Donohue, S. M., & Heywood, J. S. (2005). Job satisfaction and gender segregation. Oxford Economic Papers, 57(3), 479–496. Bertolini, P., Savioli, R., & Solinas, G. (2008). Le assistenti familiari. Condizioni e qualità del lavoro, Quaderno di ricerca, 7, Progetto Equal ‘‘Fuori Orario’’ (Equal II Fase, IT-G2-EMI023, Rif. PA 0342/RER 2004). Berton, F., Richiardi, M., & Sacchi, S. (2009). Flex-insecurity: perché in Italia la flessibilità diventa precarietà. Bologna: Il Mulino. Bigarelli, D., Corradi, S., Fregni, C., & Solinas G. (2009). Qualità del lavoro e formazione nelle imprese artigiane nel settore del packaging. In AAVV (Ed.), I fabbisogni di apprendimento nelle imprese artigiane. Analisi su imprese e lavoratori del settore del packaging e della filiera nautica. Milano: Angeli. Blau, P. M. (1955). The dynamics of burocracy. Chicago, IL: Chicago University Press. Boeri, T. (2009). La crisi non è eguale per tutti. Milano: Rizzoli. Boyer, R., & Freyssenet, M. (2005). Oltre Toyota, I nuovi modelli produttivi. Milano: Università Bocconi editore. Boyles, C., & Shibata, A. (2009). Job satisfaction, work time, and well-being among married women in Japan. Feminist Economics, 15(1), 57–84. Brugnoli, A. (2008). L’Italia e la sfida della produttività. Milano: Egea. Burchell, B. J., & Fagan, C. (2004). Gender and the intensification of work: Evidence from the European working conditions surveys. Eastern Economic Journal, 30(4), 627–642. Burchell, B., Fagan, C., O’Brien, C., & Smith, M. (2007). Working conditions in the European Union: The gender perspective. Dublin: European Foundation of the Improvement of Living and Working Conditions. Butera, F. (2009). II cambiamento organizzativo. Roma-Bari: Laterza. Causarano, P., Falossi, L., & Giovannini, P. (2008). Mondi operai, culture del lavoro e identità sindacali. Il Novecento italiano. Roma: Ediesse. Ceri, P. (2009). Come è cambiata l’identità sociale del lavoro? Formazione e lavoro, 1, 145–150. Chinoy, E. (1955). Automobile workers and the American dream. Boston, MA: Beacon. Clark, A. E. (1997). Job satisfaction and gender: Why are women so happy at work? Labour Economics, 4, 341–372. Clark, A. E. (2005). Your money or your life: Changing job quality in OECD countries. British Journal of Industrial Relations, 43(3), 377–400. Clark, A. E. (2009, 6–7 November). Work, jobs and well-being across the millennium. Paper presented at Aiel-Child-Labor International Conference ‘‘Labor Market and the Household: Theory and Application’’.
258
T. Addabbo and G. Solinas
Dalton, M. (1950). Men who manage. New York: Wiley. Davis, L. E., & Taylor, J. C. (1972). Design of jobs: Selected readings. Harmondsworth: Penguin. Diprete, T. A. (1988). The upgrading and downgrading of occupations: Status redefinition vs deskilling as alternative theories of change. Social Forces, 66(3), 808–826. European Communities. (2008). Employment in Europe 2008. Luxembourg: Office for Official Publications of the European Communities. Facchinetti, G. (2002). Ranking functions induced by weighted average of fuzzy numbers. Fuzzy Optimization and Decision Making, 1(3), 313–327. Facchinetti, G., Giove, S., & Pacchiarotti, N. (2002). Optimisation of a non linear fuzzy function. Soft Computing, 6(6), 476–480. Facchinetti, G., & Pacchiarotti, N. (2006). Evaluation of fuzzy quantities. Fuzzy Set and Systems, 157(7), 892–903. Gallie, D. (1994). Patterns of skill change: Upskilling, deskilling or polarization? In R. Penn, M. Rose & J. Rubery (Eds.), Skill and occupational change. Oxford: Oxford University Press. Gallino, L. (1993). Dizionario di Sociologia del Lavoro. Torino: Utet (I ed. 1978). Gallino, L. (2001). Il costo umano della flessibilità. Roma: Laterza. Gouldner, A. W. (1954). Industrial bureaucracy. New York: Free Press. Green, F. (2004). Work intensification, discretion, and the decline in well-being at work. Eastern Economic Journal, 30(4), 615–625. Green, F., & Tsitsianis, N. (2005). An investigation of national trends in job satisfaction in Britain and Germany. British Journal of Industrial Relations, 43(3), 401–429. Herbst, P. G. (1974). Socio-technical design. London: Tavistock. Isfol. (2004). La qualità del lavoro in Italia. Isfol: Collana Temi & Strumenti. Jahoda, M. (1982). Employment and unemployment. Cambridge, MA: Cambridge University Press. Johnston, W. B., & Packer, A. H. (1987). Workforce 2000: Work and workers for the 21st Century. Indianapolis, IN: Hudson Institute. Kaiser, L. (2005). Gender-job satisfaction differences across Europe: An indicator for labor market modernization. IZA Discussion paper, 1876. Kern, H., & Schumann, M., (1991). La fine della divisione del lavoro. Produzione industriale e razionalizzazione. Torino: Einaudi. (I ed. Des Ende der Arbeitsteilung? Rationalisierung in der industriellen Produktion, München: Verlag Bech’sche, 1984). La Rosa, M. (1983). Qualità del lavoro e teorie socio-lavoriste. In M. La Rosa (Ed.), Qualità della vita, qualità del lavoro. Milano: Angeli. La Rosa, M. (1993). Il lavoro nella sociologia. Roma: La Nuova Italia Scientifica. Lazzarini, P. (2009). Cattolici al lavoro: valori religiosi e scelte lavorative. Formazione e Lavoro, 1, 89–98. Leoni, R. (Ed.). (2008). Economia dell’innovazione. Disegni organizzativi, pratiche lavorative e performance d’impresa. Milano: Angeli. Leoni, R., & Albertini, S. (2009). Innovazioni organizzative e pratiche del lavoro nelle imprese industriali del Nord. Milano: Angeli. Leschke, J., Watt, A., & Finn, M. (2008). Putting a number on job quality? Constructing a European job quality index. Brussels: European Trade Union Institute for Research, Education and Health and Safety (ETUI-REHS). MacInnes, J. (2005). Work-life balance and the demand for reduction in working hours: Evidence from the British social attitudes survey 2002. British Journal of Industrial Relations, 43(2), 273–295. Manghi, B. (2009). L’esperienza del lavoro di fronte alle trasformazioni. Formazione e Lavoro, 1, 99–108.
12
Non-Standard Employment and Quality of Work
259
Merllié, D., & Paoli, M. (2001). Third European survey on working conditions–2000, European Foundation for the improvement of living and working conditions. Luxembourg: Office for Official Publications of the European Communities. Milgrom, P., & Roberts, J. (1992). Economics, organization and management. Englewood Cliffs, NJ: Prentice Hall. Miller, E. J., & Rice, A. K. (1967). Systems of organizations. Londra: Tavistock Institute. Mumford, E. (1967). The computer and the clerk. London: Routledge and Kegan Paul. Musso, S. (2002). Storia del lavoro in Italia dall’Unità a oggi. Venezia: Marsilio. Paci, M. (2005). Nuovi lavori, nuovo welfare. Bologna: Il Mulino. Piegat, A. (1999). Fuzzy modelling and control. New York: Physica Verlag. Reich, B. R. (2008). Supercapitalismo. Come cambia l’economia globale e i rischi per la democrazia. Roma: Fazi Editore. Rice, A. K. (1963). Productivity and social organization. London: Tavistock. Roberts, J. (2004). The modern firm, organizational design for performance and growth. Oxford: Oxford University Press. Roberts, K. (2007). Work-life balance—the sources of the contemporary problem and the probable outcomes: a review and interpretation of the evidence. Employee Relations, 29(4), 334–351. Rose, M. (2005). Job satisfaction in Britain: Coping with complexity. British Journal of Industrial Relations, 43(3), 455–467. Saraceno, C. (2005). Le differenze che contano tra i lavoratori atipici. In S. Bertolini & R. Rizza (Eds.), Atipici? (pp. 15–24). Milano: Angeli. Sarchielli, G. (2008). Psicologia del lavoro. Bologna: Il Mulino. Seifert, H., & Tangian, A. (2009). Index quality of labour: Nordic Countries and Germany in comparison. WSI-Mitteilungen 01/2009. Semenza, R. (2000). Le nuove forme del lavoro indipendente. Stato e mercato, 58, 143–168. Sennet, R. (1999). L’uomo flessibile. Milano: Feltrinelli. Sestito, P. (2002). Il mercato del lavoro in Italia. Com’è, Come sta cambiando. Roma-Bari: Laterza. Sloane, P. J., & Williams, H. (2000). Job satisfaction, comparison earnings and gender. Labour, 14, 473–502. Sousa-Poza, A., & Sousa-Poza, A. A. (2000). Well-being at work: A cross-national analysis of the levels and determinants of job satisfaction. Journal of Socio-Economics, 29(6), 517–538. Sousa-Poza, A., & Sousa-Poza, A. A. (2003). Gender differences in job satisfaction in GreatBritain, 1991–2000: Permanent or transitory? Applied Economics Letters, 10, 691–694. Sousa-Poza, A., & Sousa-Poza, A. A. (2007). The effect of job satisfaction on labour turnover by gender: An analysis for Switzerland. Journal of Socio-Economics, 36(6), 895–913. Spenner, K. I. (1995). Technological change, skill requirements and education: the case for uncertainty. In D.B. Bills (Ed.), The new modern times: Factor reshaping the world of work. Albany, NY: Suny Press. Tangian, A. (2007). Is flexible work precarious? A study based on the 4th European Survey of Working Conditions 2005. WSI Discussion paper, (p. 153). Dusseldorf. Tarantelli, E. (1986). Economia politica del lavoro. Torino: Utet. Taylor, F. W. (1911). The principles of scientific management. New York: Norton. Trist, E. L., & Bamforth, K. W. (1951). Some psycological consequences of the Longwall method of coal getting. Human Relations, 4, 3–38. Tronti, L. (2009). Impresa, comunità e flessibilità del lavoro: per un nuovo scambio politico. In A. Gravina Ridolfi (Ed.), I valori del nuovo. Laicità, lavoro, costituzione. Scritti per Gian Paolo Meucci. Firenze: Leo S. Olschki. UNECE (2007). UNECE/ILO/Eurostat Seminar on Quality of Work (18–20 April 2007). http:// www.unece.org/stats/documents/2007.04.labour.htm. Van der Linden, M. (Ed.) (1993). End of labour history? International Review of Social History, 38(Supplement 1).
260
T. Addabbo and G. Solinas
Walker, C. R., & Guest, W. H. (1952). The man of the assembly line. Cambridge, MA: Harvard University Press. Warren, T. (2003). Class- and gender-based working time? Time poverty and the division of domestic labour. Sociology, 37(4), 733–752. Zadeh, L. A. (1965). Fuzzy sets. Information and Control, 8, 338–353. Zuboff, S. (1988). In the age of the smart machine. New York: Basic.
Chapter 13
Is There a Trade-Off Between Labour Flexibility and Productivity Growth? Some Evidence from Italian Firms Federico Lucidi
13.1 Introduction The Italian economy has been experiencing a serious slowdown of labour productivity growth since the end of the 1990s, when it started to diverge from the trend of other industrialized countries, and has become even negative in most recent years. Worryingly, this phenomenon does not appear to be short-term or due to the negative economic cycle; rather, it seems to be caused by structural factors. Indeed, various indicators (for example, Italy’s declining share of world trade, and poor performances in terms of R&D spending and patenting activity) point to a general deterioration in Italy’s competitive position, as recognized by some authors (Faini and Sapir 2005; Barca 2005). The anatomy of the productivity slowdown is evinced by Fig. 13.1, where the average growth of value added per equivalent labour unit has been decomposed, using a shift-share procedure,1 into three components (labour productivity growth within sectors, reallocation of the labour force to higher productivity industries, and a residual or ‘interaction’ effect). It emerges that the average annual increase of labour productivity (the sum of the three components) slowed down from 1.9% in the period 1992–1996 to 0.9% in the years 1996–2000, and then to virtually zero (0.02%) in the last interval, 2000–2004. It should be noted that, while the
I wish to thank Area Studi Capitalia for generously giving me access to their firms’ database. I am grateful to Marcella Corsi, Alfred Kleinknecht, Paolo Piacentini, Andrea Salvatori and Leonello Tronti for useful comments and discussions. 1
For details, see Appendix A. F. Lucidi (&) Fondazione Giacomo Brodolini, viale di Villa Massimo 21, 00161 Rome, Italy e-mail:
[email protected]
T. Addabbo and G. Solinas (eds.), Non-Standard Employment and Quality of Work, AIEL Series in Labour Economics, DOI: 10.1007/978-3-7908-2106-2_13, Springer-Verlag Berlin Heidelberg 2012
261
262
F. Lucidi
Fig. 13.1 Decomposition of labour productivity growth (2-digit sectors, 1984–2004) Source Own calculations on Istat data. See Appendix A for details
reallocation effect remained positive and did not change excessively during the last three periods, the intra-industry component, which should reflect technological advance inside sectors, seems to be the one most affected by the slowdown and even exhibits a negative value in 2000–2004. Surprisingly, this trend followed a decade of intense reforms which shifted the Italian model from a ‘managed economy’, where public intervention was pervasive, to a ‘common sense’ market model aimed at increasing competition and efficiency by reducing rigidities in both the labour and goods markets. In particular, the labour market underwent radical transformations during this period. First of all, wage bargaining institutions were profoundly changed by the tripartite agreements of 1992 and 1993, which marked the end of the automatic wage indexation system (scala mobile) which dated back to the mid-1970s. The new bargaining arrangements gave national labour contracts the sole purpose of maintaining the purchasing power of real wages, while the distribution of company-level productivity increases to workers was left to decentralized firm-level agreements (not compulsory for firms, however). At the same time, an increase in labour flexibility was pursued through the laws 196/1997, 368/2001 and 30/2003, which deregulated the adoption of fixed-term contracts, allowed the use of temporary agency workers, and introduced new ‘atypical’ contractual arrangements. In the mainstream view, the decentralization of wage-setting institutions and greater flexibility, in both wages and numerically, were needed to cope with the unemployment crisis of the early 1990s, whose causes (according to the dominant paradigm of ‘supply-side’ policies) were considered to be mainly structural. Undoubtedly, the conditions of the Italian labour market significantly improved in the last 10 years, with the unemployment rate decreasing from 11.2% in 1995 to 7.7% in 2005, and the employment rate increasing in the same period from 51.8 to 57.5% (but still largely falling short of the Lisbon target). Nonetheless, this improvement occurred at the expense of real wage growth: a phase of significant wage moderation followed the changes made to contractual arrangements, causing real wages to increase on average less than labour productivity and leading to a decline of the labour share on national income (Tronti 2007). The magnitude of the wage restraint period also appears significant in light of an international
13
Is There a Trade-Off Between Labour Flexibility and Productivity Growth
263
Fig. 13.2 Labour productivity growth versus share of fixed-term workers (1993–2004) Source Istat. (Note: a break in LFS series occurred in 2004)
comparison which ranks Italy last among industrialized countries for real wage growth during the decade 1992–2002 (Zenezini 2004). Employment growth has been accompanied by the increased use of fixed-term and atypical jobs; in particular, the share of fixed-term employees over total employees (excluding freelance workers and other ‘atypical’ categories, which in the labour force survey are classified as self-employment) increased by five percentage points from 7.3% in 1995 to 12.3% in 2005.2 Therefore, wage moderation and the increase of external labour flexibility appear to be among the main institutional changes that have occurred in the Italian labour market over the last 15 years. However, if their role in boosting employment has been emphasised (on the quantitative side; conversely, the debate on the ‘quality’ of the new jobs is still ongoing), their impact on the productive system has often been neglected, from both the theoretical and empirical viewpoints. But in the present context, labour market reforms do not appear to have been neutral with respect to the productivity slowdown. In fact, various theoretical arguments suggest that the prolonged period of wage moderation and increased numerical flexibility curbed the incentives of firms to innovate, while providing incentives to compete by adopting ‘low road’ practices: cutting labour costs, maintaining lowproductive jobs, and reducing the scope for training and high quality human resource management practices. A general idea of the potential trade-off between flexibility and productivity growth is provided by Fig. 13.2, where the correlation between the slowdown of labour productivity growth and the increased incidence of fixed-term contracts is evident. Empirical testing using firm-level data appears the most adequate strategy to provide robust evidence on these relationships. This paper therefore presents results of productivity growth estimates using data from the ninth ‘‘Indagine sulle 2
The values are not fully comparable owing to a break in the Istat Labour Force Survey (LFS) in 2004.
264
F. Lucidi
imprese manifatturiere’’, conducted by the Capitalia Bank Research Centre (formerly Mediocredito Centrale) in the period 2001–2003, where ‘flexibility’ indicators are included among the factors explaining labour productivity growth at firm level. The structure of the paper is as follows. Section 13.2 provides a theoretical background for the linkage among wage moderation, labour flexibility, innovative activity and productivity growth. The dataset and the variables used for the empirical estimation are described in Sect. 13.3, while the model is presented and the results are discussed in Sect. 13.4. A concluding section summarizes and draws some policy implications.
13.2 Theoretical Background Various kinds and sources of ‘labour flexibility’ can be identified. In particular, it is common practice to divide labour flexibility practices into three categories (Beatson 1995): ‘numerical’ (external) flexibility, which enables firms to easily adjust the number of employees in response to changes, and depends on the strictness of legislation on hiring and firing, fixed-term contracts and working hours; ‘functional’ (internal) flexibility, which concerns the possibility to reorganize the workforce internally by means of internal training and the development of multi-skilled employees; ‘wage’ flexibility, which concerns the responsiveness of wages to external shocks (however, this is commonly perceived as ‘downward’ flexibility), largely depending on the features of the wage-setting institutions. The reforms undertaken in Italy over the past 15 years have concerned the first and the last of these aspects. However, whilst the effects of labour market reforms on increased numerical flexibility are unambiguous, the issue of wage flexibility is more complex. The changes made to the wage-setting institutions at the beginning of the 1990s aimed at increasing wage flexibility at the firm level by explicitly providing for the company-level distribution of labour productivity increases to workers. However, decentralized bargaining was little used (in particular among smaller firms), so that the Italian wage-setting system remained highly centralized (at least for inflation compensation). What one observes 15 years later is therefore a long period of ‘institutional’ wage moderation (Zenezini 2004; Tronti 2007; Brandolini et al. 2007). Another key explanation for the wage restraint period resides in the new climate of industrial relations, which, after a long period of conflict, gradually became more cooperative from the early 1990s onwards. This new industrial-relations environment fostered wage moderation during the 1990s by allowing the systematic underestimation of the forecast inflation rate, which is a pillar of the new bargaining system (Brandolini et al. 2007). Hence, wage moderation was institutionally grounded, and can only partly be explained by increased wage flexibility at firm level (Zenezini 2004).
13
Is There a Trade-Off Between Labour Flexibility and Productivity Growth
265
With these concepts in mind, the rest of the paper will focus on external flexibility and wage moderation as possible explanations for the poor productivity performance of Italian firms.3 There is no bias in considering these two causes as a priori positively correlated: from a theoretical point of view, the correlation can take place in both directions. If temporary contracts are used mainly with lowskilled or low-productivity workers, or entail lower social security contributions for employers, then the increase in external flexibility will be correlated with a reduction in firms’ wage bills. But the reverse case may occur if workers accepting temporary jobs are likely, ceteris paribus, to demand a risk premium to off-set the chance of not being hired permanently on expiry of their temporary contracts (compensating differentials theory). Nonetheless, empirical evidence from other countries (Segal and Sullivan 1995; Sànchez and Toharia 2000; Booth et al. 2002; McGinnity and Mertens 2004; Addison and Surfield 2005; Kleinknecht et al. 2006) and from Italy (Picchio 2006) shows that fixed-term workers, on average, earn less than regular workers even after observed and unobserved personal characteristics have been controlled for. On these bases, it seems possible to identify four theoretical headings under which to explain the direct and indirect effects of wage moderation and external labour flexibility on productivity growth. These four transmission channels involve effects on firms’ innovative activity, internal training, workplace cooperation, and patterns of aggregate demand. Under the first heading, according to Sylos Labini (1984, 1993, 1999), wage increases (in particular with respect to the price of ‘machinery’) are a major stimulus for the adoption of technological innovations intended to save labour both in absolute terms (by increasing workplace efficiency) and relative ones (by dynamically substituting labour with capital4). This process is influenced by firms’ market power, in particular by the capacity to transfer labour cost increases onto prices by means of mark-up pricing. In a competitive environment, therefore, entrepreneurs will have a greater incentive to enhance labour productivity in order to preserve their profit share. Bhaduri (2006) has recently proposed a model of endogenous growth built on an mechanism of this kind. Moreover, from a Schumpeterian perspective, one can argue that high real wage growth and labour market rigidity may foster, to a certain extent, the process of creative destruction and favour the adoption of innovations by firms (Kleinknecht 1998; Naastepad and Kleinknecht 2004; Kleinknecht et al. 2006). On this view,
3
Functional flexibility is not considered in this paper owing to the absence of information in the dataset used for the empirical estimation. However, various studies (Bassanini and Ernst 2002; Michie and Sheehan 2003; Kleinknecht et al. 2006; Cristini et al. 2003) suggest that the linkage between functional flexibility and firm performance may act with an opposite sign, i.e. it provides a favourable environment for productivity increases. 4 Note that the dynamic substitution between capital and labour, in this context, differs from the static substitution, with constant technology, implied by the neoclassical theory as a response to the relative variation in the prices of factors. The former, in fact, involves technological change incorporated in new capital goods (Sylos Labini 1993).
266
F. Lucidi
innovative firms compete better in a context of higher costs (both labour costs and adjustment costs due to stricter regulation). By contrast, looser regulation and (downward) wage flexibility can be considered as a ‘grant’ to low-productive firms competing through ‘low-road’ practices (for example, passive price strategies, to be achieved by cutting labour costs, and limited innovative activity: Antonucci and Pianta 2002 and Pianta 2003). The outcomes of innovation are uncertain, whilst the gains in competitiveness deriving from a cut in labour costs—both explicit (due to wage moderation) and implicit (due to the reduction of firing costs)—are immediate and unquestionable, even if presumably short-lived. Hence slack labour market regulation may be an important incentive to entrepreneurs with short timehorizons to follow the ‘low road’ path, preferring cost scrapping to innovation. Ramazzotti (2005) has conducted a detailed analysis of different firm strategies and choices in this context. The standard view on this matter generally suggests an alternative position: namely that greater labour market rigidity may have negative effects on productivity because it hampers the reallocation of labour ‘‘from old and declining sectors to new and dynamic ones’’ (for a review of the effects of labour market institutions on economic performance see Nickell and Layard 1999). However, while this effect may be apparent at a higher level of aggregation, it does not seem relevant when the performance of individual firms within a given sector is considered. Some authors also argue that, in rigid labour markets, the adjustment costs arising from the adoption of a new technology may inhibit the innovative process itself (Scarpetta and Tressel 2004). As regards wage bargaining institutions, a similar mechanism may be at work if decentralized unions appropriate the rents deriving from productivity gains (this is the classical hold-up problem: for literature surveys see Metcalf 2002 and Menezes-Filho and Van Reenen 2003). Nonetheless, the real occurrence of these effects depends closely on the extent to which the wage-setting actors are centralized and coordinated, on the nature of industrial relations, on whether employees can be internally reassigned (‘functional’ flexibility), and whether internal activities can be contracted out. This position has also been challenged by part of the neoclassical literature: for example, a model that explicitly links the presence of firing costs with greater scope for process innovation has been presented by Saint-Paul (2002). Turning to the second heading, it appears obvious that labour flexibility impacts on training and human capital accumulation. If labour relationships are expected to be short-lived, there is little incentive for firms to invest in both the general and specific training of their workforces (firms need an adequate pay-back period in order to recoup their investment costs). Workers, for their part, will be reluctant to acquire firm-specific skills if they do not feel a long-term commitment to their employers. Similar conclusions hold if we hypothesise that higher labour flexibility (in particular, along the wage dimension) reduces the compression of the wage structure (both within and between firms), which is one of the main reasons for the provision of firm sponsored training (Acemoglu and Pischke 1999; Agell 1999). The result of higher labour flexibility could therefore be an underprovision of on-the-job training, with potentially negative effects on productivity
13
Is There a Trade-Off Between Labour Flexibility and Productivity Growth
267
growth. Empirical evidence of a negative correlation between fixed-term employment and the probability of receiving work-related training has been provided for the UK by Arulampalam and Booth (1998) and Booth et al. (2002). As to the effects of labour flexibility on productivity via workplace cooperation, a strand in the literature supports the idea that productivity-enhancing effects ensue from ‘high trust’ or ‘high road’ human resources management practices, and from cooperative labour relations (Huselid 1995; Buchele and Christiansen 1999a; Lorenz 1999; Michie and Sheehan 2001, 2003; Naastepad and Storm 2005). According to these theories, higher on-the-job protection and subsequent cooperative relationships between management and employees may positively affect firm performance, encouraging innovative activity and promoting efficiency gains. Finally, it is important to note that wage moderation and labour flexibility may have negative effects on aggregate demand, both directly and indirectly (for example, through increases in precautionary saving by employees with temporary jobs). And, through this channel, they may negatively affect labour productivity growth. Various theories suggest that there is a direct link among demand growth, innovation and labour productivity growth, both in the context of dynamic increasing returns (via the so-called ‘Verdoorn–Kaldor law’) and on a demand-pull hypothesis concerning innovative activity (Schmookler 1966; Brouwer and Kleinknecht 1999). A number of empirical analyses have been conducted on the effects of labour flexibility on productivity dynamics (both measured as labour productivity and TFP) or on the innovative activity of firms. However, the majority of these studies focus on the country or sectoral level (Buchele and Christiansen 1999a, b; Nickell and Layard 1999; Bassanini and Ernst 2002; Scarpetta and Tressel 2004; Auer et al. 2005; Naastepad and Storm 2005),5 and only a few report firm-level evidence. In particular, Michie and Sheehan (2001, 2003) studied the impact of various flexibility practices on innovation indicators for British firms, evidencing a negative effect of external flexibility and a positive effect of functional flexibility. Similar results have been obtained with reference to labour productivity growth by Dekker and Kleinknecht (2004) and Kleinknecht et al. (2006) for Dutch firms. Finally, Arvanitis (2005) found a positive relationship between functional flexibility and labour productivity for a sample of Swiss companies, but a not significant effect of external flexibility.
5
Most of these studies observe a positive effect of employment protection (measured by the Oecd index or other indicators) on productivity growth or innovation indicators. Auer et al. (2005) find a positive (though decreasing) relation between job stability, measured as average tenure, and labour productivity. Scarpetta and Tressel (2004), however, report a negative effect of employment protection, mainly in countries with sectoral and uncoordinated wage bargaining. The distinction among different industrial relations models is also considered by Bassanini and Ernst (2002), who conclude that Epl (employment protection) strictness is significantly correlated to technological specialization in countries with coordinated relations.
268
F. Lucidi
13.3 The Data The data used for the empirical analysis are taken from the ninth ‘‘Indagine sulle imprese manifatturiere’’, a survey of manufacturing firms conducted by Capitalia Bank Research Centre and covering the period 2001–2003. The Capitalia sample, which includes 4,289 firms, is representative of Italian manufacturing companies with more than 10 employees. Firms are selected by means of a stratification method by industry, geographic area and firm size. The survey provides information under different headings. In particular, it includes full information on workforce composition by contract type (full-time or part-time, permanent or temporary), hirings and lay-offs, sales, investments in fixed capital, R&D expenditure and other innovation indicators. Unfortunately, it does not contain information on working hours, while information on temporary agency workers and freelance workers is only available for 2003, and therefore could not be used for my analysis.6 Using fiscal codes as identifiers, the Capitalia survey was then merged with balance sheet data from the Bureau Van Dijk AIDA dataset (which contains balance sheet data for firms with turnovers higher than 500,000€), to obtain some variables necessary for the empirical analysis (in particular, value added and labour costs). The number of firms with complete balance sheet data for the 3 years was 3,351. As evident from Table 13.1, this operation did not produce significant modifications in the composition of the sample by size class, geographic area and sector (according to the Pavitt taxonomy). However, when account was taken of the presence of missing values in some of the variables of interest, the total number of firms reduced further, according to the different specifications of the model, to around 2,600 firms when using a full specification (see Sect. 13.4 for details). In order to estimate productivity regressions, monetary variables were standardized by the number of employees (value added, investments, labour costs) or by the amount of sales (R&D expenditure) and deflated using the appropriate price and total labour turnover) were created. Extreme and unreliable values were cleaned from variables by using a trimming procedure which excluded observations falling outside the first and last 0.5 percentiles (an analogous method has been used on the Capitalia survey data by, for example, Parisi et al. 2006 and Benfratello et al. 2005). A complete list of the variables, with detailed information and descriptive statistics, is given in the Appendix.
6
No attempt was made to merge with the previous wave of the survey (which covered the period 1998–2000) because, given the high incidence in this survey of missing values in flexibility variables, the number of firms for which full information was available over the 6 years was remarkably low (around 350).
13
Is There a Trade-Off Between Labour Flexibility and Productivity Growth
269
Table 13.1 Percentage composition of the sample by size class, geographic area and sector Full sample (n = 4,289) With balance sheets (n = 3,351) 11–20 employees 21–50 employees 51–250 employees 251–500 employees More than 500 employees Total North–West North–East Centre South Total Traditional sectors Scale sectors Specialized supply High technology Total
20.9 30.9 37.0 5.2 6.1 100.0 35.9 30.1 17.6 16.4 100.0 51.2 17.6 27.1 4.1 100.0
20.2 31.7 37.1 5.0 6.0 100.0 35.3 32.1 18.0 14.7 100.0 51.1 17.5 27.5 3.8 100.0
According to the Pavitt taxonomy
13.4 The Model The approach used to estimate productivity equations was a modification of Sylos Labini’s (1984, 1993, 1999) equations for the determinants of labour productivity growth, adapted for use with microdata and amended with the inclusion of variables not considered in the original model (in particular, two indicators of external labour flexibility). Differently from models based on the production function, Sylos Labini’s model explains productivity growth by means of three components. Specifically, productivity increases depend on: the variation of wages relative to the price of investment goods (‘Ricardo effect’); the growth of aggregate demand, in order to verify the occurrence of dynamic increasing returns [‘Smith effect’, corresponding to the Verdoorn law (Verdoorn 1949)]; investment expenditures, in order to consider the impact of new technology embodied in new fixed capital and not captured by the other factors. The original model was estimated (at macro level) with different lag structures in order to consider delayed effects of the explanatory variables on productivity growth. My application of this model to microdata involved three considerations. Firstly, monetary variables had to be standardized in order to take firm size into account (as explained in Sect. 13.3). Secondly, the absence of information on working hours required the use of value added per employee (and not per
270
F. Lucidi
worked hour, as would have been suitable) as a measure of labour productivity.7 Thirdly, and most importantly, right-hand variables comprised a large degree of endogeneity, which appears to be critical for wages (at least theoretically, since in Italy productivity-related wage increases seem rather moderate and are likely to occur only in medium–large firms). In view of the shortness of the panel, the solution chosen to minimize endogeneity problems was simply to carry out OLS cross-section regressions for the growth rate of labour productivity in the period 2001–2003 using lagged values (taken in 2001) of the regressors. This solution appeared to be valid in the absence of serial correlation of the residuals, which, however, cannot be tested in a cross-section framework. Considering productivity growth over 3 years (2001–2003) also allowed partial control to be made for the volatility of labour productivity at the firm level (labour hoarding is likely to occur in this context, since firms are unable or unwilling to adjust their workforces in the very short term). Moreover, this approach allowed account to be taken of a lag before the effects of right-hand variables on productivity growth become observable.8 The empirical specification of the model modified Sylos Labini’s original equation in various respects. Firstly, the initial level of value added per employee was included in order to allow for technological catch-up among firms.9 With regard to ‘cost-driven’ increases in productivity, given the impossibility of constructing an index of wage costs relative to capital costs at firm level, an indicator of real labour costs per employee was used instead.10 This variable was included both in lagged levels (taken in 2001) and in growth rates over the period 1998– 2000 (taken from balance sheet data). As regards ‘demand-driven’ effects on productivity growth, the growth of sectoral value added at 2-digit level (taken from Istat national accounts) was included in order to consider the effects of markets 7 This limitation would be more serious if I had been trying to explain the determinants of the labour productivity level, instead of its growth rate. However, attempts to take account of the number of part-time employees in measuring labour productivity (e.g. considering a part-time worker to be half a full-time worker) did not produce significant changes in the coefficient estimates. Therefore, in order to avoid too demanding assumptions, the straightforward indicator of value added per employee was used for the estimations. 8 Unfortunately, the short time horizon (3 years) and the presence of the lagged dependent variable among the regressors precluded the use of panel methodologies to take individual firms’ effects into account. 9 The inclusion of lagged productivity, which was always highly significant, probably also allowed control for the (unobserved) variation in the utilisation of productive capacity during the period. This consideration can be easily explained. If, for instance, a firm has an abnormally low (or high) productivity level at the beginning of the period for transitory reasons (e.g. restructuring, temporary difficulties, etc.), and then returns to its ‘normal’ level, one might erroneously infer that its productivity has strongly increased (decreased), whereas the variation has been mainly induced by the fluctuation in the utilisation of its productive capacity. The inclusion of lagged productivity, at least theoretically, could allow one to control for this phenomenon. 10 Sylos Labini (1984, 1993) considers both indicators in the theory, while excluding one of them in the empirical analysis owing to their high collinearity.
13
Is There a Trade-Off Between Labour Flexibility and Productivity Growth
271
expansion11 on firm level productivity (Smith–Verdoorn effect). This variable was considered instead of industry dummies in order to avoid multicollinearity problems. As for the ‘investment-driven’ effects on productivity growth, investments in equipment and machinery per employee were considered. The baseline model was completed by including R&D expenditure over sales as an indicator of innovative activity inside firms. For the specific purposes of the analysis reported here, two indicators of external labour flexibility were included among the right-hand side variables, namely the share of employees on temporary contracts, and a measure of total labour turnover (the sum of hirings and layoffs divided by the number of employees). Finally, dummies for industry, size class, geographic area and firm age were included as controls. The empirical specification of the model was therefore: D0103 ln pijt ¼a þ b1 ln pijt2 þ x0ijt2 b2 þ flex0ijt2 b3 þ b4 D0103 ln yjt þ d0i c þ eijt ; where the growth of value added per employee between 2001 and 2003 in firm i belonging to sector j (measured as a logarithmic difference) is explained by the lagged level of (log) value added per employee, by a vector of lagged variables xijt2 (including investments per employee, R&D/sales and one of the two indicators for real labour costs per employee), by a vector of lagged flexibility indicators flexijt2 (the share of employees on fixed-term contracts and the indicator of total turnover), by the growth of sectoral value added D0103 ln yjt (rather than industry dummies) and by a vector of firm-specific dummies di Robust standard errors were calculated (the White/Koenker statistic always rejected the null of no heteroskedasticity). In a second step, I allowed for heterogeneity in the flexibility coefficients by making them interact with a dummy for firms performing R&D activity (as in Kleinknecht et al. 2006) in order to verify different response patterns to flexibility practices in innovating versus non-innovating firms.12 In fact, it is possible to hypothesise that in innovative, more dynamic firms, external flexibility is not necessarily a ‘low-road’ practice with detrimental effects on productivity. Indeed, in the case of R&D firms, more flexibility is likely to ease the acquisition of
11
This variable was considered to be exogenous, assuming as irrelevant the effect of the single firm’s performance on value added at industry level. However, this may be false for larger firms and in more concentrated sectors. Nonetheless, all the specifications were also estimated without this variable in order to make comparison possible. 12 Different innovation indicators were tested in order to include these interaction terms: in particular, two dummies for firms declaring that they had introduced process or product innovations during the reference period. However, the dummy for firms performing R&D activities appears to discriminate the coefficients under the two headings better, maybe because it more closely reflects the attitude to continuous innovative activity inside firms.
272
F. Lucidi
knowledge workers (for whom mobility may not necessarily be a ‘bad’), as opposed to non-R&D firms, where it appears functional to saving on labour costs. Several specifications of this model were estimated. Table 13.2 shows the outcomes of estimations using the level of real labour costs per employee among the regressors. The first column reports the coefficients of the baseline model, without R&D and flexibility indicators, in order to have a benchmark with the highest number of observations (more than 3,000). As will be seen from Table 13.2, all the coefficients exhibit the expected sign and are strongly significant. In particular, there appears to be a negative effect of the initial productivity level, which suggests that firms lagging behind at the beginning of the period are actually growing faster, i.e. some sort of catch-up process may be at work. The effect of investments in fixed capital (expressed in thousands of euros, without taking logarithms in order to avoid the exclusion of non-investing firms) is positive as expected, and so is the effect of the initial level of labour costs per employee. The latter finding indicates that firms facing higher labour costs at the beginning of the period display, on average, higher productivity growth, consistently with the theoretical considerations previously outlined. In regard to the other controls, significant scale effects appear to operate: firm size (over 50 employees) matters for labour productivity growth, with increasing magnitude along the firm dimension. There is also strong evidence that younger firms (under 25 years old) perform significantly better than older ones, suggesting their (expected) higher dynamism in acquiring market shares as opposed to the incumbents. Coefficients on regional dummies (four macro areas) are never reported because of their insignificance at the conventional levels,13 while industry controls (also not reported) show an outperforming result for refinery, chemicals and pharmaceuticals, rubber and plastic products, and basic metal companies against the reference industry (food products). The baseline model was then completed with R&D and flexibility indicators. As evident from Table 13.2, R&D expenditure over sales never exerts a significantly positive effect on labour productivity growth. However, this may simply stem from bad measurement of this variable (firms may include other kinds of expenditures in their declared R&D), or from the fact that higher R&D expenditure by itself does not necessarily imply the adoption of innovative processes or products (i.e. ‘successful’ R&D).14 As regards labour flexibility variables, on the other hand, the estimates show that both the incidence of fixed-term contracts and total labour turnover, when taken separately, are negatively (and significantly) correlated with productivity growth. However, when both variables are included, the effect of the fixed-term share seems to dominate that of labour turnover, which becomes insignificant; this 13
Regional controls are indeed significant in explaining labour productivity levels, where Southern regions rank behind Northern and Central ones. 14 Moreover, if R&D expenditures are idiosyncratic to sectors rather than to individual firms, their effect could be captured by industry controls: in fact, on removing industry dummies, the effect of R&D turns positive, yet not statistically significant.
Age: more than 40 years (reference: less than 25)
Size: more than 500 employees (reference: less than 21) Age: 25–40 years
Size: 251–500 employees
Size: 51–250 employees
Growth of sectoral value added (2001–2003) Size: 21–50 employees
Share of employees on fixed–term contracts (2001) Total labour turnover (2001)
R&D/Sales (2001)
Log labour cost per worker (2001)
Investment per worker (2001)
Log value added per worker (2001)
0.021 (0.016) 0.047*** (0.015) 0.063** (0.027) 0.113*** (0.028) –0.045*** (0.015) -0.057*** (0.017)
-0.323*** (0.026) 0.003*** (0.001) 0.129*** (0.035)
0.012 (0.016) 0.041*** (0.016) 0.059** (0.029) 0.104*** (0.036) –0.042*** (0.016) -0.052*** (0.018)
-0.337*** (0.028) 0.003*** (0.001) 0.149*** (0.037) -0.033 (0.419) -0.178*** (0.059)
0.014 (0.016) 0.042*** (0.016) 0.059** (0.029) 0.106*** (0.037) –0.043*** (0.016) -0.053*** (0.018)
-0.068** (0.032)
-0.335*** (0.028) 0.003*** (0.001) 0.148*** (0.038) -0.054 (0.406)
0.013 (0.016) 0.042*** (0.016) 0.059** (0.029) 0.106*** (0.037) –0.042*** (0.016) -0.052*** (0.018)
-0.335*** (0.028) 0.003*** (0.001) 0.143*** (0.038) -0.012 (0.420) -0.159** (0.062) -0.038 (0.033) 0.918*** (0.112) 0.011 (0.017) 0.037** (0.016) 0.067** (0.029) 0.110*** (0.037) –0.045*** (0.016) -0.048*** (0.018)
-0.280*** (0.028) 0.003*** (0.001) 0.115*** (0.036) 0.256 (0.434) -0.159*** (0.060)
Table 13.2 Determinants of labour productivity growth (value added per employee) between 2001 and 2003 (1) (2) (3) (4) (5)
-0.068** (0.032) 0.901*** (0.112) 0.013 (0.017) 0.038** (0.016) 0.066** (0.029) 0.112*** (0.037) –0.045*** (0.016) -0.049*** (0.018)
-0.278*** (0.028) 0.003*** (0.001) 0.113*** (0.037) 0.221 (0.421)
(6)
Is There a Trade-Off Between Labour Flexibility and Productivity Growth (continued)
-0.278*** (0.028) 0.003*** (0.001) 0.110*** (0.037) 0.271 (0.435) -0.138** (0.065) -0.040 (0.035) 0.918*** (0.113) 0.012 (0.017) 0.038** (0.016) 0.066** (0.029) 0.112*** (0.037) –0.044*** (0.016) -0.048*** (0.018)
(7)
13 273
0.890*** (0.094) Yes Yes 3,017 12.12 (0.00) 94.31 (0.00) 0.18
(1) 0.857*** (0.100) Yes Yes 2,661 10.45 (0.00) 82.43 (0.00) 0.19
(2) 0.860*** (0.101) Yes Yes 2,654 10.39 (0.00) 84.64 (0.00) 0.19
(3)
(4) 0.874*** (0.102) Yes Yes 2,639 10.05 (0.00) 82.86 (0.00) 0.19
Robust standard errors in parentheses; *significant at 10%; **significant at 5%; ***significant at 1%
Sector dummies Regional dummies Observations F-test (p-value) White/Koenker (p-value) R-squared
Constant
Table 13.2 (continued) (5) 0.666*** (0.081) No Yes 2,661 14.10 (0.00) 62.54 (0.00) 0.14
(6) 0.667*** (0.081) No Yes 2,654 13.90 (0.00) 64.51 (0.00) 0.13
0.678*** (0.082) No Yes 2,639 12.97 (0.00) 63.02 (0.00) 0.13
(7)
274 F. Lucidi
13
Is There a Trade-Off Between Labour Flexibility and Productivity Growth
275
effect may be explained by the high level of correlation between the two variables (rho = 0.37). This finding, however, provides support for the existence of a tradeoff between external flexibility and productivity growth at the firm level. The model was then re-estimated replacing industry dummies with the growth of sectoral value added (the last three columns of Table 13.2). The coefficients suggest a very strong effect of market expansion on firm-level productivity dynamics. This is evidence for the existence of dynamic increasing returns in manufacturing, in accordance with the Verdoorn law (or ‘Smith effect’, as Sylos Labini puts it). However, some considerations counsel caution on the magnitude of the Verdoorn coefficient, which may be biased due to endogeneity, or may simply capture short-term variations in the degree of utilisation of productive capacity (testing the hypothesis of dynamic increasing returns would probably require a longer time dimension). This consideration applies in particular when one considers the modest growth of value added in the 2001–2003 period. However, the inclusion of sectoral demand growth did not affect the estimates on the other variables, in that it induced only a slight reduction in the size of the coefficients on lagged productivity and labour cost per employee; the R&D coefficients became positive, yet not significant (this was probably due to the removal of industry dummies, given the high specificity of R&D expenditure by sectors). Table 13.3 shows the results of estimation of the same regressions using past growth rates of labour cost per employee (in the period 1998–2000) among the right-hand side variables in place of its level in 2001. It should be pointed out that this variable grounds uniquely on balance sheet information (from the AIDA dataset), so that the number of employees (used as denominator) may not be fully comparable with that declared by firms in the Capitalia survey.15 In this case, too, the results display a positive effect of past ‘wage push’ on productivity growth. The inclusion of this variable does not affect the coefficients on lagged productivity level and investments per worker; moreover, the negative impact of fixedterm share and labour turnover appears to be stronger, although the latter is still not significant in the full specification. Replacing industry dummies with sectoral demand growth lowers the effect of past wage increases, but essentially leads to the same conclusions as above. Finally, Table 13.4 reports a re-estimation of the model with lagged labour costs where flexibility indicators were interacted with a dummy for firms performing R&D activities during the period 2001–2003.16 The results show that the negative effect of numerical flexibility on productivity growth seems to be restricted to non-R&D firms: in these firms, indeed, the use of flexibility seems more likely to be functional to the exploitation of ‘low road’ strategies providing lower incentives to productivity growth.
15
The number of employees in AIDA is counted as an average over the whole year, while in the Capitalia survey it measures the stock at the end of the year. 16 The model with growth rates of labour cost per employee yielded comparable results.
Age: more than 40 years (reference: less than 25)
Size: more than 500 employees (reference: less than 21) Age: 25–40 years
Size: 251–500 employees
Size: 51-250 employees
Growth of sectoral value added (2001–2003) Size: 21–50 employees
Share of employees on fixed-term contracts (2001) Total labour turnover (2001)
Growth of labour cost per employeea (1998–2000) R&D/Sales (2001)
Investment per worker (2001)
Log value added per worker (2001)
0.006 (0.017) 0.048*** (0.017) 0.045 (0.041) 0.161*** (0.030) -0.039** (0.017) -0.046** (0.020)
-0.303*** (0.024) 0.002*** (0.001) 0.033** (0.013)
-0.004 (0.018) 0.041** (0.017) 0.040 (0.042) 0.134*** (0.032) -0.036** (0.018) -0.041** (0.020)
-0.314*** (0.026) 0.003*** (0.001) 0.037*** (0.014) 0.276 (0.440) -0.271*** (0.065)
-0.001 (0.018) 0.041** (0.017) 0.039 (0.043) 0.139*** (0.032) -0.037** (0.018) -0.043** (0.021)
-0.091** (0.038)
-0.312*** (0.027) 0.002*** (0.001) 0.035** (0.014) 0.276 (0.433)
-0.002 (0.018) 0.041** (0.017) 0.041 (0.043) 0.137*** (0.032) -0.035* (0.018) -0.041** (0.021)
-0.316*** (0.027) 0.003*** (0.001) 0.035** (0.014) 0.294 (0.441) -0.248*** (0.069) -0.047 (0.039) 1.020*** (0.100) -0.002 (0.018) 0.035** (0.018) 0.048 (0.043) 0.130*** (0.031) -0.036* (0.018) -0.032 (0.021)
-0.274*** (0.025) 0.003*** (0.001) 0.030** (0.015) 0.604 (0.456) -0.231*** (0.065)
Table 13.3 Determinants of labour productivity growth (value added per employee) between 2001 and 2003 (1) (2) (3) (4) (5) (6)
-0.087** (0.037) 0.907*** (0.124) 0.001 (0.018) 0.035** (0.018) 0.048 (0.043) 0.135*** (0.032) -0.037** (0.019) -0.033 (0.021)
-0.272*** (0.026) 0.003*** (0.001) 0.028* (0.015) 0.599 (0.448)
(7)
(continued)
-0.275*** (0.026) 0.003*** (0.001) 0.028* (0.015) 0.610 (0.457) -0.207*** (0.071) -0.048 (0.039) 0.924*** (0.125) -0.001 (0.018) 0.035* (0.018) 0.048 (0.043) 0.133*** (0.032) -0.035* (0.019) -0.032 (0.021)
276 F. Lucidi
a
1.018*** (0.120) Yes Yes 2,396 9.18 (0.00) 83.25 (0.00) 0.21
(1) 1.223*** (0.125) Yes Yes 2,153 7.84 (0.00) 85.80 (0.00) 0.22
(2) 1.054*** (0.158) Yes Yes 2,145 7.72 (0.00) 88.04 (0.00) 0.22
(3)
(4) 1.237*** (0.127) Yes Yes 2,135 7.58 (0.00) 85.24 (0.00) 0.22
(5) 0.925*** (0.124) No Yes 2153 11.35 (0.00) 73.24 (0.00) 0.17
(6) 1.016*** (0.101) No Yes 2,145 11.13 (0.00) 75.20 (0.00) 0.17
From balance sheet data (AIDA). Robust standard errors in parentheses; *significant at 10%; **significant at 5%; ***significant at 1%
Sector dummies Regional dummies Observations F-test (p-value) White/Koenker (p-value) R-squared
Constant
Table 13.3 (continued) 1.027*** (0.102) No Yes 2,135 10.51 (0.00) 72.91 (0.00) 0.17
(7)
13 Is There a Trade-Off Between Labour Flexibility and Productivity Growth 277
Size: more than 500 employees (reference: less than 21)
Size: 251–500 employees
Size: 51–250 employees
Size: 21–50 employees
Growth of sectoral value added (2001–2003)
Labour turnover (2001)* Non R&D firm
Labour turnover (2001)* R&D firm
Fixed-term share (2001)* Non R&D firm
Fixed-term share (2001)* R&D firm
R&D/Sales (2001)
Log labour cost per worker (2001)
Investment per worker (2001)
Log value added per worker (2001)
0.011 (0.016) 0.040*** (0.016) 0.058** (0.029) 0.102*** (0.036)
-0.337*** (0.028) 0.003*** (0.001) 0.149*** (0.037) -0.097 (0.422) -0.057 (0.088) -0.240*** (0.074)
0.013 (0.016) 0.040** (0.016) 0.057* (0.029) 0.103*** (0.037)
-0.031 (0.035) -0.095** (0.043)
-0.336*** (0.028) 0.003*** (0.001) 0.148*** (0.038) -0.164 (0.414)
0.012 (0.016) 0.040** (0.016) 0.057* (0.029) 0.103*** (0.037)
-0.335*** (0.028) 0.003*** (0.001) 0.143*** (0.038) -0.120 (0.427) -0.050 (0.106) -0.210*** (0.079) -0.022 (0.041) -0.055 (0.045) 0.923*** (0.112) 0.011 (0.017) 0.036** (0.016) 0.065** (0.029) 0.108*** (0.037)
-0.280*** (0.028) 0.003*** (0.001) 0.115*** (0.036) 0.171 (0.437) -0.010 (0.089) -0.235*** (0.077) -0.027 (0.035) -0.097** (0.044) 0.907*** (0.112) 0.013 (0.017) 0.036** (0.016) 0.064** (0.029) 0.109*** (0.037)
-0.278*** (0.028) 0.003*** (0.001) 0.113*** (0.037) 0.097 (0.430)
Table 13.4 Determinants of labour productivity growth (value added per employee) between 2001 and 2003, with interaction terms (1) (2) (3) (4) (5)
(continued)
-0.278*** (0.028) 0.003*** (0.001) 0.109*** (0.037) 0.144 (0.442) 0.007 (0.108) -0.206** (0.084) -0.028 (0.042) -0.056 (0.046) 0.926*** (0.113) 0.012 (0.017) 0.036** (0.016) 0.064** (0.029) 0.108*** (0.037)
(6)
278 F. Lucidi
-0.043*** (0.016) -0.052*** (0.018) 0.856*** (0.101) Yes Yes 2,661 10.26 (0.00) 82.79 (0.00) 0.19
(1) -0.043*** (0.016) -0.054*** (0.018) 0.862*** (0.101) Yes Yes 2,654 10.14 (0.00) 84.99 (0.00) 0.19
(2) -0.042*** (0.016) -0.052*** (0.018) 0.875*** (0.102) Yes Yes 2,639 9.60 (0.00) 83.32 (0.00) 0.19
(3)
Robust standard errors in parentheses. *significant at 10%; **significant at 5%; ***significant at 1%
Sector dummies Regional dummies Observations F-test (p-value) White/Koenker (p-value) R-squared
Age: more than 40 years (reference: less than 25) Constant
Age: 25–40 years
Table 13.4 (continued) -0.045*** (0.016) -0.047*** (0.018) 0.669*** (0.081) No Yes 2,661 13.51 (0.00) 62.92 (0.00) 0.14
(4) -0.046*** (0.016) -0.049*** (0.018) 0.670*** (0.081) No Yes 2,654 13.16 (0.00) 64.90 (0.00) 0.14
(5)
-0.045*** (0.016) -0.048** (0.018) 0.682*** (0.082) No Yes 2,639 11.81 (0.00) 63.48 (0.00) 0.14
(6)
13 Is There a Trade-Off Between Labour Flexibility and Productivity Growth 279
280
F. Lucidi
13.5 Concluding Remarks and Policy Implications The results from the previous section provide quite robust evidence for the existence of a firm-level trade-off between external labour flexibility and labour productivity growth. In particular, firms exhibiting a higher share of temporary workers in their workforces and a higher rate of labour turnover achieved a slower growth of value added per worker in the period 2001–2003. This evidence seems to be stronger for non-R&D firms, where the use of flexible practices may be functional to cost scrapping. Moreover, lagged labour costs per employee (both in levels and in changes) have a significant role in explaining labour productivity growth during the period: a decreasing wage bill makes it economically convenient to preserve low-productive jobs and labour-intensive productive processes, curbing incentives for their modernization and the adoption of new technologies. From a Schumpeterian viewpoint, the process of ‘creative destruction’ is thus hampered: even if positive effects may ensue in the short run (increased employment, albeit through the creation of low-quality jobs), the long-term negative effects on economic growth are evident. These considerations raise some doubts concerning the sustainability of both labour-market reforms and continued wage restraint policies in the long run. On the one hand, revision of the 1992 and 1993 agreements on the wage-setting institutions appears necessary in order to re-distribute productivity increases to employees, and to take advantage (from a growth perspective) of the dual nature of labour incomes, which represent both a stimulus to technological innovation and the main component of aggregate demand. It is arguable, however, whether a change of the wage-setting institutions should involve their decentralization, for example by making firm-level agreements compulsory, or reinforcement of the role of centralized bargaining (e.g. leaving the task of redistributing the average sectoral increase in productivity to industry-level agreements).17 Whatever the case may be, such policies should be tempered by increased competition in the goods markets in order to prevent the occurrence of wage-price spirals. On the other hand, the evidence on labour flexibility suggests that employment protection legislation should be revised so as that temporary jobs are not less costly to firms than permanent ones, at least in terms of social contributions. Fixed-term contracts should be designed to help firms deal with production peaks (due to the seasonality or cyclicality of certain activities) or as a means to screen entrants with a view to their future inclusion in the stable workforce. Instead, recent labour market reforms increasing flexibility ‘at the margin’ without changing the rules for core workers have created the conditions for the persistence of workers in precarious employment, without providing them with the bargaining power that they 17
According to the Schumpeterian paradigm, the second solution may be preferable in inducing technological laggards (whose performance in terms of productivity is worse) to adopt innovations so that they are not forced out of the market.
13
Is There a Trade-Off Between Labour Flexibility and Productivity Growth
281
need to negotiate higher wages (as a risk premium to compensate for the chance of not being hired permanently on expiry of the temporary contract). In many cases, therefore, the use of fixed-term contracts (and, more specifically, of ‘atypical’ arrangements) provide employers with three sorts of saving: not only on firing costs, but also on wage costs and on social contributions.18 To quote Sylos Labini, ‘‘(…) when labour market is too rigid there are troubles, but troubles of a different kind can arise when flexibility is unlimited. Here, too, there is a problem of achieving an optimal level’’ (Sylos Labini 1999, p. 265). The task of finding an ‘optimal’ level of labour flexibility is still an open challenge for policy-makers. However, the results of this analysis emphasise the need to accelerate this process, and to evaluate (as a direction for future research) the impact of labour market reforms not only in a short run perspective.
Appendix A: Labour Productivity Decomposition This appendix explains the decomposition of labour productivity growth presented in Fig. 13.1. Since labour productivity, at an aggregate level, is a weighted average of sectoral productivity levels with weights corresponding to the employment shares of each sector, its trend depends both on the variation of productivity in each sector and on the variation of the sectoral composition of employment. Therefore, productivity growth between two periods can be algebraically decomposed into a between component, a within component, and a residual. The first component represents the contribution to productivity growth made by the reallocation of labour from low-productive to high-productive industries, corresponding to the increased productivity that would be observed maintaining productivity levels constant within sectors. The second identifies the growth of productivity due merely to intra-sectoral increases, in the absence of labour reallocation. Finally, the residual captures the interaction effects between productivity and employment at the industry level, taking a positive sign if the two variables are positively correlated, a negative one in the opposite case. The formula used for the decomposition is the following: n P
pt p0 i¼1 ¼ p0 ¼
pit qit
n X i¼1
18
n P
pi0 qi0
i¼1 n P
pi0 qi0
i¼1
qit qi0 pit pi0 ðpit pi0 Þðqit qi0 Þ pi0 þ qi0 þ p0 p0 p0
On the individuals’ side, in particular for young people, this also translates into lower expectations for future pensions.
282
F. Lucidi
Table 13.5 Decomposition of average annual labour productivity growth 1984– 1988– 1992– 1996– 1988 1992 1996 2000 Structural change Intra-industry productivity growth Interaction effects Total
2000– 2004
1.21 0.92
0.51 0.69
0.27 1.63
0.45 0.61
0.15 -0.05
-0.27 1.86
-0.13 1.08
-0.04 1.87
-0.11 0.95
-0.08 0.02
Source Own computation on Istat data
Table 13.6 Descriptive statistics (full sample, 2001–2003) Variable Mean Median Std. Dev. Min Value added per worker (thousands of euros) 46.094 40.905 23.785 Growth of value added per worker (2001–2003) -0.025 -0.017 0.336 Investment per worker (thousands of euros) 5.197 2.210 8.162 Labour cost per worker (thousands of euros) 26.403 25.341 8.535 R&D/Sales (thousands of euros) 0.006 0.000 0.016 Share of fixed-term contracts 0.032 0.000 0.098 Total labour turnover 0.143 0.095 0.197
Max
3.432 195.180 -2.444 2.772 0.000 65.944 4.267 74.022 0.000 0.134 0.000 0.944 0.000 1.875
where pt is aggregate labour productivity at time t, pit is labour productivity in sector i at time t, and qit is the share of employed in sector i at time t. The first term in the square brackets is the ‘structural change’ (between) effect in sector i; the second term is the ‘productivity growth’ (within) effect in sector i; finally, the last term is the interaction effect in sector i. The decomposition between 1984 and 2004 (by 5-year intervals) was performed using value added per equivalent labour unit (at constant prices) as the labour productivity index, and sectors were selected according to the Ateco 2002 classification at 1-digit level (sections and subsections, 30 sectors). Table 13.5 summarizes the results of the analysis.
Appendix B: Variables Definitions and Descriptive Statistics Value added per worker: value added was calculated as the value of production (net sales ± variation of inventories, + capitalized costs) less net consumption (materials ± variation of inventories) and services. It was deflated using value added deflator disaggregated at 2-digit level, and divided by the number of workers declared by firms in the questionnaire. Firms reporting negative or zero value added were not considered.
13
Is There a Trade-Off Between Labour Flexibility and Productivity Growth
283
Investment per worker: investment in equipment and machinery as declared in the questionnaire, deflated with the gross investment deflator at 2-digit level of disaggregation and divided by the number of workers. Labour cost per worker: labour costs deflated with the consumer price index and divided by the number of workers. When considered in the interval 1998– 2000, the average number of employees during the year (taken from balance sheet data) was considered. R&D/Sales ratio: R&D expenditure divided by the amount of sales, both derived from questionnaires Table 13.6. Share of fixed term contracts: percentage of workers (both full-time and part-time) on fixed-term arrangements. Total labour turnover: sum of hirings and layoffs divided by the number of workers.
References Acemoglu, D., & Pischke, J. (1999). Beyond Becker: Training in imperfect labour markets. The Economic Journal, 109(453), 112–142. Addison, J. T., & Surfield, C. J. (2005). Atypical work and compensation, IZA discussion paper, 1477 Agell, J. (1999). On the benefits from rigid labour markets: Norms, market failures, and social insurance. The Economic Journal, 109(453), 143–164. Antonucci, T., & Pianta, M. (2002). Employment effects of product and process innovation in europe. International Review of Applied Economics, 16(3), 295–307. Arulampalam, W., & Booth, A. L. (1998). Training and labour market flexibility: Is there a tradeoff? British Journal of Industrial Relations, 36(4), 521–536. Arvanitis, S. (2005). Modes of labour flexibility at firm level: Are there any implications for performance and innovation? evidence for the Swiss economy. Industrial and Corporate Change, 14(6), 993–1016. Auer, P., Berg, J., & Coulibaly, I. (2005). Is a stable workforce good for productivity? International Labour Review, 144(3), 319–343. Barca, F. (2005). Italia frenata. Roma: Donzelli. Bassanini, A., & Ernst, E. (2002). Labour market institutions, product market regulations, and innovation: Cross-country evidence. OECD economics department working paper, 316. Beatson, M. (1995). Labour market flexibility. Sheffield employment department research series, 48. Benfratello, L., Schiantarelli, L., & Sembenelli, A. (2005). Banks and innovation: Microeconometric evidence on Italian firms. Boston College Working Papers in Economics, 631. Bhaduri, A. (2006). Endogenous Economic growth: A new approach. Cambridge Journal of Economics, 30(1), 69–83. Booth, A. L., Francesconi, M., & Frank, J. (2002). Temporary jobs: Stepping stones or deadends? Economic Journal, 112, 189–213. Brandolini, A., Casadio, P., Cipollone, P., Magnani, M., Rosolia, A., & Torrini, R. (2007). Employment growth in Italy in the 1990s: Institutional arrangements and market forces. In N. Acocella & R. Leoni (Eds.), Social pacts, employment and growth. Heidelberg: Physica. Brouwer, E., & Kleinknecht, A. (1999). Keynes plus? Effective demand and changes in firm-level R&D: An empirical study. Cambridge Journal of Economics, 23, 385–391.
284
F. Lucidi
Buchele, R., & Christiansen, J. (1999a). Employment and productivity growth in Europe and North America: The impact of labour market institutions. International Review of Applied Economics, 13(3), 313–332. Buchele, R., & Christiansen, J. (1999b). Labor relations and productivity growth in advanced capitalist Economies. Review of Radical Political Economics, 31(1), 87–110. Cristini, A., Gaj, A., Labory, S., & Leoni, R. (2003). Flat hierarchical structure, bundles of New work practices and firm performance. Rivista Italiana degli Economisti, 2, 313–341. Dekker, R., & Kleinknecht, A. (2004). Flexible labour and productivity growth: An empirical study. In C. Nova-Kaltosouni & M. Kassotakis (Eds.), Proceedings of the conference ‘‘promoting new forms of working organization and other cooperative arrangements for competitiveness and employability’’. Athens: University of Athens. Faini, R., & Sapir, A. (2005). Un modello obsoleto? Crescita e specializzazione dell’economia italiana. In Boeri et al. (Eds.) Oltre il declino. Bologna: Il Mulino Huselid, M. (1995). The impact of human resource management practices on turnover, productivity and corporate financial performance. Academy of Management Journal, 38, 635–670. Kleinknecht, A. (1998). Is labour market flexibility harmful to innovation? Cambridge Journal of Economics, 22, 387–396. Kleinknecht, A., Oostendorp, M. N., Pradhan, M. P., & Naastepad, C. W. M. (2006). Flexible labour, firm performance and the Dutch job creation miracle. International Review of Applied Economics, 20(2), 171–187. Lorenz, E. H. (1999). Trust, contract and economic cooperation. Cambridge Journal of Economics, 23(3), 301–316. McGinnity, F., & Mertens, A. (2004). Wages and wage growth of fixed-term workers in east and west Germany. Applied Economics Quarterly, 50(2), 139–163. Menezes-Filho, N., & Van Reenen, J. (2003). Unions and innovation: A survey of the theory and empirical evidence. CEPR discussion paper, 3792. Metcalf, D. (2002). Unions and productivity, financial performance and investment: International evidence. CEP discussion paper, 539. London: School of Economics and Political Science Michie, J., & Sheehan, M. (2001). Labour market flexibility, human resource management and corporate performance. British Journal of Management, 12, 287–306. Michie, J., & Sheehan, M. (2003). Labour market deregulation, ‘flexibility’ and innovation. Cambridge Journal of Economics, 27, 123–143. Naastepad, C. W. M., & Kleinknecht, A. (2004). The Dutch productivity slowdown: The culprit at last? Structural change and economic dynamics, 15, 137–163. Naastepad, C. W. M., & Storm, S. (2005). The innovating firm in a societal context: Productivity, labour relations and real wages. In R. Verburg, J.R. Ortt, & W. Dicke (Eds.), Management of technology: An introduction. London: Routledge. Nickell, S., & Layard, R. (1999). Labour market institutions and economic performance. In O. Ashenfelter & D. Card (Eds.), Handbook of Labour Economics, Vol. 3c. Amsterdam: North Holland/Elsevier. Parisi, M., Schiantarelli, F., & Sembenelli, A. (2006). Productivity, innovation and R&D: Micro evidence for Italy. European Economic Review, 50(8), 2037–2061. Pianta, M. (2003). Innovation and employment. In J. Fagerberg, et al. (Eds.), Handbook of innovation. Oxford: Oxford University Press. Picchio, M. (2006). Wage differentials between temporary and permanent workers in Italy. Quaderni del Dipartimento di Economia dell’Università Politecnica delle Marche, 257 Ramazzotti, P. (2005). Labour market flexibility and technological innovation or, desperately seeking a trade-off. Dipartimento di Istituzioni Economiche e Finanziarie dell’Università di Macerata, Temi di discussione, 25 Saint-Paul, G. (2002). Employment protection, international specialization, and innovation. European Economic Review, 46, 375–395. Sànchez, R., & Toharia, L. (2000). Temporary workers and productivity: The case of Spain. Applied Economics, 32, 583–591.
13
Is There a Trade-Off Between Labour Flexibility and Productivity Growth
285
Scarpetta, S., & Tressel, T. (2004). Boosting productivity via innovation and adoption of new technologies: Any role for labor market institutions? World Bank Policy Research Working Paper, 3273 Schmookler, J. (1966). Invention and economic growth. Cambridge: Harvard University Press. Segal, L. M., & Sullivan, D. G. (1995). The temporary labor force. Economic Perspectives, 19(2), 2–20. Sylos Labini, P. (1984). The forces of Economic growth and decline. Cambridge, MA: MIT Press. Sylos Labini, P. (1993). Progresso tecnico e sviluppo ciclico. Bari: Laterza. Sylos Labini, P. (1999). The employment issue: Investment, flexibility and the competition of developing countries. BNL Quarterly Review, 210, 257–280. Tronti, L. (2007). The July protocol and economic growth: The chance missed. In N. Acocella & R. Leoni (Eds.), Social pacts, employment and growth. Heidelberg: Physica. Verdoorn, P. J. (1949). Fattori che regolano lo sviluppo della produttività del lavoro. L0 Índustria, 1 (A. P. Thirlwall, Trans.).In L. Pasinetti (Ed.) (1993) Italian Economic Papers (Vol. 2). Oxford: Oxford University Press. Zenezini, M. (2004). Il problema salariale in Italia. Economia e Lavoro, 38(2), 147–181.