Challenges for Europe
Edited by Hugh Stephenson
Challenges for Europe
This page intentionally left blank
Challeng...
35 downloads
1240 Views
797KB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Challenges for Europe
Edited by Hugh Stephenson
Challenges for Europe
This page intentionally left blank
Challenges for Europe Edited by
Hugh Stephenson Centre for Economic Performance London School of Economics and Political Science
Editorial matter and selection © Hugh Stephenson 2004 Individual chapters © contributors 2004 All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with written permission or in accordance with the provisions of the Copyright, Designs and Patents Act 1988, or under the terms of any licence permitting limited copying issued by the Copyright Licensing Agency, 90 Tottenham Court Road, London W1T 4LP. Any person who does any unauthorized act in relation to this publication may be liable to criminal prosecution and civil claims for damages. The authors have asserted their rights to be identified as the authors of this work in accordance with the Copyright, Designs and Patents Act 1988. First published 2004 by PALGRAVE MACMILLAN Houndmills, Basingstoke, Hampshire RG21 6XS and 175 Fifth Avenue, New York, N.Y. 10010 Companies and representatives throughout the world PALGRAVE MACMILLAN is the global academic imprint of the Palgrave Macmillan division of St. Martin’s Press, LLC and of Palgrave Macmillan Ltd. Macmillan® is a registered trademark in the United States, United Kingdom and other countries. Palgrave is a registered trademark in the European Union and other countries. ISBN 1–4039–3659–5 This book is printed on paper suitable for recycling and made from fully managed and sustained forest sources. A catalogue record for this book is available from the British Library. Library of Congress Cataloging-in-Publication Data Challenges for Europe / edited by Hugh Stephenson. p. cm. Includes bibliographical references and index. ISBN 1–4039–3659–5 1. Europe—Economic conditions—21st century. 2. Europe— Economic policy. I. Stephenson, Hugh, 1938– HC240.C48 2005 330.94—dc22 10 9 8 7 6 5 4 3 2 1 13 12 11 10 09 08 07 06 05 04 Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham and Eastbourne
2004052198
Contents List of Tables
vi
List of Figures
vii
Preface
ix
Notes on the Contributors
x
1
1
What’s Wrong with Europe’s Economy? Adair Turner
2 Assessing the Euro: Expectations and Achievements Peter B. Kenen
31
3 The Return of Deflation: What Can Central Banks Do? Willem Buiter
46
4 The Problem of Inequality Paul Krugman
75
5 Education Matters Alan B. Krueger
90
6 On the Edge: The Uneasy Boundaries Between Public and Private Sectors John Kay
125
7
141
Demographics, Economics and Social Choice Adair Turner
8 Europe: Pillar of the World Economy or Just an Appendix? Norbert Walter
168
Index
179
v
List of Tables 1.1 1.2 1.3 1.4 1.5 1.6 3.1 5.1 7.1 7.2 7.3 7.4 7.5 7.6
US and EU growth compared Labour productivity per hour worked Unemployment rates, 1973–2000 Hours worked per person in employment Labour productivity per hour worked Labour productivity by type of industry Total, goods and services inflation in the UK Cumulative CAT gains by SES Support ratio dynamics Support ratio forecasts, 2000–50 Support ratio dynamics Demographic conditions and investment attractiveness Immigration-only responses: UN migration scenario Fertility intentions of women: England and Wales
vi
3 3 4 7 10 13 59 116 150 151 152 155 156 164
List of Figures 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 1.10 1.11 1.12 1.13 1.14 1.15 3.1 3.2 3.3 3.4 3.5 3.6 4.1
4.2 5.1 5.2 5.3 5.4 5.5
Real GDP per capita growth, 1993–2000 Employment rates, 2000 Productivity, labour input and prosperity: France and US, 1999 Labour productivity per hour worked: US Relative output per hour worked by sector, 1999 US productivity per hour, 1973–2002 Industry contribution to labour productivity gap Labour productivity growth by sector, 1995–99 Employment rates, 2000 Average hourly wage costs, 2000 General government total outlays as percentage of GDP, 2001 GDP per capita growth and public expenditure, 1960–95 Exports have sustained Eurozone growth Italian gross government debt General government structural balances, 2000 Key economic trends in Japan: 1950 to 2002 Key economic trends in USA: 1950 to 2002 Key economic trends in EU: 1950 to 2002 Key economic trends in UK: 1950 to 2002 Price level and inflation: UK, 1800–1914 Bank rate, inflation and £/$ exchange rate, 1817–1914 Income and population shares for top 0.01%, 1% and 10% of US population: 1913, 1929, 1970 and 2000 US federal tax rates Earnings rise with education, February 1990 Difference in log earnings versus difference in schooling; Twinsburg identical twins sample Years of education, by quarter of birth Season of birth and years of schooling Mean log weekly wage, by quarter of birth vii
5 5 6 8 9 10 11 14 17 18 19 19 22 23 24 48 49 50 51 52 53
77 82 92 94 96 97 98
viii List of Figures
5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13 5.14 5.15
5.16 5.17 5.18 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9 7.10 7.11 7.12
Increases in education lead to GDP growth over long periods, 97 countries Ratio of wages of college⫹ to high school graduates Most of national income is due to education and experience: United States, 1996 Years of schooling by birth cohort: US native-born residents at age 35 Returns to cognitive skills: hourly earnings by test scores IALS scores by year of birth: United States IALS scores by year of birth: United Kingdom IALS scores by year of birth: Sweden Hanushek’s class-size results for 277 estimates Average percentage of estimates positive, negative or unknown sign, by number of estimates taken from study Sample starting in kindergarten Maths CAT gains over the school year, by family socioeconomic status (SES) Maths CAT gains over the summer, by family socioeconomic status Life expectancy at birth, male Life expectancy at 60, male Total fertility rates: Europe and North America, 1950–2000 Total fertility rates: Asian countries, 1950–2000 Total fertility rates: Iran, Turkey and Brazil, 1950–2020 From pyramids to columns PAYG pension systems: key ratios Demographic change in the UK and China, UN Medium Variant projection Rising longevity, fixed retirement age and stable support ratios Population density: US and Europe Italy’s population structure, 1970–2050 European fertility rates, 2001
101 103 104 105 107 107 109 109 111
111 113 115 115 143 143 145 146 146 147 148 154 157 160 164 165
Preface In 2002, the Centre for Economic Performance at the LSE was awarded a Queen’s Anniversary Prize for Higher and Further Education for its research on unemployment and inequality. To mark this award, the Centre ran a series of Queen’s Prize public lectures by some of the world’s leading academic economists and commentators on the theme of the main current challenges for Europe. This volume consists of an edited version of those eight public lectures delivered at the LSE in the order they appear here between February 2003 and January 2004. The CEP is the leading research centre for applied economics in Europe. The citation for the Queen’s Anniversary Prize said that the CEP is ‘recognized as a centre of national and international excellence in the application of economic theory and rigorous empirical analysis to the issues of unemployment, productivity, education and international trade’. These lectures provide stimulating and controversial contributions to many of the key debates about the future of Europe, drawing extensively on American experiments and experience to throw light on both possibilities and pitfalls that face us. HUGH STEPHENSON
ix
Notes on the Contributors Willem Buiter is Chief Economist at the European Bank for Reconstruction and Development. John Kay is Visiting Professor at the Centre of Economic Performance at the London School of Economics and Political Science and author of The Truth about Markets. Peter B. Kenen is Walker Professor of Economics and International Finance at Princeton University. Alan B. Krueger is Bendheim Professor of Economics and Public Affairs at Princeton University and a Research Associate at the National Bureau of Economic Research. Paul Krugman is Professor of Economics and International Affairs at Princeton University. Adair Turner is Vice-Chairman of Merrill Lynch Holdings and a Visiting Professor as the London School of Economics and Political Science. Norbert Walter is Chief Economist of Deutsche Bank Group.
x
1 What’s Wrong with Europe’s Economy? Adair Turner
My title for this lecture clearly indicates my view that something is wrong. In 2001, there was much confidence about Europe’s prospects, relative to the US. The euro had been successfully launched. Its impact on intra-European trade and industrial restructuring was already apparent and the difficult period of pre-launch fiscal consolidation lay in the past. The US, with greater personal exposure to stock market investment, appeared more vulnerable to the dotcom bust. All major forecasts in January 2001 foresaw more rapid Eurozone than American growth in 2001 and 2002. And since then the US has suffered more unexpected shocks – the potential impact of September 11 on consumer confidence and more accounting and corporate governance scandals – than has Europe. But it is Europe’s performance that has been poor and America’s surprisingly robust. The 15 countries of the European Union experienced growth of about 0.7 per cent in 2002, the US 2.4 per cent. Almost all forecasts suggested a similar differential in America’s favour in 2003. Eurosceptics find this unsurprising. It reflects, they argue, deeprooted rather than short-term factors. Europe has grown more slowly than the US for 20 years; it has failed to achieve a productivity spurt from the application of information technology; and it is uncompetitive because of high taxes, barriers to free competition and inflexible labour markets. Deep structural reform is therefore essential, say the Eurosceptics, certain that the required reform will never occur. But Europhiles say the same, believing that the needed reform can be achieved. The debate at European summits – Lisbon, Stockholm, 1
2
Adair Turner
Barcelona – has focused on how far Europe is willing to embrace the challenge of structural reform. But I want to question this conventional wisdom, or at least to deconstruct it. The phrase ‘structural reform’ – in relation to Europe as also to Japan – is now used in such an all-embracing, vague and polemic sense that it confuses our understanding and can misdirect appropriate policy focuses. Specifically I want to argue five propositions: ●
●
●
●
●
First, that much of the difference in economic performance between the US and Europe reflects not differences in economic efficiency, but in individual and social choice, which are unlikely to be influenceable by policy initiatives. Indeed, we should perhaps not attempt to change them, since the purpose of liberal economics is to maximize choices available and not to tell people how they should exercise that choice. Second, that the product market liberalization agenda of the Lisbon, Stockholm and Barcelona summits, while positive, is of marginal importance to Europe’s future prosperity. Third, that the idea that Europe must cut taxes and welfare spending in order to be more competitive is largely wrong or, at the very least, overstated and too general. Fourth, that the problems now facing the Eurozone economy (and, in particular, Germany and Italy) result primarily from deficiencies in the Eurozone macro-policy framework. Fifth, however, that Europe does face one huge long-term structural challenge: the sustainability of its pensions systems in the face of an ageing population. This is a challenge with little relevance to economic prospects this year and next, but with huge implications for European society over the next 50 years.
My overall theme, therefore, is that European policy discussion often fails to focus on the most important issues. I shall not pretend here to provide a full description of Europe’s economic problems or their solutions or indeed of Europe’s strengths. There is no discussion here of skills and the science base, or creative industries and entrepreneurship, or industry clusters, or the management of public services, or other important issues. My aim here is simply to challenge some specific common beliefs about the ‘EU versus US’ comparison as a stimulus to debate. I also want to suggest useful themes within
What’s Wrong with Europe’s Economy? 3
that debate – macro challenges versus micro; social choice versus economic efficiency; and under what conditions different social choices are sustainable. Let us begin with some facts. What long-term picture are we trying to explain? As Table 1.1 shows, Europe has grown more slowly than the US since the 1980s: 2.2 per cent a year versus 3.1 per cent in the US. But note that almost all of that difference disappears if we look, not at absolute GDP growth, but at GDP growth per capita. The difference in absolute growth is almost entirely explained by the fact that the US population is still growing at about 1 per cent a year, the European at about 0.3 per cent. As machines for delivering increasing prosperity to individuals, these two continental economies are very similar. Almost exactly equivalent over the period 1980–95, the US moved a bit ahead in the late 1990s. A still clearer pattern emerges when we look at productivity per hour worked (see Table 1.2) Until the early 1990s, the EU was increasing productivity and, as a result, hourly wage rates far more rapidly than the US. But the late 1990s
Table 1.1 US and EU growth compared 1980–95 US EU15
US EU15
1995–2001
1980–2001
3.6 2.2
3.1
1980–95
1995–2001
1980–2001
1.9 1.8
2.3 2.1
2.0 1.9
2.9 2.1
2.4
Sources: OECD Historical Statistics; OECD Economic Outlook.
Table 1.2 Labour productivity per hour worked (% per annum growth)
1990–95 1995–2000
US
EU15
1.1 2.2
2.4 1.5
Source: van Ark et al. (2002).
4
Adair Turner
Table 1.3 Unemployment rates, 1973–2000 Unemployment rate (%)
US France Germany EU 15
1974
1983
1996
2000
5.5 2.8 2.1 2.8
9.5 8.3 7.9 9.8
5.4 12.3 8.8 10.8
4.1 9.6 9.5 8.4
Source: OECD Historical Statistics (1973–2000).
saw a turnaround, with the US enjoying a strong productivity spurt and European productivity slowing, though still faster than America in the 1980s and early 1990s. As for unemployment, Table 1.3 shows a mirror image of the productivity picture. The US’s low productivity growth in the 1980s and early 1990s was matched by a long and sustained fall in unemployment, while the EU’s high productivity growth in the 1980s and early 1990s was accompanied by very poor employment creation. European unemployment then fell in the late 1990s, with strong employment growth in some major countries – Spain and France for instance, though notably not in Germany. But Europe’s unemployment rate was still appreciably above US levels. So, overall, we are not trying to explain a sustained and general failure of the European economy over a number of decades. We are trying to explain, first, why the US achieved a productivity spurt in the late 1990s while Europe did not, and, second, why Europe has tended to combine high productivity growth with low employment growth. There must however be a caveat even on this summary. Europe’s economic performance was not homogeneous. Figure 1.1 shows how several European economies since 1993 (Ireland, Finland, Spain, Sweden, the Netherlands and Denmark) achieved per capita growth rates similar to or higher than the US. Europe’s labour market performance is hugely diverse, as shown by Figure 1.2, with several economies, notably in Scandinavia, achieving employment rates as high as the US. Against that, some countries, notably the Mediterranean ones, had very low rates. But overall and on average, there are some Europe versus US distinctions worth explaining.
5
Ireland
8.4
Finland
4.4
Spain
3.1
US
3.0
Sweden
3.0 2.8
Netherlands
2.6
Denmark
Figure 1.1 Real GDP per capita growth, 1993–2000 (% per annum) Source: OECD Historical Statistics (1973–2000).
78
Norway Denmark
76
US
75
Sweden
73
Netherlands
72
Portugal
72
UK
71
Austria
69
Finland
68
Germany
67
Ireland
66
France
62
Belgium
60
Spain
57
Italy
55
Greece
55
Figure 1.2 Employment rates, 2000 (% of population aged 15–64)
6
Adair Turner
GDP per hour worked US France
102 100
GDP per employee US France
121 100
US France
Hours worked per employee US France
118 100
GDP per capita 140 100
Employees per head of population US France
115 100
Figure 1.3 Productivity, labour input and prosperity: France and US, 1999 (France ⫽ 100) Sources: O’Mahony and de Boer (2002); OECD Labour Force Statistics (1981–2001).
Let us begin with productivity. Before we look at the recent past, it is important to get the bigger picture clear. This reveals a big difference between the US and Europe: not in efficiency, but in social choice. Figure 1.3 compares France and the US in 1999. The US was then something like 40 per cent better off in GDP per capita, as measured by OECD purchasing power parities. This difference derived from two factors: 21 per cent higher output per person employed and 15 per cent higher employment per head of population. But higher output per person employed was in turn explained by American workers working 18 per cent more hours per employee than French. The difference in productivity per hours worked was trivial. Overall, all of the prosperity difference between the US and France is explained by a difference in total hours worked per head of population and none by measured efficiency per hour worked. This means that, if these lower hours are freely chosen by French people, then the difference between French and US GDP per capita is a product of choice, not of efficiency. French people are choosing a different trade-off between leisure and additional income. But are those lower hours freely chosen? Some of them are clearly not. Those which result from involuntary unemployment, or from involuntary early retirement, are not a product of social choice, but of labour market
What’s Wrong with Europe’s Economy? 7
inefficiency. France’s higher unemployment rate would suggest that at least five of the 15-point difference in employees per head of population are involuntary. But the best evidence from French public opinion surveys and from, for instance, their reaction to the 35-hour-week legislation is that the greater part of this difference in work/leisure tradeoff is freely chosen. If French people are happier with more leisure but less income than Americans, then no liberal economist should criticize them for that choice. The aim of liberal economies is to ensure that the economy achieves efficient frontiers of production and utility preference functions, but not to tell individuals where along those utility preference functions they should make their trade-offs. What is true of France is true of the rest of Europe. Table 1.4 shows that, in almost all European countries, hours worked per employee are below US levels and are falling. The only exceptions are the Scandinavian countries, where hours worked are stable but were far below American levels even twenty years ago. In America no such fall occurred. Why this major divergence in social preference exists is an interesting issue. On the basis of almost any theory of utility preference, it is the American behaviour that is more difficult to explain: a society getting steadily richer, but totally focused on more income, not more leisure. For my purposes here, the implication is simple: the fact that Europe’s GDP per capita growth has been marginally below the US’s for 20 years is not a sign of inferior economic performance, but reflects the fact that Europeans have chosen to turn increased productivity into a mix of greater leisure and greater income and Americans to turn it solely into greater income. Table 1.4 Hours worked per person in employment 1979
1990
2001
West Germany France Italy
1 732 1 806 1 715
1 583 1 657 1 674
1 446 1 532 1 606
UK Ireland
1 815 –
1 767 1 922
1 711 1 674
Sweden Denmark
1 517 –
1 549 1 492
1 603 1 482
US
1 838
1 838
1 821
Source: OECD Employment Outlook, July 2002.
8
Adair Turner
120.00
100.00
80.00
60.00
40.00
20.00
19 5 19 0 5 19 2 5 19 4 5 19 6 5 19 8 6 19 0 6 19 2 6 19 4 6 19 6 6 19 8 7 19 0 7 19 2 7 19 4 7 19 6 7 19 8 8 19 0 8 19 2 8 19 4 8 19 6 8 19 8 9 19 0 9 19 2 9 19 4 9 19 6 98
0.00
US
UK
France
Total Germany
West Germany
Japan
Figure 1.4 Labour productivity per hour worked: US (1996 ⫽ 100) Source: O’Mahony and de Boer (2002).
There is a second important ‘big picture’ point. Figure 1.4 shows productivity trends over the last 50 years (as set out in Mary O’Mahony and William de Boer’s latest study for the NIESR) for the US, three leading European economies and Japan. The story is one of catch-up by France and West Germany between the 1960s and the 1990s, as they closed the gap with the US, but noticeable absence of catch-up by the UK. Despite a slight closing of the gap in the 1980s and 1990s, British productivity remained well below US, French and West German levels. The modest rate of catch-up shown would take 60 years or so to close the gap. It is worth spelling this out, because in itself it should throw a big bucket of cold water over some of the conventional wisdom about Europe’s problems. Continental Europe, it is claimed, needs to copy the successful model of the UK’s liberalized product and labour markets in order to catch up with the US leader. Figure 1.4 would suggest that the answer is a good deal more complex.
What’s Wrong with Europe’s Economy? 9
But the picture is not as favourable to France, Germany or, indeed, Britain as Figure 1.4 might suggest. For whereas it shows France and West Germany on a par with the US, and a US acceleration in the late 1990s so slight that we cannot even make it out visually amid these overlapping lines, these figures are for the whole economy. If we narrow down our focus to the private market sector of the economy, we find a more significant US advantage, both in terms of level and in terms of recent growth. Thus the O’Mahony and de Boer figures for relative output per hour worked, shown in Figure 1.5 suggest substantial American advantages in major sectors such as manufacturing, transport and communications and distributive trades. They also suggest a major acceleration of US productivity growth in the market sector of the economy in the late 1990s (see Table 1.5), an acceleration not matched in Europe. The issue, therefore, remains: first, why does the US achieve higher productivity levels than all Europe countries in some specific marketoriented sectors; and, second, what happened to US productivity in the 1990s. Why do the Bureau of Labor Statistics figures in Figure 1.6, for instance, show the trend growth rate accelerating from 1.4 per cent in the 20 years prior to 1995 to 2.4 per cent since then? Why did
Manufacturing
UK
Germany*
France
US
Distributive trades
100
UK
129 Germany*
132
155
France
US
Transport and communications
100
UK
112
150
161
Germany*
100
88
France
US
101
113
Figure 1.5 Relative output per hour worked by sector, 1999 (UK ⫽ 100) Note: * German figures are for all Germany. Separate West German figures are unavailable on a comparative basis after about 1996, but would show a significantly higher productivity. Source: O’Mahony and de Boer (2002).
10
Adair Turner
Table 1.5 Labour productivity per hour worked (% per annum growth)
1979–89 1989–95 1995–99
US
France
UK
Germany*
0.97 1.15 1.92
2.94 1.42 1.16
2.41 2.28 1.37
1.92 3.13 1.98
Note: * West Germany prior to 1989. Note that high German productivity growth in the 1990s is strongly driven by rapid increases in East German productivity from very low levels. Source: O’Mahony and de Boer (2002).
120
Trend: 2.4% p.a.
110 100 90 Trend: 1.4% p.a. 80
1973 1975 1977 1979 1982 1984 1986 1988 1991 1993 1995 1997 2000 2002 Figure 1.6 US productivity per hour, 1973–2002 (Index 1992 ⫽ 100) Note: Index on logarithmic scale. Source: US Bureau of Labor Statistics.
Europe not share this acceleration? Is this evidence of sclerotic, overtaxed, uncompetitive Europe unable to grasp the potential of IT investment? One way to answer these questions is to break down the differences, both in level and in growth rate, by sector. A sectoral breakdown suggests once more that differences in social choice may be as important as any European failure in absolute economic efficiency. Figure 1.7 sets out, on the left, O’Mahony and de Boer’s breakdown of the productivity gap between the US and the UK. On the right, it
What’s Wrong with Europe’s Economy? 11
USA versus UK Primary and construction Utilities Manufacturing Distributive trades
USA versus Germany Primary and construction
1
Utilities
2
Manufacturing
10 14
Finance and business services Other* Total market services
6 3 36
3 2 (3)
Distributive trades
12
Finance and business services
1
Other* Total market services
2 17
* Transport, Communication + Personal services
Figure 1.7 Industry contribution to labour productivity gap (% points of difference) Source: O’Mahony and de Boer (2002).
shows my own estimates (using their figures) for the US–Germany gap.1 From both comparisons, one sector leaps out: the distributive trades, both retail and wholesale. Lower productivity in retailing and wholesaling accounts for almost 40 per cent of the total productivity gap between the UK and the US and for 70 per cent of the gap between Germany and the US. The dominance of distributive trades in productivity gap comparisons holds true for Europe in general and, indeed, for Japan. Other advanced countries get reasonably close to US productivity in manufacturing, in utilities and even in finance and business services, but almost all fall well short of the US in retailing and wholesaling. The reason for this gap is rooted in deeply physical facts. It is simply far easier to run retailing and wholesale distribution at a high level of efficiency in a lightly populated, dispersed physical environment, such as the US, than in more densely populated Europe or Japan. Physical space has a pervasive impact on many aspects of productivity, but above all in those sectors concerned with the physical movement of goods. In-store retail management is more efficient if you have aisles wide enough to allow restocking without consumer service degradation. Back-of-store handling efficiencies are far easier to achieve if you have large trucks, coming at predictable times, having
12
Adair Turner
travelled along uncongested freeways, and turning round in wide spaces than with smaller trucks, arriving at unpredictable times, and queuing to use constricted unloading bays. Productivity is increased if average pack size is bigger because consumers have bigger fridges, located in bigger kitchens. Size and physical space matter. In all sorts of subtle but inherent ways, America gets a productivity advantage from the fact that only a small proportion of its population lives in the kind of dense urban clusters that dominate the economic geography of England, the Netherlands, Germany, Italy or Japan. The only European country that comes close to America in measured productivity in the retailing and wholesaling sectors is France, because France has lower population density and, as a result, has been more willing to allow large out-of-town retailing developments of the sort common in the US. France has close to US levels of retailing productivity because the outskirts of French towns look more like the outskirts of American towns than elsewhere in Europe. Physical space is a major driver of the difference in the level of European and US productivity, not only in distributive trades but also in some subsectors of manufacturing, where extensive plant layout is also helpful in raising productivity. The evidence suggests that batch size and absolute scale is often the key factor in superior US manufacturing productivity, made possible by easier long-distance transportation and thus by the ability to serve a larger market from one plant location. But, while physical space could explain an inherent difference in productivity level, it is less obvious that it could explain the difference in recent growth rates. Why has US productivity growth accelerated? Much attention has focused on the impact of IT, both in its own production industries and in its use. Thus work by van Ark et al. (2002) (see Table 1.6) shows that the acceleration of total US productivity from 1.1 per cent to 2.2 per cent during the 1990s was explained by two factors: ●
First, a slight acceleration in productivity growth in IT-producing industries from 6.1 per cent to 6.5 per cent, an effect amplified by the fact that these IT-producing sectors (the manufacture of semiconductors, PCs, mobile phones, etc.) were growing as a percentage of the total economy.
What’s Wrong with Europe’s Economy? 13
Table 1.6 Labour productivity by type of industry (% p.a. growth rates) United States 1990–95 1995–2002
European Union 1995–2002
Total economy
1.1
2.2
1.5
ICT-producing industries
6.1
6.5
8.5
ICT-using industries
1.4
4.2
1.3
Non-ICT industries
0.4
0.4
1.0
Sector drivers of growth differential Retailing: 55% Wholesaling: 26% Securities trading: 20% Other sectors: Nil
Source: van Ark et al. (2002).
●
Second, a major acceleration in productivity growth in sectors intensive in their use of ICT from 1.4 per cent to 4.2 per cent a year.
Those sectors of the economy that were neither ICT producers nor intensive ICT users showed no acceleration at all. Van Ark et al. then compared their findings with European figures, with two interesting results. First, the Europeans have achieved quite as fast growth in ICT production. (Nokia’s mobile phone factories are quite as good as Motorola’s.) Second, however, Europe has failed to achieve accelerated productivity growth in those sectors that are intensive users of ICT. But, when they then looked at which specific sectors are failing to achieve this ICT-based productivity leap, they arrive at the startling result that 80 per cent of the entire difference in productivity growth rates between Europe and the US is located in retailing and wholesaling. Since the end of the 1990s, US productivity in retailing and wholesaling has been growing at a dramatic 6 per cent a year, while typical European growth rates have been only 1 per cent or so (see Figure 1.8). This fact alone seems to account for the vast majority of the entire difference in growth performance that we are trying to explain. So why has US retailing and wholesaling achieved such a productivity spurt? The answer seems to lie in a combination of ICT-based redesign facilitated by use of physical space, that is, by ease of
14
Adair Turner
USA Utilities
France
US – France 3.2
2.3
(0.9)
Manufacturing
4.2
2.7
1.5
Transport and communications
4.2
2.5
1.7
Distributive trades Financial and business services Non-market services
1.1
5.8 2.4
(0.1)
4.7
(0.3)
2.7 0.8
(0.9)
Figure 1.8 Labour productivity growth by sector, 1995–99 (% per annum) Source: O’Mahony and de Boer (2002).
greenfield development. All studies of the US retailing productivity phenomenon stress the importance of new build of very large discount stores by Wal-Mart, Home Depot, Best Buy, Circuit City and the like. One major study indeed finds that all of the productivity growth in US retailing in the 1990s is attributable to the new entry of new, more efficient retailing establishments, new stores on new sites, and that none is attributable to productivity improvement within existing stores.2 Big new stores on greenfield sites achieved huge productivity improvements, partly from the use of ICT, partly from exploiting the inherent physical advantages of large-site operation, and partly from the combination of the two. Barcode-based logistics systems dramatically cut inventory levels in circumstances where replenishment time can be predicted because of freely moving roads and uncongested back-of-store unloading. In Europe this does not seem to have happened and, if these studies are right, that fact alone explains the vast majority of Europe’s widening productivity gap versus the US. Productivity did not spurt because large out-of-town new entry is irrelevant to retailing competition in the densest urban environments, because planning restrictions limit large out-of-town development where it is potentially relevant, and because our roads are congested. Even in France, where the 1970s and 1980s saw an explosion of an earlier generation of outof-town developments, the freedom to open new still larger stores was restricted in the 1990s. French retailing productivity, while higher than elsewhere in Europe, is now growing as slowly.
What’s Wrong with Europe’s Economy? 15
What implications follow for policy? Some would say that they are clear. Get rid of these planning restrictions and seize this productivity potential. Let neither ‘nimby’ opposition to hypermarkets in Hampshire nor Italian small shop owners lobbying against supermarkets on the edge of Florence stand in the way of prosperity. Certainly, the link between planning rules and productivity (and between planning rules and ease of competitive entry) is a fertile policy issue. But Europe’s planning restrictions are not there for arbitrary reasons; they express consumer preferences for preserved countryside and preserved urban communities. They are more restrictive than in the US because there is less space than in the US. Again, we face an issue not just of economic efficiency, but of consumer and social choice, and one where the difference between the US and Europe is easy to understand. The more crowded the environment, the more value citizens will place on environmental preservation and the lower, everything else being equal, will be the utility-maximizing productivity level in any sector where there exists a trade-off between measured productivity and environmental quality. Both the productivity gap between Europe and the US and the increase in that gap in the late 1990s may, therefore, be significantly determined by social choices, which it is unlikely that we could change, or perhaps should seek to change. Given this analysis, how well targeted and how relevant is the liberalization agenda being pursued by Europe and being urged upon Europe by the British government? The answer, I believe, is that the product market liberalization on which much attention is focused is useful and worth pursuing, but unlikely to make more than a marginal difference to the gap in productivity levels between the US and Europe. The Spanish presidency defined five priorities for the Barcelona summit in March 2002: strengthening European networks, liberalizing energy and financial service markets, reforming European labour markets, and improving Europe’s education and science base. Within these five priorities, most attention was focused on energy liberalization. The rhetoric surrounding this summit stressed the vital character of these reforms. Progress on economic reform in general, but on energy liberalization in particular, it was argued, was essential to ‘kick-start the European economy’. If Europe failed on this, due to French intransigence, doubts would be cast on the credibility of the entire reform programme and Europe’s growth would suffer.
16
Adair Turner
These statements strike me as largely nonsense for two reasons. First, I think that they confuse short-term conjunctural effects with long-term effects. No supply-side structural change ever kick-starts the short-term growth prospects of an economy. Second, they vastly overstate the importance of the energy sectors, even in the long term. The importance of the energy sectors to the productivity gap between Europe and the US is trivial. The correlation across Europe between low energy prices to industry and degrees of energy market liberalization is mildly positive, but only mildly. The total benefit to the European economy of energy market liberalization, in a sector accounting for no more than 2–3 per cent of GDP, could not possibly amount to more than a small fraction of 1 per cent of GDP. Even such a benefit would accrue only over the long term and would have no impact on the growth rate this year or next. The case for energy market liberalization is on balance a good one, but this is no more than a marginal issue in the economic history of Europe. Financial service liberalization and airline competition, the other product market changes on the Barcelona agenda, are somewhat more important. Financial services productivity does show up as a significant driver of the productivity gap for some European countries and there are major inefficiencies in European financial service delivery which fiercer competition might help to shake out. But what is still striking, looking at the totality of this agenda, is how slight it is relative to the rhetoric of far-reaching liberalization, or relative to the size of the productivity gap that opened up between the US and Europe in the late 1990s. I do not believe that this is due to some failure of imagination or will on the part of European politicians. Instead it reflects two facts: ●
●
First, most of the potential for spurring productivity through product market liberalization has already been seized, through the single market programme of 1986–92, through the privatizations and domestic liberalizations pursued by governments across Europe, and through the telecoms and postal services liberalizations already agreed at European level. Second, the key drivers of the remaining productivity gap arise in sectors where physical constraints are important and where the key policy decisions to be made, for instance on land planning, need to reflect trade-offs at local level, which are and should be matters of national or regional, rather than European-level, competence.
What’s Wrong with Europe’s Economy? 17
Norway
78
Denmark
76
US
75
Sweden
73
Netherlands
72
Portugal
72
UK
71
Austria
69
Finland
68
Germany
67
Ireland
66
France
62
Belgium
60
Spain
57
Italy
55
Greece
55
Figure 1.9 Employment rates, 2000 (% of population aged 15–64)
One of the five items on the Barcelona economic agenda – reforming labour markets – is, however, clearly important. Europe’s productivity failure has been greatly overstated and is concentrated in particular sectors with particular characteristics. But some European labour markets do suffer from a more pervasive and longerlasting problem, a tendency towards a high level of unemployment even during periods of rapid growth. Europe as a whole, however, does not have a labour market problem. Some countries in Europe have different national problems. As Figure 1.9 (repeating Figure 1.2 on page 5) shows, Italy has one, Denmark does not. So their cause cannot be Europe-wide factors, such as relatively high taxes, or legislation introduced by the European Union. Instead, labour market problems arise from complex combinations of different factors: the level of minimum wages, the ease of hire and fire and the ease of part-time working, the nature of wage-fixing processes, and the generosity and, crucially, the conditionality
18
Adair Turner
Wages
Employer’s social security
Germany
20.8
7.7
US
20.2
5.3
28.5
25.5
Figure 1.10 Average hourly wage costs, 2000 (Euros) Source: Eurostat Yearbook 2002, p. 213.
of unemployment benefit. Those combinations are nationally specific and so, too, must be the solutions. But it is clear that in some countries there is a major problem and that its solution will require political courage. In this area, at least, difficult structural reforms are needed. But such changes in European labour markets do not need to be accompanied by wholesale dismantling of European welfare states to allow tax cuts in order to make European economies competitive. The belief that such changes are needed is, however, a mainstay of journalistic commentary and of business lobby group belief. Europe (and Germany in particular), the argument goes, simply has too high labour costs, partly because of higher wages, but primarily because of high non-wage labour costs. As Figure 1.10 shows, German labour costs are 10 per cent higher than American, because of 50 per cent higher social security charges. Therefore, European economies must cut taxes to become competitive. But these arguments are at the very least greatly overstated. There must, of course, be some level of tax burden which so distracts incentives as to undermine prosperity, but the empirical evidence does not support the idea that there is an inverse correlation between the tax burden and growth, or between the savings rate and growth, or between employment levels and growth, at around the rates of tax observed in European and other OECD countries. Germany and Italy are Europe’s current economic growth laggards, with tax burdens of almost 46 per cent (see Figure 1.11) But Europe’s recent growth
What’s Wrong with Europe’s Economy? 19
Sweden
52.5
Denmark
50.8
Germany
45.9
Italy
45.7 44.6
Finland
41.7
Netherlands
38.5
Spain
Figure 1.11 General government total outlays as percentage of GDP, 2001 Source: OECD Economic Outlook, June 2002.
GDP per capita % growth
5 Japan
4.5 4 3.5 3 2.5 2 1.5 1 25
30
35
40
Public expenditure (% GDP)
45
50
55
y = −0.0173 x + 3.3374 R 2 = 0.0347
Figure 1.12 GDP per capita growth and public expenditure, 1960–95 Source: OECD Historical Statistics.
successes include Sweden and Denmark with much higher burdens, Finland with roughly the same and the Netherlands and Spain with somewhat less. Over a longer time period, the correlation of growth with tax burden is weak. The best fit on this 35-year correlation has an R2 of just 0.03, and the slightest of downward slopes (see Figure 1.12).
20
Adair Turner
(Indeed, this becomes a slight upward slope, if the one plot point for Japan is removed.) So I am deeply suspicious of the argument that there is any generalized correlation between tax burdens and economic performance. However, I do think that the high levels of German total labour costs may illustrate an important macroeconomic problem created by the transition to the euro. High payroll taxes need not (and usually do not) create a competitiveness problem, for two reasons. The first is that, at least if labour markets are reasonably flexible, the true incidence of payroll taxes falls not on employer costs, but on the wages of employees, workers receiving a lower wage rate than they would if payroll taxes were lower. In flexible labour markets, payroll taxes, though nominally falling on employers, are no different in their effective incidence from income taxes. The fact that, say, Sweden has higher payroll taxes than Britain should, in theory, mean simply that Swedish wages are lower than they otherwise would be, Swedish citizens receiving more of their remuneration in collective government expenditure and less in a direct wage form. Indeed, not only is that true in theory, but the fact that Sweden is a highly competitive economy, with a robust external position and achieving a high rate of employment, suggests that it is also true in practice. But, even if this were not the case, the second reason for doubting this ‘competitiveness’ argument would still apply. In a floating exchange rate system, any imbalance produced by a combination of high payroll taxes and labour market inflexibilities can be offset by a decline in the exchange rate. The idea that, with an externally floating exchange rate, the whole Eurozone can suffer a ‘competitiveness’ problem is just a theoretical confusion. The concept of high taxes creating such a competitiveness problem is just a specific subset of that confusion. But, clearly, it is possible for a specific country within the fixed exchange rate regime of European monetary union to be, at least for a transitional period, in a state of competitive disequilibrium. This is particularly so if it has relatively inflexible labour markets in which downward adjustments of nominal wages will be resisted. It is, however, possible that Germany is in this position, having entered the euro at too high an exchange rate. How far that is the case is debatable. But certainly it is a possible result of entry into a fixed exchange
What’s Wrong with Europe’s Economy? 21
rate regime, a risk inevitably present in the initial creation of a monetary union. So it is possible that German total labour costs are now too high and one way to reduce them would indeed be to reduce non-wage labour cuts, that is, to cut payroll taxes. But another way would be to reduce wages. This illustrates the point that, if Germany does now have a problem of uncompetitively high labour costs, the problem is not essentially driven by the burden of tax, but by an exchange rate fixed too high. Finland (in the euro) and Sweden (outside it) have equally high nonwage labour costs, but no evidence of a competitive disequilibrium of the sort which may be facing Germany. If there is a German labour cost problem, it does not prove the argument that European tax burdens are unsustainably high.3 But it may point us towards a significant problem in the macroeconomics of the eurosystem. Let us now look therefore at macro demand policy issues. My focus up to this point has been on long-term supply-side differences between Europe and the US. But it is differences in demand and in the macro-policy framework that explain the divergent growth since 2001. This point needs to be spelled out clearly, precisely because – bizarrely – it is frequently denied. Since 2001, Eurozone growth has severely disappointed expectations, showing only 1.6 per cent in 2001 and 0.7 per cent in 2002, with the performance in the zone’s largest economy, Germany, lower still at 0.6 per cent and 0.3 per cent. The US, meanwhile, continued to grow far more robustly. The reason for Europe’s disappointing performance is not a failure of competitiveness, nor structural problems, but inadequate domestic demand. Figure 1.13 shows how Eurozone exports, for instance, have actually performed better than in America or the UK. Net exports have continued to make a positive contribution to growth. But the domestic demand contribution to Eurozone growth has plummeted, with household consumption, household borrowing, retail sales, car sales – whatever measure of domestic demand you choose – far less dynamic in the Eurozone than in the US. What is true for the Eurozone as a whole is even truer of Germany, the slowest-growing of the large European economies. German export performance, for all the talk of a competitiveness problem, has been remarkably resilient since 2001, but consumption expenditure has fallen, bringing down real domestic demand. Germany, of course, has
22
Adair Turner
20
5
15
4
Eurozone
10
3
5
2
0
GDP growth
Domestic demand contribution
1 UK
−5
0
−10
US
−15
Net exports contribution
−1 −2
1998
1999
2000
2001
2002
1998
1999
2000
2001
2002
Figure 1.13 Exports have sustained Eurozone growth Source: Merrill Lynch, European Economics, January 2003.
some long-term structural problems, but these are not the cause of its present slow growth. German nominal GDP has risen only 2 per cent a year since 2001, US nominal GDP by 3.5–4.0 per cent, the UK’s by almost 5 per cent. This, not structural problems, is the fundamental reason why the German economy has grown slowly of late. That slow growth in nominal demand derives from the combination of fiscal and monetary policy and is rooted in the macro-policy rules of the euro system and the fiscal positions of countries at the time of entry.4 To many people, the demon here is the Stability and Growth Pact (SGP), which limits public deficits to 3 per cent of GDP and which is, therefore, forcing Germany to tighten fiscal policy, a move which will make Germany’s recession worse and which could provoke a dangerous spiral of falling prices and falling demand – a classic deflation trap. There are good arguments for recasting the Stability Pact, basing it on cyclically adjusted measures. But I do not think that we should fool ourselves that the problems derive solely from an illdesigned set of rules. For some set of rules, either on deficits or on accumulated debt levels, is essential to ensure the solvency of European governments within a monetary system where nations have surrendered the ability to inflate their way out of debt traps. Almost any set of rules would curtail today’s freedom of fiscal manoeuvre of major Eurozone countries, given the positions with which they entered euro. The Maastricht limit on accumulated stock of debt (less than 60 per cent of GDP) and the Stability Pact guidance
What’s Wrong with Europe’s Economy? 23
140 124%
Percentage of GDP
120
107%
100 90% 80 60
Maastricht criterion
40 20 0 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Figure 1.14 Italian gross government debt (as % of GDP) Source: OECD Historical Statistics (1973–2000).
on structural deficits (ideally to be about zero) were not and are not pointless masochism, but actually good Keynesian guidelines. If they had been met, they would now give countries the freedom to allow automatic fiscal stabilizers to operate. The problem is not that those rules were set, but that they were not met. Italian gross government debt at 107 per cent and, perhaps, actually rising this year (see Figure 1.14) is far above the Maastricht 60 per cent guideline and will severely limit Italy’s ability to provide fiscal stimulus for many years to come, whatever adjustments to the Stability Pact are now agreed. The fact that the major Eurozone countries, as shown by Figure 1.15, entered the global economic slowdown with significant structural deficits was bound to limit their fiscal response and thus to depress nominal demand. We should certainly reform the SGP. But we have to face the facts. Failure fully to meet the Maastricht convergence criteria before going ahead with the euro has introduced a potentially dangerous constraint on fiscal policy response. This is a key cause of the Eurozone’s poor performance since 2001. If fiscal policy is constrained, the burden of policy response inevitably falls more on monetary policy. Whatever the fiscal constraints, monetary policy (provided it is pursued aggressively enough) should always be able to maintain nominal demand growth at
24
Adair Turner
1.5%
UK
1.3%
US
(1.0%)
Eurozone
Germany
Italy
France
(1.3%)
(1.4%)
(1.7%)
Figure 1.15 General government structural balances, 2000 (% of GDP) Source: OECD Economic Outlook (June 2002).
reasonable levels, either via interest rate policy, or (if the zero bound of zero interest rates is reached) via government debt monetization. But this potential capability needs to be guided by appropriate targets and there is an overwhelming case that the European Central Bank’s (ECB) current inflation target of 0–2 per cent is significantly too low. There is no science to setting an optimal inflation target, but it should be high enough to provide some lubrication to real wage flexibility and to give the central bank the option, when needed, of setting interest rates at a negative real level while also being low enough to prevent a cycle of increasing inflationary expectations. Something like 2–3 per cent is almost certainly better than 0–2 per cent. There are two specific reasons for nudging the target higher still in the specific conditions of the Eurozone. The first is the danger that entry exchange rates could be fixed wrong, which, as we have seen, is the real issue in relation to German labour costs. This argues for ensuring a reasonable level of inflation even in the lowest-inflation countries. If Germany needs to adjust real wages down, it will do so more easily with mild inflation, avoiding the need for nominal wage cuts. The second is that, for reasons to do with different productivity growth rates between traded and non-traded sectors of the economies,
What’s Wrong with Europe’s Economy? 25
countries in a process of catch-up towards higher income levels (such as Spain, or Portugal, or in the future Poland) need within a fixed exchange rate regime to have measured inflation rates above the level appropriate for the richest countries. This is the so-called Balassa–Samuelson effect and applies even to the current Eurozone, let alone the Eurozone after enlargement. My calculations suggest that the optimal inflation rate for Spain, if it is to converge to German standards of living over, say, a ten-year period with a 4 per cent real growth rate, is something like 4 per cent. This implies both that the ECB should not overreact to that sort of inflation rate in the catch-up countries and that its overall Eurozone inflation target needs to be about 0.25 per cent higher to reflect these effects within the existing Eurozone and, perhaps, 0.5 per cent higher post-enlargement. It also implies that any attempt to make the new accession countries meet the Maastricht convergence criteria on inflation before entry to the euro will further depress the European economy. Overall, therefore, the case for changing the ECB’s target and for making it explicit is overwhelming. It is only macro-policy changes, such as a changed ECB approach and greater flexibility within the SGP, that have the ability to make a difference to Eurozone growth rates over the next few years. To sum up, there is a conventional wisdom that Europe’s recent poor performance is a sign of deep structural problems, which must be addressed by product market liberalization, by labour market reform and by the reduction in uncompetitive tax burdens. The agendas set out at Lisbon, Stockholm and Barcelona define crucial priorities in that economic reform process. But, against that conventional wisdom, I have argued that: ●
●
●
First, the Eurozone’s poor growth is caused by macro-policy problems and rules, which must be changed. Second, taxation burdens are not in some general and structural sense unsustainable. Third, when we look at longer-term differences between European and American prosperity and productivity, we need to recognize the impact of physical environment and social choice. Within this context, while the European Union’s product market liberalization agenda is positive and sensible, it is unlikely to have more than a marginal impact.
26
●
Adair Turner
Fourth, however, some European Union labour markets are seriously inefficient and should be reformed.
Provided such a policy mix is pursued, there is no reason why Europe should not continue to grow GDP per capita at an attractive rate, delivering increasing prosperity and increasing employment in an already rich continent. Its absolute GDP per capita will almost certainly remain permanently below America’s, because of the social choices that Europeans make,5 but it will grow as far as fast. Its total GDP growth, however, will continue to lag the US’s, simply because of different population growth rates. The issue that then arises is whether these choices – to sacrifice some productivity potential in order to protect rural and urban environments and to take some of the benefits of productivity growth in increased leisure – are unsustainable? Some people believe they are, assuming that Europe cannot choose to trade off income for leisure, or income for protected environment, because such choices make Europe ‘uncompetitive’ in the global economy. But these arguments are in general quite wrong. National and continental economies do not compete with one another in the normal sense of the word and societies have wide degrees of freedom to make their own social and economic trade-offs. If Europeans choose both to produce less and to consume less, that has no consequences for competitive sustainability. Provided Europeans understand the consequences of their trade-offs – for instance, that shorter hours mean lower GDP per capita – there is nothing unsustainable about that choice. But choices could become unsustainable, if based on inconsistent assumptions: for instance, on decisions to do less work without accepting the consequences of lower income. And there is certainly one European social choice that is unsustainable – the current combination of birth rates, retirement ages and explicit or implicit pension promises. Europe’s rapidly ageing population will, unless average retirement ages rise significantly, produce soaring dependency ratios and will lead to increases in public pension spending that would push tax levels to unsustainable levels6 and would place huge strains on corporate pension provision, or leave many pensioners with inadequate provision in old age. On pensions, European citizens are making decisions and assumptions which are unsustainable. The possible solutions
What’s Wrong with Europe’s Economy? 27
to this huge challenge are beyond the scope of this lecture. But the scale of the challenge is clear. The argument that deep and difficult structural reform is required to the European mode is often overstated, but not in relation to retirement ages and pension provision. All of this implies a very different ordering of Europe’s economic priorities than is implied by the agenda of economic reform summits. At these summits, the Eurozone macro framework receives limited attention and the ageing population only the most general discussions. Yet those are the most important issues we face: the one shortterm, the other long-term. Labour market reform is discussed, but in general rather than specific terms. By contrast, product market liberalization receives top billing, even though the concrete proposals under discussion could make only a minor difference to long-term prospects and almost none to growth this year or next. The reason for this disconnection is clear. European Union summits talk about product market liberalization, because the European Union has a specific competence in that area. They talk only in general terms about labour markets, because the specific actions required are for national governments. They avoid macro issues, because the Maastricht Treaty gives the ECB independence as to specific objectives, as well as independence in pursuit of those objectives. And they largely avoid the pension issue, because that, too, is a national issue, extremely long-term and very sensitive. The European Union can only do what it is empowered to do; its agendas reflect its competence and, of necessity, always will. But it is important to change the priorities, if not necessarily of European summits, then of the wider economic reform debate. Macro-policy constraints need to be addressed. In the 1960s and 1970s, policy discourse overrated the power of macro demand policies and underrated supply-side structural factors. The danger today in Japan and Europe is the inverse. There is a real risk, if we do not address EMU’s macro risks and deficiencies, that these will outweigh the long-term supply-side benefits of the euro project. The pension issue needs to be brought centre stage, as the most difficult structural issue Europe faces. Labour market reform needs to be recognized as a more crucial challenge than are the remaining steps of product market liberalization. And, if physical constraints really are the major driver of the productivity gap versus the US, we need to recognize that fact and to
28
Adair Turner
consider carefully the trade-offs involved in land-use planning. We need to ensure that underinvestment in transport does not make the productivity penalty worse and we need to avoid fooling ourselves that legal liberalizations, such as in energy markets, are going to make more than a marginal impact. But we also need to consider how, if Europe will never be as productive as the US on the basis of expansive land use, it can better exploit the potential advantages of its economic geography, ensuring that its cities are vibrant centres of tourism, education, science and creative industries. Europe needs a vision of a vibrant future that is not simply a carbon copy of the American mode. But it needs to start with a more subtle and detailed understanding of its economic performance, absolute and relative to the US, than the easy mantra of ‘structural reform’ expresses. Let me end, however, with one final thought on the issue of social choice. We need to recognize the consequences of our social choices, even if, having recognized them, we stick with them. I have in this lecture been fairly positive towards Europe’s social choices: its preference for a mix of leisure and income, its preference for preserved rural and urban environments over the final four percentage points of productivity potential. They seem to me perfectly natural preferences for rich societies. The measure of economic success should be income per hour worked, rather than absolute economic growth. And sensible societies may well sacrifice even some income per hour worked in favour of environmental objectives. But these choices do have consequences for absolute economic growth and absolute economic growth has consequences for geopolitical power. The choices that Europe is making – on its birth rate, on immigration, on leisure and hours worked, and on physical planning, quite apart from on defence expenditure as a percentage of GDP – will continue to drive a major shift in the relative economic and geopolitical power of the US and Europe. Europe will probably continue to grow GDP per hour worked quite as fast as the US, but it is almost inevitable that American absolute GDP will continue to grow significantly faster than Europe’s, unless America’s population growth slows to European levels and Americans begin to make the same income/leisure trade-offs as Europeans. Europe is, I believe, making social choices that are rational and natural for human beings in mature, already rich and peaceful societies.
What’s Wrong with Europe’s Economy? 29
The only doubt I leave hanging is whether the world can be peaceful enough to sustain these desirable choices, to allow us to focus on the liberal objective of maximizing individual choice, rather than the statist objective of maximizing national wealth. I hope it can be. I actually believe it can be. But it is a debatable issue.
Notes 1. These figures combine the impact of different productivity levels within a sector with the impact of the relative size of the sector (‘Total industry contribution’ as per Appendix E-1 in O’Mahony and de Boer, 2002). The figures for ‘level’, which exclude sector share effects, show a slightly less pronounced but still strong dominance of the distributive trades sector. 2. Foster et al. (2002), quoted in Robert Gordon, High-Tech Innovation and Future Productivity Growth. 3. But note that, while the general case that tax burdens (and specifically non-wage labour costs) are too high appears either overstated or wrong, there are good arguments for believing that high payroll taxes on, specifically, lower-wage workers can be harmful, since it is likely that at low wages the incidence shift from employers to employees does not occur. See Turner, ‘Choosing the State We Want’, for the detailed argument. 4. See Turner, ‘The Monetary Policy Framework – Guarding against Deflation’, Aspen Italia, 2003, for a more detailed discussion of these issues and of Balassa–Samuelson calculations. 5. The experience since the mid-1990s has been that the size of the GDP per capita gap has increased, with lower European productivity growth rates driving lower GDP per capita growth. But the driving factors are likely to be one-off – the potential benefits of physically expansive retailing increasing over a number of years, but not limitlessly. The best guess of future trend is that European GDP per capita will remain below US in level, but with roughly similar growth rates. 6. The unsustainability would derive not from ‘competitiveness’ effects, but from popular opposition to rising tax rates, expressed either in political rejection of governments that imposed it or in tax avoidance and evasion.
References Foster, L., J. Haltiwanger and C.J. Krizan (2002) The Link Between Aggregate and Micro Productivity Growth: Evidence from Retail Trade, NBER Working Paper No. 9120. Gordon, Robert (2002) High-Tech Innovation and Future Productivity Growth: Does Supply Create its Own Demand, NBER Working Paper No. 9437.
30
Adair Turner
O’Mahony, Mary and William de Boer (2002) Britain’s Relative Productivity Performance: Updates to 1999, NIESR March. Turner, Adair (2000) ‘Choosing the State We Want’, Lecture at the LSE, October. Turner, Adair (2003) ‘The Monetary Policy Framework – Guarding against Deflation’, Paper delivered at Aspen Italia conference, February. van Ark, B., R. Inklaar and R.H. McGuckin (2002) ‘Changing Gear: Productivity, ICT and Service Industries: Europe and the United States’, Paper Presented to Brookings Workshop, May.
2 Assessing the Euro: Expectations and Achievements Peter B. Kenen
There are some things that I shall not be discussing in this lecture. I shall not try to assess the quality of the monetary policy followed by the European Central Bank (ECB), the quality of its policy-making framework, or the merits of the rigorous way it quantifies price stability. I shall not explain why I believe that Britain should adopt the euro, although some of the issues raised in this lecture bear directly on that issue. I shall, instead, revisit issues debated during the run-up to monetary union, drawing on subsequent experience and more recent research. I shall ask whether they were the right questions and whether we went about answering them in the most useful way. These are the questions on which I will concentrate: ●
●
●
●
Do the members of the European Union (EU), or some subset of them, come sufficiently close to being an optimum currency area? Should monetary union be accompanied by some sort of fiscal union? Will monetary union contribute strongly to the further integration of goods and asset markets and does the single market need a single currency? Will the euro challenge the dollar as the main international currency?
I shall argue that we did not give very good answers to the first question, because we ignored the way in which a monetary union alters the incidence of exogenous shocks. I shall seek to explain why 31
32
Peter B. Kenen
a monetary union need not be accompanied by a fiscal union and why its members should not be constrained by restrictive fiscal rules. I shall discuss the recent research suggesting that a monetary union can have surprisingly strong effects on goods market integration. And I shall explain why the euro is not likely to displace the dollar, but why that should not matter much. When Europe began moving towards monetary union in the late 1980s, economists turned to the theory of optimum currency areas – OCA theory for short – as the appropriate framework for assessing the costs and benefits of monetary union. But we did not fully appreciate the restrictive nature of the assumptions on which OCA theory was based. These assumptions were very commonly used in the early 1960s, when Robert Mundell wrote the paper that gave OCA theory its name.1 We assumed that: ●
●
●
Each country’s wages and prices were rigidly fixed in domestic currency, so that changes in nominal exchange rates would be fully reflected in real exchange rates. There was no capital mobility, so that countries adopting fixed exchanges rates could still pursue independent monetary policies. Mundell himself assumed explicitly that two countries adopting a fixed exchange rate would not merge their central banks; he was dealing in effect with a simple currency union (what Max Corden once called a pseudo monetary union2) rather than a full-fledged monetary union with a supranational central bank like the ECB. Countries confront two types of long-lasting shocks: expenditurechanging shocks (that is, exogenous changes in consumption or investment) and expenditure-switching shocks (that is, exogenous shifts of demand between home and foreign goods).
This framework led directly to a straightforward pairing of exogenous shocks and policy responses. Consider a world comprising two countries, East and West. Let each one begin in internal balance, which we can define as the highest output level consistent with long-run price stability. Let both countries also begin in external balance, which is necessarily defined in this context as balanced bilateral trade, because there can be no other crossborder flows in the absence of capital mobility. Now let East experience an expenditure-raising shock – a long-lasting increase of aggregate
Assessing the Euro: Expectations and Achievements 33
demand. As it raises Eastern output and imports, and thereby raises Western output, it drives both countries away from internal and external balance. But a tightening of Eastern monetary policy can offset this shock, restoring internal balance in each country and external balance between them. Hence, national monetary policies can be assigned to combat expenditure-changing shocks – Eastern policy to Eastern shocks and Western policy to Western shocks. Consider, however, an expenditure-switching shock – a shift of expenditure from Western to Eastern goods. It raises Eastern output, reduces Western output and causes East to run a current account surplus. National monetary policies are not helpful here, but a real appreciation of the Eastern currency can offset the shock by raising the relative prices of Eastern output and switching demand back to Western goods. A real appreciation can be achieved by a nominal appreciation when, as here, goods prices are rigidly fixed in domestic currency. Hence, the nominal exchange rate can be assigned to combat expenditure-switching shocks. Suppose now that East and West form a currency union by undertaking to fix irrevocably the nominal exchange rate connecting their currencies. Without capital mobility, the countries can still use their monetary policies to offset expenditure-changing shocks, but they have no way to offset expenditure-switching shocks. Mundell pointed out, however, that a switch of expenditure to Eastern goods would create an excess demand for Eastern labour and an excess supply of Western labour. If labour were perfectly mobile between the two countries, unemployed Western workers would move to the East and thus restore internal balance in both countries. In effect, labour migration would enlarge the Eastern economy and shrink the Western economy. It would thus accommodate the switch in demand from Western to Eastern goods. Furthermore, it would eliminate the imbalance in East–West trade, restoring external balance. When workers migrate from West to East, their demand for Western goods is externalized: it becomes part of Eastern import demand rather than part of Western domestic demand, thus raising Eastern imports. Conversely, their demand for Eastern goods is internalized: it becomes part of Eastern domestic demand, rather than part of Western import demand, thus reducing Western imports. Accordingly, Mundell concluded that perfect labour mobility would obviate the need for exchange rate flexibility and that the domain of
34
Peter B. Kenen
labour mobility should, therefore, be used to define an optimum currency area – the set of countries that can safely adopt irrevocably fixed exchange rates connecting their national currencies. Mundell’s analysis had enormous influence on subsequent research, long after economists had ceased to employ the strong assumptions on which it was based. It also influenced the European debate about the advisability of monetary union. It drew attention, for example, to the equilibrating role of labour mobility and thus the need for more of it within and between EU countries, if they were to forgo exchange rate flexibility. Unfortunately, Mundell’s analysis led us to pay too much attention to expenditure-switching shocks. My own early paper, for example, argued that industrial diversification would contribute importantly to the optimality of a monetary union.3 I took this view because diversification reduces the macroeconomic effects of industry-specific shocks. These are the shocks featured in OCA theory as posing the gravest threat to the viability of a monetary union. Recently, moreover, there has been a debate about the endogeneity of the criteria defining an optimum currency area, which likewise reflects the importance attached to industry-specific shocks. On one side, Krugman and others have warned that a monetary union may become less optimal after it is formed, because it may promote inter-industry trade, reducing domestic diversification and leaving the members more vulnerable to industry-specific shocks. On the other side, Frankel and Rose have argued that a monetary union may become more optimal after it is formed, because it may promote intra-industry trade and thereby avoid any reduction in domestic diversification.4 We should have paid far more attention to another issue – how a full-fledged monetary union can cope with expenditure-changing shocks. Mundell swept this issue aside by assuming that each member of a currency union can pursue an independent monetary policy. When capital mobility is high, however, there can be only one monetary policy in a currency union. This holds whether it is the national monetary policy of a leading country in a simple currency union, or the monetary policy of a supranational institution like the ECB in a full-fledged monetary union. Elsewhere, I have analysed formally the key difference between these two regimes. It turns on the difference between their policy domains.5
Assessing the Euro: Expectations and Achievements 35
Consider a stylized representation of the previous European Monetary System (EMS), where the monetary policy of the German Bundesbank was mimicked by the central banks of the other EMS countries. Suppose (as was true) that the Bundesbank sought to maintain price stability in Germany without concern for the effects on other EMS countries. Under this leader–follower regime, an expenditure-raising shock in Germany would cause the Bundesbank to tighten its monetary policy by an amount that would necessarily depress economic activity in other EMS countries. The reason is simple. An expenditure-raising shock in Germany would raise aggregate demand in Germany by more than it would raise aggregate demand elsewhere in the EMS. Therefore, a tightening of Bundesbank policy strong enough to stabilize aggregate demand in Germany would, through its impact on its partners’ policies, actually reduce aggregate demand elsewhere in the EMS. Conversely, an expenditure-raising shock in another EMS country would cause the Bundesbank to tighten its monetary policy by an amount that would not be enough to stabilize aggregate demand in that other country. The reasoning involved is perfectly symmetrical. In general, a leader–follower regime is always optimal for the leader, but never optimal for the followers. For the followers, expenditurechanging shocks are asymmetric: ● ●
They do not occur uniformly in all countries. Their effects are not felt uniformly in the country where they originate and the countries to which they are transmitted.
Note, in passing, that this leader–follower regime also provides a crude characterization of unilateral dollarization in Latin America and unilateral euroization in Central Europe. Consider, instead, a stylized representation of the European monetary union, where the ECB pursues a monetary policy aimed at maintaining price stability in the Eurozone, by stabilizing aggregate demand in that whole zone. Under this supranational regime, an expenditure-raising shock in any single Eurozone country will cause the ECB to tighten its monetary policy by an amount insufficient to stabilize aggregate demand in that single country, but by too much to stabilize it elsewhere in the Eurozone. The joint effects of the shock and the policy response will thus resemble the effects of an
36
Peter B. Kenen
expenditure-switching shock in Mundell’s framework, which raised Eastern aggregate demand and reduced Western aggregate demand. To put the same point differently, the one-sized monetary policy of a supranational central bank can never fit all of its members’ needs perfectly when they experience different expenditure-changing shocks. Nevertheless, a supranational regime is manifestly superior to a leader–follower regime, when viewed from the followers’ standpoint, because the followers experience smaller changes in output, employment and prices. This is not because the supranational regime gives greater weight to the followers’ policy preferences: different policy preferences play no role in my story. It is because a supranational central bank has a union-wide policy domain, whereas the Bundesbank, or any other national leader, necessarily has a national policy domain. What are the chief implications of my story? Let me mention three: ●
●
First, the main cost of monetary union may not derive, as we once thought, from forgoing reliance on exchange rate changes to offset expenditure-switching shocks. It may, instead, derive from forgoing reliance on national monetary policies to offset expenditure-changing shocks, wherever they originate, thus partaking of a monetary policy that never be perfectly optimal for any single member. Faced with high capital mobility, no country can pursue an independent monetary policy if it fixes its exchange rate or extinguishes its exchange rate by joining a full-fledged monetary union or opting for formal dollarization. Second, it may have been wrong to do what many of us did during the run-up to monetary union, when we compared the characteristics of shocks besetting subnational regions in a single country such as the United States with the characteristics of shocks besetting European countries. The subnational regions of a single country necessarily partake of that country’s single monetary policy, and that policy presumably reduces the size of the expenditure change in the region beset by an expenditure-changing shock, but can be expected to reverse the sign of the induced expenditure change in every other region. If, then, two regions’ expenditure-changing shocks were truly uncorrelated, the
Assessing the Euro: Expectations and Achievements 37
calculated correlation obtained by conventional methods could be strongly negative. If, instead, the regions’ shocks were perfectly correlated, the calculated correlation could be very low. This last observation suggests that comparisons made in the early 1990s, which showed that European countries were further from being an optimum currency area than were US regions, may have been overly generous to the European countries, because the true correlation between US regional shocks may have been even higher than the calculated correlations.6 To obtain the true correlations for the US regions, it would have been necessary to disentangle the shocks besetting those regions from the effects of the policy responses of the Federal Reserve System. No one sought to do that, because OCA theory had paid no attention to the change in the domain of monetary policy that occurs when countries form a fullfledged monetary union. Consider a third implication of my story about the interplay between expenditure-changing shocks and the single monetary policy of a monetary union. It pertains to the need for using national fiscal policies to cope with the imperfect fit between that single monetary policy and the needs of the individual members. At one time, I believed that a monetary union should be accompanied by some sort of fiscal union. Indeed, I was the first to say so. In my 1969 paper on OCA theory, I made two points. First, the income-based federal tax system in the United States cushions the effects of expenditure-switching shocks, because it leads automatically to fiscal transfers from households and firms in prosperous regions to households and firms in depressed regions. Second, this method of cushioning the effects of shocks is clearly superior to built-in or discretionary stabilization at the regional level. Interregional transfers via the federal fiscal system are not likely to have large effects on the fiscal stance of the federal government. Therefore, they are not likely to cause significant changes in the stock of government debt. If regions or states had to conduct contracyclical fiscal policies, however, states in depressed regions would have to borrow and issue debt, and they might find it hard to do so when faced with a long-lasting shock.7 Today, moreover, the case for a fiscal union would be reinforced by concerns about the intertemporal effects of an increase in
38
Peter B. Kenen
government debt. If households believe that an increase of debt today must inevitably lead to an increase of taxes tomorrow, in order to service the increase of debt, they may choose to save more today and thus cut back their spending today. If they do that, of course, they will thereby offset some or all of the income-stabilizing impact of a state government’s deficit spending. And the strength of the argument is also reinforced by my earlier observation that expenditureswitching shocks are not the only ones leading to opposite signed changes in the incomes of individual regions or countries. When the regions or countries involved have a single monetary policy, expenditure-changing shocks also lead to opposite signed changes in incomes. A fiscal union is not needed for the effective functioning of a monetary union. It can be helpful, however, in compensating for the imperfect fit of the union’s single monetary policy. It can do that without confronting the constraints faced by the individual members’ governments when they attempt to use fiscal policy to stabilize aggregate demand. In the Eurozone, of course, the responsibility for fiscal policy will reside with the national governments for the foreseeable future. Let me, therefore, comment briefly on the fiscal policy problem that faces the Eurozone governments. I have never been impressed by the arguments made on behalf of the fiscal constraints built into the Maastricht Treaty – the 3 per cent limit on budget deficits and the 60 per cent limit on government debt – which trigger the excessive deficit procedure. With the benefit of hindsight, moreover, I have come to doubt the wisdom of the main argument made for them during the run-up to monetary union: that they are required to underpin the integrity of the euro and the operational independence of the ECB. Indeed, if many people believe them, they may spell trouble for the euro. I understand the political and economic rationale for the Stability and Growth Pact, given the fiscal constraints contained in the Treaty. National governments cannot run persistent budget deficits if they need room for deficit spending to cushion the effects of large expenditure-reducing shocks without breaching the fiscal constraints imposed by the Maastricht Treaty. But this is the wrong time for them to pursue the ‘near to balance’ goal containing in the Stability and Growth Pact, even if granted additional time in which to reach
Assessing the Euro: Expectations and Achievements 39
that goal. The governments that have already adhered to the original timetable are understandably annoyed by those that have fallen behind. They have, nevertheless, to ask themselves how their own countries’ economies would be affected if the laggards were forced to catch up under present circumstances. Continuing stagnation, or outright deflation, in one or more of the three big countries could have very large adverse effects on the smaller countries. I come now to the third subject on my list: the impact of monetary union on economic integration in the Eurozone. It was widely and rightly agreed that monetary union would foster the integration of European financial markets. By banishing exchange rate risk completely, it would enhance the substitutability of the securities issued by the governments and firms of the individual Eurozone countries. It would also relax conventional and prudential restrictions on crossborder holdings of financial assets. Investors that have to limit their holdings of foreign currency assets can now invest freely in eurodenominated assets, regardless of the country in which they originate. The results have been impressive, especially in bond markets. The broadening and deepening of those markets has led to a very large increase of bond issues, not only by Eurozone borrowers but also by foreign borrowers. The initial depreciation of the euro, however, suggests that the increase in the supply of euro-denominated debt exceeded the increase in demand resulting from the introduction of the euro. McCauley anticipated this possibility, drawing on one of the main results derived from portfolio balance models.8 In models of that sort, a change in relative supplies of financial assets denominated in different currencies is offset endogenously by a change in the exchange rate connecting the relevant currencies. An excess supply of euro-denominated debt would thus induce a depreciation of the euro, which would serve to reduce the dollar value of the stock of euro-denominated debt relative to the dollar values of all other stocks of debt. It would thereby eliminate the excess supply of eurodenominated debt. This is, indeed, the most plausible explanation of the events that followed the introduction of the euro. Expectations regarding the impact of monetary union on goods market integration were somewhat more modest. Some economists, especially Eichengreen, argued that monetary union would protect the single market from pressures for covert protection. Pressures of
40
Peter B. Kenen
that sort emerged after the EMS crisis of 1992–93, when there was a large increase of exports from countries such as Italy, whose currencies had depreciated sharply.9 Furthermore, many economists argued that the introduction of the euro would promote price transparency and reduce cross-country differences in various goods prices. But no one expected monetary union to lead to a very large increase in trade inside the Eurozone, as few economists believed that exchange rate risk is a significant barrier to trade. Three years ago, however, Andrew Rose published a paper showing that the members of a currency union trade much more with their partners than do other country pairs, and his results were ratified by several later studies.10 Like many economists, I was at first critical of Rose’s results, partly because they pertained mainly to currency unions involving very small developing countries. Rose himself warned that the Eurozone countries might not experience a comparable increase in their bilateral trade. But subsequent research suggests that monetary union has substantially increased trade within the Eurozone.11 All of these results, moreover, are thoroughly consistent with another finding: that international trade tends to be much smaller than interregional trade.12 In other words, national borders matter and the existence of national currencies may be an important reason. The most extravagant claim for monetary union was the promise (or threat) that the euro would come to rival the dollar as the main international currency and even, eventually, to displace it. To understand why this claim was extravagant, we have to look separately at the various roles of the dollar in the international monetary system. It is especially important to distinguish between those roles that can change incrementally, as the result of actions taken individually by governments or private participants in international financial markets, and those roles that can only change discretely, as a result of actions taken collectively or, at least, simultaneously. There is nothing to prevent an individual central bank from investing some of its foreign exchange reserves in euro-denominated assets. Indeed, some Asian central banks appear to have done that recently. They have not been switching massively from dollars to euros, but have apparently been buying euros with some of their newly acquired reserves. Nevertheless, the most recently available data from the International Monetary Fund show little change in the currency composition of total reserve holdings.
Assessing the Euro: Expectations and Achievements 41
In 1998, industrial countries held 67 per cent of their foreign exchange reserves in dollar assets and 17 per cent in Deutschmark and other Eurozone currencies (the so-called legacy currencies). Four years later, the dollar holdings of those countries had risen to 70 per cent of their foreign exchange reserves. Their euro holdings were not much larger than their previous holdings of the legacy currencies, amounting to only 21 per cent of their foreign exchange reserves. It can be objected that these numbers are distorted by a simple arithmetic side effect of monetary union. The introduction of the euro did not merely involve the conversion of Deutschmark holdings into euro holdings. It also involved the exclusion thereafter of those euro holdings from the currency reserves of the central banks of the euro system. The Bank of France previously treated its Deutschmark holdings as reserve assets, but cannot now count its euro holdings as reserve assets, because no central bank can hold its own currency as a reserve asset. For this and other reasons that need not detain us, it is more useful to look at the currency composition of the developing countries’ reserves, as they can count their euro holdings as reserve assets and are, indeed, freer to optimize the composition of their currency reserves than are the major industrial countries. In 1998, developing countries held 65 per cent of their foreign exchange reserves in dollar assets and 14 per cent in Deutschmark and other legacy currencies. By 2002, their dollar holdings had fallen by only four percentage points and their euro holdings exceeded their previous holdings of legacy currencies by only three percentage points. These are not very large changes, although there may be more to come, as some of the Asian countries with very large reserves begin to diversify their currency holdings. It is equally easy for individual investors to buy euro-denominated assets and for individual borrowers, including governments, to issue euro-denominated debt. I have already mentioned the large increase in the size of the market for euro-denominated bonds. Some economists, such as Portes and Rey, have suggested that this increase in bond market activity could induce a major change in the functioning of foreign exchange markets.13 By causing a very large increase of trading in euros, it could reduce the cost of trading and lead to much wider use of the euro in the foreign exchange market. It might even lead to use of the euro as a vehicle currency.
42
Peter B. Kenen
What do we mean by a vehicle currency? Holders of small countries’ currencies cannot readily move directly from one such currency to another – from, say, Mexican pesos to Korean won. Those countries’ foreign exchange markets deal primarily in one foreign currency, usually the dollar, so someone seeking to sell pesos for won does that by selling pesos for dollars and then selling dollars for won. The dollar thus serves as a vehicle currency in foreign exchange markets. But Portes and Rey have suggested that greater use of the euro, due to an increase of trading in euro-denominated assets, could lead to the use of the euro in transactions of this sort. Thus far, however, the introduction of the euro has not led to any significant fall in the cost of foreign exchange trading. In fact, the cost of dollar/euro trading rose initially, compared with the previous cost of dollar/Deutschmark trading. It is important to note, moreover, that something more must happen before the euro can become a widely used vehicle currency. New bilateral currency markets must come into being. The dollar is used as a vehicle for moving from pesos to won because Mexico and Korea trade heavily with the United States. There is thus the need for a market in which Mexican and Korean firms can buy and sell dollars against their own national currencies. It is not the demand for dollar assets that sustains those bilateral markets, but rather the volume of merchandise trade. There is, of course, much trade between Mexico and the Eurozone countries. It is cheaper, however, to exploit the vehicle role of the dollar, not merely when moving between the peso and won, but also when moving between the peso or won, on the one hand, and the euro, on the other. If many Mexican firms decided tomorrow that they would prefer to trade pesos directly for euros, they might induce a sufficient number of Mexican foreign exchange dealers to open a new bilateral market and trade pesos directly for euros at a cost competitive with the current cost of trading pesos for dollars. But no single firm, or foreign-exchange dealer, can create that market. No single dealer can do so, because a dealer needs other dealers in order to engage in the ‘wholesale’ trading required to manage exchange rate exposure. That is why any such change in the monetary system would require coordination. There have been some interesting changes in foreign exchange markets. In Hungary, for example, bilateral forint/euro trading appears to account for a larger fraction of total currency trading than did
Assessing the Euro: Expectations and Achievements 43
bilateral forint/Deutschmark trading before the advent of the euro. It also accounts for a larger share than does forint/dollar trading. In the Czech Republic, the share of the euro is not much larger than the previous share of the Deutschmark, but it is much larger than the share of the dollar. In the Russian market, however, the dollar is by far the most important currency, due in part to the dominant role of oil in Russia’s foreign trade. Like many other primary products traded on organized commodity markets, oil is priced in dollars. In Asia, including Japan, moreover, the dollar is still dominant, and that is also true in Latin America, although the market share of the euro is somewhat larger than the previous share of the Deutschmark. In short, there is little evidence thus far of a significant change in the functioning of foreign exchange markets or, for that matter, the functioning of international commodity markets. I have not used this lecture to provide a comprehensive assessment of EMU. I have tried, instead, to focus selectively on issues that figured prominently in the debates and research during the run-up to EMU. I cannot conclude, however, without a more general comment. The ECB has been heavily criticized for paying too much attention to monetary aggregates, for adopting an unduly tough definition of price stability and for being too slow to make interest rate changes. I agree with the critics. I am especially worried about the definition of price stability – an inflation rate no higher than 2 per cent. This does not allow sufficiently for the strong probability that rapidly growing Eurozone economies will experience more inflation than more mature ones. At one time, most economists believed that the adoption of a single currency and single monetary policy would greatly reduce and even eliminate persistent differences in national inflation rates. The differences have persisted, and they will get bigger after the accession countries have joined the Eurozone. Nevertheless, the ECB deserves admiration. In five short years, it has created an institutional culture. It has made the Eurozone into an operational concept and not a mere statistical construct. Those achievements have exceeded expectation.
Notes 1. See R.A. Mundell, ‘A Theory of Optimum Currency Areas’, American Economic Review, vol. 51 (1961), pp. 657–65.
44
Peter B. Kenen
2. W.M. Corden, Monetary Integration, Essays in International Finance 93, Princeton, NJ: International Finance Section, Princeton University. 3. P.B. Kenen, ‘The Theory of Optimum Currency Areas: An Eclectic View’, in Mundell and Swoboda (eds), Monetary Problems of the International Economy, Chicago: University of Chicago Press, 1969, pp. 41–60. 4. See P. Krugman, ‘Lessons of Massachusetts for EMU’, in Torres and Giavazzi (eds), Adjustment and Growth in the European Monetary Union, Cambridge: Cambridge University Press, 1993, pp. 41–261; and J.A. Frankel and A.K. Rose, ‘The Endogeneity of the Optimum Currency Area Criteria’, Economic Journal, vol. 108 (1998), pp. 1009–25. 5. See P.B. Kenen, ‘Currency Unions and Policy Domains’, in Andrews, Henning and Pauly (eds), Governing the World’s Money, Ithaca, NY: Cornell University Press, 2002, pp. 78–104. 6. For calculations of this sort, see, for example, T. Bayoumi and B. Eichengreen, ‘Shocking Aspects of European Monetary Integration’, in Torres and Giavazzi (eds), Adjustment and Growth in the European Monetary Union, Cambridge: Cambridge University Press, 1993, pp. 193–229. Note, however, that these and similar calculations are subject to another bias that runs in the opposite direction. Bayoumi and Eichengreen extract the shocks besetting each country or region by running a separate regression equation for each country or region, but they do not include on the righthand side the current or lagged values of income in other relevant countries or regions, although those variables are bound to affect the income of the country or region under study. Hence, the spillover effects of income changes in other countries or regions are treated by default as shocks originating in the country or region under study, and the cross-country correlations of the measured shocks are apt to exceed the cross-country correlations of the true shocks. This bias is likely be stronger for regions than countries, because interregional integration is probably tighter than international integration and the spillover effects are thus larger. 7. See Kenen, cited in note 3. The argument was revived in the MacDougall Report (Report of the Study Group on the Role of Public Finances in European Integration, Brussels: Commission of the European Communities, 1997) and was reintroduced in X. Sala-i-Martin and J. Sachs, ‘Fiscal Federalism in Optimum Currency Areas: Evidence for Europe from the United States’, in Canzoneri, Grilli and Masson (eds), Establishing a Central Bank: Issues in Europe and Lessons from the US, Cambridge: Cambridge University Press, 1992, pp. 195–219. Others have argued that Sali-i-Martin and Sachs failed to distinguish clearly between income-stabilizing transfers and redistributive transfers: see, for example, J. von Hagen, ‘Fiscal Arrangements in a Monetary Union: Evidence from the US’, in de Boisseau and Fair (eds), Fiscal Policy, Taxes, and the Financial System in an Increasingly Integrated Europe, Deventer: Kluwer, 1992, pp. 337–59. But von Hagen draws that distinction badly when he defines an income-stabilizing transfer as one that lasts only for a short time. Blanchard and Katz have shown that a region may take several years to recover from an income-reducing shock: see O.J. Blanchard and L. Katz, ‘Regional Evolutions’, Brookings Papers on Economic Activity 1, 1992, pp. 1–61.
Assessing the Euro: Expectations and Achievements 45
8. R.N. McCauley, ‘The Euro and the Dollar’, Essays in International Finance 205, Princeton: International Finance Section, Princeton University, 1997. 9. B. Eichengreen, Does Mercosur Need a Single Currency?, Working Paper 6821, Cambridge: National Bureau of Economic Research, 1998. 10. A.K. Rose, ‘One Money, One Market: The Effect of Common Currencies on Trade’, Economic Policy, vol. 30 (2000); for a survey of subsequent research, see A.K. Rose, ‘The Effects of Common Currencies on International Trade: A Meta-Analysis’, forthcoming in Alexander, Mélitz and von Furstenburg (eds), Monetary Unions and Hard Pegs, Oxford: Oxford University Press, 2004. 11. See A. Micco, E. Stein and G. Ordoñez, The Currency Union Effect on Trade: Early Evidence from the European Union, Discussion Paper, Washington: Inter-American Development Bank, 2003. 12. See J. McCallum, ‘National Borders Matter: Canada-US Regional Trade Patterns’, American Economic Review, vol. 85 (1995). 13. R. Portes and H. Rey, ‘The Emergence of the Euro as an International Currency’, Economic Policy, vol. 26 (1998).
3 The Return of Deflation: What Can Central Banks Do? Willem Buiter
Introduction After an absence of almost half a century, the spectre of deflation is once again haunting the corridors of central banks and finance ministries in the industrial world. The great deflations of the nineteenth century and 1930s made way for the post-Second World War era of persistent inflation – low to moderate in the advanced industrial countries, moderate to high (with occasional bursts of hyperinflation) in developing and transition countries and in emerging markets. The recent renewed concern with deflation is due in part to the historical association, at least during the interwar years, of deflationary episodes with financial crises, recession, stagnation and even depression. It is also prompted by the fear that, in deflationary conditions, nominal interest rates may come close to their lower bound of zero, at which point conventional monetary policy is thought to lose most if not all of its effectiveness. Here I define deflation to be a sustained decline in the general price level of current goods and services, that is, a persistently negative rate of inflation. In principle, the price index is the ideal cost of living index. Real-world approximations include the Consumer Price Index (CPI) in the US, the Retail Price Indexes (RPI, RPIX and RPIY) in the UK and the Harmonized Index of Consumer Prices (HICP) in the EMU area, now called the CPI in the UK. For many practical and policy issues, the distinctions between these indices are important. For the purpose of this lecture, they are irrelevant.1 What is important is that deflation, as defined here, refers to a declining general price level for current goods and services. It does not 46
The Return of Deflation 47
refer to asset price deflation – a fall in the prices of existing stores of value, either real or financial. Asset price movements are an important part of the transmission mechanism of monetary and fiscal policy. They also transmit other developments and shocks, domestic and foreign. They may even, when asset prices depart from fundamental valuations, be a source of extraneous shocks themselves. They often complicate the task of the monetary and fiscal authorities and prevent the simultaneous achievement of price stability, full employment and a balanced structure of production and demand. Asset price deflation may at times precede, be associated with, or even cause downward movement in the general price level of goods and services. Asset price deflation is, however, conceptually quite distinct from deflation in the sense I use in this lecture. The timing of the renewal of political concern with (and scholarly interest in) deflation is not surprising. As shown in Figure 3.1, in Japan the central bank discount rate has been at 50 basis points or less since 1995, raising concern about the zero lower bound on nominal interest rates, at least at the short end of the maturity spectrum. Until 2003, when GDP grew at 3.1 per cent, Japan was in a protracted economic slump, which started in 1992. Money wages declined in four of the five years from 1998; the GDP deflator declined in each of the five years and the CPI in four of them (CPI inflation in 2003 was ⫺0.3 per cent). Short-term nominal interest rates in Japan are near zero. A number of observers have concluded that Japan is in a liquidity trap, that is, monetary policy of any kind is incapable of stimulating aggregate demand. The risk of the zero lower bound becoming a binding constraint on monetary policy is (or should be) a concern also in Western Europe and in the United States. In the US, with the federal funds target rate in February 2003 at 1 per cent, the Fed continues to show some concern about the possibility that monetary policy could become constrained by the zero lower bound on nominal interest rates. During 2003 CPI inflation was 2.3 per cent and the real economy was once again expanding quite briskly. As early as the autumn of 1999, the Fed organized a conference to discuss the ‘zero bound problem’ and more recently its staff have produced a thorough study of Japan’s experience in the 1990s and the lessons this holds for preventing deflation. 2 Figures 3.1 and 3.2 provide in summary form the key data for Japan and the USA over half a century.
48
Willem Buiter
30
25
(%)
20
15
10
5
0
1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
−5
Real GDP growth (%)
CPI inflation (%)
Money wage growth (%) (1950 = 100)
20
3,500
18 3,000 16 2,500
14
(%)
12
2,000
10 1,500
8 6
1,000
4 500 2 0 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
0
Short-term interest rate
Long-term interest rate
Equity index (TOPIX, deflated by CPI, 1950 = 100, second axis)
Figure 3.1 Key economic trends in Japan: 1950 to 2002 Sources: UN Common Database and European Commission; UN Common Database and Bloomberg.
In the euro area, inflation (on the HIPC measure) averaged 2.1 per cent (year on year) in 2003. At the time of writing (February 2004), the ECB’s repo rate stands at 2.0 per cent. The euro is at its strongest since its introduction on 1 January 1999. The real economy in Euroland continues to be weak. This raises the question as to whether
The Return of Deflation 49
30
25
20
(%)
15
10
5
0
1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
−5
Real GDP growth (%)
CPI inflation (%)
Money wage growth (%)
500
18
450
16
400
14
350
12
300
10
250
8
200
6
150
4
100
2
50
0
0 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
(%)
(1950 = 100) 20
Short-term interest rate
Long-term interest rate
Equity index (S&P 500, deflated by CPI, 1950 = 100, second axis)
Figure 3.2 Key economic trends in USA: 1950 to 2002 Sources: UN Common Database and European Commission; UN Common Database, Federal Funds and Bloomberg.
a margin of 200 basis points provides enough insurance against a slump in aggregate demand. In February 2004, the UK’s repo rate was 4.0 per cent, 50 basis points above the July 2003 trough of 3.50 per cent (its lowest level since 1955), with HICP inflation for 2003 at 1.4 per cent.3 The UK therefore
50
Willem Buiter
30
25
20
(%)
15
10
5
−5
1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
0
Real GDP growth (%)
CPI inflation (%)
Money wage growth (%)
500
18
450
16
400
14
350
12
300
10
250
8
200
6
150
4
100
2
50
0
0 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
(%)
(1950 = 100) 20
Short-term interest rate
Long-term interest rate
Equity index (DAX, deflated by CPI, 1976 = 100, second axis)
Figure 3.3 Key economic trends in EU: 1950 to 2002 Sources: IMF and European Commission; UN Common Database, Bundesbank and Bloomberg.
appears to be more at risk of experiencing deflation than of hitting the zero bound for the repo rate. Figures 3.3 and 3.4 provide comparable data for the 15 members of the pre-2003 European Union and for the UK separately.
The Return of Deflation 51
30
25
20
(%)
15
10
5
−5
1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
0
Real GDP growth (%)
RPI inflation (%)
Money wage growth (%) (1984 = 100) 400
20 18
350
16 300 14 250
(%)
12 10
200
8
150
6 100 4 50
2
0 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
0
Short-term interest rate
Long-term interest rate
Equity index (FTSE, deflated by RPI, 1976 = 100, second axis)
Figure 3.4 Key economic trends in UK: 1950 to 2002 Sources: UN Common Database, Wolfbane Cybernetic and European Commission; UN Common Database, Bank of England and Bloomberg.
Are deflationary periods typically periods of stagnation and depression? The historical record suggests (and this should not come as a surprise to those who recall their first lecture on identification in econometric models) that deflations have at times been associated
52
Willem Buiter
20
15
10
5
0 1800 1806 1812 1818 1824 1830 1836 1842 1848 1854 1860 1866 1872 1878 1884 1890 1896 1902 1908 1914
−5
−10
−15
−20 RPI
RPI (annual % change)
−25
Figure 3.5 Price level and inflation: UK, 1800–1914
with booms and at other times with depression.4 The longest deflationary episode for which we have acceptable quality data is also the one that is probably most relevant for today. It is the great deflation of the nineteenth century. As Figure 3.5 shows for the UK, the average rate of inflation over this 115-year period was slightly negative (certainly if we start our count at the end of the Napoleonic Wars) and the variability of the inflation rate was high. Figure 3.6 shows that the British bank rate did not fall below 2 per cent throughout the 115 years preceding World War the Second.5 The UK got through a deflationary century without encountering the zero lower bound constraint on nominal interest rates, let alone any liquidity trap. The deflationary periods between the two world wars are less relevant to our current experience. Although the failure to deal effectively with deflation no doubt
The Return of Deflation 53
15
Inflation (left-hand scale)
11 Bank rate (left-hand scale) 10
10 5
9
0
8
–5
7
–10
£/$ exchange rate (right-hand scale)
–15 –20 1817
6 5
1827
1837
1847
1857
1867
1877
1887
1897
1907
4
Figure 3.6 Bank rate, inflation and £/$ exchange rate, 1817–1914
prolonged and deepened the Great Depression of the 1930s, deflation then was the result of a catastrophic collapse of aggregate demand, not the cause of it. A number of studies, both historical–descriptive (for example, Bordo and Redish, 2003; Bordo et al., 2004) and statistical (for example, Atkeson and Kehoe, 2004), show that past deflationary episodes are as likely to have been periods of above-normal economic activity as periods of below-normal economic activity. In their study of deflation and growth in the US and Canada during the Gold Standard period (1870–1913), Bordo and Redish (2003) conclude that there are good and bad deflations. Good deflations occur when positive supply shocks cause potential output to grow faster than nominal aggregate demand. They will be characterized by rising employment and output growth, robust profits and booming stock markets. Bad deflations occur when negative shocks to aggregate demand cause nominal demand growth to fall below the growth rate of potential output. They will be characterized by falling employment and output growth, weak profits and declining stock markets. Actual deflations during the Gold Standard period tended to be the product of both negative nominal demand shocks (modelled as monetary shocks) and positive supply shocks.6 While not part of the Bordo and Redish
54
Willem Buiter
study, the Great Depression of 1929–33 and the post-First World War recession of 1919–21 were undoubtedly bad deflations.
What’s special about deflation? Can policy-makers prevent deflation, or eliminate it once it has taken hold, simply by reversing the policies that have been proven to be effective in preventing or eliminating inflation? Or is there a discontinuity at zero? Some of the costs and benefits of deflation (and of eliminating deflation) are not qualitatively different from the costs and benefits of inflation (and of eliminating inflation). For instance, menu costs (costs of changing prices) apply symmetrically to price increases and to price cuts.7 Anticipated inflation causes welfare losses due to shoe-leather costs of cash management, if the opportunity cost of holding cash (the risk-free short nominal interest rate) increases with the expected rate of inflation. Deflation reduces the opportunity cost of holding non-interest-bearing cash. The optimal quantity of money theorem (advanced by Bailey (1956) and by Friedman (1969)) is the proposition that welfare is maximized when the opportunity cost of holding money is zero, that is, when the riskfree nominal interest rate is zero. If welfare is maximized when the expected rate of inflation equals minus the short-term real interest rate (and if the short-term real interest rate is positive), the optimal monetary rule is characterized by deflation. In a Bailey–Friedman world, deflation is not a problem; it is bliss. Unanticipated inflation redistributes wealth from creditors whose contracts are nominally denominated and not index-linked to debtors. Unanticipated deflation redistributes wealth from (nominal) debtors to creditors. If and to the extent that higher inflation is associated with greater uncertainty about relative prices, it increases the ‘noiseto-signal’ ratio of the price signals sent and received by households and enterprises and impairs the efficiency of the price mechanism. The same applies if the rate of deflation increases in absolute value. There are four reasons why deflation is not just inflation with the sign reversed. First, there is the problem of a zero lower bound on all nominal interest rates caused by the existence of stores of value with a risk-free zero nominal interest rate. These are coin and currency and commercial bank reserves with the central bank.8 The zero nominal interest rate on base money (also called high-powered money or the
The Return of Deflation 55
monetary liabilities of the central bank) sets a zero floor under nominal interest rates for all other stores of value, private and public. If, in order to stimulate demand, lower real interest rates are required but nominal interest rates are already at their zero lower bound, conventional monetary policy is powerless. Nominal interest rates are more apt to hit the zero floor when there is deflation. Second, redistributions from debtors to creditors associated with unexpectedly high deflation in a world with imperfectly index-linked debt contracts are more likely to lead to default, bankruptcy and other forms of financial distress than redistributions from creditors to debtors associated with unexpectedly high inflation. Default, bankruptcy and corporate restructuring are not just mechanisms for redistributing ownership and control of assets. These processes also destroy real resources. ‘Debt deflation’, the increase in the real value of nominal debt caused by a falling general price level, was considered an important source of financial distress by the great monetary economists of the nineteenth century and the first half of the twentith century (although much of this literature failed to distinguish between anticipated and unanticipated deflation). Irving Fisher (1932, 1933a) went as far as arguing that the interaction of deflation and large accumulations of private nominal debt could account for every major recession in the US. Borrowers with short-maturity nominal liabilities and illiquid and/or real or foreign currency-denominated assets are especially vulnerable to deflationary shocks. Commercial banks fit that description and the incidence of banking crises and bank defaults during the Great Depression of the 1930s and other severe recessions are consistent with a role for (unanticipated) debt deflation in the propagation of the business cycle. Homeowners with mortgages, or households with significant outstanding unsecured consumer debt, have similar vulnerabilities in their portfolios, as do highly indebted enterprises. Third, there is a widely held view that there exists an asymmetry in nominal wage and price adjustment. According to this view, the degree of downward rigidity in some nominal prices (and especially in money wages) is not matched by a similar degree of upward nominal rigidity. This means that disinflation, the process of bringing down the rate of inflation through a reduction in the growth rate of nominal demand, will be more costly in terms of output and
56
Willem Buiter
employment forgone (that is, the sacrifice ratio will be higher) when the inflation rate falls into the negative range than when it remains in the positive range. Fourth, in living memory, there has been considerable experience of inflation (and even of hyperinflation) while there has been only limited experience of deflation (and none of hyperdeflation). This fourth point will turn out to be relevant also to the interpretation of the third point. The proximate cause of deflation is the failure of nominal demand to grow at least at the rate of growth of potential output. It may, therefore, be descriptively correct that recent deflationary episodes have been the result of faster than expected productivity growth in some (significant) parts of the world – the USA and emerging Asia (China, India etc.) are often mentioned in this context. Even if, in an accounting sense, a reduction in inflation is associated mainly with an increase in real GDP growth driven by higher productivity growth, rather than with a reduction in nominal GDP growth, such a diagnosis does not absolve the monetary and fiscal policy authorities. Whatever the supply side of the economy may generate by way of a growth rate of potential output, it is always possible to use monetary and fiscal policy to generate any growth rate of nominal demand and, therefore, any rate of inflation in the medium term.9 Sustained deflation is, therefore, either a policy choice, or the result of policy failure. This leads us to three key questions. 1. Why don’t we see negative nominal interest rates? The nominal interest rate on any financial instrument, i, say, cannot be lower than the risk-free nominal interest rate on the safest and most liquid financial instrument (base money, that is, coin and currency and commercial bank balances with the central bank), iM, say, that is, i ⱖ iM. If the nominal interest rate on base money is zero, nominal interest rates in general cannot be below zero. If we wish to lower the zero floor for nominal interest rates in general, the authorities must be able to pay a negative nominal interest rate on base money. Why doesn’t this happen? Financial instruments can be divided into two categories: bearer securities and registered securities. Registered securities are financial instruments for which the identity of the owner is known to the issuer and can be verified by third parties. Bearer securities are financial
The Return of Deflation 57
instruments for which the owner is anonymous – the identity of the owner is unknown to the issuer and cannot be verified by third parties. Paying interest at a positive or a negative rate on registered securities is a simple task. Take, for instance, current or deposit accounts. The bank knows the owner of each account. Payment of interest at any rate – positive, zero or negative – is administratively straightforward. The bank just periodically credits or debits the account.10 With bearer securities, paying any non-zero interest rate is not administratively trivial. If the interest rate is positive, care must be taken that the interest due is paid only once during a given payment period. Since the identity of the owner is unknown to the issuer, the same security could be presented multiple times during any given payment period, either by the same holder or by a sequence of different holders. The way around this is to identify, label or mark the security rather than the owner. The security in question is marked in a verifiable manner by the issuer or his agent, whenever the security is presented for payment of interest due. Historically, bearer securities had coupons attached to them, which were cut off (clipped) one at a time whenever an interest payment was made.11 Other ways of identifying bearer securities as being ‘ex-interest’,12 such as stamping, or more high-tech identification methods can no doubt be thought of. If the interest rate on the bearer security is negative, the issuer faces the opposite problem of the holder not presenting himself to pay the issuer the negative interest due on the security. The solution is to find a way, first, of identifying the bearer security as being ex-interest and, second, of ensuring that securities that are not ex-interest are unattractive to potential owners. The reason for the second condition becomes clear when one considers the bearer security that is of special interest to this lecture: currency, that part of the monetary liabilities of the central bank that is generally accepted as means of payment and medium of exchange in the central bank’s jurisdiction.13 Today, currency is fiat money. It has no intrinsic value as a consumer good, a capital good or an intermediate input other than the value of the paper it is printed on. It has value today if, and only if, the public believes it will have value tomorrow – valuable fiat money is a gigantic, and socially most beneficial, confidence trick. For the issuer (the central bank) to put an expiry date on a bank note would be ineffective if the public chose
58
Willem Buiter
to ignore it. To make paying negative interest on currency possible it must be possible (a) to identify bank notes as being ex-interest and (b) to attach a sufficiently severe penalty to holding money that is not ex-interest after the date the interest is due. Fines and, in the limit, confiscation or worse would be required to enforce negative interest on currency. The idea of taxing currency is not new. It goes back at least to Gesell (1949) and the Social Credit movement in the second and third decades of the twentith century. No less an economist than Irving Fisher (1933b) viewed the idea sympathetically. It has recently been revived by Panigirtzoglou and myself (Buiter and Panigirtzoglou, 2001, 2003) and by Goodfriend (2000). Taxing currency by paying negative interest on it would be a costly administrative exercise. These costs must be set against the cost of being stuck at the zero nominal interest rate floor, or the cost of pursuing a sufficiently high inflation target to minimize the risk of the zero nominal interest rate floor becoming a binding constraint. 2. Do asymmetric, downward nominal price and wage rigidities make deflation particularly costly? Conventional economic theory has a rather easy time explaining real rigidities, but a hard time explaining any kind of nominal rigidities, let alone asymmetric nominal rigidities. Empirically, there appear to be important nominal rigidities, but mainstream economic theory does a poor job explaining why the numeraire matters. A fortiori, mainstream economic theory has little to say about asymmetries in the incidence and/or severity of nominal rigidities. For instance, menu costs do not generate asymmetries between upward and downward price adjustments, although they can account for the spike in the frequency distribution of individual price changes at zero. Neither do other state-contingent or time-contingent contracting stories.14 Nominal price rigidity, symmetric or asymmetric, has never been attributed to nominal asset prices, or to the prices of freely traded homogeneous commodities. Its domain has never been argued to encompass more than the money prices of highly processed goods and services and to money wages. With nominal price cuts becoming more frequent in the low inflation environment of the 1990s (see, for example, the divergent behaviour of prices for goods and prices for
The Return of Deflation 59
Table 3.1 Total, goods and services inflation in the UK
CPI inflation* CPI goods* CPI services* Services–goods inflation, 1990–97 Services–goods inflation, 1990–2002
UK
US
Euro area
Japan
1.9 ⫺1.1 4.6
1.7 ⫺0.1 3.1
2.2 1.4 3.3
⫺0.9 ⫺1.6 0.0
1.6
1.6
1.6
1.3
2.3
1.8
1.3
1.3
Note: * Year to August 2002. Inflation rates are calculated as the total increase in the price index over the indicated period, based on monthly data, expressed as a twelve-month growth rate. Sources: King (2002); original sources ONS (for UK) and Thompson Financial Datastream for data on US, Euro area and Japan.
services in the UK, shown in Table 3.1), those who argue for the importance of asymmetric downward nominal rigidity are focusing mainly on the labour markets. There is no coherent theory of asymmetric nominal rigidity in the labour market. Arguments based on fairness, justice and morale miss the point, since fairness, justice and morale should concern real wages and/or relative real wages over time and across reference groups, not money wages. The observation, painstakingly documented in hundreds of interviews by Bewley (1999), that both workers and managers will strongly resist money wage cuts can plausibly be attributed to the fact that the interviewees (in the 1990s) had known only positive inflation rates during their working lives. When nominal prices and wages have on average been rising for more than forty years, a nominal wage cut is also likely to be a real wage cut. Resisting a cut in money wages is a pretty good first stab at (indeed almost certainly a necessary condition for) resisting a real wage cut and, in decentralized labour markets, a relative wage cut. Nickell and Quintini (2003), using a unique UK micro-data set on nominal wage changes for the period 1975–99, found that the proportion of individuals whose nominal wages fall from one year to the next is large (reaching 20 per cent in periods of low inflation). They also found that there is evidence of some rigidity at a nominal wage change of zero. However, while this causes a statistically significant
60
Willem Buiter
distortion in the distribution of real wage changes, the magnitude of the impact is ‘very modest’.15 The policy relevance of even this ‘very modest’ estimated impact is contingent on the degree of downward nominal wage rigidity being ‘structural’, that is, invariant under changes in the long-run rate of inflation. With high and even moderate inflation becoming a thing of the past throughout the industrial world, any nominal rigidities and asymmetries in downward and upward nominal wage rigidity (due to memories acquired and mental reference frames constructed during inflationary episodes) will become less important as time passes. Finally, the spike in the empirical frequency distribution of contract wage changes at zero is, at most, evidence of nominal wage rigidity, not of asymmetric nominal wage rigidity. I conclude that, while there is convincing empirical evidence that nominal price and wage rigidities exist, there is no evidence that these nominal rigidities are asymmetric.
3. How can we disentangle the effects of monetary and fiscal policy? The distinction between monetary and fiscal policy and monetary and fiscal policy instruments is an unimportant definitional issue. It sometimes gets tangled up with important issues involving the institutional arrangements for the decentralization and delegation of the fiscal, financial and monetary management activities of the state, such as central bank independence (both operational independence and combined operational and target independence). To understand the economic fundamentals that determine how monetary and fiscal policy affect aggregate demand one should think of the central bank and the general government sector as a single, consolidated unit – the state, or the sovereign. The balance sheet, budget constraint and solvency constraint that matter are the balance sheet, budget constraint and solvency constraint of the consolidated general government and central bank. This consolidation makes sense, because the central bank is generally majority-owned or wholly owned by the Treasury (Ministry of Finance). Central bank profits go to the Treasury and the Treasury has an (implicit) obligation to ‘stand behind’ the central bank, that is, to recapitalize the central bank should it take a dangerous hit on its balance sheet.
The Return of Deflation 61
When we consider the practical, operational aspects of implementing certain policies in a specific country, it is indeed helpful to consider the particular institutional arrangements in that country. Often we have to focus on the central bank as a separate agency of the state, with a distinct legal personality, charged with the management of the legal tender liabilities of the state and frequently also with the management of the official international foreign exchange reserves. Monetary policy could be defined as ‘whatever the central bank does’, but a slightly more restrictive definition will turn out to be more useful in framing the analysis and organizing the argument. I consider four potential monetary instruments: one of them unconventional, the other three conventional.16 The unconventional monetary instrument is the nominal interest rate on base money (conventionally zero on coin and currency, but not necessarily zero on the other component of the monetary base, commercial bank reserves held with the central bank). The three conventional monetary policy instruments are: (1) the short-term risk-free nominal interest rate on non-monetary financial claims (in the UK this would be the two-week repo rate; in the US the federal funds rate, although the Fed does not peg that rate exactly); (2) the stock of base money; and (3) the nominal spot exchange rate (the relative price of foreign currency in terms of domestic currency). If there is unrestricted international capital mobility and the country is small in global capital markets, out of the short nominal interest rate, the quantity of base money and the exchange rate only one can be chosen independently by the authorities.17 In practice, countries either have a managed exchange rate (the exchange rate is the policy instrument), or they use the short nominal interest rate itself as the monetary instrument. I know of no country that uses (or used) the monetary base as the policy instrument, although in principle it is possible. I am here defining a conventional monetary policy action as any change in the quantity of base money, in the short nominal interest rate or in the exchange rate that, at given prices and activity levels, does not change the financial net worth of the state (the consolidated general government and central bank), now or in the future. Conventional monetary policy is, therefore, a subset of the state’s financial portfolio management. This includes the sale and purchase
62
Willem Buiter
of long-dated government debt instruments, financed by matching changes in shorter-maturity instruments, and changes in the currency composition of the government’s financial assets and liabilities (including sterilized and non-sterilized foreign exchange market intervention, changes in the mix of nominal and index-linked debt, public debt retirement financed through privatization of state assets, swaps, trading in contingent claims markets, etc.). Monetary policy involves a subset of such asset swaps. For our purposes, it always includes issuance or retirement (contraction) of base money, financed through the purchase or sale of government interestbearing debt (generally of a short maturity) or of foreign exchange reserves. Sterilized foreign exchange market intervention (purchases or sales of foreign currency with the impact on the monetary base neutralized by sales of purchases of non-monetary liabilities of the government) also are generally conducted by the central bank. Note, however, that in the UK most of the foreign exchange reserves are owned by the general government (the Treasury), although the Bank of England manages them as the government’s agent. Most debt management operations not involving changes in the monetary base are no longer conducted by the Bank of England. Fiscal policy includes any change in public spending or tax rules, regardless of whether it alters, at given prices and activity levels, the sequence of net financial balances of the state.
Policies to stimulate demand To guide the discussion of the conditions under which monetary and fiscal policy (conventional and unconventional) can or cannot stimulate aggregate demand, I will appeal to some well-known properties of a standard formal model of aggregate demand (see Buiter, 2003a, 2003b). Monetary (and fiscal) policy ineffectiveness concerns the inability of monetary and fiscal policy to influence nominal aggregate demand. The precise question I will ask is how monetary and fiscal policy can be used to stimulate aggregate demand, holding constant (that is, for given) current and future prices, wages, exchange rates, employment, output and initial stocks of financial and real assets. How the change in nominal aggregate demand is translated into changes in real GDP, and/or in the general price level, depends on the details of the specification of the ‘supply side’ of the
The Return of Deflation 63
economy and is not relevant in this discussion of how policy can affect demand. All that matters is whether a given policy (or combination of policies) can boost nominal aggregate demand.18 When policy effectiveness is in question, both higher real output and a higher price level are welcome. The key building blocks of the aggregate demand model are as follows: the demand for real domestic output, DH , is the sum of private consumption demand for domestic output,CH, private investment demand for domestic output, IH, government demand for domestic output,GH, and export demand, X: DH ⫽ CH ⫹ IH ⫹ GH ⫹ X
(3.1)
Export demand and the shares of total private consumption, total private investment and total government spending allocated to domestic output depend on the relative price of imports and domestic goods. Holding constant total private consumption, private investment and government spending as well as the domestic currency price of exports and the foreign currency price of imports, aggregate demand for domestic output will increase when the exchange rate is devalued if the Marshall–Lerner condition is satisfied. If the initial trade balance is zero, the Marshall–Lerner conditions are satisfied if the sum of relative price elasticities of import and export demand is greater than one in absolute value. From now on, the exchange rate is treated as given. Private investment is modelled as an increasing function of Tobin’s q, the ratio of the present discounted value of future profits over the current reproduction cost of capital. Government spending is exogenous, as is export demand at a given exchange rate and domestic and foreign prices. Total private consumption is the sum of consumption by ‘Keynesian’ consumers, who always spend their current disposable income, and ‘permanent income’ consumers who smooth consumption over the life cycle by borrowing and lending freely in efficient financial markets. A fraction of households is Keynesian. The real wage bill is W and real labour income tax payments are T. Total real consumption is C, real financial wealth is A and H is human wealth, the present discounted value of current and future labour income of households currently alive. The marginal propensity to consume out
64
Willem Buiter
of the comprehensive (financial plus human) wealth of permanent income consumers is . The consumption function can therefore be written as: C ⫽ [A ⫹ (1⫺)H] ⫹ (W ⫺ T )
(3.2)
Note that private financial wealth, A, includes stock market wealth, the present discounted value of current and future profits. Conventional monetary policy For the moment assume that iM, the nominal interest rate on base money, is zero and that i, the short nominal interest rate on nonmonetary financial claims, is above zero. Conventional monetary policy attempts to stimulate demand through a cut in the short nominal interest rate. At given current and future expected prices, a cut in the short nominal interest rate, i, is a cut in the short real interest rate, r ≡ i ⫺ , where is the rate of inflation. A cut in the short real interest rate will boost Tobin’s q because it lowers the discount factor applied to near-term future profits. It may also boost investment through other channels, such as the credit channel (including the bank-lending channel). A lower short-term nominal interest itself may affect consumption, because the marginal propensity to consume out of comprehensive wealth, , may depend on current and future nominal interest rates as well as on current and future real interest rates. A lower short-term nominal interest rate will also affect consumption, at given current and future prices, because it lowers the current short real interest rate. This works through three channels: (1) the substitution effect; (2) the income effect; and (3) the valuation effect. The substitution and income effects of a change in the current real rate of interest both work through . The substitution effect of a lower real interest rate will boost current consumption. The income effect depends on whether the household is a net debtor, in which case a lower real interest rate boosts current consumption, or a net creditor, in which case it lowers consumption. If the intertemporal substitution elasticity is 1, the substitution effect and income effect of a change in real interest rates cancel out and is a constant. Valuation effects of a lower real interest rate work through A, by lowering the discount factors applied to current and future profits, and through H, by
The Return of Deflation 65
lowering the discount factors applied to current and future after-tax wages. Cutting the nominal (and real) interest rate for just the current period is unlikely to do much for private consumption and investment demand. However, the effect is strengthened if a cut in the current interest rate leads to expectations of future cuts in nominal interest rates. None of this works, however, if the short nominal interest rate cannot be cut because it is at the (zero) floor set by the (zero) nominal interest rate on base money. In that case unconventional monetary policy is called for. I start with the least unconventional policy: open market purchases of anything and everything. Unconventional monetary policy Fundamentally, as long as there is any nominal interest rate, at any maturity, that is positive, monetary policy has not exhausted its capacity to stimulate demand. The central bank can purchase government securities, long-dated and short-dated, in principle until all government interest-bearing debt is held by the central bank. It can engage in purchases of foreign exchange, sterilized or non-sterilized. It can purchase other public or private foreign currency assets. It can purchase private domestic securities, financial or real, including bonds, stocks, real estate and options. It can also ease the eligibility requirements for securities acceptable as collateral in repo operations and it can expand the list of eligible counterparties. Clearly, there is a downside to the central bank purchasing private securities on a significant scale. There is risk of moral hazard and adverse selection on the part of the private counterparties to these transactions. To avoid favouritism the central bank would have to ‘buy the market’, that is, buy a representative, value-weighted index fund. There is risk of back-door socialization of the means of production, distribution and exchange and of central bank interference in the management of the enterprises in which they obtain a significant shareholding. None of these problems apply to the public debt, however. As long as there is any public debt held outside the central bank, monetary policy is not out of ammunition. What to do when the ‘foolproof’ way is not foolproof? Now consider the case where nominal interest rates at all maturities, from zero to infinity, have been brought down to their (zero) floor.
66
Willem Buiter
In this case there is no escape using conventional monetary policy, including devaluations, or even using generalized open market purchases of everything. Svensson (2000, 2003) proposes a ‘foolproof’ way out of a liquidity trap and deflation consisting of three key elements: (1) an explicit central bank commitment to a higher future price level; (2) a concrete action (concretely a devaluation or intentional depreciation of the currency) that demonstrates the central bank’s commitment, induces expectations of a higher future price level and jump-starts the economy; and (3) an exit strategy that specifies when and how to get back to normal. It is immediately clear that Svensson’s ‘liquidity trap’ refers to the rather benign case where only the current short-term nominal interest rate is at its lower bound.19 In that case the ‘foolproof’ way and many others would work – they would stimulate demand. My definition of ‘liquidity trap’ is rather more demanding. It refers to a situation where the current short nominal interest rate and current expectations of future short nominal interest rates over the indefinite future are zero, as well as current interest rates on longer-dated securities with every maturity, from zero to infinity. Under these conditions, the ‘foolproof’ way is not foolproof. Announcing a planned, expected or desired future price level (or announcing an inflation target) is spitting in the wind. There is no instrument, now or in the future, to drive the future price level (or inflation rate) to its announced target. When all current and future short nominal interest rates and longterm interest rates at all maturities are zero, negative nominal interest rates on base money are one way of lowering the floor for the nominal interest rates on non-monetary financial instruments. Paying a negative nominal interest rate on commercial bank balances with the central bank is administratively trivial. Paying a negative nominal interest rate on currency through a ‘carry tax’ on currency is, as argued on p. 57, feasible but costly. If that is not possible, fiscal policy has to come to the rescue. Fiscal policies to stimulate aggregate demand It is clear from the consumption function in equation (3.2) that a debt-financed or money-financed tax cut will always stimulate aggregate demand if there are Keynesian consumers, as these will simply consume all of their tax cut. If ⫽ 0, and there are only permanent
The Return of Deflation 67
income consumers, debt-financed current tax cuts will raise aggregate consumption demand if they raise the permanent income of households currently alive (by raising H in equation (3.2)). This can happen, even if households are fully rational and realize that, holding constant government spending, a tax cut today must imply that the present discounted value of future taxes has gone up by the full amount of the tax cut. A necessary and sufficient reason for aggregate demand to be stimulated by a debt-financed tax cut is that the tax cut (together with the future tax increases) redistributes resources from households with a low marginal propensity to consume (the young and, a fortiori, those who have not been born yet) to consumers with a high marginal propensity to consume (the old). Only if the existing generations are linked to all future generations through an operative intergenerational motive will there be ‘Ricardian equivalence’ or ‘debt neutrality’. In all other cases, postponing taxes while keeping the present discounted value of current and future taxes constant will boost aggregate demand. When there is debt neutrality, debt-financed tax cuts do not boost demand. Under these conditions, a temporary increase in public spending on goods and services will boost demand. It will make no difference in this case whether the public spending increase is financed with current taxes or with future taxes (by borrowing today). The negative effect on permanent income and thus on current private consumption of the tax increase is smaller, in the short run, than the positive effect on demand of the temporary increase in public spending. A permanent increase in public spending would have no effect. There is a non-Keynesian tax policy that could boost aggregate demand in the short run even when there is debt neutrality. The idea (advocated by Feldstein, 2003) is to use non-lump-sum (distortionary) taxes on consumption (such as VAT) to twist the pattern of consumption over time towards earlier consumption. By cutting VAT today and raising it in the future (ensuring that the present value of current and future taxes is not affected), the intertemporal terms of trade are turned in favour of current consumption. Because the tax measure leaves the present value of taxes unchanged, the income effect of the change in the intertemporal terms of trade is (approximately) neutralized and the substitution effect boosts consumption today.
68
Willem Buiter
Last but not least, there is the combined monetary–fiscal policy action of a tax cut (or increase in government transfer payments) financed by issuing (printing) base money. This is the humdrum version of Milton Friedman’s famous helicopter drop of money. It is different from a debt-financed tax cut, even if nominal interest rates are zero, because base money is irredeemable or inconvertible: the individual private holder views it as an asset, but to the issuer it does not constitute a liability in any meaningful sense. With zero interest bonds, the principal still has to be redeemed. With base money, the principal is irredeemable: a £10 banknote gives the holder no claim on the issuer for anything else than the £10 banknote itself. Therefore, if a tax cut is financed by issuing base money and if the public does not expect the central bank to reverse this action in the future, consumption demand can increase. The magnitude of the increase is given by the marginal propensity to consume out of comprehensive wealth, , times the change in the present value of the terminal (very long run) money stock (see Buiter, 2003c). An attractive feature of the money-financed tax cut or transfer payment is that it can bypass the banking sector completely. If sending cash by mail is deemed dangerous, households could be sent a cheque drawn on the government’s account with the central bank. The cheque could be cashed at any commercial bank, post office or benefit office in the country. In a country like Japan, which until 2003 was mired in a persistent recession, the ability of the authorities to bypass the banking system when implementing a moneyfinanced tax cut would have been especially welcome in view of the precarious state of its banking system.20 How much can the central bank do on its own? As regards open market purchases of ‘everything’, it may be the case that existing constraints on the range of financial claims the central bank can hold in its portfolio have to be relaxed. Can the central bank implement a helicopter drop of money on its own? In the UK this would mean that the Governor, Mervyn King, would send a cheque for, say, £1,000 to each man, woman and child, drawn on the Bank of England. If suspicious recipients decided to save the money, more cheques could be sent out, each one with a larger number of zeros before the decimal point. At some point the consumer would cry uncle and spend at least some of the money.
The Return of Deflation 69
For better or worse, by law, central banks are not generally permitted to act as fiscal agents capable of making transfer payments to private recipients.21 Implementing a helicopter money drop therefore requires the cooperation of the Treasury. The Treasury would cut taxes or increase transfer payments and borrow from the central bank to finance this. If direct borrowing by the government from the central bank is not permitted (which is the case in the EMU), then the government could borrow in the markets (issue new debt) and the central bank could buy a corresponding amount of government debt in the secondary market.
Conclusions First, although preventing or combating deflation poses some unique difficulties for the monetary authorities – difficulties that are not present in preventing or combating inflation – deflation can always be prevented and, if it has taken hold, can always be overcome by coordinated actions of the monetary and fiscal authorities. Second, monetary policy alone cannot always prevent or cure deflation, if we restrict ourselves to conventional monetary policy, that is, reductions in the risk-free short-term nominal interest rate, or a devaluation of the nominal exchange rate. Monetary policy alone is likely to prevent or cure deflation, if one is willing to consider the monetization (if necessary without limit) of the public debt (short, long, nominal or index-linked) and/or open market purchases of a wide range of foreign and private domestic securities. If one is willing to contemplate an even more unconventional monetary instrument – the payment of negative nominal interest rates on base money through the imposition of a carry tax on currency – the zero lower bound on nominal interest rates disappears as a constraint on monetary policy. To say that deflation can be prevented or cured using conventional monetary and fiscal policy is not to say that all the economic problems faced by the most prominent recent example of a deflation-afflicted economy – Japan – could have been solved this way. The Japanese banking sector is paralysed by a massive overhang of bad debt. Other financial intermediaries, especially insurance companies, are suffering the cumulative impact on their balance sheets of the most spectacular asset boom and collapse in modern history – the stock market and real estate boom of the 1980s and its unravelling since 1989.
70
Willem Buiter
Cleaning up the balance sheet of the Japanese banking sector, reducing the size of the banking sector and recapitalizing key viable non-bank intermediaries will be a painful and protracted process. However, this inevitably painful structural adjustment was not facilitated by the deflationary monetary policies of the previous Governor of the Bank of Japan, Mr Hayami. Deflation is a problem that can be avoided. If it has taken hold, it is a problem that can be solved. A tax cut (or transfer payment) directly aimed at households and financed by increasing the stock of base money will surely boost aggregate demand. So will a basemoney-financed increase in public spending. Inflation-reducing policies involve tax increases or public spending cuts and, therefore, tend to be politically unpopular. Anti-deflationary policies involve tax cuts, increased transfer payments or higher public spending on goods and services. They will tend to be politically popular. It is, therefore, somewhat of a mystery why a policy programme that makes economic sense and should be politically popular does not get implemented. The clue to the mystery probably lies in the fact that, with conventional monetary policy nearly exhausted, further effective antideflationary policy requires the cooperation of the central bank and the Ministry of Finance in the design and implementation of a coordinated monetary and fiscal stimulus. Some of the central banks that acquired operational independence only recently have interpreted central bank independence to require no cooperation, no coordination and at times even no communication with the fiscal authorities. An atmosphere of mutual distrust between operationally independent monetary and fiscal authorities is probably the root cause of the observed failure to address the deflation problem. Milton Friedman argued that inflation (and by implication deflation) is always and everywhere a monetary phenomenon – and he was right.22 Sargent and Wallace (1981) have told us that, in the longrun, money (and, therefore, inflation and deflation) is always and everywhere a fiscal phenomenon – and they too were right. Structuralists and political economists inform us that excessive public sector deficits are the result of unresolved social and political conflict about the size and composition of state spending and about who should pay for it. They, too, may well be right. I would like to add to this sequence of truths the proposition that persistent unwanted deflation is always and everywhere evidence of
The Return of Deflation 71
unnecessary, avoidable macroeconomic mismanagement. Governments through the ages have demonstrated their ability to create inflation, often to undesirable and at times disastrous levels. It is hard to believe that the very simple analytics and attractive politics of making inflation have been forgotten in the new millennium.
Notes 1. There is a widely held view that real-world price indices present us with systematically upward-biased estimates of true inflation. This is important for issues ranging from cost-of-living indexation to the choice of an appropriate inflation target by the monetary authority. A commission headed by Michael Boskin studied the CPI bias and presented the results of its report in December 1996. It concluded that CPI inflation in the US was likely to overestimate true inflation by about 1 to 2 per cent a year. The sources of this bias in CPI inflation identified by the Boskin Commission were: (1) substitution bias (0.2–0.4 points per year); (2) outlet bias (0.1–0.3 points); (3) quality changes (0.2–0.6 points); (4) new products (0.2–0.7 points); (5) formula bias (0.3–0.4 points) (see Boskin et al., 1996, 1998). While not every aspect of the methodology used by the Commission, or the magnitude of the bias it found, have been universally accepted, there is widespread agreement that there was a significant upward bias. Changes made since then by the Bureau of Labor Statistics have probably reduced the magnitude of the bias. 2. The proceedings of the conference were published as Journal of Money, Credit and Banking (2002). 3. UK HICP inflation rates are between 0.5 and 0.75 per cent a year below its RPIX inflation rate. 4. The empirical correlation between deflation and real economic activity is the macro-level analogue to the empirical correlation between price and quantity in models of a single competitive market. When demand shifts, the association is positive; when supply shifts the association is negative. 5. The temporary collapse in the external value of the US dollar starting in 1861 reflected the exceptional circumstances of the American Civil War and its aftermath, the ‘greenback period’. 6. A key identifying restriction in their trivariate (inflation, real output growth and nominal money stock growth) VAR is that, in a small open economy with a fixed exchange rate, non-monetary aggregate demand shocks have no long-run effect on the price level. This is only correct if there are no non-traded goods. If there are non-traded goods, the fixed exchange rate only pegs the long-run domestic currency price of traded goods. Any non-monetary shock to aggregate demand that changes the long-run relative price of non-traded and traded goods will have a longrun effect on the price level. The Balassa–Samuelson effect is one example. 7. Menu costs do not, of course, attach only to changes in the prices of goods and services included in the CPI or GDP deflator. They presumably
72
8. 9.
10.
11. 12.
13.
14.
15.
16.
17.
18.
Willem Buiter
apply also to changes in money wages and intermediate goods and services and even to changes in the prices of existing assets. There are countries where commercial bank reserves with the central bank are remunerated, sometimes with close-to-market interest rates. Governor Hayami of the Bank of Japan gets part of the way to a correct understanding of the relationship between technological change, changes in market structure, relative price changes and inflation in the following quote from a speech given at the Keizai (Economic) Club on 29 January 2002: ‘The recent price decline is attributable to various factors such as technological innovation, deregulation and an increase in lowpriced imports. But, above all, the major factor is that Japan’s economy was not able to achieve a full-scale recovery in the 1990s and that the negative output gap expanded due to lack of demand.’ Positive nominal interest rates on bank accounts have been common for decades. During the 1970s, the Swiss authorities taxed non-resident holders of Swiss bank accounts by paying a negative nominal interest rate. After allowing for bank fees, the net nominal rate of return on many current accounts with a (low) positive nominal interest rate is frequently negative. This would be problematic for a bearer perpetuity, such as British Consols. I use ‘ex-interest’ analogously with ‘ex-dividend’ for common stock. A security is ex-interest for a given payment period if the interest due on it (positive or negative) has been paid. The monetary liabilities of the central bank consist of currency in circulation and commercial bank balances with the central bank. Banks’ balances with the central bank are registered securities. The identity of the owner is known to the issuer. Any interest rate, positive or negative, can be charged on it with negligible administrative expense. (Sometimes, of course, currencies are accepted outside their central bank’s jurisdiction, as with the US dollar and the euro today.) For instance, surveys on price setting by supermarkets and other firms show a marked bunching of the frequency distribution of price changes at zero, but such observations tell us nothing about the existence of asymmetries in the degrees of downward and upward nominal price stickiness or rigidity. Nickell and Quintini calculate that, if long-run inflation were to rise from 2.5 to 5.5 per cent a year, the equilibrium unemployment rate would fall from around 6 to around 5.87 per cent. I restrict the analysis to monetary and fiscal policy in reasonably wellfunctioning market economies. Quantitative credit controls and other government-imposed forms of credit rationing are not considered. Changes in deposit reserve requirements are best viewed as fiscal measures (changes in the taxation of deposit-taking activities). A ‘small’ country in a particular market is a price-taker in that market. All countries, except the US, can for practical policy purposes be treated as small in global capital markets. In Buiter and Panigirtzoglou (2001, 2003), a simple continuous time, closed endowment economy (representing an agent version of the aggregate demand and money demand model developed here) is combined
The Return of Deflation 73
19.
20.
21.
22.
with an old Keynesian (in Buiter and Panigirtzoglou, 2001) and a new Keynesian (in Buiter and Panigirtzoglou, 2003) Phillips curve. Or at most, the current short-term nominal interest rate, current expectations of future short-term nominal interest rates over some limited period and nominal interest rates on current longer but finite maturity securities are at their zero lower bound. In late 1989 Japan experienced the beginning of the largest stock market collapse and debt deflation in modern history. From the early 1990s until 2003, the country experienced recession and, since 1998, deflation. In the USA, the Treasury could implement a money drop on its own, without the assistance of the Federal Reserve Board, if it were willing to issue more ‘Treasury notes’. So-called US notes, issued by the Department of the Treasury since the Legal Tender Act of 1862 (as gold or silver certificates) are part of the stock of US currency. Like Federal Reserve notes (authorized by the Federal Reserve Act of 1913), they are non-interestbearing irredeemable bearer notes and constitute legal tender. They were issued until 21 January 1971. Those that remain in circulation are obligations of the US government. ‘It follows from the propositions I have so far stated that inflation is always and everywhere a monetary phenomenon in the sense that it is and can be produced only by a more rapid increase in the quantity of money than in output. However, the reason for the rapid increase in the quantity of money may be very different under different circumstances. It has sometimes reflected gold discoveries, sometimes changes in banking systems, sometimes the financing of private spending, sometimes – perhaps most of the time – the financing of governmental spending’ (Friedman, 1973).
References Atkeson, Andrew and Patrick J. Kehoe (2004) Deflation and Depression: Is There an Empirical Link, NBER Working Paper No. 10268, January. Bailey, Martin J. (1956) ‘The Welfare Costs of Inflationary Finance’, Journal of Political Economy, vol. 64, no. 2, pp. 93–110. Bewley, T.F. (1999) Why Wages Don’t Fall During a Recession, Cambridge, Mass.: Harvard University Press. Bordo, Michael D. and Angela Redish (2003) Is Deflation Depressing? Evidence from the Classical Gold Standard, NBER Working Paper No. 9520, February. Bordo, Michael D., John Landon Lane and Angela Redish (2004), Good Versus Bad Deflation: Lessons from the Gold Standard Era, NBER Working Paper No. 10329, February. Boskin, Michael J., Ellen Dulberger, Rolet Gordon, Zvi Griliches and Dale Jorgenson (1996) Toward a More Accurate Measure of the Cost of Living: Final Report to the Senate Finance Committee, Washington, DC. US Government Printing Office, for the US Senate Committee on Finance, December.
74
Willem Buiter
Boskin, Michael J., Eller R. Dulberger, Rolet J. Gordon, Zvi Griliches and Dale Jorgenson (1998) ‘Consumer Prices, the Consumer Price Index and the Cost of Living’, Journal of Economic Perspectives, 12(1), Winter 1998, pp. 3–26. Buiter, Willem H. (2003a) Deflation Prevention and Cure, NBER Working Paper No. W9623, April. Buiter, Willem H. (2003b) Appendix to Deflation: Prevention and Cure, European Bank for Reconstruction and Development mimeo, 13 July, available at http://www.nber.org/~wbuiter/defap.pdf . Buiter, Willem H. (2003c) Helicopter Money: Irredeemable Fiat Money and the Liquidity Trap, NBER Working Paper No. W10163, December. Buiter, Willem H. and Nikolaos Panigirtzoglou (2001) ‘Liquidity Traps: How to Avoid Them and How to Escape Them’, in Wim F.V. Vanthoor and Joke Mooi (eds), Reflections on Economics and Econometrics: Essays in Honour of Martin M.G. Fase, Amsterdam: De Nederlandsche Bank, pp. 13–58. Buiter, Willem H. and Nikolaos Panigirtzoglou (2003) ‘Overcoming the Zero Bound on Nominal Interest Rates with Negative Interest on Currency: Gesell’s Solution’, Economic Journal, vol. 31, no. 490, pp. 723–46. Feldstein, Martin (2003) ‘The Role for Discretionary Fiscal Policy in a Low Interest Rate Environment’, in the 2002 Federal Reserve Bank of Kansas City Annual Conference volume, Rethinking Stabilization Policy, Kansas City: Kansas City Federal Reserve Bank. Fisher, Irving [1932] Booms and Depressions, New York: Adelphi Company. Fisher, Irving [1933a] ‘The Debt-Deflation Theory of Great Depressions’, Econometrica, March, pp. 337–57. Fisher, Irving [1933b] Stamp Scrip, New York: Adelphi Company. Friedman, Milton (1969) ‘The Optimum Quantity of Money’, in The Optimum Quantity of Money and Other Essays, Chicago: Aldine, pp. 1–50. Friedman, Milton (1973) Money and Economic Development, the Horowitz Lectures of 1972, New York: Praeger. Gesell, Silvio (1949) Die Natuerliche Wirtschaftsordnung, Nuinberg: Rudolf Zitzman Verlag, available in English as The Natural Economic Order, London: Peter Owen Ltd, 1958. Goodfriend, Marvin (2000) ‘Overcoming the Zero Bound on Interest Rate Policy’, Journal of Money, Credit and Banking, vol. 32, no. 4, Part 2, pp. 1007–35. Journal of Money, Credit and Banking (2002) Monetary Policy in a Low-Inflation Environment, vol. 32, no. 4, Part 2. King, Mervyn (2002) ‘The inflation target ten years on’, Bank of England Quarterly Bulletin, vol. 42, no. 4, pp. 459–74. Nickell, S. and G. Quintini (2003) ‘Nominal Wage Rigidity and the Rate of Inflation’, Economic Journal, vol. 113, no. 490, pp. 746–62. Sargent, Thomas J. and Neil Wallace (1981) ‘Some unpleasant monetarist Arithmetic’, Quarterly Review, Fall, Federal Reserve Bank of Minneapolis. Svensson, Lars E.O. (2000) ‘The Zero Bound in an Open Economy: A Foolproof Way of Escaping from a Liquidity Trap’, Monetary and Economic Studies (Special Edition), February, pp. 277–316. Svensson, Lars E.O. (2003) Escaping from a Liquidity Trap and Deflation: a Foolproof Way and Others, NBER Working Paper No. 10195, December.
4 The Problem of Inequality Paul Krugman
All the main things that are most directly on people’s minds now in the developed world involve macroeconomics. They involve the issues of deflation, or the approach of deflation, and monetary policy that seems to be reaching its limits. Liquidity traps are no longer just something that happened in the 1930s, or something that happens only in Japan. Budget deficits are part of the story I am going to tell. But I am not going to talk about them from the usual point of view, which is concerned with the crowding out of investment and related issues. Instead, I am going to talk about the extraordinary change that is taking place in income distribution in the United States and, now, elsewhere. With it we are seeing, in my view, a change in the shape of our political economy and, possibly, a change in the shape of our society. This is something that people in the United States are reluctant to talk about. They somehow feel that they are being excessively partisan in raising the topic, or that there are overtones of envy when they do so. People in the United States get quite intimidated by the mere mention of class warfare and shy off the subject. But something is happening and, if it looks like class warfare, perhaps class warfare is part of the story. Let me start with a potted history of two interlinked developments: first, income distribution and, second, the welfare state (the system of government taxes and transfer payment programmes). Once upon a time, before the Depression and the Second World War, western societies were extremely unequal: in income distribution and 75
76
Paul Krugman
in wealth distribution, with very few mitigating factors. The social safety net, when it existed at all, was very close to the ground. None of the programmes that have emerged since then existed. It was a society that was unequal in ways that are hard for an economist to grasp, but clearly social inequality went along with this economic inequality, which government did little or nothing to mitigate. Then during the Depression and, especially, during the war, throughout the whole of the Western world, we saw a drastic narrowing of inequality, combined with the emergence of welfare states of varying extensiveness. Claudia Goldin at Harvard calls what happened in the US during this period the ‘Great Compression’. Over a relatively short period there was an astonishing narrowing of income differentials, which was not reversed. For the United States, along also came social security and, much later, Medicare. In Western Europe, of course, we saw much more extensive welfare state programmes than in the US. That was the pattern for about 30 years after the Second World War. Then problems began to emerge. The welfare state seemed to be undermining incentives and this provoked ‘reform’ programmes of varying degrees, ranging from Reaganism and Thatcherism in the US and the UK to almost nothing in Germany. Since then the United States has led the way, but the UK and other countries, notably in Europe, have followed. The result has been a substantial increase in income inequality. Until quite recently, people were still trying to deny that there has been any real increase in inequality. Even now, they tend to talk about it in terms of how the top quintile is doing relative to lower quintiles; or in terms of the ratio of the income of the 90th percentile to the 10th (the 90/10 ratio); or, possibly, in terms of how well the top 5 per cent have done. Here I must become a bit technical. It is a fact that the upper tail of the income distribution follows a ‘Pareto’ distribution. This basically means that the 100th richest man is to the 10th richest man as the 10th richest man is to Bill Gates. In other words, the income distribution has a constant proportionality. We do not have a good theory of why this should be true, but a similar kind of ratio also seems to apply in samples of city sizes. That is, it is one of those things that works in practice but not in theory. But in any case, it is true. It is not sort of, kind of, true. Within limits of measurement error, it is true.
The Problem of Inequality 77
The way to see that it is true is if you look at the income shares and population shares as in Figure 4.1, which plots natural logs of population share for the top 0.01 per cent, top 1 per cent and top 10 per cent against natural logs of income share for the same groups. The Pareto distribution tells us that this is a straight line. And it really is a straight line for the top income groups, which are all I show here. These data come from a remarkable long-term study, based on income tax records, done by Thomas Piketty of the Commissariat Général du Plan in France and Emmanuel Saez of the University of California at Berkeley, and published in the Quarterly Journal of Economics. Up here where the diamonds are, there are just two data points, which is 1913 data, and the longer line passing through that is the 1929 data. So pre-war, there is no change. This upper tail in income distribution was the same in 1929 as it was in 1913. The lower line is the income distribution in 1970. So as you can see, the top 10 per cent had a somewhat smaller share of income than they did in 1929, significantly smaller, actually. The top 1 per cent, is a more drastic reduction, and so on. So this equalizing movement
–4.5 4 Income share (natural logs)
3.5 3 2.5 2 1.5 1 0.5 0 –5
–4
–3
–2
–1
–0.5
0
1
2
3
–1 Population share (natural logs)
Figure 4.1 Income and population shares for top 0.01%, 1% and 10% of US population: 1913, 1929, 1970 and 2000
78
Paul Krugman
which took place, the great compression, if you look at it in broad measures, you see something. But if you think about what the real economic elite is getting, it is really quite drastic. In the pre-war period, the top 0.01 per cent, the 100th of 1 per cent, received about 3 per cent of total income. In 1970 they received about half a per cent of total income. So there is a really sharp reduction. So that is the great equalization. Well, I didn’t tell you something about Figure 4.1 There are actually three lines at the top. They are right on top of one another, so you can’t tell them apart. The one at the bottom is 1970. The ones at the top are 1913, 1929 and 2000. For the 2000 line, by the way, I’ve used the data that exclude capital gains, so that is not the stock market boom. That is ex-capital gains income. So we see that the United States, over the past 30 years, has returned to an income distribution that is every bit as unequal as it was before the Great Depression. The same trend is clear in the UK; but you have not gone the whole way back, because income distribution in the UK in 1913 was even more unequal than it was in the US. It is difficult to convince the media or the general public that we are not here talking about the top 5 per cent. We are talking about huge gains at the very top. Let me give an example of how hard it is to get the point across. Last year I had a longish piece on inequality in the New York Times magazine. To illustrate my point about the growing wealth of the wealthy, they included what they thought was a picture of a ‘mansion’. But it was not a picture of a mansion. There are several houses just like the one that they illustrated going up in Princeton, costing about $3 million and with about 8,000 square footage. That is nothing these days. Houses of 50,000, even 100,000, square feet are going up all over the place, costing $40 million or $60 million. In terms of income distribution, we really are back to the era of J.P. Morgan and The Great Gatsby and all that went with it. We have returned, in many respects, to a pre-Great Compression income distribution. If you look for other indicators, there has been a substantial growth again in the employment of personal servants. They had almost vanished from the scene, but now they are back. But to some degree that understates what is happening. Much of what was considered personal service in 1913 has been outsourced in the modern world. So you have a cleaning service and you have restaurant caterers, rather
The Problem of Inequality 79
than a cook in your house (although many people now have the cook as well). We have not returned to a pre-Great Compression wealth distribution, at least not yet. There are, probably, two reasons for this. One is that the process takes time; the change in income distribution is there and wealth will follow. The other is that, until recently, accumulation of large personal fortunes has been inhibited by inheritance taxes. Without making any value judgement, this is an amazing transformation. The United States did not start as a society that you could describe as middle-class. We were a society with a dominant economic elite. We became a middle-class society and thought that we had reached a stable state. But we were wrong, because we have now moved right back to what we were before. Why has this been happening? The short answer is that we don’t know. Once you start to focus on what is going on at the top, the simple explanations tend to fall apart. You could argue that growing international trade has devalued unskilled workers in advanced countries, resulting in a relative income distribution effect. But that would not explain why the top 0.01 per cent should have became so much richer compared with the average for the top 10 per cent. You could suppose that skill bias and technological change have increased the value of formal education and technological skills. But, as others have pointed out, CEOs on average have approximately the same number of years of education as high school teachers, but their income has not exactly evolved in the same path. You can talk about technical skills, but CEOs are probably well behind much of the rest of the population in their ability actually to use computers and the Internet. Most people who contemplate these questions tend to get all ‘fuzzy’. That certainly goes for Piketty, Saez and myself. As Robert Solow of MIT once said, in trying to explain Britain’s poor productivity performance, it all ‘ends in a blaze of amateur sociology’. You find yourself talking about institutions and norms. I am all for discussing institutions and norms, but we do not have very good economic models of them. A case in point here, obviously, is executive compensation. It is clear that executive compensation has little to do with marginal productivity and everything to do with insider advantage. Something has happened to the norms that used to inhibit full exploitation of
80
Paul Krugman
that insider advantage. In the US in 1970 the top 100 CEOs were paid on average about 40 times the salary of the average worker. In 2000, they were paid approximately 1,000 times as much as the average worker. This really is an incredible shift. For want of better explanation for it, I end up putting a lot of weight on fuzzy sociological factors and, possibly, also on political factors. At all events, something dramatic has happened. The question that follows from this is why should we care. Certainly the phenomenon is of research interest and it is a story that we would like to understand. But is there any reason why we should really care about this kind of change? The obvious reason to care is that these changes in income distribution are big enough to represent a substantial slice of the economic pie. People are inclined to say that the rich may be getting richer, but that this cannot be enough to make a difference to what is available to the rest. That was true when you were looking at income distribution movements in the early 1970s. But, today, the changes are so large that they really are significant. Between 1970 and 2000 the share of the top 10 per cent in total national income rose from 32 per cent to 44 per cent. This means that the share of the other 90 per cent fell from 68 per cent to 56 per cent. In other words, there was a reduction of not far short of 20 per cent in the share of the economic pie left over for the bottom 90 per cent of the population. Again, you can see why, when people look at 90/10 ratios to measure inequality, they are missing much of the story. In fact, the income share of the 80 to 95 percentiles has been flat. All the gain has been concentrated in percentiles 95 and above and much of that just in the top 1 percentile. If you go on to ask the question,‘how can it be that measured median family income in the US is barely higher than it was 30 years ago?’, one answer is that the Consumer Price Index is a problematic measure. But the conclusion cannot be avoided that a very large share of the gains of the increase in output per worker over this period has gone to a relatively small number of people. That is certainly an economically significant fact. But, more importantly, it is an interesting and, I would say, alarming fact in terms of the political economy. In political theory, everybody’s favourite construct (because it is easy to use) is the ‘median voter’. More and more in the United States
The Problem of Inequality 81
it seems that you can produce a map of Congressmen’s political positions in terms of a single political issue: the progressivity of the tax system and the generosity of welfare transfer programmes. The map is drawn on a simple left/right axis. In terms of median voter theory (with everyone lined up on that axis), the further up the income distribution people are the more they would tend to prefer a low-tax, low-redistribution policy. The further down the income distribution they are, the more they tend to prefer a high-tax, highredistribution policy. So, if you saw a radical increase in income inequality and much greater concentration of income at the top of the scale, you would expect the voters to vote for politicians who proposed to soak the rich. Voters should vote for politicians who propose more progressive taxation, generally higher taxation and more generous social programmes. The median family would have more to gain from a more redistributive government and less to lose than with a more equal pre-tax income distribution. That has not happened. If you look at the political evolution in the United States and in other western countries, what you actually seem to see is the reverse correlation. As pre-tax, pre-transfer income inequality has increased, our tax systems have become less progressive and our transfers less generous. If you look across countries, you discover that the United States, which has by far the most unequal income distribution, also has by far the least generous welfare state. Figure 4.2 is a rather crude chart of federal tax rates, showing what has happened to the US tax system. Average tax rates are just estimates of those faced by groups of people in categories that are available – quintiles, top 10 per cent, top 1 per cent. Each point represents an income-average tax rate pair; where incomes are measured in thousands of 2000 dollars. The upper line in the figure shows the 1979 average federal tax rates. The lower line is the 2000 average federal tax rates, that is still pre-Bush II. The picture is one of lower tax rates in general. But the big changes are at the higher levels of incomes. Clearly, over this period, the US tax system became substantially less progressive. The only counter move was that, somehow in the process, we got a little bit more generous towards the working poor, with tax rates at the very bottom going down. The two Bush tax cuts that we have had so far clearly push much further in the same direction. Let me mention two facets of them. What really tilted the first (2001) tax cut towards the upper end was
82
Paul Krugman
40 1979 35
2000
Tax rate
30 25 20 15 10 5 0 0
200
400
600 800 Income ($1,000)
1,000
1,200
1,400
Figure 4.2 US federal tax rates
the repeal of the estate tax. What the law actually says is that the estate tax is to be phased out completely between now and 2010, after which it will spring back to its full 2000 levels. If you actually believed that things would happen as the law is written, you would expect a large number of elderly wealthy people to experience fatal accidents in December 2010! But everyone assumes that the present law is just intended to force permanent repeal at that point. The estate tax in the United States already had a very large exemption. Before 2001 there already was an exemption for each individual of $550,000. (For a married couple that doubled up to $1.1 million.) Only about 2 per cent of US estates paid any tax. In 1999 about half the total estate tax was paid by just 3,300 estates. So the effect of estate tax repeal is something that is really concentrated at the very top end of the scale. The second (2003) tax cut was originally supposed to be an elimination of the tax on dividend income. As enacted, it anyway represents a very sharp reduction in the tax rate involved. Again, rich people are much more likely to have dividend income. But this alone does not capture the full extent to which the change is tilted towards the top. This is because we already have tax-sheltered retirement accounts, with an upper limit on the amount that can be invested.
The Problem of Inequality 83
For most middle-class people, most of their dividend income already flows into tax-sheltered accounts. So the sharp reductions in tax rates on dividend income favour those people who have stock holdings above retirement account limits, particularly those people who have high marginal tax rates. According to the estimates from the Urban Brookings Joint Tax Policy Centre, about 19 per cent of the benefit of this second tax cut goes to people with incomes of more than $1 million a year, which roughly corresponds to that 0.01per cent of the income distribution. We are talking about no more than some 130,000 people in the US, who already have about 3 per cent of the total national income. This represent an extraordinary tilt towards the top. Tax policy is not leaning against the dramatic increase in inequality. It is reinforcing it. Why should the Congress have agreed to such a tax change? By way of answer, I can only again present a mixture of some serious modelling and some fluffy speculation. Roland Benabou, my colleague at Princeton, in a nice analysis in the 2000 American Economic Review, showed that the median voter is a lot higher up the income distribution than the median family, because voter participation rates are much higher at higher incomes. The median voter in the US is to be found in something like the 70th to 80th percentile of income distribution. This in itself suggests that the median voter, having a well-above-average income, might well conclude that he or she would lose out from more redistribution. This suggested to Roland that there ought to be a negative relationship between income inequality and the amount of taxation and redistribution that the median voter favours. This analysis would make some sense of what we see happening in the political sphere. But something more than that is going on. If the median voter is at, say, the 80th percentile of the income distribution, it has to be said that recent policies do not seem to be favouring him. They seem to be favouring the voter at the 99th percentile, or higher, which raises some questions about how the political process works. At this stage we start to get a little bit fluffy. We start to talk about the role of money in politics and about the role of large incomes in financing think tanks and all those other institutions that influence the political discourse. We may even start to talk about media ownership and the fact that all five major news sources in the US are parts of large corporations with other interests, and so on. You can go too far in
84
Paul Krugman
that direction, but it is clear that people of substantial wealth have a disproportionate influence in the political process. Here the story of how the repeal of the estate tax was sold in the United States is revealing. There is a decent economic argument against inheritance taxes, even if I don’t happen to agree with it. Basically, it is an incentive argument. It says that inheritance taxes penalize people who are successful and, therefore, reduce their incentive to take risks. Those who might take great risks in the hope of leaving a great fortune to their heirs may not do so if they feel that 55 per cent of it will be taxed away when they die. But that was not at all the way it was sold politically. It was sold on the basis of sympathy. You were supposed to feel sorry for people who have their estates taxed away. It was sold on the basis of stories about family farms and small family businesses broken up because of the estate tax. Somebody leaked the scripts for an ad campaign to be run against Senator Tom Daschle of South Dakota, the ranking Democrat in the Senate, which included segments in which a barber was bemoaning the fact that Daschle had opposed the repeal of the estate tax, which would mean that he would not be able to leave his business to his heirs. Remembering that $1.5 million exemption, that implies quite some barber shop! Yet the stories of family farms, in particular, being broken up as a result of estate tax are pervasive in the discussion. As it happens, in the US there is even a special higher exemption for family farms. In fact, no one has yet actually been able to find a family farm that was broken up because of the estate tax. Where do these stories come from? The answer is usually the Heritage Foundation. And where does the Heritage Foundation come from? The answer is the Owen Foundation, the Smith-Richardson Foundation, the Scaife Foundation. It is not simply that money buys influence. It also has the ability to shape the debate, even in the absence of any real evidence to support the argument. At this point, you could become quite cynical. After this latest round of tax cuts, we have reached a point where dividends and capital gains are not just taxed at rates below the top marginal rate for ordinary income, but taxed at rates sharply below several other income brackets. Essentially, a wealthy coupon clipper who draws no salary will end up paying a lower overall federal tax rate than a successful upper-middle manager.
The Problem of Inequality 85
Although we now have a tax system that is no longer strictly progressive, the US welfare state as a whole still remains a substantial redistributive transfer system. Although people prefer not to emphasize it, there are three big components of the US welfare state: Medicaid, which is explicitly means-tested and, therefore, redistributive; Medicare, which is financed out of a flat tax on earned income, but provides the same care to everybody and, therefore, is in effect redistributive; and social security, which is designed to simulate a retirement account but uses formulas that are in fact quite redistributive. So long as these three transfer systems are in place, we are not really back to the pre-Great Compression days. But it is becoming increasing plausible to argue that the policies now in place are designed to put great pressures on those programmes. The United States is running the biggest budget deficit in the history of the world. So what, you may ask – the US, being so big, always has the biggest everything. But the budget deficit is currently running at about 4 per cent of GDP and the great bulk of it is structural, not cyclical. We have no reason to expect it to go away, if and when the economy recovers. Roughly speaking, the US federal government, given the programmes in place, has (largely incompressible) expenditure of about 20 per cent of GDP. The tax laws, as now written after this last tax cut, give us revenue of 17 per cent of GDP, or a bit less, even with a full economic recovery. The required sums to maintain the present expenditure system would rise from 20 per cent to perhaps 24 per cent over the next 15 years or so, partly because of demography and partly because of the seemingly inexorable rise of medical costs. So the tax law, as now written, is radically inconsistent with the programmes that we now have in place. It could always be changed. The US is under-taxed compared with other OECD countries. For example, the overall local, state and federal tax take in the US is a full 10 percentage points lower than in Canada. So it could be easy to raise enough revenue to maintain all of those programmes. But clearly the intention of the tax-cut policy is to create facts, to make it very difficult for any future government to reverse course. In this case, something has to give. Some, at least, on the right are quite explicit that their aim is to provoke a fiscal crisis that will force sharp cuts in the basic middle-class social programmes. Certainly, as it now stands, the fiscal crisis looks locked in.
86
Paul Krugman
In 2002 I spoke at the LSE about the irresponsibility of US fiscal policy. All I can say is that it has surpassed all expectations at this point. But that may be part of the plan. Obviously, the possible future bankruptcy of the social welfare system is a troubling idea. But what is really scary here is the notion that this may be a cumulative process. Again, I am very much indebted to Roland Benabou for his thoughts, though he bears no responsibility for my enlargement on them. His argument is not only that there is, over some range, a tendency for an unequal income distribution to lead to a smaller, less generous system of redistribution in the political system, but that there is also a reverse causation. Reducing the size of the redistributionist system and, in general, reducing taxation and the size of government tends to make the income distribution more unequal. You can think of this happening through several channels. One is simply that, if you reduce the levels of taxes on high-income individuals, they are able to accumulate more assets. Of course, that is particularly true if you reduce or eliminate inheritance taxes. If wealth distribution is not back to the pattern of pre-Depression America or Edwardian Britain, it may just be a question of time. With estate tax repealed, and with the underlining flow of income extremely unequal, over time we will start to see wealth distribution also becoming extremely unequal to an even greater degree than it is now, which would in turn tend to reinforce the inequality of incomes. That would be a direct positive feedback channel. Indirectly, at least to some extent, public services are important for the ability of the people who are not born to wealth to earn high incomes. Here the most obvious public service is the public school system (or state education system in the UK) If you have a system in which the political dominance of people who are wealthy enough to send their children to private schools leads to financial starvation of the public/state school system, then you are going to find that the people who are not so wealthy will not get a good education. You would expect this to make the income distribution more unequal still. The last point regarding the change in the income distribution in the United States relates again to the explosion in executive compensation. How did it come about? At least part of the answer must be a change in the political climate. In the 1970s there was a sense
The Problem of Inequality 87
that there were certain things that you didn’t do, like paying the CEO of an unsuccessful company a salary of $35 million a year. If you did, the federal government would find some way to punish you! But, as you move to a situation in which the government is essentially composed of unsuccessful CEOs, that becomes less likely, which tends to feed on itself. Obviously, there are always limits; or at least I think there are limits. There has got to be some point at which counter forces come into play – Herbert Stein’s law that ‘things that can not go on forever, don’t’. There must be a point at which Assistant Vice-Presidents suddenly realize that this thing has gone so far that they are actually on the losing end; when people making $200,000 a year realize that we have a system that is working against the interests of everybody making less than $500,000 a year. That point may be reached in the United States when people start to understand about the alternative minimum tax. But that is another story. This would not mean, however, that we had to go back, or actually to reverse the process. It would just mean moving to another equilibrium. There is the sense that we may be in the process of transition from a relatively egalitarian equilibrium back to the quite unequal equilibrium of previous generations. There is the sense, also, that what we had in the 30 years after the war may have been the aberration. Certainly, the relatively egalitarian society that we had in 1970 may have been a kind of one-off event, caused by the Depression and the war, which we have now lost. It is a source of concern that we seem to be involved in a continuing process that is leading towards increasing oligarchy. What are the implications of all this for policy? The US is quite different from the rest of the advanced world. Our level of income inequality is completely off the scale compared to the rest of the advanced world. The smallness of our welfare state is also off the scale. So you might say that the United States is already in a quite different equilibrium from other industrial countries. But the United States has, also, been a successful economy. It has created a lot of jobs. We used to say that the US economy delivered job creation but low productivity performance. In the 1990s all of that changed. So, in many respects, the United States has been seen as a model to emulate. I wouldn’t completely reject that. It is clear that I am not an American triumphalist, but I would say that the US
88
Paul Krugman
lesson does suggest that it is possible to have too generous a welfare state and to have excessive regulations. The UK here, as in most things, appears to be located roughly in the middle of the Atlantic Ocean. It has more inequality, and more growth in inequality, than the continental European countries, but well short of the US. Some of the Thatcherite reforms bear some resemblance to US policies, although in modified form, and showed considerable pay-offs. But the policy lesson here is that you need to think carefully about the impacts of inequality. We can no longer dismiss income distribution as a minor issue. In the United States it is now of the same order of magnitude as overall economic growth in determining the standard of living of ordinary families. You need to think ahead about the impacts of your policies via income distribution (and possibly other channels as well) on future political economy. If you are planning a programme of wholesale economic reform, you would want to ask: are we risking the creation of US-style dynamic, exploding inequality and increasingly harsh dismantling of the social safety net as part of the consequence? You need to think a couple of steps ahead and ask whether the policy changes under consideration might in fact lead in a decade or so to what is happening in the US. For what it is worth, the US right does think that way. Earlier this year the Wall Street Journal wrote an editorial complaining about the fact that the working poor pay too little in taxes. It gave the hypothetical example of someone earning $12,000 a year who was a ‘lucky ducky’, who doesn’t pay any income tax. (The WSJ conveniently overlooked the fact that such a person does actually pay payroll tax.) The reason the editorial did not like this was that the failure of the lucky duckies to pay any income tax means that their blood is not ‘boiling with rage’ against the tax. So you have to have a system that wants people with low incomes to hate the government and to hate taxes so that its future agenda on taxes can be sustained. The infamous ‘lucky duckies’ debate may have been an embarrassment to the Journal. Nonetheless, you have to say that it has a programme for the future! If you do not like where the US is going, then you have to be equally foresighted in your planning. Some of what I have been talking about is intellectually fascinating for an economist. The reality is
The Problem of Inequality 89
that I am extremely upset and kind of scared about what is happening to my country. We once thought that we made a lot of social progress. We thought that we had become a much more decent, much more egalitarian, place: a society in which the indignities of extreme inequality had been greatly reduced. But it looks from the numbers that we are already back to that extremely unequal society that our grandfathers lived in. It is hard for me not to believe that the political and social feel of the country will change to reflect that as well. And that is not a country I particularly want to live in.
References Benabou, Roland (2000) ‘Unequal Societies: Income Distribution and the Social Contract’, American Economic Review, vol. 90, pp. 96–129. Piketty, Thomas and Emmanuel Saez (2003) ‘Income Inequality in the United States, 1913–1998’, Quarterly Journal of Economics, vol. 118, no. 1, pp. 1–39.
5 Education Matters Alan B. Krueger
The topic of my lecture is a particularly fitting one given all the research that has been done on education at the Centre for Economic Performance. What I thought I would do is give a fairly broadranging – you could call it discursive – lecture on various topics from the economics of education and the various lessons that I think emerge from research in the economics of education. I want to begin by emphasizing that, in the study of economics, the ‘human capital’ model has been one of the leading success stories. One thing that emerges clearly from such research – and from education research generally, not just that by economists – is that the amount of time spent studying and in school is a key determinant of learning and of the amount of skills that workers possess. Not a surprising conclusion, you might say, but there aren’t many short cuts to producing human capital. I shall be looking at some of the trends in human capital acquisition in both the US and the UK. They show similarities and, in many respects, policy reactions have also been similar on both sides of the Atlantic. I also think some of the lessons from research are also relevant to both countries. I shall end with some reflections on ways to enhance human capital. Because I shall cover a lot of ground, I want to mention some of the major themes of my lecture upfront. First, human capital is valuable. Indeed, it is even more valuable now than it was in the past in most developed countries. Second, research by myself and others into the time devoted to schooling has shown that returns to human capital are at least as great for those at the bottom of the income or the skills distribution as they are for 90
Education Matters 91
those at the top. This finding is something that surprises those who believe that the pay-off from education only exists for an elite group. If anything, the evidence points to a higher return for those at the bottom of the distribution of abilities. Third, there is no free lunch. Unfortunately, there are no magic bullets that I am aware of when it comes to producing human capital. There is no real short cut for improving the efficiency or output of the education sector. Some argue that there is a lot of waste in the sector. If we could just get rid of it, they say, we would improve efficiency. I don’t question that there is some waste, but not nearly as much as is commonly argued. Others maintain that we just have not yet harnessed the available technology to improve productivity in education; others argue that, if we had more competition in the sector, including voucher schemes, we would see a much greater payoff from investment in human capital. I have to say that I am sceptical of these arguments. This leads me to my fourth theme: I have become more and more convinced that having independent, randomized evaluations in education is really important and could lead to steady productivity growth in the education sector. Such evaluation techniques have revolutionized medicine and I think that they can revolutionize education as well. They will not do so overnight; but I think that, if one wants steady progress, that is the place to look. Let me start with a quotation from The Wealth of Nations. It expresses human capital theory in a nutshell. Adam Smith wrote in 1776: A man educated at the expense of much labour and time to any of those employments which require extraordinary dexterity and skill, may be compared to one of those expensive machines. The work which he learns to perform, it must be expected, over and above the usual wages of common labour, will replace to him the whole expense of his education, with at least the ordinary profits of an equally valuable capital. The basic idea is that, by investing in skills, workers can improve their job prospects and receive higher wages; and that the return on such investment will be at least as great as the ordinary return for making that same investment in machinery. That theory holds up remarkably well. Figure 5.1 shows that there is a strong correlation
92
Alan B. Krueger
Mean log hourly wage
3.4
JD, MD, etc. PhD
3.2 MA
3
BA
2.8 2.6 2.4 7– 8
2.2 2
7
9
10 11 9
11
Some college HS graduate 12(no degree)
13 15 17 Mean years of education
Slope = ‘Rate of return to education’ ~10% per year 19
21
Figure 5.1 Earnings rise with education, February 1990 Notes: Average earnings after adjusting for experience and race; sample of men. Source: Card (1999).
between earnings and years of schooling. Whether you plot years of schooling or degrees achieved, the relationship with future earnings is close to log-linear. Of course, there is a lot of variability across individuals. But, at the level of aggregation shown here, the fit is quite good. You come up with a rate of return to education of about 10 per cent higher earnings for each additional year of schooling. This has been known for decades, but what does it mean? Maybe those who stay in education for 18 or 20 years just have more inherent ability; maybe they have more motivation. The same motivation that drove them to get a high level of education might be what propels them to earn higher salaries when they get into the labour market. Perhaps they have better connections in the world that helped them to get into a university, which also helped them get a better job. It is always possible that we have not properly measured these omitted variables. However, the literature gives a remarkably strong affirmative answer to the question: ‘Are earnings and education causally linked?’ Early studies tried to control as best they could for an individual’s characteristics, like IQ or family background. They tended to find that some of that upward-sloping relationship is due to these omitted factors, but most of it is not. Most of the relationship between earnings and years of schooling looks like a true return to schooling.
Education Matters 93
Another set of studies tried to look for natural experiments, involving reasons why some students might get more education than others for reasons having nothing to do with their inherent ability or family connections. If you happen to live near a community college, for example, it is clearly easier to go on to college. Geographic proximity to colleges has been exploited in research on the pay-off to education. Such studies suggested, again, that there is a substantial pay-off to completing more schooling, even when the additional schooling is just a result of geographic proximity. Another line of research, largely initiated by the late Paul Taubman, looked at identical twins. He related differences in their schooling to the differences in their earnings, and found what he believed was a substantial upward bias in previous estimates: that is, he concluded that much of the measured return to schooling was in fact a result of family background or genetic characteristics because when he related differences in education and earnings between twins, he found little relationship. I have become sceptical of that conclusion, in part because of research I did with my Princeton colleague, Orley Ashenfelter. Figure 5.2 shows one of our findings after we had collected data on twins ourselves. We went on a field trip with a group of graduate students to the world’s largest twins festival – at Twinsburg, Ohio. We asked the twins about their schooling level and about their sibling’s schooling level. We did this because identical twins tend to have similar schooling levels, but people make mistakes when they report their education. Sometimes they exaggerate, or they forget some years that they don’t think are relevant. On the face of it, there is no reason to think that apparent differences in schooling that are just a result of reporting errors are going to have much impact on earnings. We tried to get a better measure of education by asking each twin separately how much schooling they had and, then, how much schooling their sibling had. When we analysed the data we found that the two sets of answers were reasonably highly correlated. The correlation between ‘what I say my education is’ and ‘what my identical twin says it is’ was about 0.9. But, at the same time, the correlation between a twin’s reported level of schooling and his or her actual level of schooling was also high, around 0.6. So, for differences in the level of schooling between twins, the reporting errors were a high proportion of the observed variability. When we made a simple adjustment for the measurement
94
Alan B. Krueger
Difference in log hourly wage
2
1
0
–1
–2 –7 –6 –5 –4 –3 –2 –1 0 1 2 3 4 Difference in years of education
5
6
7
8
Figure 5.2 Difference in log earnings versus difference in schooling; Twinsburg identical twins sample Source: Ashenfelter and Krueger (1994).
error (just taking the average of what ‘I said’ and what ‘my twin said’), we ended up with an estimate of the return to a year of schooling that was a little bit higher than the ordinary least squares approach would show. As Figure 5.2 shows, we ended up with about 11 per cent higher earnings for each additional year of education. The scatter diagram shows the difference in the log hourly wage of the twins against differences in their years of education. Clearly, there is a lot of variability; people’s earnings fluctuate from year to year; some twins happen to be lucky and land a better job than others regardless of their education, and so on. Nevertheless, there is an upward-sloping relationship that is highly statistically significant. So I conclude that, also with twins, the relationship between earnings and education is a causal one. Another strand of evidence comes from studies of compulsory schooling. Joshua Angrist at MIT and I published a paper in the Quarterly Journal of Economics (Angrist and Krueger, 1991) that looked at the impact of compulsory schooling in an indirect way, using the following natural experiment. The date that someone is born
Education Matters 95
determines how old they are when they start school. In the United States it used to be common to have a 1 January cut-off, that is, you needed to turn six by 1 January to start first grade. Otherwise, you had to wait a year. That deadline has been moved around a bit, but for the period we are looking at 1 January was the most common cut-off. So, if you happened to be born on 2 January, you had to wait a whole year before you started school, making you one of the oldest students in your class. However, the end of compulsory schooling in the US is usually your sixteenth birthday (though some states have the seventeenth birthday and some the eighteenth). Thus the compulsory schooling system requires students to stay in school for differing lengths of time, depending upon the day of the year they were born. Those born at the end of the calendar year are required to have more schooling than those born at the beginning. Few students drop out at the compulsory schooling level in the United States, certainly not any more. Even in the period we were looking at, only about 10 per cent of the students left once they reached the compulsory schooling age. (The UK is quite different in this respect, with a much higher proportion of students leaving school once they have satisfied the compulsory schooling requirements.) So this situation gave us a natural experiment, because – unless you believe in astrology – someone’s date of birth is quite likely random and unrelated to their inherited ability, their family connections, or their intrinsic motivation. We could, therefore, look at how schooling differences vary with individuals’ month of birth, and then relate that to their earnings. Figure 5.3 shows a remarkable picture. It uses data from the US Census, so it is based on a very large sample. We calculated average years of education by quarter of birth. The result is this regular, jagged pattern where the individuals born in the beginning of the calendar year (Quarter 1) tended to spend a little less time in school. They are the ones who were the oldest when they started school and so they could drop out at a lower-grade level. The students born at the end of the calendar year (Quarter 4) tended to spend a little bit more time in school. We looked at 30 birth cohorts. Figure 5.4 presents the truly remarkable regular pattern that we found. The little blips below zero for the students born at the beginning of the year and the little blips above for students born at the end of the year showed up in 29 of our 30 birth years. We are talking here about small differences in education,
96
Alan B. Krueger
Years of completed education
13.2 4
13.0 341
4 3
3
12.8 3
4 3
12.6
4 3
12.4
4 1 2
4
1
3
2
2
2
4
4
2
4 3
12 3
1 2
2
2 3 1
1
1
2 1
12.2 30
32
34 36 Year of birth
38
40
13.9 2
Years of completed education
4 3 23
13.5
1 3
3
4
2 1
3 41 2 3 4
2 4 1
4 1
4 2
1
3
4
3 1 2
2
34
4
2
1
3 1
1 2
13.0 40
42
44 46 Year of birth
48
50
Figure 5.3 Years of education, by quarter of birth
because only about 10 per cent of the students were dropping out as soon as they reached the compulsory schooling age. So the magnitude of these blips between the first quarter and the fourth quarter represents, on average, about one-tenth of one year of schooling. This is a difference so tiny that you could not possibly pick it up without an enormous dataset. Fortunately, we had a dataset from the census with over a million observations.
Education Matters 97
0.2 3
4
3
4
0.1 3
3
4
–0.1
3
2 2
1
4
4
2 4
1
2 2
3
4 23
0 1
3
4
1
3 12
1
12
1
4 2 4 3
1 –0.2 32
30
34
Schooling differential
0.2 0.1
2
3
3
34
2
4
0 –0.1
38
4
4
2 1
36
3 1
2
4 1
3
3 2
2
1
1
3
1
4
2 1
4 2
2 3
1
41
1
34
34 2
–0.2 40
42
44
46
48
0.2 0.1
4 2 3 41
0 –0.1
23 23
1
1
3 2 3 2 3 4 2 4 23 2 2 2 3 3 4 4 4 4 4 1 1 1 1 1 1 1
–0.2 50
52
54
56
58
Year of birth Figure 5.4 Season of birth and years of schooling
Next we looked at earnings, shown in Figure 5.5 as the average of the logarithm of earnings. The patterns that we saw in years of schooling are mirrored in the earnings of the same individuals. We see these little blips down for those born in the first quarter and up for those born in the fourth. The data are for men born in the 1930s and 1940s. Of course, earnings rise with work experience, producing
98
Alan B. Krueger
Log weekly earnings
5.9
5.8
5.7
5.64 30
35
40 Year of birth
45
50
Figure 5.5 Mean log weekly wage, by quarter of birth (all men born 1930–49, 1950 census)
the upward-sloping pattern. But if you abstract from that, you see that the first-quarter births are earning a little bit less than average and the fourth-quarter births are earning a little bit more. If you took the ratio of the earnings blips to the education blips – which we do using an instrumental variables estimator – you come up with a rate of return to education of about 10 per cent per additional year of schooling completed. This finding has been rather controversial. John Bound at the University of Michigan and collaborators wrote a paper arguing that, if you have weak instruments, the results will tend to replicate simpler estimates, for example, ordinary least squares (Bound et al., 1995). However, I do not think that is what is going on with these data. For one thing, the patterns I have just described are too strong to conclude that they result from weak instruments. Another possible criticism is that somebody’s month of birth does say something about them; it might be related to their potential for some unknown reason. So we went on to look at people who had more than the minimum compulsory level of schooling and found that there was no relationship between date of birth and earnings, or date of birth and education, for this group. In other words, the chance of getting a PhD, for example, is pretty much unrelated to your date of birth!
Education Matters 99
This led us to conclude that we were picking up the effect of compulsory schooling. Other studies point in the same direction. Harmon and Walker (1995) looked at the impact of raising the compulsory schooling age in the UK from 15 to 16 and found an even larger rate of return to schooling than we did. There is also a recent paper by Costas Meghir and Marten Palme looking at compulsory schooling in Sweden, where the compulsory schooling age varies between counties, which found a similar result. The conclusions of this literature are really rather remarkable. You take a group of students who don’t want to be in school and would drop out if it were not for the compulsory schooling age, and you constrain them to stay in school an extra year, yet it seems that the extra year of schooling raises their earnings by about 10 per cent. This seems to be evidence of schooling having a rather large impact on the income of the bottom of the education distribution. Moreover, if you look directly at income, it is lower-income students who are affected by the compulsory schooling law. One interpretation of these results is that there is a higher pay-off to the investment in schooling for those at the bottom of the income distribution. This might be the case because individuals from lower-income families, or students that would otherwise get a low level of education, have high discount rates. They tend to think mainly about the present and not much about the future. They leave school as soon as they can, in spite of having a high pay-off from staying in school longer, because of their high discount rates. I suspect also that individuals who leave school at a low level of education tend to find schooling rather unpleasant. If you think about this amenity aspect of schooling, this may be another reason why they have a higher discount rate and why they get a higher rate of return to their schooling. While this finding may seem surprising on one level, at another it is not. Indeed, Adam Smith more or less anticipated these results in The Wealth of Nations, arguing: The difference between the most dissimilar characters, between a philosopher and a common street porter, for example, seems to arise not so much from nature, as from habit, custom and education. To summarize, I think that there is no longer much controversy about what the micro literature shows on the returns to schooling.
100 Alan B. Krueger
A more interesting question now is what we find when we look across countries? What do I mean by this? Most of the studies that look across countries find that changes in education are not highly related to economic growth, but that the initial level of a country’s education is related to its subsequent growth. This is a conclusion reached by Robert Barro and Xavier Martin at Harvard, by Paul Romer at Stanford and by a number of other economists who study economic growth. They think of this as a ‘growth externality’ to education. This idea goes back at least to Nelson and Phelps, who argued that the initial level of schooling was relevant for the adoption and development of technology and thus for growth (Meghir and Palme, 1999; Barro and Martin, 1995; Romer, 1990; Nelson and Phelps, 1966). I am sceptical of this interpretation. I say this, in part, because of work I did with Mikael Lindahl (Krueger and Lindahl, 2001). We tried to reconcile why it is that changes in education are not related to GDP growth across periods of time. Most of the literature looks at very short periods of time, say five years. When you look across countries in this way, their education level does not change very much, because average years of education do not change very much over such a short period. Here, as with the twins, I think we are just dealing with reporting errors. There are mistakes in the way education is calculated for different countries and, especially when you look at changes over short periods, these mistakes are going to dominate the variability in the data. In my view, most of the variability we see will be the result of mistakes in the data, because we do not have good data on average education. However, if you look over longer periods, you do see an upwardsloping relationship. Figure 5.6 shows data for a 20-year period. The countries that have a larger increase in schooling do tend to have faster per capita GDP growth, even though there is a lot of variability around the line. I am not saying that education is the only thing that matters for economic growth, because clearly that is not the case. You can find many examples of countries where education levels increased and GDP was relatively stagnant. But, looking across countries as a whole, there does seem to be an upward-sloping relationship. In fact, the slope looks pretty similar to what one finds from the earlier cross-sectional micro-evidence on earnings and education. There looks to be a pay-off of about 10 per cent in terms of additional GDP for an increase in schooling of about one year. This suggests to
Growth of real GDP per capita, 1965–85
Education Matters 101
1.5
1
r = 0.32
0.5
0
–0.5 0
1 2 3 Change in average years of schooling, 1965–85
4
Figure 5.6 Increases in education lead to GDP growth over long periods, 97 countries Source: Krueger and Lindahl (2001).
me is that there may be externalities from education, but that they are probably not enormous. If you were looking for externalities from education, you would expect that the social return would be greater than the private return. However, as a first approximation, I think that the social return is probably about the same order of magnitude as the private return, except for those at the bottom of the skill distribution, where education is associated with reduced crime and less welfare dependence. But, by and large, I tend to focus on the private return, because I think that the externalities are not all that large. What about the growth externality, the effect of the initial level of education on GDP? If you are interested in the econometric details of the argument, I refer you to the Krueger and Lindahl article in the Journal of Economic Literature. However, the effect of the initial level of education on growth does not seem to be very robust. Instead, it seems to be a result of the quirky way in which the data are aggregated in cross-country regressions. For the average country I would not expect a very strong relationship between the initial level of education and subsequent growth.
102 Alan B. Krueger
The other argument about education is that, even if it is not generating higher income because of an improvement in human capital, it may be acting as a signal or credential. However, the evidence from compulsory schooling suggests that education is not just a signal of someone’s inherent ability. If someone’s schooling goes up because it is mandated, that does not convey much information about that person’s inherent ability. The fact that even additional compulsory schooling generates a fairly high return weighs against the signalling model. On the other hand, you might say that it is hard for employers to figure out if someone got extra schooling just because it was compulsory. But here the evidence from the UK is a bit stronger, because there was such a big upward shift in the education distribution when the school leaving age was raised. Likewise, the finding that gains in schooling at the national level are associated with GDP growth also militates against the view that schooling levels are just a signal of ability. I do not want to make too much of the international evidence, however, because this relationship is not so strong and might be a result of reverse causality. Maybe what is going on here is that, when countries get richer, they invest more in education. Education could be just an aspect of consumption. What I think is missing in the international evidence is the reliance on these exogenous changes that labour economists have exploited in the cross-individual evidence. With this cross-individual evidence, I want to emphasize that the pay-offs to education have increased over time. Richard Freeman wrote an important book in 1976 on the ‘overeducated American’, at the end of which he boldly predicted a rebound in the return to education, though I suspect that even he has been surprised by the strength of that rebound. Figure 5.7, showing the ratio of wages of college graduates to high school graduates, is a simple way of looking at the pay-off to education. You see the same kind of pattern more generally comparing earnings differences associated with other levels of education. My impression is that a similar pattern also holds for the UK: a very rapid increase in the pay-off to education in the 1980s, followed by more of a flattening in the upward trend. But what is clear is that skills are more valuable now than they were before. If, as I have suggested, we are looking at an increase in the mainly private return to education, why should the government care about that? One answer is that human capital is important for the overall level of GDP, even if the externalities may not be so great. More
Education Matters 103
1.9 1.8
Ratio
1.7 1.6 1.5 1.4 1.3 1950
1960
1970
1980 Year
1990
2000
Figure 5.7 Ratio of wages of college⫹ to high school graduates Note: Figures adjust for educational distribution of college graduates, experience, region of residence, sex and race.
importantly, I think of this as something of an internality. Many individuals that drop out before they reach the compulsory schooling level, or before they complete high school, do not appreciate how high the earnings returns available to them are. In particular, if they have myopic preferences and if they are discounting at too high a rate, then they are making mistakes. I suspect that this is something like the literature on smoking, where, if the individuals make a mistake, the internalities are much bigger than the externalities. That is one reason why public policy should care. Another reason is that human capital is important if you want to understand GDP. I once tried to decompose the labour share of national income into a part that is due to just raw labour and a part that is due to human capital. I took the standard Mincer log-linear model seriously and said: let us try to predict earnings from education and experience and then calculate what would be the earnings for someone who had zero years of experience and zero years of education. I called that ‘raw labour’, but probably a better term would be ‘intercept labour’, because it is just the intercept from this regression. The rest I attributed to human capital. The result (shown in Figure 5.8) is that most of national income relates to human capital. You could do similar types of calculations
104 Alan B. Krueger
Physical capital’s share (24%)
Raw labour’s share (6%)
Human capital’s share (70%)
Figure 5.8 Most of national income is due to education and experience: United States, 1996 Source: Krueger (1999).
with, say, the minimum wage. If all workers had some component of their output due to raw labour, which was rewarded at the minimum wage, the rest would be largely due to human capital. You can, if you like, think of human capital as the dog that wagged a tail. Human capital really is an important force in the economy, certainly over long periods of time, and is thus important to any understanding of economic growth, or levels of income. Here the evidence in Figure 5.9 is interesting and disturbing. It is from a paper by Bradford DeLong, Claudia Goldin and Larry Katz at Harvard. They tried to compute average years of schooling in the US by year of birth from 1876 to 1975, stringing together data from various censuses. They looked at people when they were around age 35. They show a fairly steady expansion of educational attainment in the United States until the cohort born in the early 1950s. After that you see a distinctly flatter trend. At the outset I emphasized the importance of time on task – and I think of ‘time in school’ as largely reflecting time on task – in the development of human capital. This suggests that, since the 1950s, the expansion in human capital has been slower than it was in the past, but I don’t want to be alarmist about the trend. Often in the
Education Matters 105
14
Years of schooling at age 35
13 12 11 10 9 8 7 1876
1890 1900 1910 1920 1930 1940 1950 Year of birth
1965 1975
Figure 5.9 Years of schooling by birth cohort: US native-born residents at age 35 Source: Delong, Goldin and Katz (2003).
United States people will assert that current generations of students are not learning as much as the previous generation: that skills are atrophying. This is not the case. It is simply that there is a slowing in the rate of increase in education and in human capital. Here I want to switch gear and talk about test scores: the measurement of cognitive ability. This is one area where economists (myself included) could really make a major contribution by learning more psychometrics. We tend simply to take the output of the standardized tests that are administered by others and just run with them. In all other areas economists are insistent that we should look at revealed preference and actual behaviour. But, when it comes to test scores, we just seem to accept what comes out of the testing agencies. There are many subtle issues involved here: how tests are scaled, how they are administered, how much weight is given to various topics. Tests are rarely validated against the kinds of real outcomes that economists are interested in. At Princeton, besides the university, is
106 Alan B. Krueger
located the Educational Testing Service (ETS). The ETS administers the largest college entrance exam, known as the SAT exam, and also the Graduate Record Exam (GRE). When I gave a talk there a while ago about education and the economy, they were truly surprised to see that their tests had anything to do with economic outcomes. When they design these tests, they do not have economic outcomes in mind at all. What I think we have learned in this area is that both cognitive and non-cognitive abilities – the cognitive ones being measured by the standardized tests – are correlated with subsequent income. The cognitive test scores seem to work primarily through educational attainment. It is not so much that being smarter by itself is important. What seems to be the case is that test scores are strongly correlated with income because they are correlated with education. Once you condition on education, the relationship is much weaker between test scores and income. Nevertheless, it is the case, as Christopher Jencks and Meridith Phillips (1999) have shown, that if you look at how students’ test scores increase (say, from tenth grade to twelfth grade in their study), the increase is related to their subsequent income. That suggests to me that the test scores are a relevant measure for economists to understand and explain. Figure 5.10 shows the raw relationship between test scores and earnings, for various quartiles of the test-score distribution. You can see that there is a monotonic relationship here. Test scores predict earnings. Another reason why test scores are an unavoidable outcome measure is that, in the education area, it is vital to examine evaluation data before it is too late. Those in the policy world cannot wait another 25 or 30 years for evidence as to how school resources related to earnings and whether schools were working well in 2004. We need to work with an intermediate measure. That measure tends to be test scores. Figure 5.11 presents some new evidence on how test scores have varied over time in the US. One of the major problems in this area is that getting consistent data on test scores over time is next to impossible. We have data called the National Assessment of Educational Progress, which is a standardized test intended to be comparable over time and which began in the early 1970s. But, if you want to look back over a longer period, there is nothing comparable. So, with the help of an undergraduate at Princeton, I have tried to put together a longer series by looking at the International Adult
107
Avg. hourly wage, 1992 dollars
14 12 10 8 6 4 2 0 Maths
Paragraph comprehension
Bottom quartile 2nd quartile
3rd quartile Top quartile
Figure 5.10 Returns to cognitive skills: hourly earnings by test scores Note: Results for 28 year olds.
Literacy scale score (0–500)
Source: US Department of Labor, National Longitudinal Study of Youth data.
330 320 310 300 290 280 Quantitative 270 260 Prose 250 Document 240 230 220 1925 1930 1935 1940 Prose
1945 1950 Year of birth
1955
1960
Document
Quantitative
1965
1970
Figure 5.11 IALS scores by year of birth: United States (five-year moving average) Source: International Adult Literacy Survey (1994).
108 Alan B. Krueger
Literacy Survey (IALS). This is a cross-sectional survey that tests people at a point in time, but those tested are at different ages, that is, they were born in different cohorts. The problem of looking at those in the IALS by year of birth, of course, is that individuals are being tested at different ages. Of those tested in 1991, say, some are in their early 60s, some late 50s, some as young as 20. Maybe, as people get older, they forget things; or, maybe, they learn more. But, rather surprisingly, we did not find an independent effect of ageing, at least not for the ages that we are looking at. (You do find one if you are looking at older ages.) Even if you are understandably sceptical about this assertion, however, the impact of ageing on measured ability does not bias comparisons across countries, as long as ageing has the same effect on ability in all countries. For the US, as Figure 5.11 shows, you see a kind of gradual improvement in quantitative literacy, prose literacy and document literacy – all three measured by exams. This improvement peaks for the cohort born in the early to mid-1950s and has been flat since then, with perhaps a slight dip most recently. The results of such other standardized tests as exist (for example, the National Assessment of Educational Progress) tend to mirror this picture, which gives me some additional confidence in these findings. Figure 5.12 shows the interesting long-run pattern for the UK: little or no change until the 1930s cohort, then a rather sharp increase, followed by stagnation again from the late 1940s right up to our most recent cohorts. The UK pattern, particularly for the cohorts born after the 1950s, is similar to that in the US: a fairly stagnant trend in test scores. Other international tests point in the same direction, which again gives me some confidence in these data. One possible explanation is that compulsory schooling is part of this growth, because we are using individuals that were tested as adults. The 1930s to the late 1940s were years of big increases in compulsory schooling, but so far we do not have strong evidence of such a link. Figure 5.13 shows data for a third country, Sweden, which had higher average levels on the IALS and also enjoyed a more continuous improvement. So what can we do to improve human capital? The first point to be clear about is that improving human capital is not going to be cheap. One of the main costs is, as the Mincer model emphasizes, the opportunity cost of student time: students in school could be working and earning wages. However, as an aside, I think it is probably more
Literacy scale score (0–500)
109
330 320 310 300 290 280 270 260 250 Quantitative 240 Prose 230 Document 220 1925 1930 1935 1940
1945 1950 Year of birth
Prose
Document
1955
1960
1965
1970
Quantitative
Figure 5.12 IALS scores by year of birth: United Kingdom (five-year moving average)
Literacy scale score (0–500)
Source: International Adult Literacy Survey (1996).
330 320 310 300 290 280 270 260 250 240 230 220 1928
Document Prose Quantitative
1933
1938
1943
Prose
1948 1953 Year of birth Document
1958
1963
1968
Quantitative
Figure 5.13 IALS scores by year of birth: Sweden (five-year moving average) Source: International Adult Literacy Survey (1994).
110 Alan B. Krueger
appropriate to think of school as also providing a day care service for very young children. While their children are in school, most mothers can be out working. In fact, if you put a value on this day care service to mothers, it swamps any opportunity costs to the students, certainly until they reach the age of about age 11 or 12. (Admittedly, there is also the opportunity cost of other family members’ time, for example, giving help with homework, which I recognize as being important as well.) On the side of direct costs, teachers’ salaries are the largest component of school budgets. Therefore, having more teachers makes schooling more costly. Thus, the pupil/teacher ratio is an important component of schooling costs. Some people have argued that a low pupil/teacher ratio involves a great deal of inefficiency. In 1997 my friend Eric Hanushek at the Hoover Institute at Stanford argued that some ‘400 studies’ of student achievement demonstrate that there is no strong or consistent relationship between student performance and school resources. On this basis, perhaps, by getting rid of this waste we can improve human capital on the cheap. I am sceptical of that assertion for a couple of reasons. One is that I have my doubts about what these 400 studies actually show, if indeed there are 400 studies. Hanushek took 59 studies in all, which looked at class size, and extracted multiple estimates from some of them. In this way he came up with 277 estimates of class-size effects from 59 separate published articles, and classified these as to whether they found a positive impact of small classes, a negative impact, or an impact of unknown sign. Remarkably, 20 per cent of his results had an impact of unknown sign (see Figure 5.14). You rather begin to wonder about a study if it did not report the sign of the coefficient involved. According to Hanushek, the literature indicates it is just about as likely to get a positive effect from smaller classes as a negative one. I think that Hanushek has come to a misleading result because of the way that he extracted his estimates. As many as 24 estimates were taken from two studies by the same authors. Only one estimate was taken from a 1992 study that David Card and I published in the Journal of Political Economy. The two studies from which 24 estimates each were taken used exactly the same dataset. So you might ask: are these really 48 independent estimates? The first set of bars in Figure 5.15 is based not on Hanushek’s reworked estimates, but on each of the 17 studies from which he took only one estimate. These studies
111
Per cent of estimates
60 50 41.5% 40
38.6%
30 19.9%
20 10 0 Positive
Negative
Unknown sign
Figure 5.14 Hanushek’s class-size results for 277 estimates (per cent positive, negative and unknown sign) Notes: The 277 estimates are drawn from 59 studies. P-value is for a two-tailed test of equal number positive and negative. Source: Based on data from Hanushek (1997).
80 70
Average %
60 50 40 30 20 10 0 1 estimates Positive
2–7 estimates Negative
8+ estimates Unknown
Figure 5.15 Average percentage of estimates positive, negative or unknown sign, by number of estimates taken from study Note: Arithmetic averages of per cent positive, negative and unknown sign are taken over the studies in each category. Source: Based on data from Hanushek (1997).
112 Alan B. Krueger
overwhelmingly found positive impacts from small classes. There were other studies from which he took between two and seven estimates. They are summarized by the middle set of bars. They also disproportionately found positive effects of small classes. There were nine studies from which eight or more estimates were taken. They accounted for 44 per cent of Hanushek’s 277 total estimates. These were the ones which showed a negative impact. So the whole conclusion that class size is irrelevant boils down to these nine studies. If you look at the substance of these nine studies, you ask: how can you make policy based on them? Many of them were not even interested in class size at all, because they were looking at the effect of family income on student outcomes. They just estimated a ‘kitchen sink’ regression that controlled for expenditures per student, as well as the pupil/teacher ratio. This does not mean that the studies were bad, just that they are not relevant for this issue. This work by Hanushek has had a major impact in public policy circles both in the US and the UK. I have become sceptical about his conclusion. All studies are not created equally. Instead of tabulating studies by the number of positive or negative estimates, I would argue that the most weight should be placed on the best research. In this field, the best-designed study is the ‘Tennessee STAR Experiment’. This was an actual experiment where class size was reduced for randomly selected classes in Tennessee, beginning with a cohort of students who started school in 1986. The normal size class then was 22 students, though there was the usual range around that norm. Average class size was then reduced to 15 for four years for selected students in this one wave of students. Teachers were randomly assigned to these classes as well. About 80 schools participated. Each school had at least one small class, one normal size class, and one normal size class with a teacher aide. Again, I should underscore that random assignment was used to set up the classes, so there is no reason to be concerned that differences in student performance across classes, on average, was a result of anything other than differences in class size or the presence of a teacher aide. Figure 5.16 shows results for the wave of students who started in kindergarten, the first year of schooling in the US. At the end of Year 1, the children in the experimental small classes were scoring about 5 percentile points higher than those in the larger classes. By the end of Year 2 of school they scored about 6 points higher. Interestingly,
Education Matters 113
58
Average percentile
56 54 52 50 48 46 44 K
1st
2nd
3rd
Grade Small
Regular
Reg/Aide
Figure 5.16 Sample starting in kindergarten Source: Krueger (1998), Table 5, column 4 model.
this improvement then fell back to 5 points. This finding is in terms of percentile ranks, which suggests that that gap was fairly stable for the next couple of years when the students were in small classes. I have done a long-term study of these students, tracking them through the end of high school. Some of the gains seem to fade as the gap, which was 5 or 6 points, falls to maybe 3 percentile points. But it still persists. High school graduation was higher for the students who were in the smaller classes. Aspirations, measured by where the students applied to go to college, were also higher. Achievement on the college entrance exams, ACT or SAT, was higher. However, the effects of smaller class sizes were not uniform. We found larger effects for the students who qualified for free lunches, that is, children with family income up to 150 per cent of the poverty level. Minority students, which in Tennessee means African-American, benefited the most. Inner-city students benefited more, and boys benefited more than the girls, at least in the short run. One interpretation I reached from all this is that the smaller classes may have had a ‘socialization’ effect. In other words, the students who would have had the most difficulty in school were the ones that benefited the most. They may have been helped to ‘become students’.
114 Alan B. Krueger
If, instead of starting in a kindergarten class of 30 where discipline was a problem, these students learn what it means to be a student, the pay-off could be enormous. The reason why I speculate that we might be looking at such a socialization effect comes from the fact that the gain in achievement was across the board on their tests; it showed for reading, mathematics and all the other tests that the students were given. So something that affected the learning process more or less equally across subjects seems to have been happening. Obviously, what goes on outside of school is also important for human capital acquisition. Lack of family support for education is clearly a significant problem in many households. Gallup did a survey in July 2000 to find out how many parents read Harry Potter to their children and how many children read the books themselves. Gallup’s results were that virtually all children were reading Harry Potter. But, when I crunched the numbers, I found that children from poor families were not reading Harry Potter during the summer. During the summertime children from poor families were getting much less enrichment than children from wealthier families. Figure 5.17 is taken from a study by Doris Entwistle, Karl Alexander and Linda Olson (1997), who are sociologists at Johns Hopkins. They looked at a group of students from some 20 schools in Baltimore that had entered first grade in 1982. They tested them at the beginning and end of the school year, using what is called the California Achievement Test (CAT). They then broke the students down into those from low and those from high socioeconomic-status families. The gains in test scores are about the same during the school year as a whole for both high and low socioeconomic-status students. It seems as if the gains are the biggest in the earlier years, that is, at the younger ages. This may be a real phenomenon, or it may be a result of the way these tests are scaled. I cannot tell, but educational researchers believe that students make bigger gains at the younger ages. However, Figure 5.18 for maths scores emphasizes the point I want to make here. It shows that the performance of children from the high socioeconomic-status families actually rises in the summer months, while students from the low socioeconomic-status families fall behind. Table 5.1 does the same for both maths and reading combined. During the course of the school year as a whole the children from the low and high socioeconomic-status families are making essentially the same gains. There is a gap, because they start their first
115
Average CAT score gain
60 50 40 30 20 10 0 −10 1982
1983
1984
Low SES
1985
1986
High SES
Figure 5.17 Maths CAT gains over the school year, by family socioeconomic status (SES) Note: Sample consists of 498 Baltimore public school students who entered first grade in 1982. Source: Entwisle et al. (1997), Table 3.1.
Average CAT score gain
20 15 10 5 0 −5 −10 1982
1983 Low SES
1984
1985
1986
High SES
Figure 5.18 Maths CAT gains over the summer, by family socioeconomic status Note: Sample consists of 498 Baltimore public school students who entered first grade in 1982. Source: Entwisle et al. (1997), Table 3.1.
116 Alan B. Krueger
Table 5.1 Cumulative CAT gains by SES SES
School year
Summer
Maths
Low High
185.8 186.3
⫺8.0 24.9
Reading
Low High
193.3 190.9
0.8 46.6
Note: Sample consists of 498 Baltimore public school students who entered first grade in 1982. Source: Entwisle et al. (1997), Table 3.1.
year in school with one, but that gap is not expanding during the course of the school months. It is expanding (and expanding dramatically) over the summer months. I suspect that is related to what I call the ‘Harry Potter divide’. The children from the low socioeconomic-status families are not reading much in the summer. (Again, ‘time on task’ is coming through.) I wrote a piece about these results for the New York Times that proposed vouchers for summer schools. The idea was to give children from low-income families a voucher for summer school or other vacation enrichment activities that they could not afford. It would make a good experiment with vouchers and the feedback from the article was positive, but I don’t know of any community that has tried it yet! At this point I want to introduce some thoughts about ‘magic bullets’ in education. In 1999, John Chambers, the President and CEO of Cisco, predicted that the next big killer application for the Internet would be education. It would make email usage, he said, ‘look like a rounding error’. He may be a bit less confident of his judgement now than he was in 1999, because Cisco’s stock has crashed since then. Nevertheless, it is worth asking: how has ‘e-education’ been working out? In 2000 Peter Navarro of the University of California Graduate School of Management at Irvine wrote a paper published in the Journal of Economic Perspectives based on a survey of people that use the Internet for teaching. He found that: 92 per cent of instructors experienced technical problems; 92 per cent of instructors spent more time developing a cyber-course than a conventional one; and 67 per cent of the instructors spent more time teaching their course when using the Internet.
Education Matters 117
You might suppose that these problems were just ‘set-up costs’. But Navarro makes clear that the marginal time/effort of e-education is also higher. Unless student learning was much greater, it is not obvious that Internet-based instruction increased productivity. Although it is early in its history, I do not think that Internet-based instruction is going to lead to a major efficiency gain for traditional primary, secondary or college instruction. If there is an efficiency gain to be had by e-education, it likely will be for adults who want to go back for additional training. For them, the flexibility of being able to do their training when they are not at work is a potentially huge benefit. This could be the case, for example, for school teachers. The whole area of computer-assisted instruction is controversial, in part because so much money has been poured into it with such little evidence that it works. Indeed, there is little credible research on the effectiveness of computer-assisted instruction. The best study so far is probably one on Israel, published in the Economic Journal (2002) by Joshua Angrist of MIT and Victor Lavy of the Hebrew University in Jerusalem. I think that there are limitations to their paper, as with all research, but it finds very little positive impact of the introduction of computers into schools in Israel. I could also plug other research that I have been doing with Cecilia Rouse, a colleague at Princeton, into a programme called ‘Fast Forward’, which was developed by a group of neuroscientists at the University of California at San Francisco and Rutgers University. Fast Forward is a family of programmes of computer-based training developed based on the latest brain research. It is intended to produce rapid gains in language and reading. It is produced commercially by a company called Scientific Learning Corporation (SLC). The company claims that the programme is for anyone: from children struggling with basic language skills to adults wanting to improve reading, comprehension and organizational abilities. It has something for everyone; it has won awards; and it is growing quickly in the US. Fast Forward was initially developed for children with severe learning disabilities and auditory processing problems. The interesting theory was that children with difficulty reading probably also have difficulty hearing. So, when the teacher reads something, the student may not pick up the nuances or subtlety in the speech. Now in comes computer technology. Students put on a headset and the computer
118 Alan B. Krueger
reads to them, but it slows down and emphasizes certain sounds to facilitate reading and language skills. The initial evidence for this theory was provided by the developers of the product from brain scans. I learned about the research from a school superintendent, who wanted to sink $3 million into the programme in his district, a major urban school district in the northeast. The researchers showed him evidence from a small number of students that the part of the brain used by normal children when reading was not activated before they started this programme, but was firing in the normal way afterwards. The superintendent said he was intrigued, but he wanted to know if their actual reading scores improved as well. I suggested that, before he spent $3 million, he should let us randomly select students to go through the programme and others to serve as a control group. We could then compare their reading scores. Since the superintendent himself gets evaluated on state reading tests for his schools, he agreed and was supportive throughout our research. The company SLC also makes other claims. In addition to the evidence from brain scans, they administered some very basic reading tests on students before and after they went through the Fast Forward programme, which took on average only six weeks to complete. The average student scored 80 (well below normal) before the programme and about 95 after it. Tremendous progress! But one of my themes is the importance of independent evaluation. A commercial company doing its own evaluation does not have the right incentives. Even in this case, where I think that the scientists involved in setting up the project are very distinguished, I would worry that the profit motive might interfere. For one thing, there was no control group in their evaluation that I mentioned. Maybe the improved scores just reflected the normal maturation process. As the students got older, they became more familiar with the tests and just did better. There is another problem with the evaluation of this particular programme in that the students are tested at the end of each week. If they pass the test, they go on to the next level. Only students who passed the tests and went on to the next level were included in the company’s first evaluation. To their credit, the company also did a controlled experiment in which they randomly assigned students in the group for the treatment. But again they only included students in the treatment group if they had
Education Matters 119
completed the programme. In my view that is not rigorous science, although the results have been published in journals. In our evaluation, Cecilia Rouse and I worked with four of the schools in this school district and randomly selected 513 students, all in years 4 to 7. They all scored in the bottom 20 per cent in the statewide reading exam. We gave them the CELF exam, a test of basic language skills, and some other tests. As in SLC’s evaluation, we found that the scores were higher for participants after participation in the programme than before. But the gain was the virtually the same for the control group, which did not have access to training on the programme. The difference between the two groups was not statistically significant. Geoffrey Borman and Laura Rachuba (2001) did another study of the Fast Forward programme at about the same time as us. They looked at 415 students in eight schools in Baltimore, Maryland. They used scores on the Comprehensive Test of Basic Skills, a widely used standardized test, as an outcome measure, and found no significant difference in terms of the improvement between those in the programme and those in the control group. The promoters of the Fast Forward programme are claiming that it has an enormous impact, perhaps a full standard deviation. If you could raise students’ reading skills by a full standard deviation, the impact on their lifetime earnings would be huge and would more than justify the $1,000 a student that it costs. Even if the impact was one-tenth of a standard deviation, the economics of the programme would suggest it is a good investment. Finally, I want to talk about another magic bullet, much discussed in some circles in the United States: school vouchers. Just to be clear, I am talking about providing money for families to send their children to independent schools, or schools that charge a fee. The US Supreme Court recently decided by a five to four vote that, despite the separation between church and state in the US constitution, local districts could use their funds to give money to families to send their children to religious schools. In the United States, 90 per cent of those students who go to private school go to religious schools. It is widely assumed that this decision by the court will open the door to vouchers. President Bush described the decision as being as significant as Brown vs Board of Education, the 1954 Supreme Court ruling against racial segregation in public schooling. In fact, Brown
120 Alan B. Krueger
vs Board of Education had little or no impact for 15 years. There was no real change as a result of that decision until the late 1960s. This Supreme Court ruling may have the impact that the President expects, but I doubt if it will have much effect over the next decade. Nevertheless, vouchers are often proposed as a solution to complaints about the current school system. This is an extremely complicated topic, not only because of the church and state issue. If you go back to what Milton Friedman said about vouchers (or even further back to what Tom Paine said about them) their justification was not that they would improve cognitive abilities, but that they would create more competition and that more competition would help both publicly funded and privately funded schools. The research to date into vouchers has looked primarily at whether students who attend fee-paying schools have an advantage in terms of the skills that they are learning. The main evidence at this point comes from a series of three randomized experiments done by Paul Peterson of Harvard. His best study was for New York City, but all his results are very similar. For New York, in 2002 Professor Peterson conducted the experiment together with Mathematic Policy Research (MPR), a private research company. This major experiment involved 2,600 students who had applied for a voucher in all: 1,300 of them were randomly selected and offered a voucher that they could use to go to private school, and 1,300 served as a control group. The voucher was worth up to $1,400 a year for three years, 1997 to 2000. That figure represents about half of the cost of the average fee-paying school. These schools, remember, are mainly Catholic and other religious schools, so they are not too expensive. Many of the schools also chipped in additional money for the children to attend there. The rest of the money was contributed by the families themselves. The students were followed up for the next three years. Interestingly, the relatively small amount of money offered caused 80 per cent of the students involved to switch to private schools. (Among the control group students, only 10 per cent went to private school.) These were all ‘poor’ students, since every student who participated qualified for a free school lunch. Their average score on the Iowa Basic Skills Test is at the 25th percentile, so we are talking about the bottom of the distribution. For the sample as a whole, Peterson and MPR found no detectable difference between the treatment
Education Matters 121
group and the control group. That finding is very relevant for the debate on vouchers in the UK so I will repeat it: if you look at the full sample, there was no difference in achievement. In the US we are particularly concerned about minority populations. So the results were broken down for African-American and Hispanic students. The claim was that the programme had a positive impact on African-American students. In fact, the headlines on the news stories about the study were frequently ‘Vouchers work’. William Safire wrote a piece in the New York Times, in the middle of the 2000 presidential election, claiming that the study showed that vouchers caused a ‘stunning reversal’ of the fortunes of the AfricanAmerican students and emphasized that George Bush supported vouchers. That finding interested me since, as Richard Freeman had pointed out to me, if the average effect of the programme was zero, some other group of students must have lost out as a result of it. It turned out that the study had found a small negative impact for the Hispanic students, which cancelled out the African-American gain. This made me sceptical about the results for the African-American students. It turns out that there are some problems with the way in which the experiment classified children as African-American or nonAfrican-American. Student’s race was not directly collected. Instead, it was inferred from mother’s race, and in a non-standard way. If a mother wrote down Black/Hispanic as her race, her child was classified as neither Black nor Hispanic. If information on a mother’s race was missing, the student was not classified as Black if his or her father reported his race as Black. In addition, a large number of students who had missing baseline data were excluded from the analysis. But since this experiment used random assignment to put students in the treatment and control group, it was not necessary to control for any variables. I discovered that if you use a broader, more comprehensive sample, you find no significant difference in performance between those offered a voucher and those placed in the control group at conventional levels of significance. The difference in scores between the group that was offered a voucher and the group that was not is so small (around 1.4 percentile points) that it easily could have occurred by chance. To put this in context, the difference is only about 0.05 as large as the standard deviation of test scores across students nationwide. This is also the conclusion that emerges from much of the other literature: hardly any effect of vouchers for private school
122 Alan B. Krueger
attendance on student achievement. Certainly, when you look at the population of students as a whole in the US, there is not much reason to believe that the schools that charge a fee are performing much better than state-funded schools. This conclusion also appears to hold for minority groups. This does not mean that vouchers would be irrelevant. It may be that competition provoked by them could have a bigger impact than the meagre difference implied by the comparison of achievement of statistically identical students who attended state or fee-based schools. However, I am a bit concerned that, if we do have a system with much more choice, schools that exist to make a profit may exaggerate the quality of what they produce. They also would be likely to cater to the particular interests of their submarket, which could be based on religion, sports or ideology, for example – but not necessarily academic achievement. On the other hand, it may be the case that the publicly funded schools would react to the extra competition and innovate in ways that we cannot anticipate, thus increasing their productivity. My personal guess is that such a result is unlikely. I also think that there is little reason for families to switch to private schools, if what they are seeking is higher skills attainment for their children, because the evidence suggests that there is very little gain. If this is the case, then the competition would likely not take place over academic achievement, but instead over ancillary outputs, such as religious education. To sum up, in this lecture I have emphasized that human capital is important. Without wishing to sound shrill, I think that the current trends in human capital accumulation are a cause for some concern. There has been a period of slower growth and, in some respects, stagnation in terms of skill attainment in both the US and UK. I do not see a bargain basement solution to this problem. If there is one out there, we haven’t found it yet. In the past, the policy of expanding years of schooling and of expanding the length of the school year certainly produced results. We stopped this ‘time-intensive’ strategy in the United States in the early 1950s. If I was to give advice to policy-makers now, it would be why don’t you continue with a time-intensive strategy until we figure out what works, or what might be more efficient. In this process there will be a high payoff to randomized experiments analysed by independent evaluations. We should remain highly sceptical about evaluations conducted by the companies that stand to gain from a particular programme.
Education Matters 123
The US today is moving in this direction. In 2001 the US Congress passed a Bill entitled No Child Left Behind. This piece of legislation can be described as ‘the largest unfunded mandate in United States history’, because it mandates certain levels of educational outcome levels that I do not believe are attainable, or, at least, are not attainable on the basis of fair and accountable measurements. The Act requires that educational interventions be assessed by scientifically based research, but the Act is vague as to what that means. My suspicion is that an evaluation of the kind done on Fast Forward by its founders would classify as scientifically based research for these purposes. Even when evaluations are done by academics, I think we need to worry about the strength of the evidence in education. The New York voucher study was a real eye-opener to me. I do not like ending on a sour note, but it left me disillusioned. Reports like the one produced by Paul Peterson and MPR look like the kind of research that you would really hope to see forming the basis of public policy. It used randomized evaluations; the samples were reasonably large; attrition wasn’t bad and an attempt was made afterwards to model the nature of sample attrition and to adjust the sample weights for it. But, when you look at the actual data, you find that 40 per cent of the available observations were not used! This was never mentioned – and adding those students turns out to have a substantial impact on the results for the African-American students. So my main conclusion is that you should not trust a study until a number of eyes have seen it and others have had an opportunity to independently analyse the data. I was not so cynical in the past. But I have come to the belief that what is important for the future of education – and social science more generally – is that there should be much more of an ethic of sharing data.
References Angrist, Joshua and Alan Krueger (1991) ‘Does Compulsory Schooling Affect Schooling and Earnings?’, Quarterly Journal of Economics, vol. 106, no. 4, pp. 979–1014. Angrist, Joshua and Victor Lavy (2002) ‘New Evidence on Classroom Computers and Pupil Learning’, The Economic Journal no. 112, October, pp. 735–65. Ashenfelter, Orley and Alan Krueger (1994) ‘Estimates of the Economic Return, to Schooling from a New Sample of Tasks’, American Economic Review, 84(5), pp. 1157–73. Bound, John, David A. Jaeger and Regina M. Baker (1995) ‘Problems with Instrumental Variables Estimation when the Correlation between the Instruments and the Endogenous
124 Alan B. Krueger
Explanatory Variable is Weak’, Journal of the American Statistical Association, 90(430), pp. 443–51. Card, David (1999) ‘The Causal Effect of Education on Earnings’, in Orley Ashenfelter and David Card (eds), Handbook of Labor Economics, Vol. III, Amsterdam: Elsevier. Card, David and Alan Krueger (1992) ‘Does School Quality Matter? Returns to Education and the Characteristics of Public School in the United States’, Journal of Political Economy, 100(1), February, pp. 1–40. DeLong, Bradford, Claudia Goldin and Larry Katz (2002) ‘Sustaining U.S. Economic Growth’, in H. Aron et al., Agenda for the Nation, Washington DC: The Brookings Institution, pp. 17–60. Entwisle, Doris, Karl Alexander and Linda Olson (1997) Children, Schools, and Inequality, Boulder, Colo: Westview Press. Freeman, Richard (1976) The Overeducated American, New York: Academic Press. Goldin, Claudia and Lawrence Katz (2003) ‘The “virtues” of the past education in the first hundred years of the new republic’, National Bureau of Economic Research Working paper No. 9958. Hanushek, Eric (1997) ‘Assessing the Effects of School Resources on Student Performance: Update’, Educational Evaluation and Policy Analysis, 19(2) pp. 141–64. Harmon, Colm and Ian Walker (1995) ‘Estimates of the Economic Return to Schooling for the UK’, American Economic Review, vol. 85, no. 5, pp. 1278–86. Howell, Williams and Paul Peterson (2002) ‘The Education Gap: Vouchers and Urban Schools’, Brookings Institution Press. Jencks, Christopher and Meredith Phillips (1999) ‘Aptitide or Achievement: Why do Test Scores Predict Educational Attainment and Earnings?’ in Mayer, Susan and Paul Peterson (eds), Learning and Earning: How Schools Matter, Brookings Institution Press. Krueger, Alan (1998) ‘Experimental Estimates of Education Production Functions, ‘Quarterly Journal of Economics, 114(2), May. Krueger, Alan (1999) ‘Measuring Labor’s Shape’, American Economic Review: Papers and Proceedings, vol. 89, May. Krueger, Alan (2000) Education Matters: A Selection of Essays on Education by Alan B. Krueger, Cheltenham: Edward Elgar. Krueger, Alan and Mikael Lindahl (2001) ‘Education for Growth: Why and For Whom?’ Journal of Economic Literature, vol. 39(4), December, pp. 1101–36. Krueger, Alan and Diane Whitmore (2001) ‘The Effect of Attending a Small Class in the Early Grades on College-Test Taking and Middle School Test Results: Evidence from Project STAR’, Economic Journal, vol. 111, pp. 1–28. Krueger, Alan and Pei Zhu (2004) ‘Another Look at the New York City School Voucher Experiment’, American Behavioral Scientist, vol. 47(5), January, pp. 65–98. Martin, Xavier (1995) Economic Growth, MIT Press. Meghir, Costas and Marten Palme (1999) ‘Assessing the Effects of Schooling on Earnings Using a Social Experiment’, Stockholm School of Economics Working Paper No. 313, March 25. Meghir, Costas and Marten Palme (2003) ‘Ability, Parental Background and Education Policy: Empirical Evidence from a Social Study’, Institute of Fiscal Studies Working Paper No. WO3/05, London. Navarro, Peter (2000) ‘Economics in the Cyberclassroom’, Journal of Economic Perspectives 14(2), Spring. Nelson, R.R. and R. Phelps (1966) ‘Investments in Humans, Technological Diffusion and Economic Growth’, American Economic Review: Papers and Proceedings, vol. 56, May. Romer, Paul (1990) ‘Human Capital and Growth: Theory and Evidence’, Carnegie Rochester Conference Series on Public Policies, 32(0), Spring, pp. 251–86. Safire, William (2000) ‘Are School Vouchers the Answer?’ New York Times, 31 August.
6 On the Edge: The Uneasy Boundaries Between Public and Private Sectors John Kay
It is not often in social sciences that we can conduct controlled experiments. There was a dramatic one, however, in Germany after the end of the Second World War. One country was divided into two zones. One was run on the basis of central planning and coordination. The other was run as a decentralized market economy. We all know what happened. After 15 years the East Germans had to erect a wall to keep their experiment going and to stop their citizens fleeing to West Germany. Twenty-five years later the citizens of the two zones, taken together, demolished that wall. The difference in productivity between what had emerged in the West and East in those 25 years was something of the order of two or three to one – a very clear experimental result, even by the standards of the natural sciences. What have we learned from that experiment? For some, the lesson is simply that the private sector is rather good at managing things and the public sector rather bad. Therefore as many functions as possible ought to be transferred from the public to the private sector. If you have had some of my experiences of public sector activities, you might be tempted to accept such a proposition, but I think that such an analysis is simplistic. I want here to look at the question in another way. What should the boundaries be between the private sector and the public sector and how can we, by regulation, law and contract, define proper functions for one and the other and determine the relations between them? I want to look at the question of what lessons we can learn from the successes of the private sector that are relevant to the public 125
126 John Kay
sector. Then, instead of simply drawing a boundary between the public and private sectors, I want to ask how we should manage the wide range of hybrid institutions that fall somewhere in between those managed by some government agency under political direction and those subject to the ordinary processes of competitive markets. We are bound to have a large intermediate range of this kind and our real problem here is how to find a combination of public sector virtues and private sector disciplines in relation to them. The traditional differences between public and private sector management are to do with issues of legitimacy and authority. Traditionally, the state used to wage war, adjudicate disputes and maintain order. It levied taxes to finance military activities, justice, policing and other like functions. Government still does these things, but they are no longer the principal activities that we expect from it. Today we look to government for the provision of education and a transport infrastructure, for a guarantee of medical treatment and security in old age, to collect our rubbish and to make sure that we have a reliable supply of electricity. In all these and other areas, we are defining a role for government in the provision of goods and services, rather than simply in the exercise of authority. If we are just concerned with the exercise of authority, then its legitimacy and the propriety of the process by which it is exercised are fundamental requirements. We want our judges and our policemen to follow the law, our soldiers to obey orders and our tax inspectors to implement the tax code. We are not looking for innovation, imagination and originality from these people. In the delivery of services, however, our primary concern is not with the process, but with the outcome. We want to send our children to good schools. We want the buses and the trains to arrive on time. We want the rubbish to go and the lights to go on. We want to face retirement with confidence and to be made to feel better when we are ill. Our concern with the mechanisms by which these results are delivered is secondary. Most people are no more interested in how their hospital is run than they are in how their supermarket is run. Their concern in both cases is that the goods and services they want are actually delivered. A large part of today’s problems with the management of public services has to do with insufficient recognition of the transfer of the government’s role from that of traditional authority to that of the
The Boundaries Between Public and Private Sectors 127
delivery of goods and services. The requirements relating to the delivery of goods and services in the public sector are very similar to those that we impose on the private sector. The key issue for us today is that process-orientated mechanisms, which we need for the proper exercise of authority, are inappropriate or irrelevant to the effective delivery of goods and services. We know that judges could reach their conclusions more quickly, or that tax inspectors could collect revenues much more cheaply. But we do not want them to do so, because we are concerned just as much with the propriety of how the result is achieved as with the actual result itself. In relation to the delivery of goods and services, we think differently. Here, efficiency and effectiveness are the criteria we properly apply. This issue of legitimacy bedevils everything that government does. Indeed it bedevils every exercise of authority in a free society. Actions are legitimate only if there is a good and a widely accepted answer to the question: ‘What gives them the right to do that?’ That is the basic question of legitimacy in a democracy. We address it just as much to private corporations as we do to public agencies. Indeed, working-class organizations in politics came into being at the end of the ninteenth century largely around this central issue of the legitimacy of the private exercise of economic power. Trade unions and political parties challenged the basis on which private entrepreneurs exercised seemingly arbitrary authority over individuals and used their political power to reinforce that authority. For over a hundred years of politics that class structure has defined the roles of political parties and the nature of political rhetoric. These arguments about the legitimacy of private economic power do not have the same resonance in today’s rich western societies with their geographic mobility and more flexible patterns of consumption, work and credit. All these things mean that competitive markets today perform many of the functions that once seemed to require political action. If you don’t like your job or its pay and conditions of employment, you have the opportunity to find another one. If the milkman waters the milk or the brewer salts the beer (as people were traditionally concerned that they would), then in a competitive market you can take your custom elsewhere. While political organization and the regulation of economic activity did historically help to improve working conditions, to secure better product quality and to secure fairer prices, the primary thing that
128 John Kay
brought about these results was in fact competitive markets. This is the central part of the story of the superior economic performance of a competitive market system over a politically directed economic organization. To the chagrin and ultimately the defeat of socialism as a political doctrine, the market economy was at least as effective in achieving the economic aims that people sought. This is why today, when the public sector is so largely concerned with the provision of goods and services, we face the issue that it does not seem able to produce the pace of improvement and the responsiveness to people’s changing demands that the consumers of private sector goods and services have come to take for granted. So far as private economic organizations are concerned, the answer to the legitimacy question is that businesses won it through success. We accept (and even welcome) the authority of Sainsbury’s and Tesco in delivering our groceries, not because of the process by which they do it, but because of the manifest success with which they have done it in the past. As a result, we now have this unsettling paradox that people trust supermarkets more to secure the safety of their food than they trust government itself. We like what these businesses do and, if we don’t, we have alternatives. In a modern service economy, this is what has made the exercise of private economic power legitimate. In such an economy, what is legitimate is what works. This is the way in which we need to look at issues in both the public sector and the private. We do not want imagination from our tax inspectors, our judges or our soldiers. We want them to lay down clear and fair procedures and rules. But we do want imagination from our teachers and doctors. We do want innovative capabilities in the people who create and manage our transport infrastructure. The principal reason why the private sector has a much better record of innovation than the public sector is not a difference in objectives. The delivery of goods and services is central to both and the same kinds of people work in both public and private organizations. The difference comes from the ways in which private and public sector organizations are structured. By way of illustration let me focus specifically on two examples of how planned economies fail. Nikita Khrushchev succeeded Stalin as First Secretary of the Soviet Communist Party in 1953. Five years later, on a visit to the United States, he was in Iowa and famously went to a supermarket. It is said that he and his party went back to
The Boundaries Between Public and Private Sectors 129
Moscow believing that the shelves of the supermarket had been specially stocked for their arrival, as indeed would have been the case if the President of the United States had gone to see a supermarket in the Soviet Union. However, what really made an impression on Khrushchev in Iowa were the luxuriant fields of maize growing over the prairies. Not even Soviet planners could fake maize stretching as far as the eye could see. Khrushchev returned to the Soviet Union convinced of the economic potential of maize for the future of Soviet agriculture. So, over the next two years, very large tracts of Soviet wheat production were converted to maize. The experiment was not a success. One of Khrushchev’s acolytes said: ‘Under socialism maize can be grown anywhere.’ It turned out that he was wrong. Production fell and the economic setback that followed was one of the principal reasons why Khrushchev was ultimately toppled from power in 1964. Khrushchev had simply made a mistake. Maize was not in fact more suitable in the Ukraine than wheat. There were good reasons why that traditional crop was the one that was grown there. But that is the kind of error which people in business quite routinely make. Anyone who has experience of a large corporation has seen some senior executive go on a trip, visit another company, read a book, attend a seminar or whatever and come back with some bee in his bonnet about how the company can be transformed. Indeed, they often get the same sycophantic reaction that Khrushchev got when he brought his maize idea back from the United States. It was the context of Khrushchev’s mistake that turned the kind of misjudgement that we all make in economics into a national disaster for the Soviet Union. Khrushchev did not have (and could never have had) the information required to exercise the massive centralized economic power that he enjoyed. In the Soviet Union, mechanisms of decision-making were centralized and personalized. Outcomes were implemented on a huge scale. People who reported on the consequences of the experiment wished neither to hear nor to deliver bad news. They were concerned to protect their own positions and to win approval from their superiors. The result was that the party leaders were only slowly, if at all, accountable for the failures of the system. Let us be clear that this story is not the product of ‘public’ versus ‘private’ structures. It is the product of centralized control structures
130 John Kay
as against decentralized experimental structures. The problem hinges on the absolutism of authority. In the Soviet Union, as in many large private companies today, decision-making is personalized and undemocratic. Things would surely be better if we were only to do them through more rational and democratic processes? Should we not get wise people together in a single democratic institution to assemble evidence, to consider it dispassionately and then to set the direction of the industry? This was how the British electricity industry was managed after nationalization. Let us recall what happened. Back in the mid-1960s, just after Khrushchev had fallen, power blackouts in Britain deprived homes and businesses of electricity. This followed immediately after the election of Labour to power in 1964, using in its campaign a successful slogan about embracing ‘the white heat of technology’. This was the heyday of central planning, not just in the communist bloc but also in Western societies and large corporations. So the new Labour government decided on a large programme of investment in new power stations. After debating a variety of options, it was concluded that five new advanced gas-cooled nuclear reactors (AGRs) should be built on the basis of a small prototype design that had been constructed by the UK Atomic Energy Authority. The Minister of Power at the time was Fred Lee, a trade unionist who had reached the pinnacle of his not particularly distinguished political career. Lee announced the government’s decision with pride. He talked not just about cheap power, but about how Britain was going to develop a series of export industries around these new technologies. ‘I’m quite sure we’ve hit the jackpot this time’, he said. In the event, building these reactors was probably the worst single commercial decision ever made, anywhere, in the history of the world. On average, the construction time of the five reactors (incredibly the programme was actually extended to seven) turned out to be 20 years. It was not until the mid-1990s that they were operating to something close to their planned capacity. Dungeness B took 22 years to build. Hartlepool B took 23 years from cutting the first sod on site to the day when the first electricity was produced. It was almost 30 years before the programme’s output matched its originally planned capacity. Over those 30 years the total building costs for the programme totalled (at current prices) something like £100 billion. That represented about 10 per cent of Britain’s national output.
The Boundaries Between Public and Private Sectors 131
At the beginning of 2003 the entire investment in the AGRs was finally written off, when British Energy, the privatized business that owned these seven reactors, reached an agreement with its creditors, who essentially got all the equity of the business. We know that even this was only possible because the government agreed to pick up the ultimate decommissioning liabilities for the plants. So we spent £100 billion for assets which today have a negative value. That was the result of central planning in Britain’s electricity industry. Let us now look at the mechanisms of decision-making. It is pretty obvious who made decisions in the Soviet Union about maize. In Britain, it is much harder to be sure that the decision to build advanced gas-cooled reactors was made by anybody at all. There were four groups involved in the decision: the Atomic Energy Authority, which was responsible for research; the Central Electricity Generating Board, which until 2001 was the British nationalized electricity industry; the civil servants; and the ministers. In a formal sense, the decision to go ahead was given by the hapless Fred Lee. But Lee was a mouthpiece for the decision, not its arbiter. The Atomic Energy Authority was, as you might expect, in favour of the proposal. The senior staff of the Central Electricity Generating Board was divided as to whether it was a good idea. Among the civil servants there were probably two key figures: Burke Trend, who was Secretary of the Cabinet at the time, and Edwin Plowden, who occupied a variety of Whitehall roles from the 1940s to the 1970s, including being the first Chairman of the Atomic Energy Authority from 1954 to 1959, and who continued to be an influential figure behind the scenes and a strong advocate of nuclear power. I suspect that if we were to place responsibility for the AGR disaster anywhere, most of it lies with Lord Plowden. He chaired at least five official committees to do with industrial matters. His wife, Lady Plowden, chaired or was a prominent member of numerous committees and advisory bodies on social and educational matters. Between them, this pair probably did more damage to the British economy than any other two figures over the last half-century. One of Plowden’s committees, which reported in 1975 (halfway through the process of building the AGRs), was responsible for another round of restructuring of state-owned electricity. The problem, Plowden concluded, was that there was not enough centralization. His
132 John Kay
report said: In many of its dealing with other bodies, and particularly with government, the industry finds it difficult to speak with a single voice. The Department of Energy told us that the government seeks to obtain one considered and consistent view from the industry on a large number of topics, many of which may be interrelated. But the Chairman often cannot speak with authority on behalf of the whole industry. At most he may be able to state a range of views or give those of council members.1 That paragraph encapsulates in clear language the elegant nonsense that is characteristic of large bureaucracies. The concern is entirely with administrative process and administrative tidiness at the expense of outcome or result. Why on earth should one expect the totality of people in the electricity business to speak with a single voice? Rather than the Chairman speaking with authority on behalf of the whole industry, would it not be appropriate to hear a range of views? After all, if one is planning the British electricity industry for the next fifty years, it is certainly plausible that there might be a range of different views on how it might develop. Plowden’s words are the reason why these key civil servants should bear more than anyone else the responsibility for that dire decision. They reveal the way in which they approached this issue. Increasingly, as matters developed, the industry did speak with a single voice. It was the single voice of those in favour of the agreed policy, the AGR programme. People who expressed doubts found their career progression blocked or terminated. Immediately, Christopher Hinton, the Chairman of the Central Electricity Generating Board, was pushed out of office and made a life peer. Those who stayed learned the lesson that anyone who provided negative feedback on this remarkably unsuccessful programme would be treated in much the same way. The purge of dissidents in Britain was more subtle than the one that took place in Russia. But, in its own terms, it was more successful. Khrushchev’s failure did lead to him being toppled in 1964. Lord Plowden, on the other hand, continued to be a respected figure, not just in British society, but in the specific areas in which his expertise had been deployed so disastrously up to the end of his life.
The Boundaries Between Public and Private Sectors 133
To this day there has been no enquiry into the AGR programme, no audit of the total costs which were involved (the numbers I have cited can only be gathered from a variety of different sources) and no attempt to learn the lessons. Even today there are people who would have us embark on another nuclear programme framed in a remarkably similar mode. In the British electricity case, decision-making had the appearance of high rationality, but the consequences were similar to those we observed in the Soviet Union: uniformity of opinion in the short run leading to economic failure in the long run. The single voice is the characteristic of centralized decisionmaking, whether it is the voice of an individual like Khrushchev or the synthetic product of a process that is orchestrated and minuted by figures like Burke Trend. Commonly, a decision based on the single voice is one that is executed on a very large scale. Khrushchev’s experiment with maize in the Soviet Union was interesting and might have worked. What distinguished it (and the reason it became so disastrous) was its size. If an individual Russian farmer had visited the United States and been impressed by maize production, he might have brought back some seeds. He would have found that yields from the seeds were disappointing and that would have been the end of the experiment. If yields had been good, his neighbours might well have started imitating his particular innovation. The AGR programme, by contrast, was to build five power stations more or less simultaneously to a design that had yet only been tested in a very small prototype. If you are running a single electricity generating business for the whole of England and Wales, you have to make decisions on that scale. The point is that nobody, however talented, rational and well-informed, has the capacity to decide which technologies are going to be appropriate for British electricity twenty or thirty years ahead. The probability is high that any decision made for that period on that scale is going to be wrong. To stand any chance of success, centralized decision-making processes would have to be exceptionally sensitive to the results of their actions and exceptionally responsive to a change in environment. But in the Russian agriculture and British electricity cases there was no such sensitivity or responsiveness. Even modest men rarely tire of receiving the praise of their subordinates. To point out from below obvious failures of a policy is to label oneself as a disruptive influence in an organization that, by reference to its own values, may be
134 John Kay
performing more than satisfactorily. In the British case, the stream of misleading information about electricity only ended in 1990 because of the privatization of electricity companies. This had the incidental effect that ministers and directors became personally liable for any statements that they made to investors. When that effectively forced disclosure of the true costs of the AGR programme, the reactors were withdrawn from sale. And that was the end of the construction of any further nuclear power stations in the UK. Let us now turn from these failures of planning to some of the successes of the market economy. For many years most experts thought that computing power would be rather like electric power. (I suspect that, if Lord Plowden had been in charge, it would have been.) The theory was that you would have a few central computers into which everyone plugged, giving you massive economies of scale. As we all now know, the personal computer industry followed a very different course. But for a long time a high proportion of people in the computer industry expected that the ‘super computer’ would be the way of the future. Looking back, we can trace the change in direction to 1971, when Intel built what was called the first general-purpose microprocessor. Intel’s essential innovation was that the logic of the applications was contained not in the chip, as it had been previously, but in the memory of the machine. This gave the processor its multiplicities of processes. Previously people had supposed that you would build different machines for keeping your accounts, for writing your letters and so on. Intel led us down the route to the general-purpose machine that types our letters, keeps our diaries, values our portfolios, sends our emails and runs our lives. The first machine built around a general-purpose microprocessor was produced by Xerox in 1973. The only problem was that it took Xerox eight years to produce a commercial version of this machine and, by then, it was much too late. The first commercial machine came on the market at the end of 1974. It was called the Altair and was advertised in Personal Electronics Machines as being for hobbyists. You bought it in kit form and put the bits together at home. If you did it right, you had a functioning computer. You can imagine the kind of enthusiasts who went for the Altaire: interestingly, they included Paul Allen and Bill Gates. In fact, Gates dropped out of Harvard to write a programming language which would operate on that Altaire.
The Boundaries Between Public and Private Sectors 135
Much of the development of these small computers in the 1970s was essentially in the hands of hobbyists like Gates and Allen. It was only towards the end of the 1970s that big companies started getting in on the act. Interestingly, many of the companies that spotted the potential were from outside the computing industry. The first personal computer than I ever used was built by a company called Sirius, which was a subsidiary of Exxon. The dog that did not bark in that period was the world’s largest computer company, IBM. Through the 1970s, IBM had task forces looking at personal computers and committees looking at the future shape of the market. But they did not actually build any personal computers. It was not until 1980 that the then Chief Executive of IBM, frustrated by all the plans and proposals that never emerged from this morass of internal committees, set up a completely off-site operation with instructions to develop a personal computer. As they were completely separate from the main IBM operation, they bought in all the main components. Importantly, they bought the chip from Intel and an operating system from Microsoft. Microsoft had itself bought that basic operating system for $50,000 from an even smaller company in California. That was how IBM produced the first personal computer in 1981. The PC did not emerge within the mainstream of IBM because the people there understood (or sensed) that it was a potential threat to their jobs and positions. IBM had essentially given the main elements of the project to Microsoft and Intel. Realizing they had lost the initiative, the IBM mainstream tried to come back with its own operating system, which was then called OS2. This was technically better than the MSDOS operating system that had been bought from Gates and Allen. But it was not where the future lay. As we now know, that future lay in inventing computers that you can use without understanding anything about computers. That breakthrough was achieved with the Graphical User Interface (GUI), the icons and pointers that are the basis of the personal computers that we all now operate. Where did the GUI idea come from? Interestingly, the invention originally also came out of Xerox Parc, Xerox’s California research centre. Xerox Parc was perhaps the most innovative laboratory of its time, although few of its discoveries were effectively exploited by Xerox itself. But it was Apple who commercialized the GUI. It first
136 John Kay
came to market with mice and icons. Apple then tried to maintain this as a propriety system, which you could only access by buying Apple hardware. But the future of the industry really lay with something that was as easy to use as these systems were, but was also compatible with the existing IBM/Microsoft MSDOS operating systems. A Graphical User Interface on an MSDOS platform would sweep the world. That is, of course, the set of programmes that we now call Windows. What do we learn from this particular history? The most important lesson is the difference between what I call ‘disciplined pluralism’ and the single voice. If you had been Nikita Khrushchev or Lord Plowden, you might have asked: ‘Who is in charge of the successful development of the personal computer industry?’ The key point in the way in which the personal computer industry developed was that there was no such person. To answer that question by suggesting that it might be Bill Gates or the Chairman of IBM would completely miss the essential features of the way in which that market develops. Nobody ever knew more than a few months ahead how the personal computer industry would evolve. Perhaps Steve Jobs, the CEO at Apple who had the key idea of creating computers that you could use without understanding them, was the most important contributor to the whole process. But Jobs’s strategy for Apple was a conspicuous failure. Gates’s success derives from the fortuitous event that IBM rang him when they wanted to buy an operating system. The vast majority of the initiatives failed, even those that ultimately proved important to the development of the industry. The single most important contribution to the development of the computer industry came from Xerox, a company that is nowhere in the personal computer market today. The success of the personal computer industry is the success of disciplined pluralism over the single voice. Let me say a bit more about disciplined pluralism. We often talk about the ‘marketplace in ideas’. The metaphor of a marketplace in ideas identifies the most important characteristic of a market. It is not the jingle of cash registers; it is the effect of disciplined pluralism. In the marketplace in ideas you inhabit a world in which new ideas are constantly being floated. Most of these ideas are wrong or foolish, but all of them are subject to assessment and evaluation. If they survive the tests of assessment and evaluation, then knowledge advances.
The Boundaries Between Public and Private Sectors 137
The modern marketplace in ideas, as experienced for the past three or four centuries, devolved through the disciplined pluralism of scientific rationalism, which replaced the single voice of religious authorities. It is no accident that the development and evolution of disciplined pluralism in intellectual life coincides pretty much in time with the evolution of the pace of technological innovation and, particularly, with innovation in the development of commercial institutions. It is that co-evolution of technology and of economic, social and political institutions that has been the central dynamic of western productive economies since the Renaissance and the Reformation. Disciplined pluralism has two central characteristics. The ‘pluralism’ gives you freedom to experiment; the ‘discipline’ rejects bad ideas and leads to imitation and adoption of good ones. Democracy is itself a mechanism of disciplined pluralism. We can think of it as a marketplace in political leadership. In representative democracy we have a need for political leadership and authority, but the ideas that leaders implement and the authority that they exercise are regularly tested. Ideological diversity and personal ambition provide the pluralism and the process of elections provides the discipline. The competitive market for goods and services is, of course, the most familiar example of disciplined pluralism. It distributes economic authority. In my account of the personal computer revolution, everyone was free to make their own decisions about the way in which they wanted the market to evolve. Most of these decisions actually turned out in the end to be unsuccessful. But the few successful ones formed the basis of the industry’s development. Pluralism comes from decentralization of economic authority. Discipline ultimately comes from the capital market, which cuts off finance for experiments that fail. The health care industry gives us two quite different models of disciplined pluralism. In pharmacology, we have the competitive development of blockbuster innovations by corporations, a process that is highly regulated and protected by patents and commercial secrecy. Patent legislation creates an artificial, but lively, competition and discipline comes from regulatory authorities and the clinical judgements of experienced doctors. That is one model of disciplined pluralism. We have quite another model in surgery and treatment protocols. There you have piecemeal innovation by individual practitioners
138 John Kay
and by teams. There is almost no formal regulation of surgery. There have to be extensive clinical trials before a pill is licensed to be put on the market, but a doctor can cut you open using an experimental technique without having to go through any such regulatory procedures. Here we have a process of the open sharing of methods and results in a peer review process. Both methods of innovation work tolerably well. Indeed, the central problem we have in Britain with health care is that we do not have the organizational innovation that can keep up with the pace of technological innovation. To summarize and to tie the discussion back to the issue of legitimacy with which I began, one common feature of all these mechanisms of pluralism is their anonymity. In the marketplace for ideas, the verdict on ideas is reached by many appraisers. In the market for political leadership the outcome is the decision of many voters. In the market for goods and services the outcome is ultimately the verdict of consumers. This is the essential feature of this disciplined pluralism, as opposed to the authority of the single voice. So let me come to some conclusions from this account of how markets work that we can apply to the operation of political institutions. First, I am arguing that we should elide, rather than emphasize, the differences between public and private sectors operations. We should not be asking what should the public sector do and what should the private sector do, or how can we write regulations and contracts that define these interactions and demarcate these functions. Instead, we need to recognize that there will always be a variety of hybrid institutions in our society and that we need to learn some lessons from the functioning of the private sector that we can apply to the operations of the public sector. Second, we need to understand the limitations of contracts and targets as mechanisms of public sector management. The issue of incentive compatibility is fundamental. If the centre held the information needed to specify contracts or define targets, it would be able to direct the activities in question itself. The need for delegation arises from the inescapable limitations to the information the centre can hold. And so the traditional distinction in public administration between policy and implementation can, in economic matters, rarely be made. Third, we need to learn the lesson that legitimacy in a modern society is conveyed not just by the propriety of process but by success in
The Boundaries Between Public and Private Sectors 139
delivering the goods and services that people want. This means we need to recognize that the users of public services are primarily customers rather than voters. The legitimacy of authority comes through the successful delivery of services and not just through the ballot box. Fourth, we need to recognize ‘fuzzy government’ structures. These are structures that cannot be captured by any single voice and are, as a result, more robust to self-interest yet sensitive to market disciplines than mechanisms that are under the control of (or can be captured by) some single group of individuals. A danger in all political processes, which has led to so much ‘producer domination’ of public sector activities, is that the processes by which the managers of these activities are appointed and made accountable is one that can be captured by particular interest groups. It is only if we have multiple levels of accountability for the delivery of these kinds of services that we can get away from the dangers of a single voice in these areas. Fifth, accountability in public services should be for outcome rather than for process. You cannot effectively define in advance the criteria of accountability. This leads to all the difficulties about targets that have been widely discussed of late. The management of these kinds of services cannot be made fair in ‘process’ terms, because the constant demand for accountability and transparency and fairness in process in the public sector comes into direct conflict with the wider need to create accountability not for the process by which decisions are taken, but for the outcomes to which they lead. So we need to stop being ashamed of making judgements that are quite difficult in the public sector, like ‘I’m firing you because I think you are no good at your job’, or ‘I’m hiring you because I think you and I could work well together.’ It is because it is much easier to make such statements in private sector activities that it is so much easier to build effective teams there. The big idea of disciplined pluralism is that there should and can be no single voice. At a fundamental level there is no big idea. Disciplined pluralism means decentralized authority. It means accountability for outcome, not for process. It means mechanisms that force acknowledgement of failure and penalize it, but not to such an extent that people are afraid to make mistakes. Disciplined pluralism with these characteristics is the central dynamic of the market economy. It is the explanation of the striking success of the
140 John Kay
market economies over planned and centrally coordinated ones. The challenge now is to make disciplined pluralism the dynamic of public sector services as well.
Note 1. The Structure of the Electricity Supply Industry in England and Wales: Report of the Committee of Inquiry (the Plowden Report), Cmnd 6388, London: HMSO, 1976, pp. 15–16.
7 Demographics, Economics and Social Choice Adair Turner
Demographic trends were not till recently a major subject of public debate. Actuaries and demographers have been aware for years that something rather big is happening, but demographics rarely made headlines. That has changed rapidly over the last few years. As chairman of the UK’s Pensions Commission, I see each week’s press clippings – an endless series of headlines talking of crisis, black holes and ‘work till you drop’. Throughout Europe, pension reform (and opposition to it) has shot to the top of the political agenda. Increasingly, commentators and politicians are stressing, correctly, that pension system problems are not some technical issue, requiring just technical solutions, but raise far wider questions. Many press commentators now refer to UN estimates that Europe ‘needs’ tens or hundreds of millions of immigrants if it is to support its old people. David Willetts, the Conservative Opposition’s chief spokesman on pensions and employment affairs, recently issued a pamphlet arguing that Europe should consider policies to increase the birth rate.1 The Economist made the same argument a few weeks ago, with the cover headline ‘Work longer, have more babies – how to solve Europe’s pension crisis’.2 My purpose here is not to explore the details of pension system design. I recently addressed the details of pension system economics in a lecture to the Institute and Faculty of Actuaries,3 so here I shall skip briefly over some of those details, asserting some arguments rather than proving them. My focus here is instead on the wider economic and social issues posed by demographic change, which I
141
142 Adair Turner
consider under four headings: 1. 2. 3. 4.
The demographic trends The macroeconomic challenges Immigration-based ‘solutions’ Increasing birth rates.
The demographic trends Future demographic trends are uncertain. We know what has happened to longevity and fertility in the past, but future projections are just projections. With United Nations estimates of global population in 2050 ranging from 7.4 billion to 10.6 billion, projections in this field come with big caveats. But, despite that warning, the direction of the key trends – if not their precise pace – is fairly clear. In focusing on the impact of those trends, I shall be using the UN’s medium variant projection to illustrate my points. Europe’s demographics are changing as a result of three factors: first, increasing longevity; second, declining fertility; and, third, the existence of a ‘baby boom’ cohort, the generation born between about 1950 and 1970, which is larger than both the one before and the one that followed. My focus here is almost entirely on the first two factors and not on the baby boom, because it is these first two that define the long-term challenge for the whole world. The baby boom is important for specific national pension systems over the next 20 to 30 years, because it makes changes in the ‘dependency ratio’ more rapid than those that would be produced by gradual and one-directional shifts in longevity and fertility. But my aim here is to look beyond the next 20 to 30 years in specific countries to the overall trends and implications over, say, a 50-year period and at a global level. So what is happening to longevity and fertility? Longevity (life expectancy) is rising almost everywhere. Figure 7.1 shows how it has gone up in the UK throughout the twentieth century – rapidly in the first half, a slight deceleration in the 1950s and 1960s, acceleration since then. It has also risen across the world, though with a recent deceleration, which is largely AIDS-related. It has risen to UK levels or above in all successful rich economies – in Western Europe, the US and Japan. It is getting closer to these levels in countries like South Korea and Thailand. In all these countries it is still going up.
Demographics, Economics and Social Choice 143
80
80
Life expectancy
75
75 UK
70
70
65
65
60
60 World
55
55
50
50
45
45
0
0 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
Figure 7.1 Life expectancy at birth, male Note: These are ‘period’ life expectations, which actually underestimate the expected lifespan of a baby born in the year specified, but which are easier to calculate than the correct ‘cohort’ figures and therefore frequently used international comparisons. Sources: Government Actuary’s Department (GAD) for UK; United Nations for world.
21
21
20
20
Years left to live
France
19
19 18
18
UK
17
17
16
16
15
15
0
0 1960
1970
1980
1990
2000
Figure 7.2 Life expectancy at 60, male Note: On ‘period’ basis. Sources: Eurostat Demographic Yearbook; GAD for UK.
But for pension systems what matters is not so much life expectancy at birth (which can increase due to lower infant mortality, without the lifespan of the normal healthy person increasing much) but rather life expectancy at 60 or 65 (how long people live after the ages at which they typically think about retirement). Figure 7.2, shows how life
144 Adair Turner
expectancy at 60 for males in France and the UK has also has gone up rapidly, again with the same pattern of acceleration in the 1980s and 1990s. The same pattern can be found in all rich countries. The key question, of course, is how far and fast will these trends continue in future. Current official projections, for instance in the UK from the Government Actuary’s Department (GAD), show 30 years of continued rapid increase, but then a tailing-off, lifting male life expectancy by three or four years by 2050. This tailing-off, though, reflects a highly contested assumption. This is the ‘limit-to-life’ assumption that we can get more and more people to live to full old age – say 90 – but that we hit absolute biological limits much beyond that point. The deceleration in the rate of growth of life expectancy in the 1950s and 1960s appeared to provide support for this assumption, but it has been challenged by the acceleration of the last 20 years. During this period official projections have consistently underestimated actual improvements and the date at which slowdown is forecast to start has consistently been pushed out; it is always 25 to 30 years away, never getting closer. And there is therefore an alternative school of thought – of both demographers and biologists – that questions whether we have any sound basis for asserting a limit to life. It predicts further straight-line improvements, which would increase life expectancy at 65 by something more like eight years, rather than four, over the next half-century. From this lively debate I want to take the following assumptions. First, for the next 50 years, we shall see continued major improvements in life expectancy, both at birth and at retirement ages. Second, that those life expectancy increases may well be significantly higher than current official projections. Third, that it is possible that for the next 50 years at least, and perhaps for the next century, we may see no slowdown at all in the pace of increasing life expectancy. Let us now turn to fertility. The pattern here is that birth rates are falling to or below replacement levels wherever economic prosperity is achieved. In Europe, as Figure 7.3 shows, the total fertility rate (TFR) fell below replacement (say, 2.05 children per woman) in the mid-1970s. It is now well below replacement rates and still falling. The TFR indeed is below 2.0 in every European country, west and east, except Albania. In North America, which on the UN definition used here is the US and Canada but not Mexico, the TFR fell below replacement slightly
Fertility rate
Demographics, Economics and Social Choice 145
4.0
4.0
3.5
3.5
3.0
3.0
North America
2.5
2.5
2.0
2.0
1.5
1.5
1.0
Europe 1950–55
1960–65
1970–75
1980–85
1990–95
1.0
2000–05
Figure 7.3 Total fertility rates: Europe and North America, 1950–2000 Source: United Nations.
earlier that in Europe, but has since picked up a bit. But the US figures need interpretation. If you split them into ethnic groups (white, black, Asian-Pacific, Native American and Hispanic), you find birth rates at record lows for each individual group, and find that the most prosperous groups – white and Asian-Pacific – have rates significantly below the replacement level. The overall increase for the US is solely the result of an increasing Hispanic proportion in the population due to immigration, a weighted average effect. Once Hispanic fertility rates fall towards replacement – and they are clearly on that course – the average will come down again.4 But this is not just a European and American phenomenon. Figure 7.4 shows how in East Asia the decline is even more dramatic than in Europe. Japan has been at or below replacement rates since the 1960s and is now at only 1.2 children per woman. Korea, Hong Kong and Singapore are now at levels as low as the lowest European countries. China has declined to 1.8. While that decline is partly attributable to the one-child policy, what these other country lines tell us is that the Chinese have pulled forward into the demography of a highincome country, a trend which would have occurred in any case a few decades later with rising prosperity. Finally, Figure 7.5 shows UN forecasts of fertility rates falling below replacement levels within the next 15 years in some rather surprising countries – Brazil, Turkey and Iran. These figures suggest a striking message – the universality and, it seems, the economic determinism of what is going on here. We
146 Adair Turner
7
7 China Korea
Fertility rate
6
6
5
5
4
Hong kong
4
3
Japan
3
2
2
1
1 1950–55
1960–65
1970–75
1980–85
1990–95
2000–05
Figure 7.4 Total fertility rates: Asian countries, 1950–2000 Source: United Nations.
8
8
7
7
Fertility rate
6
6
Iran
5
5
Turkey
4
4 Brazil
3
3
2
2
1
1
1950–55
1960–65
1970–75
1980–85
1990–95
2000–05
2010–15
2020–25
Figure 7.5 Total fertility rates: Iran, Turkey and Brazil, 1950–2020 Source: United Nations.
sometimes talk of deep-seated cultural differences but the evidence is that, whenever we get three conditions – reasonable prosperity, a high level of female literacy, and a supply of legal, safe and reasonably cheap contraceptives – fertility rates come down towards replacement levels. This is as true for Catholic Italy and Brazil as it is for Lutheran Sweden and Minnesota, Confucian China and Buddhist Thailand, or Sunni Moslem Turkey and Shiite Iraq. We should always be cautious of declaring universal sociological truths, but this does seem to be one.
Demographics, Economics and Social Choice 147
This is the case even allowing for the pitfalls of fertility rate figures. TFR figures have the merit of being available rapidly and across the world, but they suffer from a key interpretation difficulty caused by the fact that the average age of childbirth is changing, that is, TFR movements exaggerate changes in completed family size. If, for instance, the average age of childbirth rises, as it is throughout the world, TFRs will for a period fall more rapidly than completed family size. Some of these TFRs trends may therefore slightly overstate the pace of change. But the direction is clear and analysis of the better but more difficult measure – completed family size – still suggests that, wherever we have economic prosperity, the trend is towards fertility rates below replacement levels. So, if the future is longevity rising, possibly without limit, and fertility in all successful economies falling to or below replacement level, what happens to pension systems and what social choices do we then face? The key here is, of course, the ratio of workers to retirees. Population structures can be set out graphically by arranging the number of people in each age band in layers. For the last two centuries, almost all countries with growing populations have had the kind of triangular structure shown on the left in Figure 7.6. The ratio of workers to retirees, for any given retirement age, is represented by
Age group 100 + 95–99 90–94 85–89 80–84 75–79 70–74 65–69 60–64 55–59 50–54 45–49 40–44 35–39 30–34 25–29 20–24 15–19 10–14 5–9 0–4
B
B
A
A
Figure 7.6 From pyramids to columns
148 Adair Turner
area A over area B, from which it is clear that, if people live longer beyond retirement age, Area B will increase and the ratio will fall. Also, if fertility rates fall to replacement so that each generation is the same size as its predecessor (with almost everyone surviving to, say, 60 and then dying at some point over the following 30 years), the shape of the distribution will change over time from a triangle to a column with a small triangle on top. (See the right-hand side of Figure 7.6.)
The macroeconomic challenges As dependency ratios change (area A over area B), so will the economics of pension systems. This applies to funded schemes as well as to pay-as-you-go (PAYG) pension systems, because the difference between the economics of funded and unfunded pension systems is far less fundamental than you might think. The way to see this is, first, to consider the classic state-run pay-as-you-go scheme and then to ask how things differ if pensions are funded. The classic article on the economics of PAYG pensions was written by Paul Samuelson in 1958.5 Although this was not the primary purpose of the article, in it he set out the crucial influence of demographics on pensions (see Figure 7.7). Think first about a pay-as-yougo pension system in a world of no population growth and no productivity growth. Each working generation – of equal size – pays a
No population growth
Average retirement income Average worker contribution
=
Working years Retired years
No productivity growth
Implicit rate of return on contributions = 0
Population growth positive
WC
No productivity growth
Rate of return = population growth rate
Population growth positive
WC
Productivity growth positive
Rate of return = population growth rate + productivity growth rate
RI
RI
=
=
WY RY
WY RY
× Population growth
× Population growth × Productivity growth
Figure 7.7 PAYG pension systems: key ratios
Demographics, Economics and Social Choice 149
percentage contribution of wages to support retirement pensions. In that world, you can easily show, the ratio of average retirement income over average worker contribution will be equal to the ratio of years spent working to the years spent in retirement. If, on average, you work 45 years and retire for 15, then a stable, soundly financed pension system could have pensions at 60 per cent of average earnings and worker contributions of 20 per cent of earnings, a 3 : 1 ratio; or it could have pensions at 30 per cent of average earnings and 10 per cent worker contributions; or 90 per cent and 30 per cent or any other 3 : 1 relationship you want. The implicit rate of return on contributions made into the PAYG scheme in this case would be nil. For every £1 of income sacrificed as a worker contribution, you would in retirement get £1 of income back, but with no return on your ‘investment’. If you add population growth to the model, but not yet productivity growth, something rather useful happens. The ratio of average retirement income (RI) to worker contributions (WC) now becomes working years (WY) over retirement years (RY) multiplied by the population growth between one generation and the next. Now, if you have 45 years’ work and 15 years’ retirement and also 1 per cent per annum population growth, you can have a stable PAYG system into which workers pay 20 per cent of income, but from which pensioners get not 60 per cent but over 80 per cent of average earnings because there are now more workers per pensioner. In this case, the implicit rate of return you get on your ‘investment’ in the social security fund is the rate of population growth. Samuelson labelled this the biological rate of return: the return you would indeed get in the very simplest retirement support system, where children look after their elderly parents and within which people would clearly be better off in retirement the more children they had. Finally, if we then add productivity growth to the equation, the implicit rate of return on your investment in a social security system becomes population growth plus productivity growth, that is, the rate of growth of the economy. It is on the population term in both the second and third cases that I want to focus, because it is that which creates the key feature of PAYG pension schemes in rising populations. With rising populations PAYG pension schemes are a bit like pyramid selling, chain letters, or double-your-money Ponzi schemes in that the benefits they promise in relation to what you put in crucially depend on there
150 Adair Turner
being more people in the next generation, the next step of the chain. And the essence of what is happening to pension systems in the face of falling fertility is that the pyramid scheme is coming to an end. On top of that, of course, comes the impact of rising longevity, which also affects the ratios. In a simple steady-state model, the support ratio is equal to working years over retirement years multiplied by population growth between the generations. It falls, therefore, if one of two things happens: either if longevity increases but not the age of retirement, so that years in retirement increase; or if fertility falls and thus population growth ceases. The striking thing is how dramatically the support ratio changes, given quite realistic changes in longevity and the fertility rate. Table 7.1 presents figures from a simple steady-state model, allowing us to look at the impact of increased longevity and falling fertility, without the real-world complexity of the baby boom. It assumes working life starts at 20 and in both columns the retirement age is assumed to be 65. Column 1 assumes a 15-year life expectancy at retirement and column 2 a 20-year life expectancy. There are three different population growth rate scenarios. It shows (top left-hand figure) that, with population growth of ⫹0.5 per cent and a retirement age of 65, the support ratio is 3.5. But it falls to 2.8 if life expectancy at 65 rises from 15 to 20 years, which is exactly the order of magnitude likely over the next five decades. The support ratio will drop even further (to 1.9) if the population growth rate falls from ⫹0.5 per cent to ⫺0.5 per cent, which is the sort of change likely to occur in Europe between 1980 and 2030. Table 7.2 gives the UN’s projections for five countries and for the world as a whole between now and 2050. The support-ratio shifts predicted for developed countries are of a similar order of magnitude to Table 7.1 Support ratio dynamics Ratio of workers to retirees, assuming working from age 20 Population growth rate % p.a. Life expectancy at 65 Retirement age
⫹0.5 Zero ⫺0.5
3.5 3.0 2.6
2.8 2.25 1.9
15 65
20 65
Demographics, Economics and Social Choice 151
Table 7.2 Support ratio forecasts, 2000–50 (ratio of 20–64 year olds to 65⫹ year olds) 2000
2050
UK Italy USA
3.7 3.4 4.8
2.1 1.4 2.8
China Korea
8.8 9.0
2.4 1.7
World
7.8
3.6
Source: UN Medium Variant projection.
those in the model. Support ratios are forecast to fall from 3.7 to 2.1 for the UK, 3.4 to 1.4 in Italy, 4.8 to 2.8 for the US. The falls are even more dramatic for countries like China and Korea, with support ratios falling from well above to well below current western levels in just 50 years. Such a shift in support ratios challenges pension systems throughout the developed world. It faces a society with essentially just three choices: ● ● ●
increased average retirement ages, that is, longer working lives poorer pensioners relative to average earnings or higher worker and employer contributions as a percentage of earnings.
And if you do not raise the retirement age, the size of the swings implied in terms of poorer pensioners or higher contributions is large. If your support ratio falls by 43 per cent, which is precisely the fall projected for the UK in Table 7.2, and if you do not increase average retirement ages, you either have to cut pensioner income by 43 per cent or to increase worker and employer contributions by 76 per cent. This means, inevitably, that average retirement ages are going to have to rise.6 A reasonable rule might be that retirement ages must rise in line with rising life expectancy, keeping the ratio of working years to retirement years constant. But this would only be sufficient to maintain support ratios if the only thing we were facing was an increase in longevity. It is not, because we face falling fertility as well.
152 Adair Turner
Table 7.3 Support ratio dynamics (⫹ 20 years’ life expectancy) Ratio of workers to retirees, assuming working from age 20 Population growth rate, % p.a. Life expectancy at 65 Retirement age
⫹0.5 Zero ⫺0.5
3.5 3.0 2.6
2.8 2.25 1.9
15 65
20 65
3.53 3.00 2.55 20 68.75
Table 7.3 adds a new column to Table 7.1 to show the effect on support ratios if life expectancy at age 65 rises to 20 years, but the average retirement age moves from 65 to 68.75, thus keeping constant the ratio of working years to retirement years. Under these conditions, if your population growth rate is steady at ⫹0.5 per cent a year, the increase in the retirement age fully offsets the increased life expectancy effect. However, if your population growth rate is ⫺0.5 per cent a year, raising the retirement age proportionately only gets you back to a support ratio of 2.55. You would have to take the retirement age to something like 73 to keep it stable at 2.6. That is the scale of the demographic challenge that Europe, the US and also now much of Asia are facing: falls in support ratios that cannot be fully offset even by retirement ages rising fully in line with increased longevity. This means that workers will have to make bigger contributions and/or pensioners will get poorer, even if retirement ages rise proportionately. But so far I have only been looking at pay-as-you-go state-run pension schemes. Don’t funded pension systems get round this demographic dilemma, since workers pile up capital assets rather than being dependent on future workers paying taxes? The short answer to that question is no. The long answer is set out in my earlier lecture,7 but here are the key points of the argument: ●
First, in any pension system (funded or non-funded) today’s pensioners are dependent on the output produced but not consumed by the current generation of workers. What differs between funded and non-funded systems is how the resource transfers flow, and the risks associated with the pensioner’s claim on future output. A PAYG claim is a claim on future tax resources and is subject to
Demographics, Economics and Social Choice 153
●
●
●
political risk. A funded claim involves the ownership of a capital asset, which gives you a right to receive future profit and rent streams and is subject to various categories of economic risk. Either way, a pension claim is a claim on future output and is dependent on future workers producing that output. Second, therefore, funded systems can only help overcome demographic challenges if they increase total national savings, total investment and thus the total future GDP or GNP on which future pensioners will be able to draw. There is no ‘free lunch’. With PAYG systems you need higher taxes, if you want to avoid poorer pensioners or rising retirement ages; with funded systems you need higher savings. And interestingly, you need higher savings not just in this generation, but in future generations as well. Either way, you have to cut workers’ consumption to make pensioners better off. Third, the return on additional savings is subject to demographic risks. An increase in savings to overcome rising longevity will (everything else being equal) produce a fall in the marginal rate of return. And a falling population (where generation 2 is smaller than generation 1) will tend to produce a period of falling asset prices, relative to the constant population alternative, since generation 2’s target asset accumulation will be smaller than generation 1’s. Fourth, as a result, demographic changes carry very similar implications for the mathematics of funded schemes to those of PAYG schemes. With a PAYG scheme, if you have a rise in longevity alone, a proportionate rise in retirement age is a sufficient policy response, removing any need for increased taxes. Similarly, within a funded scheme, if you have rising longevity alone, a proportionate rise in the retirement age removes any need for increased savings, or any danger of declining marginal returns or depressed asset prices. But, once you have a fertility decline as well, both types of system face more difficult problems, which simply show up in different ways.
So funded systems do not give us an escape route from the demographic challenges. This is true – at least in the long term – even when we take an open-economy rather than a closed-economy view. The earlier assumption about a falling population leading to a falling rate of return on savings and asset prices implied a closed economy, in
154 Adair Turner
which savings have to be matched by investment. In an open economy, the alternative exists of investment overseas, with the benefit coming back via the overseas investment income line of GNP, not via domestic output. This still requires a higher savings rate, but at least means that the higher savings will not earn a falling rate of return and that future pensioners will not have to sell accumulated assets solely to a smaller next generation of domestic workers. But it is true that the open-economy case slightly changes the argument; it does go only up to a point and only for a transitional period. This is because the countries that you would like to invest in themselves face changing demographics. Figure 7.8 illustrates the problem. Within 50 years China will have a demographic structure almost exactly in line with Britain’s. China is already debating its pension time bomb and would like to build up claims outside China to support its future retirees. It does not want to build up liabilities to support pensioners in other countries. Any British or European strategy that assumes that our pensioners will be supported by Chinese workers, via capital claims on their output, cannot in aggregate work, if China is pursuing the same strategy.
UK
China
2000
2050
21
30
60
54
19
16
10
30
65
54
25
16
60+ years
15–59 years
0–15 years
Figure 7.8 Demographic change in the UK and China, UN Medium Variant projection (% population by age band) Sources: OECD Historical Statistics; OECD Economic Outlook.
Demographics, Economics and Social Choice 155
This is not to deny the potential of overseas investment as a transitional phase, but Table 7.4 suggests that it is a transitional phase with a problem. There is a clear, though not perfect, correlation between economic success and falling fertility. Most of the countries that the capital markets would describe as attractive investment destinations are already at or near replacement-level fertility – countries like China, Korea and Thailand. The only countries we can be confident are going to generate a demographic pyramid into the late twenty-first century are countries like Bangladesh, Congo, Iraq and Pakistan – in which most people would be wary of investing their pensions. There are today some countries in the intermediate category – India in particular stands out as a country of attractive investment opportunities and a large and still expanding population. But the big picture and the long-term ‘catch-22’ still remain. Wherever you get economic prosperity, you get fertility falling to below replacement level. Therefore, the stable and successful countries in which you would like to invest will develop column-shaped demographies. If the whole world develops a stable population, with a column-shaped demography, the whole world cannot solve its support ratio problem by investing in the moon. Therefore, in the long run, you have to develop pension policies in individual countries and across the world that can cope with column-shaped demographies, rather than dream
Table 7.4 Demographic conditions and investment attractiveness TFR below 2.05 today
China Korea Russia Thailand Ukraine
TFR in 2015–20* Below 2.05
2.05–2.3
Above 2.3
Brazil Indonesia Iran Mexico Turkey Vietnam Uzbekistan
Algeria Argentina Chile India Malaysia Morocco South Africa
Bangladesh Congo Egypt Iraq Kenya Nigeria Pakistan Philippines Saudi Arabia
Note: * Projection per UN Medium Variant.
156 Adair Turner
of escaping that challenge via overseas investment designed to marry your capital to the base of someone else’s population pyramid. This creates an immediate challenge for the already ageing and low-fertility countries – for Europe in particular. Our support ratios are falling; our PAYG schemes are under pressure; funded schemes are only a partial solution and not fundamentally different; and overseas investment is at best a transitional answer. So what do we do?
Immigration-based ‘solutions’ One possibility is to try to change our demographics via either increased fertility or increased immigration. The immigration option is now extensively debated and, to a degree, is happening whether people like it or not. Immigration to Britain is now running at about 170 000 per annum, up from zero in the early 1990s. The British government has recently increased its estimate of long-term sustainable GDP growth from 2.5 per cent to 2.75 per cent, entirely because of higher labour force growth resulting, primarily, from higher immigration. The Economist, when not arguing for more fertility, is in favour of more immigration. A recent editorial was headlined ‘Net immigration into Britain is higher than it has ever been – good’. Newspaper articles now frequently refer to estimates of how many immigrants Europe needs to keep its support ratios constant. The problem is that these numbers are huge, implying not only massive immigration but an exploding size of total population. The UN has calculated (see Table 7.5) the immigration needed to keep support ratios constant. The result over 50 years would be the UK population growing from 59 to 136 million and Europe’s from 372 million to 1.2 billion. Populations would then have to rise as rapidly in the following 50 years because, Table 7.5 Immigration-only responses: UN migration scenario To keep the ratio of 15–64 year olds to 65⫹ year olds constant requires: Total population (millions) 2000 UK European Union
59 372
2050 136 1 228
Demographics, Economics and Social Choice 157
when you import immigrants to fix a support ratio problem, they themselves grow old and become pensioners and then you need more immigrants to support them. And you only have to look at these figures to realize that this scale of immigration is both undesirable and impossible. It is undesirable because even the most committed multicultural list would probably admit concerns about our ability to integrate those numbers and because almost everyone would dislike the environmental consequences of such huge and rapid growth. It is impossible for the same reason that the overseas investment route is impossible as a longterm solution, which is that if the whole world population stabilizes you cannot go on importing immigrants. If the world’s demography becomes a column, you cannot solve your ratio problems by laying claim to the base of someone else’s pyramid, either via overseas investment or by immigration. Fortunately immigration on this scale is also unnecessary, because these figures assume unchanged retirement ages. As Figure 7.9 illustrates, if you do not change retirement ages in the face of rising longevity, stable support ratios require accelerated population growth. But unchanged retirement ages is an absurd assumption: retirement ages will and should rise. We noted earlier, however, that even a proportional rise in retirement ages is not sufficient to stabilize support ratios if fertility is falling. So, while these ‘doomsday’ figures are an impediment to sensible debate, there is a more realistic choice to be debated. Looking again at Table 7.3, if the European support ratio falls from 3.5 to 1.9 because of increased longevity and falling fertility and if we
Retirement age
Initial structure
Plus rising longevity
Plus immigration to keep support ratio constant
Figure 7.9 Rising longevity, fixed retirement age and stable support ratios
158 Adair Turner
offset some of that fall by a proportional increase in the retirement age to 68.75, should we also plan to restore it further to, say, 3.5 by contemplating a population growth of, say, 30 per cent over 50 years, that is, the UK population growing from 60 to 80 million and the European Union’s from 370 to 500 million (a similar sort of population increase as is already expected in the US). This would require, for instance, an inflow into the UK at about twice the current annual level. That sort of scenario is possible and is worth debating. It poses a real social choice for Europe. Should we welcome a mass immigration into Europe and a population increase of, say, 20 to 30 per cent, in order to keep our support ratio constant or at least falling less fast? Simply from my Pension Commission point of view this would be useful. But there are far wider issues here than support ratios, issues that I will consider under four headings – political/cultural, environmental, economic and geopolitical. The political/cultural issue is whether Europe can integrate successfully, or indeed wants to integrate, this scale of immigration. It is a sensitive issue. Throughout Europe there is considerable opposition to immigration. But many people ask, if America can integrate 30 per cent in 50 years, why can’t Europe. Is the only problem the conservatism and racism of the old continent? Even among the proimmigration lobby meanwhile, there are divisions between integrationists and multiculturalists – should immigrants to Britain be required to learn English both to help them and to aid integration, or is that cultural imperialism? I cannot deal with this complex debate in detail, but I would like to make one point. The comparison with the US is facile, since it fails to recognize the far more difficult integration challenge that Europe faces. We often talk of an American melting pot, as if it is a successful mixing of quite different cultures. But, on the whole, it is not. Historically, the majority in the melting pot has been a mixture of different variants of European culture – Anglo-Saxon, Italian, German, Irish and Jewish. The vast majority of current immigration is Hispanic, involving people who speak a European language, and whose Catholic religion has for centuries been a major American religion. And they come from countries such as Mexico which, while poorer than the US, are functioning middle-income market economies, extensively influenced by American culture and attitudes. Meanwhile, the only American experience of mass immigration from a completely different culture – involuntary black immigration
Demographics, Economics and Social Choice 159
via slavery – has left a legacy, even 140 years after the abolition of slavery, of very incomplete integration, with, for instance, measures of black/white interracial marriages or mixed housing districts lower than in Britain. American does not provide us with a proven success story of the rapid integration of very large numbers of immigrants coming from very different cultural and economic starting points. But that is what Europe would have to achieve in order to fix the pension challenge via immigration. Its immigrants would not come from the nearest European cultural and economic analogue of Mexico; they will not come in any significant numbers from Eastern Europe, or even from Russia or the Ukraine, because the populations of these regions are small and falling rapidly – Eastern Europe’s population is likely to fall from 126 million today to 104 million in 2050. The immigration of even ten million people from Eastern Europe would be a vast demographic shift in Eastern Europe, leaving a social catastrophe of unsupported pensioners, but making almost no difference to Western European support ratios. Instead, to make any significant difference to support ratios, Europe needs mass immigration from Africa and Western Asia, here defined as Pakistan, through the Middle East to Turkey – regions which, with the notable exception of Turkey, sadly include the world’s greatest concentration of very poor people, of failed states, and of fundamentalist regimes and cultures. The economic and cultural distance between Europe and its major potential source of immigrants is far greater than between the US and Latin America. That does not answer the question of what level of immigration would be optimal. It just says that mass immigration is bound to be a more difficult political issue and a more difficult integration challenge for Europe than for the US. My second set of issues is environmental. Here the issue is not whether we want immigration, but whether we want population growth – via more immigrants or more babies. As Chairman of the Pension Commission, I would find population growth very helpful. As an individual concerned about our environment and as a lover of the English countryside, I would like population stability in the UK, in Europe and eventually across the world. We cannot permanently solve global environmental challenges – such as increasing demands for fresh water – unless at some time we reach population stability. In south-east England, another ten or even five million people means a destruction of countryside and a level of traffic congestion
160 Adair Turner
that arguably would degrade the quality of life quite as much as would a decline in the support ratio – and all for only a temporary solution to the problem. Because, if you keep the ratios constant with 30 per cent population growth up to 2050, you need another 30 per cent (another 25 million people in Britain) to keep them stable up to 2100, and again every 50 years thereafter. So, if at some stage we shall have to design a society which works with column-shaped demography, why not set about the problem with a population of 60 million, rather than in 50 or 100 years’ time with a population of 80 or 100 million? There could, therefore, be ‘quality of life’ reasons for Europeans to oppose population growth and immigration. This could be the welfare-maximizing choice. But, of course, that is not the choice that the US appears to be making. In the US – as in Europe – fertility rate trends would fairly soon produce population stability in the absence of immigration. But the population of the US is predicted to rise by 30 per cent over the next 50 years, almost entirely due to immigration and its knock-on effects. So why is the US different? One very simple explanation is population density. US population density, even excluding Alaska from the calculation, but keeping the expansive wastes of northern Sweden and Finland in the European comparison, is one-third that of the European Union (see Figure 7.10). Even what we think of as the densely populated parts of the US are
117
European Union US (excl. Alaska)
35
England
380
NY, PA, CT, MT, NJ
164
Florida
106
California Texas Arizona
81 29 16
Figure 7.10 Population density: US and Europe (000s per sq. km, 2000) Source: Government Actuary’s Department.
Demographics, Economics and Social Choice 161
far less densely populated than Europe. England’s population of 50 million is roughly the same as the number living in five of America’s more densely populated north-eastern states – New York, Pennsylvania, Connecticut, Massachusetts and New Jersey. But the population density of those states is 40 per cent of England’s. Further, it is not in those states that most population growth is now occurring, but in still less densely populated states like Florida, California, Texas and Arizona. Two-thirds of all US population growth over the last 20 years has been concentrated in nine southern and western states with an average population density a third of Europe’s and a tenth that of England or the Netherlands. A similar proportion of population growth is forecast for those states in future. That changes the politics and indeed the rational economics of immigration quite fundamentally. Americans who do not like the changing ethnic complexion of their neighbourhood, or Americans who just want more space, still have the option of moving to the suburbs being carved out of greenfield land on the edges of Atlanta, Dallas, San Antonio, Phoenix and Denver. Americans can still get the support-ratio benefits of immigration without as many integration difficulties and without local environmental pressures. That, rather than the inherent racism or conservatism is, I suggest, why the politics of immigration are quite different in Europe than in the US. Different population densities also make the economic impact of population growth different. For another issue raised by this population growth debate is the economic costs of population density, an issue that I believe receives insufficient attention in our comparison of US and Europe economic performance. The USA has a significant productivity advantage over Europe: that productivity advantage has increased since the mid-1990s and has been called the American ‘productivity miracle’. But detailed analysis does not reveal a generalized efficiency advantage across the economy; instead almost all the difference is concentrated in specific sectors where expansive physical layout and ease of transport flows can have a big influence on productivity – that is, in retailing and wholesale distribution. The most striking studies suggest that 60 per cent of the entire difference between US and European productivity in the last ten years is explained by the huge productivity improvements achieved by new US retail formats (the big out-of-town sheds) opened on large greenfield sites by Wal-Mart, Home Depot and the others.
162 Adair Turner
Greenfield developments are far easier in the US due to more plentiful supply of land zoned for commercial use. This has led some to argue for looser planning restrictions in Britain to unleash productivity growth. But our tighter planning controls are a natural response to population density. Welfare-maximizing planning rules trade off the potential productivity benefits of easier and more expansive development against consumer preferences for preserved countryside, less noise and less congestion. Almost any specification of a rational utility preference function would suggest that the higher the population density, the higher the value consumers place on those benefits. Such a value, by the way, has a concrete economic expression in the way that house prices vary according to location. This means that the optimal trade-off is bound to entail tighter controls in a more densely populated country and that densely populated countries should sacrifice the final few points of productivity potential in order to preserve environmental benefits. This suggests some internal contradictions within government assumptions and desires. The UK government has increased its sustainable growth estimate to reflect higher immigration; it is interested in easier planning rules to enable higher productivity; and it is worried about the high and rising level of house prices. But, if the population rises further, public demands for tight planning can only increase and house prices will go up further. Even in purely economic terms, we face more complex issues and trade-offs than support ratios alone. Finally, a comment on geopolitics. The implication of what I have said on culture, the environment and economics could be that the rational welfare-maximizing choice for Europe would be a population growth rate slower than the US and, therefore, a lower rate of immigration. Europe might be a happier place with a stable population, even if that means that either worker or pensioner consumption has to be lower than in the alternative growing population case. But, even if this were a rational choice in terms of individual welfare, it would clearly have consequences for geopolitical power, since, given equivalent productivity growth, US total GDP would grow significantly faster, in line with faster population growth. The continued rise of the US geopolitical and military power – relative to Western Europe and to Russia – is almost bound to continue. This might, in Robert Kagan’s words, be another case of Europe choosing Paradise over Power, a still wider set of issues, which I shall have to leave aside for now.8
Demographics, Economics and Social Choice 163
Increasing birth rates? If not immigration, what about increasing fertility? One problem is that higher fertility solutions carry the same total population implications as immigration. Indeed on average they require slightly larger total populations than immigration because of children. Under both, you need a permanently expanding population, because workers age into retirees and then you need more workers to support them. But at least immigrants can enter the country as workers, rather than having to grow up first. That effect aside, immigration and increased fertility are the same in terms of their implications for permanent population growth. This in turn implies that, if you do not want to keep support ratios constant by immigration because of environmental concerns, you should also be as opposed to driving fertility back up above replacement levels, even if you could actually achieve it, which you may not be able to. There has been no late twentieth-century case of a country that has fallen below replacement fertility and then risen again above it. This suggests that, while the immigration-based route to maintaining stable support ratios is debatable but at least possible, the fertility-based route is impossible. But there could still be a good argument for trying to increase fertility rates at least somewhat in Europe and also, for instance, in East Asia. If environmental considerations argue for population stability, they do not require population decline. And when you get actual population decline, the support-ratio mathematics become so extreme that it is difficult to imagine any combination of retirement age change, and increased worker taxes or savings, that can prevent large falls in pensioner living standards. Some European countries, in particular Italy and Spain, are presently on target to move not from demographic pyramids to columns, but to what we might we call a sort of beehive (see Figure 7.11 for Italy). There could therefore be major economic benefits if European birth rates returned not to the 2.5s or 3.0s of the 1950s, but up towards replacement rate, rather than staying at the much lower levels we see today. The good news is that, if that is feasible, it almost certainly depends on enabling women to make the choices they want to make rather than cajoling them to become hero mothers of the European Union as their contribution to solving our pension crisis. For it is a
164 Adair Turner
Age band
1970
2000
80–100
0.3 0.6
0.8
1.5
60–80
3.4 4.4
5.1
6.5
40–60
6.1
6.6
20–40
7.6
7.5
0–20
8.8 Male
7.4
8.9
8.3 Female
5.8 Male
2050 2.3 3.9 5.5
6.5
7.6
5.3 5.1
8.6
4.4 4.0
5.5
4.1 3.8
Female
Male
Female
Figure 7.11 Italy’s population structure, 1970–2050 (millions) Source: UN Medium Variant for 2050 projection.
Table 7.6 Fertility intentions women: England and Wales Woman aged
18–20 21–23 24–26 27–29 30–32 33–35 36–38
of
Average intended family size 2.08 2.16 1.98 2.02 1.97 1.97 1.90
}
GAD forecast of long-term trend ⫽ 1.74
striking feature of several surveys that, when you ask women in rich developed countries how many children they expect to and would like to have, the answer (see Table 7.6) is usually higher than the present actual levels and, as it happens, is close to two. In practice, what happens is that a variety of barriers – difficulties of taking career breaks, delays in forming families driven by economic considerations – prevent women (and in particular graduate professional women) from meeting those aspirations.
Demographics, Economics and Social Choice 165
TFR – children per woman France Denmark Finland Netherlands Belgium UK Sweden Portugal Greece Germany Spain Italy
1.90 1.74 1.73 1.69 1.65 1.63 1.57 1.42 1.29 1.29 1.25 1.24
Figure 7.12 European fertility rates, 2001 Source: GAD.
This implies that the modern natalist strategy, far from being one that attempts to return women to the traditional limits of hearth and home, is one which better enables them to combine working and family life, careers and children. This hypothesis is confirmed by the fact that across Europe (see Figure 7.12), while all countries have fertility rates below two, the lowest are in those countries – Italy, Spain, Greece and Germany – where levels of female participation in the workforce are low and where aspects of social organization, such as shop opening or school hours, have not been well designed to make participation easy. Perhaps we should have a strategy to encourage European birth rates to rise. If so, it must be one that reinforces rather than undermines the changing role of women, which has been in the past the driver of the fall in birth rates from historically high levels.
Summary and some implications Let me sum up and suggest some implications for public policy debate. Demographic change faces us with social choices far more fundamental than those between different pension systems. Indeed there are still wider social and cultural issues which I have not here considered at all, such as whether an ageing and low-fertility society is less innovative or entrepreneurial. But from what I have considered I draw out these key points.
166 Adair Turner
First, we probably shall experience, eventually throughout the world, increased longevity and, at most, replacement fertility rates. Overall that is good news, because people will live longer and because the world population may stabilize. Second, however, this poses a severe problem for pension systems, which have always previously operated within ‘pyramid’ demographics. All our inherited pension systems are to a degree pyramid schemes, their economics of contributions and receipts only sustainable because until now each successive generation has been larger than the one before. When the population stabilizes, the pyramid scheme comes to an end. That is a big problem that cannot be fully offset even by proportional rises in retirement age. Third, attempts to get round this problem by investing overseas or by encouraging immigration can only be transitional solutions and, in the case of mass immigration, have wider political, environmental and economic implications, which we need to debate. Fourth, therefore, we shall not be able to avoid changes in the pattern of intergenerational transfers, either by significantly increasing worker contributions or savings, or by accepting less prosperous pensioners. These adjustments could occur either as a result of public policy decisions, or via market forces encouraging individuals to adjust life-cycle consumption-smoothing decisions. Fifth, however, there are policies which clearly make sense, since they help limit the deterioration of support ratios. These policies are labour market-focused. We need to increase average retirement ages, implying the need for careful thinking about the tax and pension system changes which would encourage this and also about the labour market policies (anti-age discrimination, training, flexible working) that would make it possible for older people to be valued employees. We need to encourage high levels of employment participation by the working-age population, because that too contributes to better support ratios. But we also need women participating in the workforce to be able more easily to combine that with having children. This all implies an approach to labour market regulation, for instance to maternal and paternal leave rights, that recognizes the legitimate role of some regulation to support essential social objectives, while still preserving the flexibility which we know is required for high employment levels.
Demographics, Economics and Social Choice 167
Above all, we need to start debating the full set of social choices we face, for they are much more far-reaching than any details of pension system design.
Notes 1. David Willets, Old Europe? Demographic Change and Pension Reform, Centre for European Reform, 2003. 2. The Economist, 27 September 2003. 3. ‘The Macro-Economics of Pensions’, 2 September 2003 (available on www.pensionscommission.org.uk). 4. See US National Vital Statistics Reports, Vol. 51, No. 4, Figure 3. Interestingly, Canada, which also has significant immigration but with a greater proportion from Asia-Pacific and a smaller Hispanic proportion, had a TFR of 1.48 in 2000. 5. P. Samuelson, ‘An Exact Consumption-Loan Model of Interest, with or without the Social Contrivance of Money’, Journal of Political Economy, Vol. 66 (1958), pp. 467–82. 6. Note that average retirement ages can rise even if ‘official’ or ‘typical’ retirement ages do not, via an increase in employment rates among, for example, 50 to 60 year olds. Increases in average retirement ages may well be accompanied by an erosion of the idea that there are specific retirement ages at which people switch from full to nil employment. 7. ‘The Macro-Economics of Pensions’, 2 September 2003 (available on www.pensionscommission.org.uk). 8. Robert Kagan, Of Paradise and Power, New York: Alfred A. Knopf, 2003.
8 Europe: Pillar of the World Economy or Just an Appendix? Norbert Walter
The title of my lecture requires two warning notes. First, Europe certainly is not now (and during my lifetime will not be) a pillar of the world economy. But neither will it be an appendix. Second, you should be warned that I am quite emotional on the subject of Europe. In my long life as an economic forecaster, I have learned that there is no such a thing as an optimal distance to an object. You are either too close and cannot see, or you are too removed and cannot see. I am emotionally attached to Europe and, therefore, I am not an objective observer. That is how I was brought up. To explain what I mean, both my grandfathers, aged 25 and 27 respectively, died and are buried on the battlefield of Verdun. My father fought against the French and the Russians and came home disabled after years of imprisonment. I am the first member of my family who has lived for almost 60 years in peace and growing prosperity as a consequence of Europe becoming a different place. I am in this position because of great political figures who understood that we must not repeat the past. Today, no one understands the de Gasperis, the Henri Spaaks, the de Gaulles and the Adenauers. They have no followers or successors. The idea of ‘Europe’ as an ideal no longer exists and most people in European countries, including Germans, could not care less about it. If you held a referendum in my home country today on the euro, you would have a clear majority in favour of getting rid of the damn thing. The European Coal and Steel Community, the European Economic Community, the European Commission and the infant European Parliament have given us structures that can prevent the derailing of 168
Europe: Pillar or Appendix of the World Economy? 169
the whole process of European integration. But, if you polled the people in the EU member states today, you would never get a mandate for such European integration. In this case, you might well ask, why is it that in May 2004 no less than ten successful applicant countries joined the European Union, making it a club of 25 members? Most of these applicant countries held a referendum before entry, where high turnouts produced a clear majority in favour of joining. If at the same time in Western Europe you had asked why these countries (Hungary, Poland and the rest) voted to join the European Union, the populist answer would have been that they want our money. The reality, however, is that citizens in existing EU countries live in nation states where their government’s share of national income is close to 50 per cent. The EU budget amounts to between 1 and 1.27 per cent of GDP in these member countries. Despite this, the perception of the average European citizen is of Europe as a bureaucratic monster with huge demands. The average citizen in an existing EU member country believes that we shall be financing big transfers and subsidies to these new members. This is, of course, nonsense for that very reason that for each country the EU budget represents only around 1 per cent of its national product. On top of this, more than 50 per cent of the European budget is just for the Common Agricultural Policy, which is helping French, German and even some British farmers, but will not be of any obvious help to, say, the Baltic states. This very day [29 January 2004] we have begun preparations for the fourteenth Frankfurt European Banking Congress, which will take place as always on the third Friday in November. I suggested as possible headlines ‘Europe stuck’ or ‘Europe – heavyweight with no heartbeat’. My colleagues criticized me for being too negative. So the theme will probably be ‘Goodbye Lisbon’ and we shall probably invite the reelected President Putin as the guest of honour, who will be able to greet us with ‘Goodbye Lisbon! Hello Moscow!’ In short, Europe seems to be more attractive from the outside than if you are part of it. The last three major figures with a forward-looking European agenda were Jacques Delors, the President of the European Commission and the engine behind the Single Market programme and European Monetary Union, François Mitterrand and Helmut Kohl, who together provided EMU with their political weight. Without these three people, Europe would not have developed the
170 Norbert Walter
momentum in the late 1980s required to integrate Greece, Spain and Portugal. Greece became a member of the EU in 1981, and Spain and Portugal joined in 1986. Back then there were questions about whether these countries – non-democratic by tradition, far behind in economic terms and showing no real economic convergence – would dilute the European idea. As it turned out, they gave Europe a new momentum. In the same way, I welcome the eagerness of the new applicant countries to join Europe. Almost all of them are countries that only regained their national sovereignty from the Soviets at the end of the 1980s. For them to be willing to give up part of their national sovereignty so soon again is something that we, in the older democracies, should understand as being bold and politically difficult. If you look into what has happened in these applicant countries in terms of democratic institutions, market orientation and international integration, something truly remarkable has taken place. Where do we go from here? We are clearly a long way behind the Lisbon agenda. In 2000, the heads of government declared that by 2010 Europe should be the leading knowledge-based society of the world. When I read this, I thought that it must be April Fool’s Day because I never believed that they could come up with such a bold statement. I was happy with the content, but the boastful quality was American. It was almost like the 1980s Avis rent-a-car slogan, ‘We try harder because we’re second.’ Unfortunately, after this declaration of a target, little or nothing happened. The 2004 review found that we were still far behind the US and getting nowhere. After the ringing declaration at Lisbon the process of accepting an EU constitution ran into a stone wall of objections, particularly from Spain and Poland. We may get there in the course of 2004 but as of now we cannot even start the process of ratification. It is essential that we should get the correct answer to the basic issue of voting rights, if we are to win over the people of Europe to a revitalized project. If we want democratic legitimacy for this European entity, the European Parliament cannot possibly deviate far from the democratic rule of ‘one man one vote’. We cannot possibly have a representation in the European Parliament that gives a person in Luxembourg a vote worth eight times the vote of a person in Germany. On the other hand, it is equally obvious that Europe can only develop healthily if there is an acceptance that minority rights also
Europe: Pillar or Appendix of the World Economy? 171
have to be taken seriously. Therefore, we need representation of the interests of the smaller countries in something like a second chamber, with membership based on states. It is obvious that old Europe – the Europe of the six, the nine, the fifteen – benefited greatly from the contributions and influence of the smaller member countries. The big guys had to respect the right of the small guys. So the double voting system suggested for the new constitution is, in my view, an appropriate solution to this problem. The Treaty of Nice, which entered into force in February 2003, is the current basis for EU voting rights and the mechanisms by which coalitions can be formed to get things done. After the enlargement, the voting system of the Nice Treaty will no longer be a realistic way for Europe to move forward. Even when we were a union of 15 progress was blocked by this awkward voting mechanism. It will be even worse when we are a union of 25. By 2010 we shall be a Europe of 30. I am not even talking about Turkey’s application to join the EU. I am thinking here of the countries of the former Yugoslavia, which in my view will join in the wave after Romania and Bulgaria. Some optimists believe that either the Irish presidency in the first half of 2004 or the Dutch presidency in the second half of the year will achieve the necessary compromises and overcome the deadlock that we ran into with the Nice Treaty. Others claim that the draft convention as a whole is not appropriate and that it has to be adjusted at so many points that it would be better to use the time we have to come up with a new and better draft. For myself, I have a number of important criticisms of the existing draft and would be happy to see changes. But I am a practical man and I know that, once we open a Pandora’s box, it will be difficult to close it again. So I would rather accept most of the basic elements of the current draft than risk delaying the whole reform process for years. There are many who would prefer Europe not to develop on the basis of the Nice Treaty as a whole, but just on the basis of one element in it, namely the right of a qualified number of EU members to take the initiative to move forward at a different speed from the rest. This idea of a two-speed, or multiple-speed, or variable-geometry Europe comes up regularly. I am suspicious of it. It strikes me that those who are strongly in favour of this approach are, in fact, not interested in moving Europe forward at all. Consider the effect on Europe if each year one more
172 Norbert Walter
member were to join the Schengen Agreement. Since this would result in a necessity to adjust the maze of immigrating and emigrating passenger pathways through passport control and customs on a yearly basis, the result would be to maximize employment in the construction and equipment industries at airports. Such a programme would be about as smart as the recent Council decision to have the three committees for regulating financial markets in three different places (in Paris, Frankfurt and London, respectively) when all the people they must talk to and coordinate with are located in Brussels. This is a decision that will boost the revenues of airlines and international hotels. It will maximize the time experts spend in the air rather than in addressing the real issues. Such initiatives are evidence that simple national and parochial interests dominate. They are attempts to ruin the idea of Europe, not attempts to find solutions to its problems. Is there any sign of a new European leadership on the horizon? Where might we find the first permanent President of the European Council with the required qualities? I have to report that, as far as Germany is concerned, as of February 2004 we still have no idea how we should be represented in the next Commission, nor have we begun to form the coalitions with other European partners in order to achieve solutions that we consider appropriate. My hope is that the new members from Central and Eastern Europe will show a strong interest in getting many of the important posts in the Brussels Commission and other institutions. I have been lucky enough to meet many people from the accession countries and I admire their talent and their motivation to get things moving. I think it is very likely that we ‘old Europeans’ will be put under pressure by the quality of the people that we will soon have in European institutions from these new countries. I hope that these new people, with their fresh ideas and motivation, will criticize the old guys defending the status quo. In addition to their qualities as professionals, I sincerely hope that they will be talented communicators who can help reinject motivation and spirit into Europe. I warned at the outset that I am emotional about Europe. But it is, of course, a fact that the world’s centre of gravity in terms of political, military and economic power is the United States. The underpinnings of America’s superpower status are fully intact. The US has a birth rate of 2.1 per childbearing woman. The rate for
Europe: Pillar or Appendix of the World Economy? 173
the existing European Union is 1.4 and for the accession countries about 1. Therefore over the next decades the ageing of Europe’s population will be much more pronounced than the ageing in the United States. This implies that research and the implementation of innovation is far more natural in the United States than it is in Europe, old or new. In addition, the United States enjoys another fundamental underpinning to its economic dynamism, its entrepreneurial orientation. In the US, 40 per cent of young people want to become selfemployed; in Europe the comparable figure is about 25 per cent. This different attitude to entrepreneurship is supported in the United States by a creative venture capital market that provides the necessary risk capital. We Europeans, instead, prefer time deposits or, if we venture down the road towards higher risks, we buy bonds, preferably government bonds. This makes it much easier to provide the restructuring necessary to respond to technological change in the United States than in Europe. If all this is true, however, why did I not emigrate to the US 40 years ago, instead of becoming Chief Economist of Deutsche Bank? The answer is that: (a) I do believe that there is still a European potential; and (b) there is a need for Europe to respond, because a monopoly is never in the interests of the consumer. The American superpower needs a challenge. I do not see any region of the world capable of becoming a challenger to the United States between now and 2020 other than Europe. The next kid on the block could be China. But China has not even yet got a convertible currency or effective financial markets or a risk capital market of any kind. Nor has China got a democracy or an educational system that is up to standard. However, it is clear that, if China continues to develop in a sustained way, it could join Europe as a challenger to the US by about 2020. Why do I say that Europe has the potential to challenge the United States? It is because we Europeans underutilize our capacities to such a degree that, if we used them to the full, we could be a formidable contender to the United States. European education systems, particularly on the Continent, keep students far too long in school. We do not start elementary school until age seven. We waste a lot of time in secondary and tertiary education. In our universities people obviously have fun rather than work hard. I am not in the least against
174 Norbert Walter
people enjoying life, but there is no reason why this (or going to the opera) should be subsidized by the state. First, we need incentives to do better in education. We should start joining the labour force much earlier. I am not arguing for less time or resources to be devoted to education, but in an age when technological change is ever faster and the lifetime of any information technology is ever shorter, why is the first basic phase of education always getting longer? (This observation is, of course, truer for Continental Europe than it is for the UK.) Second, female participation rates in many of the European countries leave a lot to be desired. For the last 30 years we have been giving girls the same quality of education that we give to our boys. Some new studies even indicate that we are now giving girls a better education than boys, particularly in elementary and secondary schools where the teachers are predominantly female and male role models for boys are fewer. Still, the labour market participation rates for women in European countries are way below US levels. In Germany, with reunification in 1990, we inherited a high participation rate of women in the eastern part of the country. However, we then incentivized women to leave the labour force with unemployment benefits that were much too high and other inappropriate systems. Third, Europe has embarked upon a policy of reducing the working week that has led to intolerable results. The average Continental European works about 1,600 hours a year, as against about 2,000 hours a year for the average US worker. We incentivize early retirement by public pension systems that are impossible to sustain. Thus on all fronts Europe has made mistakes that cause labour input to be low. We ought to put in place incentives to start schooling earlier, to enter the labour market sooner, to work longer hours and to retire later. We should have government incentives and more intelligent management styles in companies to allow a fuller participation of women in higher-skilled as well as in lower-skilled areas. This is particularly important with respect to women because our economic structures are developing away from the military, agricultural and manufacturing industries towards the service industries and into internationalization. Women are much better in services, marketing, sales and languages. Just imagine how successful the Japanese would be if they allowed women to speak internationally and to be their marketing executives. Japanese men still have
Europe: Pillar or Appendix of the World Economy? 175
difficulties speaking foreign languages, while Japanese women are completely different. In order to achieve what we want, we have to change our tax structures. Slovakia, which has applied to join the European Union, has just introduced a tax system that I admire and that I believe will become a benchmark for Europe. It involves different taxes but only one tax rate. The rate is 19 per cent for value added tax, for income tax and for corporation tax. This policy is obviously not good for tax advisers and accountancy firms, but it releases people of talent to concentrate on doing something more positive and market-orientated than creating the complicated tax and auditing systems that we are all used to and suffer from. So, the expanding Europe will bring us new benchmarks. Look how we all benefited from Ireland joining the European Union. It moved from being a country of emigration to one of immigration; from being a country where there was a shortage of risk capital to being one of the places where risk capital became available from all over the world, including from Chicago and from those who had emigrated from Ireland to Chicago. I am convinced that Europe has the capacity to become a locomotive in the world economy. If we could develop such momentum, then some of the disequilibria that plague us today could be at least reduced, if not eliminated. The most important of these disequilibria are the twin deficits of the United States, which are not sustainable: the current account deficit of 5 per cent of GDP and the budget deficit of 5 per cent of GDP. If the United States addresses this problem after a presidential election that – as I expect – will be won by George Bush, fiscal policy will move from being expansive to being neutral and monetary policy will move from being expansive to being restrictive. (If the Democrat candidate wins, fiscal policy is likely to be moved more forcefully from expansionary to restrictive.) If, then, the United States ceases to be the locomotive for world economic expansion, will the world fall back into stagnation, recession or even full-blown deflation? Who could compensate for the absence of American momentum? China has potential and I forecast that its growth rate will stay at about 10 per cent a year between now and 2008. But the Chinese economy, even today, is still only one-fourth that of Japan. So China does not have the weight to
176 Norbert Walter
compensate for deceleration in the United States. It will rest upon Europe to take up the role of the locomotive. Is this a realistic scenario for Europe? Does the European Central Bank have enough room to pursue expansion by lowering interest rates when its base rate is already down to 2 per cent? Can fiscal policy become sufficiently expansionist in Europe? The main problem here is that France and Germany have unfortunately pursued socialist and protectionist policies for too long, leading to unacceptably high levels of government debt. They are not in a position to pursue expansionary fiscal policy. It is not just the Stability and Growth Pact that stands in their way. Economic logic also argues against a more expansionary fiscal stance by France and Germany. Against that, there is every reason to have structural policies to deregulate the labour market, to have more incentive-orientated tax policies than the ones that exist today and to correct an excessive burden that results from too expansive healthcare and old age provision systems. These are the issues that must be addressed and the European Central Bank is correct to stress this point again and again. In my own country today, the Social Democratic Party is in disarray and has lost public support because it has pursued policies that its supporters regard as ‘neo-liberal’. Supply-side policies, designed to deregulate the labour market, to reduce subsidies for small-scale industries and to reduce transfer payments to the elderly, are deemed to be a political betrayal. If this judgement becomes dominant, then Europe will lose its only available instruments for reinvigorating the European economy. It will be the end of the Lisbon agenda. Europe: pillar or appendix of the world economy? If there is no revolution, we shall become the appendix. We need a revolution to move Europe forward. If you are interested in attractive inner cities and in the quality of life, then we need to work together to change policies, not gradually or incrementally, but in a revolutionary way in order to get this continent going again before it is too late.
Further reading Deutsch, Klaus Günter and Norbert Walter (2004) ‘Mehr Wachstum für Deutschland’, Die Reformagenda, Frankfurt: Campus Verlag. Walter, Norbert (1991) ‘Changes in the Global Economy: Policy Consequences for Europe’, in W. Weidenfeld and J. Janning (eds), Global
Europe: Pillar or Appendix of the World Economy? 177
Responsibilities: Europe in Tomorrow’s World, Gütersloh: Bertelsmann Foundation. Walter, Norbert (1991) ‘Immigration: Policy Options for Europe’, in W. Weidenfeld and Joseph Janning (eds), Changes in the Global Economy: Political Consequences for Europe, Gütersloh: Bertelsmann Foundation. Walter, Nobert and Thomas Hanke (1997) Der Euro – Kurs auf die Zukunft, Frankfurt: Campus Verlag.
This page intentionally left blank
Index
AIDS 142 airline competition 16 Alexander, Karl 114 Algeria 155 Allen, Paul 134–5 alternative minimum tax 87 Angrist, Joshua 94, 117 Apple computers 135–6 Argentina 155 Ashenfelter, Orlen 93, 94 asset prices 47 Austria, employment rates 5, 17 baby boom 142 Bailey, Martin J. 54 Balassa–Samuelson effect 25 Bangladesh 155 bankruptcy 55 Barcelona summit 15, 25 Barro, Robert 100 Belgium 5, 17, 165 Benabou, Roland 83, 86 Bewley, T.F. 59 birth rate 163–5 as support for aged 141 Bordo, Michael D. 53 Borman, Geoffrey 119 Bound, John 98 Brazil 155 fertility rates 145–6 Buiter, Willem H. 46–74 Bush, George 119, 121, 175 Canada 85 capital mobility 32, 33, 34 effect on monetary policy Card, David 92, 110 central banks 40, 62
32, 33
and deflation 68–9, 70 see also European Central Bank central planning 125, 129, 130–1, 132 Chambers, John 116 Chile 155 China 154, 155, 173 fertility rates 145, 146 and GDP 175 Congo 155 contraception 146 crime 101 currency depreciation 40 taxing 57–8 Czech Republic 43 Daschle, Tom 84 de Boer, William 6, 8, 9, 10–11, 14 defence expenditure 28 deflation 46–71 avoidance of 70, 71 cause of 56 central bank cooperation 70 costs and benefits 54 debt 55 defined 46–7 and economic activity 53 and Gold Standard period 53 good and bad 53–4 and mismanagement 70 and policy 56, 62–8, 69 and price and wage rigidities 58–60 demand 62–4, 66–8 aggregate demand model 63 effect of public spending 70
179
180 Index
demand – continued effect of taxes 67, 68 and policy intervention 62–8 demographics and economics 141–67 Denmark 19, 165 employment rates 5, 17 GDP per capita growth 4, 5 hours worked 7 disciplined pluralism 136–8, 139–40 earnings 5, 92–4 economic growth 3 cause of problems 25–6 see also Europe economics and demographics 141–67 education 90–123 attainment over time 104–5 and class size 110–11; benefactors 113–14; durability of benefits 113 competition 91 as consumption 102 and crime 101 as day care 110 and earnings 92–4, 95, 106, 107; and month of birth 94–9; and test scores 107 and Educational Testing Service 106 European systems 173 evaluation of 91 and GDP 100–1, 102–3 human capital 90–1, 102, 103–5, 108, 122 and income inequality 79 and internet 116–19 and literacy over time 107–8, 109 low income families 99, 114 measurement of cognitive ability 105–6; and national income 104; and opportunity cost 108
as priority 15 and public policy 112 returns to 91–2, 102, 103 and socioeconomic status 114–16 Tennessee STAR experiment 112–13 time in school 90, 173, 174 twin studies 93–4 vouchers 91, 116, 119–20 and welfare dependence 101 Egypt 155 Eichengreen B. 39 electricity and Plowden 131–2, 133–4 employment rates 5 energy market liberalization 15, 16 Entwisle, Doris 114, 115 environmental concerns and population growth 159–60, 163 and social choice 15, 25, 26 estate tax 82, 86 Euro 31–43 as challenge to dollar 31, 32, 40 and convergence criteria 23, 25 and price transparency 40 replacing dollar 40–1 trading 42–3 as vehicle currency 41–2 Europe 168–76 airline competition 16 capital mobility 32, 33 competition 1 domestic demand 21 economic trends (1950–2002) 50 economy 1–29, 50; growth 3, 21, 22, 25–6 energy sector 15, 16 and EU voting 171 exogenous shocks 31–2, 34–7 farm subsidies 169 and female participation 174 financial services liberalization 16
Index 181
Europe – continued fiscal union 37–8 free trade 1 GDP 3 inflation 22, 24–5, 43, 48–9, 59 information technology 1 interest rates 24 labour market 8; inefficiency of 26; reform 25, 27 Maastricht Treaty 27, 38 macro-policy problems 25, 26 monetary policy 23–4, 36 monetary union 39 pensions 2; sustainability 26–7 population density 160 and Portugal 170 product market liberalization 2, 15, 16, 25 productivity 6, 13; gap 12, 16, 27–8 Schengen Agreement 172 social choice 2, 6, 15, 25, 26, 28 and sovereignty 170 and Spain 170 taxation 1, 2, 18, 25 unemployment 4 and United States 170, 172 welfare spending 2 European Central Bank 27, 31 inflation target 24, 25 interest rates 43, 176 and Maastricht constraints 38 and monetary policy 34, 35 price stability 43 exchange rates 20, 21 effect on wages and prices 32 risk 39 executive compensation 79–80, 86–7 exogenous shocks 31, 34–7 model of 32–3 farm subsidies 169 Fast Forward programme 117–19, 123
Feldstein, Martin 67 female see women fertility 150 and age of childbirth 147 declining 142, 144–7, 151 intentions of women 164 rates 165; by country 155 financial services liberalization 15, 16 Finland 19, 165 employment rates 5, 17 GDP per capita growth 4, 5 fiscal policy 60–2, 66–8 fiscal union 31–2, 37–8 and monetary union 31, 37, 38 Fisher, Irving 55, 58 foreign exchange markets 62 Hungary 42–3 foreign exchange reserves 41 France 165 employment rates 5, 17 GDP 6 hours worked 7, 9; and productivity growth 6, 10 protectionism 176 retailing 12, 14 socialism 176 unemployment 4, 6–7 free trade 1 Freeman, Richard 121 Friedman, Milton 54, 70, 120 Gates, Bill 76, 134–6 GDP 3, 5–6, 26, 100–1, 102–3 effect of taxes on 18 effect of transport investment on 28 and employment levels 18 Europe and US 7 physical contraints 27–8 population growth 26 and public spending 18–20 and savings rate 18 Germany 10, 18–19, 21, 165 economy 21–2 employment rates 5, 17
182 Index
Germany – continued output per hour worked 9 productivity 11, 125 protectionism 176 socialism 176 unemployment 4 wages 24 see also West Germany Gesell, Silvio 58 Goldin, Claudia 76, 104–5 Goodfriend, Marvin 58 government, legitimacy and authority 126–7 Greece 165, 170 employment rates 5, 17 Hanushek, Eric 110–12 Harmon, Colm 99 Hinton, Christopher 132 Hong Kong, fertility rates 146 hours worked 7, 9, 10 human capital 90–1, 102, 108, 122 and national income 103–5 Hungary 42–3, 169 IBM 135, 136 immigration environment 159–60, 163 numbers required 156 as pension support 141, 156–62 United States 158–9, 160 income effect 64 inequality 75–89 and population share 77 income distribution 78, 86 education and income 79 and effect of public schools 86 as equivalent to growth 88 executive compensation 79–80 historically 75–8 inheritance tax 79, 82 Pareto distribution 76, 77 personal servants 78–9 and population (US) 77
and technological change 79 and wealth distribution 86 India 155 Indonesia 155 inequality 75–89 inflation 22, 24–5, 43, 48–9, 52, 59 anticipated and unanticipated 54 and debt 55 and redistribution of wealth 54–5 see also deflation information technology 1, 10, 135–6 and productivity 12–14 inheritance tax 79, 82 innovation 128, 134–6 interest rates 24, 47 and consumption 64–5 and inflation 54–5 negative nominal 56–8, 66 zero bound problem 47, 54–5 interest and securities 56–7 Iran 155 fertility 145–6 Iraq 155 fertility 146 Ireland employment rates 5, 17 GDP per capita growth 4–5 hours worked 7 Italy 18, 165 currency depreciation 40 employment rates 5, 17 government debt 23 hours worked 7 population structure 164 spending as a percentage of GDP 19 Japan 43, 59 deflation 47, 69–70 economy 47–8 fertility rates 146 labour productivity 8
Index 183
Jencks, Christopher Jobs, Steve 136
106
Kagan, Robert 162 Katz, Larry 104–5 Kay, John 125–40 Kenen, Peter B. 31–45 Kenya 155 Khrushchev, Nikita 130, 132–3, 136 maize and central planning 128–9 Kohl, Helmut 169 Korea 42, 155 fertility rates 146 Krueger, Alan 90–124 Krugman, Paul 75–89 labour costs 18 ease of hire and fire 17 effect on exchange rate 20–1 effect of trade 79 minimum wage 17 mobility; and exchange rate flexibility 33–4; and trade imbalance 33 productivity 3, 8 unemployment benefit 18 wage-fixing 17 labour market 4–5, 59 inflexibilities 20 reform 8, 15, 17, 25, 27, 176 source of problems 17–18, 26 Latin America 43 Lavy, Victor 117 Lee, Fred 131 legitimacy of government 126–7, 138–9 life expectancy 143 Lindahl, Mikael 100–1 liquidity trap 66, 75 literacy and fertility 146 longevity 142, 150, 166 see also pensions
Maastricht Treaty 27, 38 guidelines 22–3, 25 Malaysia 155 market economies 125–8, 134 Medicaid 85 Medicare 85 Meghir, Costas 99 Mexico 42, 155, 158 Microsoft 135–6 minimum wage 17 Mitterrand, François 169 monetary policy 23–6, 36, 60–2, 64–6 and currency union 34 and deflation 46, 55 income effect 64 interest rates 61, 64–6 optimum currency areas 32 substitution effect 64 valuation effect 64 monetary union 31–2, 39 costs of 36 and economic integration 39 effect on trade 40 and fiscal union 31, 37–8 Morgan, J.P. 78 Morocco 155 Mundell, Robert 32–4 nationalization 130–1 Navarro, Peter 116–17 Netherlands 19, 165 employment rates 5, 17 GDP per capita growth 4–5 Nice, Treaty of 171 Nickell, S. 59 Nigeria 155 No Child Left Behind 123 North America, fertility rates 144–5 see also United States Norway, employment rates 5, 17 Olson, Linda 114 O’Mahony, Mary 6, 8–11, 14 output per hour worked 9
184 Index
Pakistan 155, 159 Palme, Marten 99 Panigirtzoglou, Nikolaos 58 part-time working 17 patent legislation 137 pensions 2, 27, 148–67 and age discrimination 166 Bangladesh 155 birth rate as solution 141, 163–5 China 154 Congo 155 demographic change 154 and employment 166 fertility 150, 164; rate by country 155 funded systems 152–4 future options 151 and GDP 153 immigration as solution 141, 156–62 Iraq 155 and longevity 143–4, 147–8, 150, 166 and overseas investment 154–5, 166 Pakistan 155 pay-as-you-go systems 148–9, 152–3 raising retirement age 152 retirement age 158, 166 support ratios 150–2 sustainability 2, 26–7 United States 158–9 Peterson, Paul 120, 123 Philippines 155 Phillips, Meredith 106 Piketty, Thomas 77, 79 planning restrictions 14–16, 162 Plowden, Edwin 131–2, 136 Poland, education 169 policy, structural reform 2 population density 160–1 and environmental issues 159–62
Portugal 165, 170 employment rates 5, 17 power, economic, legitimacy of 127–8 price stability 43 product market liberalization 2, 8, 15–16, 25, 27 productivity 6, 9, 13, 14 by industry 11, 13 and distributive trades 9, 11–12, 14 gap 8–12, 16 growth per hour worked 10 prosperity and fertility 146 public–private sector boundaries 125–40 accountability 129, 139 contracts and targets 138 and delivery of services 126–7 disciplined pluralism 136–40 and legitimacy 126–7, 138–9 Putin, Vladimir 169 Quintini, G.
59
Rachuba, Laura 119 Reagan, R./Reaganism 76 Redish, Angela 53 retailing 11–13 logistics 14 planning restrictions 14–15 Romer, Paul 100 Rouse, Cecilia 117, 119 Russia 43, 155 Saez, Emmanuel 77, 79 Safire, William 121 Samuelson, Paul 148–9 Sargent, Thomas J. 70 Saudi Arabia 155 Schengen Agreement 172
Index 185
September 11th 1 Slovakia 175 Smith, Adam 91, 99 social choice 2, 6–7, 28–9 and demographic changes 141–67 and environmental quality 15, 25–6 Solow, Robert 79 South Africa 155 Soviet Union 128–30 Spain 19, 25, 165, 170 employment rates 5, 17 GDP per capita growth 4–5 Stability and Growth Pact 22–3, 25, 38–9 Stalin 128 Stein, Herbert 87 Svensson, Lars E.O. 66 Sweden 19–21, 99 education 99 employment rates 5, 17 fertility rates 146, 165 GDP per capita growth 4–5 hours worked 7 taxation 1, 2, 18, 25, 176 alternative minimum tax 87 estate tax (US) 84 and inequality 84–5, 86, 88 and monetary policy 66–8 United States 81–7 technology change 79, 174 innovation 128, 134–6 Thatcher, M./Thatcherism 76 pay-offs 88 trade and monetary union 40 trade unions 127 Trend, Burke 131, 133 Turkey 155 fertility rates 145–6 Turner, Adair 1–30, 141–67 twin studies and education 93–4
Ukraine 155 unemployment 4, 6–7, 18 United Kingdom deflation 49–50, 52–3 economic trends (1950–2002) 51 economy 49 employment rates 5, 17 fertility rate 165 hours worked 7 income distribution 78 inflation 52, 59; and bank rate (1817–1914) 52–3; and price level (1800–1914) 52 nationalization 130–1 organizational innovation 138 output per hour worked 9 planning restrictions 162 productivity 8; gap with US 3, 11; growth per hour worked 10 school vouchers 121 United States 5, 49 alternative minimum tax 87 budget deficit 85, 175 economic growth 3; and behaviour 7 and economic power 162, 172–3 economic trends (1950–2002) 49 employment rates 5, 17 executive compensation 86–7 exogenous shocks 37 Fast Forward programme 117–19 federal taxes 82; and regional shocks 37 GDP 3–4, 6 hours worked 6–7 immigration 160 income distribution 77–8, 83, 86 income inequality 75–89 inflation 59 interest rates 47 labour productivity 3, 8, 13–14 and military power 162, 172–3
186 Index
United States – continued No Child Left Behind 123 output per hour worked 9 political influence 83–4, 172–3 population density 160–1 productivity 6, 9, 13–14; advantage 161–2; gap with UK 3, 11; growth per hour worked 10; share of gains 79 retailing 11–14 shocks 1, 36 social choice 7, 89 taxation 81–7; dividend and capital gains 82–4; estate 82, 86; inheritance 79 trade gap 175 unemployment 4 voter participation 83 welfare system 81, 85, 87 zero bound problem 47 Uzbekistan 155 valuation effect 64 van Ark, B. 3, 12–13 Vietnam 155
voter participation 83 vouchers, education 91, 116, 119–22 wages 24 costs 18 minimum 17 Walker, Ian 99 Wallace, Neil 70 Walter, Norbert 168–77 welfare state 75–6, 85, 88 disincentive effect 76 and income inequality 81 spending 2 United States 81, 85, 87 welfare dependence 101 West Germany hours worked 7 labour productivity 8 Willetts, David 141 women and business 174 labour market participation 166, 174 Xavier, Martin Xerox 134–6
100
165,