Income Inequality In Oecd Countries: What Are The Drivers And Policy Options? 9789814518529, 9789814518512

Citation preview

Income Inequality

in OECD Countries What are the Drivers and Policy Options?

8854_9789814518512_tp.indd 1

9/10/13 9:27 AM

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

This page intentionally left blank

b1598-fm

Income Inequality

in OECD Countries What are the Drivers and Policy Options?

Editors

Peter Hoeller Isabelle Joumard Isabell Koske OECD, France

World Scientific NEW JERSEY

•

LONDON

8854_9789814518512_tp.indd 2

•

SINGAPORE

•

BEIJING

•

SHANGHAI

•

HONG KONG

•

TA I P E I

•

CHENNAI

9/10/13 9:27 AM

Published by World Scientific Publishing Co. Pte. Ltd. 5 Toh Tuck Link, Singapore 596224 USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

Library of Congress Cataloging-in-Publication Data Hoeller, Peter. Income inequality in OECD countries : what are the drivers and policy options? / by Peter Hoeller, Isabelle Joumard, and Isabell Koske. pages cm ISBN 978-9814518512 1. Income distribution--OECD countries. I. Title. HC79.I5H645 2014 339.20917'7--dc23 2013027374

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

Copyright © 2014 by World Scientific Publishing Co. Pte. Ltd. All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher.

In-house Editors: Sandhya Venkatesh/Chitralekha Elumalai

Typeset by Stallion Press Email: [email protected]

Printed in Singapore

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-fm

CONTENTS

Disclaimer

vii

Chapter 1. Introduction Peter Hoeller and Mauro Pisu

1

Chapter 2. Mapping Income Inequality Across the OECD Peter Hoeller, Isabelle Joumard, Mauro Pisu, and Debbie Bloch

13

Chapter 3. The Distribution of Labor Income Isabell Koske, Jean-Marc Fournier and Isabelle Wanner

43

Chapter 4. Income Redistribution via Taxes and Transfers Isabelle Joumard, Mauro Pisu and Debbie Bloch

85

Chapter 5. Poverty Mauro Pisu

135

Chapter 6. Top Incomes Peter Hoeller

163

Chapter 7. The Distribution of Wealth Kaja Bonesmo Fredriksen

181

Chapter 8. Conclusion: Growth-Enhancing Policies and Inequality Isabelle Joumard and Isabell Koske

207

About the Authors

223

Index

225

v

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

This page intentionally left blank

b1598-fm

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-fm

DISCLAIMER This book represents an adaptation of the original working papers entitled below. • Hoeller, P., Joumard, I., Pisu, M. and Bloch, D. (2012). “Less Income Inequality and More Growth — Are They Compatible? Part 1. Mapping Income Inequality Across the OECD,” OECD Economics Department Working Papers, No. 924, OECD Publishing, Paris. Available at http://dx.doi.org/10.1787/5k9h297wxbnr-en • Koske, I., Fournier, J. and Wanner, I. (2012). “Less Income Inequality and More Growth — Are They Compatible? Part 2. The Distribution of Labour Income,” OECD Economics Department Working Papers, No. 925, OECD Publishing, Paris. Available at http://dx.doi.org/ 10.1787/5k9h2975rhhf-en • Joumard, I., Pisu, M. and Bloch, D. (2012). “Less Income Inequality and More Growth — Are They Compatible? Part 3. Income Redistribution via Taxes and Transfers Across OECD Countries,” OECD Economics Department Working Papers, No. 926, OECD Publishing, Paris. Available at http://dx.doi.org/10.1787/5k9h296b1zjf-en • Hoeller, P. (2012). “Less Income Inequality and More Growth — Are they Compatible? Part 4. Top Incomes,” OECD Economics Department Working Papers, No. 927, OECD Publishing, Paris. Available at http://dx.doi.org/10.1787/5k9h28wm6qmn-en • Pisu, M. (2012). “Less Income Inequality and More Growth — Are they Compatible? Part 5. Poverty in OECD Countries,” OECD Economics Department Working Papers, No. 928, OECD Publishing, Paris. Available at http://dx.doi.org/10.1787/5k9h28tlt0bs-en • Bonesmo Fredriksen, K. (2012). “Less Income Inequality and More Growth — Are they Compatible? Part 6. The Distribution of Wealth,” OECD Economics Department Working Papers, No. 929, OECD Publishing, Paris. Available at http://dx.doi.org/10.1787/5k9h28t0bznr-en All views expressed in this book are those of the authors alone and do not necessarily represent the views of the OECD or its member countries. vii

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

This page intentionally left blank

b1598-fm

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

1. INTRODUCTION Peter Hoeller and Mauro Pisu

In many Organisation for Economic Cooperation and Development (OECD) countries, income inequality has drifted up over the past decades. It appears that growth has not lifted all income boats to the same extent. In some countries, top incomes have captured a large share of the overall income gains, while income for others has risen only little. At the same time, poverty remains a pressing policy issue, not least because of the adverse effects of the recent economic crisis. There is growing consensus that assessments of economic performance should not only focus on income growth but also focus on the dispersion of income. The Stiglitz–Fitoussi– Sen Commission on economic and social performance, for instance, argued for such a change in perspective. Inclusiveness has also become one of the key issues for the OECD. This book provides new evidence on inequality and its determinants in OECD countries and draws out the policy issues. It builds on previous work in this area — Growing Unequal? Income Distribution and Poverty in OECD Countries (OECD, 2008) and Divided We Stand. Why Inequality Keeps Rising (OECD, 2011a). Against this background and widespread concerns that globalization and technical progress could raise inequality, this book highlights inequality trends and assesses different policy approaches to reduce labor and disposable income inequality. It also looks at the extremes of the income distribution: poverty and top incomes as well as the distribution of wealth. Finally, it investigates complementarities and trade-offs in pursuing redistribution strategies together with growth-enhancing policies. The remainder of this introductory chapter looks at the nexus between inequality, growth and well-being and reviews inequality trends prior and during the recent economic crisis.

1

October 21, 2013

2

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

Hoeller and Pisu

Inequality, Growth, and Well-being Inequality and Growth The well-known Kuznets (1955) hypothesis posits an inverted-U relationship between inequality and per capita income. Inequality widens during the early phase of economic development, then stabilizes, and eventually declines at a high stage of economic development. The main explanation proposed by Kuznets concerns the secular shift from the agricultural to the industrial sector, the latter being characterized by higher average income and higher inequality than the former. Although Kuznets’ curve deals with secular changes in per capita income and inequality, the lack of long time series data on inequality has for long forced researchers to test this relationship using cross-section datasets. The early empirical tests found support for Kuznets’ hypothesis (Ahluwalia, 1976; Campano and Salvatore, 1988; Papanek and Kyn, 1986). However, studies using longitudinal data find no evidence of an inverted-U relationship between inequality and the income level (Anand and Kanbur, 1993; Deininger and Squire, 1998) and a simple scatter plot for OECD countries shows no relationship between disposable income dispersion and gross domestic product (GDP) per capita (Fig. 1.1). Kuznets’ conjecture also spawned a vast theoretical and empirical literature on the broader issue of the link between inequality and growth. The empirical evidence on the impact of inequality on growth has been inconclusive, thus far. De Dominicis et al. (2008) performed a meta-analysis of more than 400 estimates of the effect of inequality on growth and showed that the estimation method, data quality, and sample affect estimates. They concluded that the evidence constitutes an empirical puzzle and that no general consensus has emerged so far. A simple scatter of inequality, redistribution, and growth also shows no link (Fig. 1.2).1 Furthermore, from a theoretical standpoint, the effect of inequality on growth is ambiguous, as different countervailing mechanisms might be at work and causation may 1 The extent of redistribution is affected by several factors. Rodrik (1998), using data from the late 1960s to early 1990s, points to the role of trade openness, as in economies more exposed to external shocks, the government plays a more important risk-reducing role. Yet, Bertola and Lo Prete (2008) and Bertola (2010) argue that although across countries larger government redistribution is associated with deeper economic integration, globalization erodes the capacity of governments to redistribute over time. Alesina and Giuliano (2009) show that preferences for redistribution greatly vary among countries. Differences in religion, culture, and macroeconomic volatility affect them.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

Introduction

3

Inequality in household disposable income 0.5 CHL MEX

0.4 USA

ISR PRT

GBR AUS

ITA

JPN CAN ESP KOR OECD DEU ISL GRC FRA

EST 0.3

POL

HUN SVK

CHE NLD

BEL SWE AUT FIN

CZE SVN

DNK

NOR

0.2 30

60 90 120 Average 2007-2009 real GDP per capita in 2005 PPP USD, with respect to the United States

Fig. 1.1: Inequality and GDP Per Capita. Note: Inequality in household disposable income (HDI) is measured by the Gini index. The Gini index ranges from 0 (perfect equality) to 1 (one individual receives all of the income and the others none). The Gini index of disposable income is for the late 2000s. Source: OECD Household Income Distribution and Poverty Database, December 2012 and OECD National Accounts Statistics (database), February 2013.

run in the other direction: • Inequality can affect growth positively through (Aghion et al., 1999): (i) a higher saving rate of rich people: as the investment rate is positively related to the saving rate, more unequal countries will experience faster growth; (ii) sunk costs and investment indivisibilities: wealth concentration favors the creation of new activities; (iii) work incentives: they are stronger in more unequal societies. • The mechanisms giving rise to a negative relationship between inequality and growth (Perotti, 1996): (i) endogenous fiscal policy: more unequal countries redistribute more, which creates distortions and lowers growth; (ii) sociopolitical instability: large inequalities foster political and social instabilities as more people engage in activities, such as crime and violent protests, which deter investment; (iii) credit market imperfections: because of such imperfections, inequality results in an underinvestment in human capital.

October 21, 2013

4

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

Hoeller and Pisu A. Inequality and Growth Inequality in household disposable income 0.5 CHL MEX

0.4 TUR USA

ISR

PRT GBR

ITA

AUS

JPN

CAN OECD

CHE

0.3

KOR EST

ESP GRC

ISL

NLD

DEU FRA

AUT

HUN

BEL

SVK

CZE

SWE NOR

DNK

POL

LUX

FIN

SVN

0.2 0

1

2

3 4 5 6 Growth of real GDP: 1994-2009 average

B. Redistribution and Growth Redistributive impact of household taxes and transfers 0.16

FIN DNK DEU

SWE

BEL

CZE

GBR 0.12

NOR ITA

FRA

AUT

AUS SVN

OECD

NLD

CAN JPN

0.08

PRT

ISR

EST SVK

LUX

POL

USA ESP

ISL

CHE 0.04

CHL KOR

0 0

1

2

3 4 5 6 Growth of real GDP: 1994-2009 average

Fig. 1.2: Inequality, Redistribution, and Growth. Note: Inequality in HDI is measured by the Gini index. The redistributive impact of taxes and transfers is defined as the difference in the concentration coefficients for income before cash transfers and taxes (i.e. household market income [HMI]) and after cash transfers and taxes (i.e. HDI). The redistributive impact and the inequality measures are for late the 2000s. Source: OECD Household Income Distribution and Poverty Database, December 2012 and OECD Economic Outlook Database, February 2013.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

Introduction

5

Galor and Moav (2004) provide a rationalization of the opposing effects of inequality on growth at different stages of development. Their explanation is based on the emergence of human capital accumulation as the prime engine of growth, replacing physical capital accumulation. In the early stages of the industrial revolution, physical capital accumulation was the most important driver of economic growth: inequality spurred development, as rich individuals had a higher propensity to save than the poor. In the later stages of development, human capital becomes the prime engine of economic growth and large inequalities, by aggravating the negative impact of credit constraints on human capital accumulation, become detrimental to growth.

Inequality and Welfare Besides its impact on growth, inequality can directly affect social welfare or well-being. The relationship between inequality and welfare or well-being has been the subject of intense scrutiny since, at least, the contribution of Pigou (1920) and Dalton (1920). Pigou underlined two channels through which inequality might affect welfare: distributive efficiency and relative versus absolute income. The distributive efficiency argument arises because of the law of diminishing marginal utility: the effect of an additional unit of income or wealth on utility is higher at the bottom than at the top of the income distribution.2 This argument has been extensively explored over the years in the welfare-based inequality evaluation literature with the specification of different social welfare functions, which under certain conditions are positively related to per capita income and negatively related to inequality (Atkinson, 1970; Sen, 1976).

2 Pigou

(1920) stated: “Nevertheless, it is evident that any transference of income from a relatively rich man to a relatively poor man of similar temperament, since it enables more intense wants, to be satisfied at the expense of less intense wants, must increase the aggregate sum of satisfaction. The old ‘law of diminishing utility’ thus leads securely to the proposition: Any cause which increases the absolute share of real income in the hands of the poor, provided that it does not lead to a contraction in the size of the national dividend from any point of view, will, in general, increase economic welfare.” See Layard et al. (2008), for recent estimates of the marginal utility of income using data on self-reported happiness.

October 21, 2013

6

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

Hoeller and Pisu

Individuals’ relative income will also affect well-being — in addition to its absolute level — if income is above the level necessary to satisfy basic needs.3 There is ample empirical evidence confirming that relative income matters. Income and wealth are for this reason two of the indicators included in the OECD’s Your Better Life Index (OECD, 2011b). Also, numerous choice experiments have shown fairness to be a personal motive affecting peoples’ behavior in some circumstances. Feelings of inequity aversion lead people to resist what they perceive as inequitable outcomes (Carlsson et al., 2005; Dawes et al., 2007; Fehr and Schmidt, 1999). The traditional welfarist approach has come under attack. More specifically, different welfare functions involve different value judgments. Choices, therefore, reflect opinions on values which lead to different policy recommendations on how to raise welfare. Sen (1992) has proposed an alternative approach based on functionings or capabilities rather than income. These can be interpreted as the opportunities that people have to achieve their goals. As such it advocates equality of opportunity. Based on this approach, even a high level of income inequality may be justified if this does not result from inequality of opportunity. Yet, inequality in income may also affect inequality of opportunity, if, for instance, access to education, health care, and other services and goods affecting present and future capabilities depend on income.4 Then, income inequality may become entrenched and an increasing share of it will be due to inequality of opportunity. Related to this issue, a large body of research has also investigated the link between income inequality and health, which is arguably related to individuals’ well-being. Wilkinson and Pickett (2010) reviewing 30 years’ research maintain that more unequal societies have worse health outcomes than more egalitarian ones, in addition to more acute social problems such as a lack of community life, violence, drug abuse, and large prison populations. Overall, an assessment of the empirical literature on the link between 3 On

this issue Pigou (1920) wrote “[A] larger proportion of the satisfaction yielded by the incomes of rich people comes from their relative, rather than from their absolute, amount. This part of it will not be destroyed if the incomes of all rich people are diminished together. The loss of economic welfare suffered by the rich when command over resources is transferred from them to the poor will, therefore, be substantially smaller relatively to the gain of economic welfare to the poor than a consideration of the law of diminishing utility taken by itself suggests.” 4 For instance, in the US the No Child Left Behind Act aims at improving the educational opportunities for low-income and minority students.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

Introduction

7

inequality and health for rich countries suggests, however, that the evidence is still inconclusive (Leigh et al., 2009). A meta-analysis of numerous peerreviewed papers (Kondo et al., 2009) concludes that a modest adverse effect of income inequality on health exists, with evidence also pointing to time lags and threshold effects. In addition, data limitations have so far prevented studies from convincingly disentangling the direction of causality between health and inequality.

Inequality Developments Before and During the Recent Economic Crisis The distribution of HDI varies widely across OECD countries. In around 2008, the Gini index ranged from somewhat below 0.25 in Slovenia to 0.5 in Chile (Fig. 2.4 in Chapter 2).5 The HDI distribution is fairly narrow in the Nordic countries, some eastern European countries as well as Austria and Belgium, but relatively high in the English-speaking countries as well as Japan, Korea, Italy, and Portugal. Income inequality is highest in the poorer OECD countries (Chile, Mexico, and Turkey). Since the mid-1990s, the disposable income distribution has widened in 12 countries, while it has narrowed in eight countries. It has come down in some of the poorer countries (Chile, Greece, Hungary, and Mexico), but it widened considerably in the Nordic countries (except in Norway) as well as in Australia, Canada, and Israel. For the OECD on average, there was little change in income inequality. Taking a longer view, income inequality has drifted up. For the 19 countries for which long data series are available, the Gini index rose by 2.1 points between the mid-1980s and the mid-1990s. This is equivalent to a transfer of 4.2% of average income from people below the median income to those above. In the countries that experienced greater inequality, the increase was often driven by rising inequality in the upper half of the income distribution, largely due to rapidly rising incomes at the very top (Chapter 6). So far the effect of the crisis on inequality and poverty has been muted. This suggests that the tax and transfer system has shielded households from the worst impact of the crisis. For most countries, the change in inequality

5 The

Gini index is a popular inequality index. It ranges from 0 (all individuals have the same income) to 1 (one individual owns all income).

October 21, 2013

8

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

Hoeller and Pisu

and poverty since the onset of the crisis has been modest. The correlation between the change in disposable income inequality and relative poverty between 2008 and the latest available year and the change in real GDP is low (Fig. 1.3).6 In Greece, the country most severely hit by the crisis, the rise in inequality was modest compared with the drop in GDP. However, poverty has risen noticeably — by more than one percentage point — but it rose even more in Spain, which suffered a less severe GDP downturn, and Mexico, which weathered the crisis better than most other OECD countries. The same pattern was found in an in-depth analysis by Jenkins et al. (2011). Top income shares have declined since 2008 (Table 1.1). Although detailed data are not available, this drop is likely to be due to lower capital gains and dividend payments, which accrue disproportionately to the richest individuals. The lower top income shares had an equalizing effect but it is difficult to quantify their impact, as the household survey datasets that are used to measure HDI inequality capture top incomes only imperfectly. From a conceptual point of view, the impact of major recessions on income distribution and relative poverty is not clear cut and this may help explain the different developments in inequality. Barlevy and Tsiddon (2006) stress that the impact of the economic cycle on inequality depends on the underlying long-term trends. Recessions tend to amplify long-run trends, that is, they make inequality increase (decrease) even more rapidly when long-run inequality is increasing (decreasing). Overall, inequality is likely to be a complex function of different macroeconomic variables and institutional settings. For instance, lower top incomes during recessions, especially during those caused by financial crises, contribute to declining inequality. However, this shift may be more or less compensated by what happens in the middle and at the bottom of the income distribution, which in turn depends on the cyclical behavior of wages and salaries, unemployment rates, the redistributive policies in place and policy makers’ discretionary responses to the crisis. These intertwined factors make it difficult to reach firm conclusions on the relationship between cyclical changes in the economy and inequality. There has been much debate on whether the rise in inequality in the US was an important factor behind the imbalances that led to the recent crisis. Rajan (2010), for instance, argues that rising income inequality in the US led to a boom in lending to the poor to buttress their consumption. 6 The

same picture emerges using the change in the output gap from the OECD Economic Outlook instead of real GDP.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

Introduction

9

A. Growth and Inequality Real GDP growth 2008-2011, annualized 4

y = -0.1377x-0.5937 R² = 0.0203

POL

3 2

SWE 1

DEU

USA

BEL NOR LUX NLD GBR FIN PRT

0 -1

AUT FRA

CZE

DNK ITA

-2

EST

HUN

ESP

IRL

SVN ISL

-3 -4 -5

GRC

-6 -4

-3

-2

-1

0

1

2

3

4

Change in Gini, 2008 to most recent year B. Growth and Poverty Real GDP growth 2008-2011, annualized 4

y = -0.1639x-0.1726 R² = 0.0178

POL

3 AUS

2

SWE

CAN

1

DEU

CHE LUX

NOR

0 GBR -1

PRT EST

AUT

NLD IRL

MEX SVK USA

FRA

CZE FIN

-2

BEL

DNK

HUN

ESP

SVN ISL

-3 -4 -5

GRC

-6 -3

-2

-1

0

1

2

3

4

Change in poverty rate, 2008 to most recent year

Fig. 1.3: The Crisis and Changes in Inequality and Poverty. Note: The Gini index and poverty rate data are not based on the OECD Household Income Distribution and Poverty Database, as data after 2008 were not available at the time of writing. The data used in these two panels, therefore, are not strictly comparable with those used in other parts of this publication. Source: Eurostat/SILC, Luxembourg Income Study, Australian Bureau of Statistics Surveys of Income and Housing, Human Resources and Skills Development Canada, the US Bureau of Census and OECD Economic Outlook Database, November 2012.

October 21, 2013

10

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

Hoeller and Pisu Table 1.1: Top Income Shares have Dropped. Top 1% income share

Australia Canadaa Denmark Finland France Ireland Italy Japan New Zealand Norway Spainb Sweden United Kingdomc United States

1980

2007

Latest data

Year of latest data

4.8 7.9 5.5 4.3 7.6 6.7 6.9 7.2 5.7 4.7 7.5 4.1 9.8 8.2

9.8 13.7 6.1 8.3 9.3 11.6 9.9 9.6 8.5 8.8 8.9 6.9 15.4 18.3

8.6 12.2 6.4 7.5 8.1 10.5 9.4 9.5 8.2 7.9 8.2 6.9 13.9 17.4

2008 2010 2010 2009 2009 2009 2009 2010 2009 2008 2010 2010 2009 2011

a Data

for Canada refer to 1982 instead of 1980. for Spain refer to 1981 instead of 1980. c Data for the UK refer to 1990 instead of 1980, due to an important break in the series with earlier data. b Data

Source: The World Top Incomes Database, http://topincomes.g-mond. parisschoolofeconomics.eu/, February 2013.

On the other hand, work by Atkinson and Morelli (2011) using a large country sample over a long time span does not find a consistent crosscountry pattern of high or rising inequality prior to banking crises.

References Aghion, P., E. Caroli and C. Garc´ıa-Pe˜ nalosa (1999). “Inequality and Economic Growth: The Perspective of the New Growth Theories,” Journal of Economic Literature, Vol. 37(4), pp. 1615–1660. Ahluwalia, M.S. (1976). “Inequality, Poverty and Development,” Journal of Development Economics, Vol. 3(4), pp. 307–342. Alesina, A.F. and P. Giuliano (2009). “Preferences for Redistribution,” NBER Working Papers, No. 14825. Anand, S. and S.M.R. Kanbur (1993). “Inequality and Development: A Critique,” Journal of Development Economics, Vol. 41(1), pp. 18–43.

April 25, 2014

14:22

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

Introduction

11

Atkinson, A.B. (1970). “On the Measurement of Inequality,” Journal of Economic Theory, Vol. 2, pp. 244–263. Atkinson, A.B. and S. Morelli (2011). “Economic Crisis and Inequality,” Human Development Research Paper, 2011/06, UNDP. Barlevy, G. and D. Tsiddon (2006). “Earnings Inequality and the Business Cycle,” European Economic Review, Vol. 50(1), pp. 55–89. Bertola, G. (2010). “Inequality, Integration, and Policy: Issues and Evidence from EMU,” Journal of Economic Inequality, Vol. 8(3), pp. 354–365. Bertola, G. and A. Lo Prete (2008). “Openness, Financial Markets and Policies: Cross-country and Dynamic Patterns,” CEPR Discussion Papers, 7048. Campano, F. and D. Salvatore (1988). “Economic Development, Income Inequality, and Kuznets’ U-shaped Hypothesis,” Journal of Policy Modeling, Vol. 10(2), pp. 265–280. Carlsson, F., D. Daruvala and O. Johansson-Stenman (2005). “Are People Inequality-Averse, or Just Risk-Averse?,” Economica, Vol. 72(287), pp. 375–396. Dalton, H. (1920). “The Measurement of the Inequality of Incomes,” The Economic Journal, Vol. 30(119), pp. 348–361. Dawes, C.T., J.H. Fowler, T. Johnson, R. McElreath and O. Smirnov (2007). “Egalitarian Motives in Humans,” Nature, Vol. 446(7137), pp. 794–796. De Dominicis, L., R.J. Florax and H.L. de Groot (2008). “A Meta-analysis on the Relationship between Income Inequality and Economic Growth,” Scottish Journal of Political Economy, Vol. 55(5), pp. 654–682. Deininger, K. and L. Squire (1998). “New Ways of Looking at Old Issues: Inequality and Growth,” Journal of Development Economics, Vol. 57(2), pp. 259–287. Fehr, E. and K.M. Schmidt (1999). “A Theory of Fairness, Competition, and Cooperation,” The Quarterly Journal of Economics, Vol. 114(3), pp. 817–868. Galor, O. and O. Moav (2004). “From Physical to Human Capital Accumulation: Inequality and the Process of Development,” Review of Economic Studies, Vol. 71(4), pp. 1001–1026. Jenkins, S., A. Brandolini, J. Micklewright and B. Nolan (2011). “The Great Recession and the Distribution of Household Income,” A report prepared for, and with the financial assistance of, the Fondazione Rodolfo Debenedetti, Milan. Available at: http://www.frdb.org/upload/file/report 1 palermo.pdf. Kondo, N., G. Sembajwe, I. Kawachi, R.M. van Dam, S.V. Subramanian and Z. Yamagata (2009). “Income Inequality, Mortality, and Self-rated Health: Meta-analysis of Multilevel Studies,” British Medical Journal, Vol. 339. Available at: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2776131/ [this is an online publication]. Kuznets, S. (1955). “Economic Growth and Income Inequality,” The American Economic Review, Vol. 45(1), pp. 1–28.

October 21, 2013

12

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch01

Hoeller and Pisu

Layard, R., G. Mayraz and S. Nickell (2008). “The Marginal Utility of Income,” Journal of Public Economics, Vol. 92(8–9), pp. 1846–1857. Leigh, A., C. Jencks and T.M. Smeeding (2009). “Health and economic inequality,” In: Salverda, W., Nolan, B. and Smeeding, T. (eds.), The Oxford Handbook of Economic Inequality, Oxford: Oxford University Press. OECD (2008). Growing Unequal? Income Distribution and Poverty in OECD Countries, OECD Publishing, Paris. OECD (2011a). Divided We Stand. Why Inequality Keeps Rising, OECD Publishing, Paris. OECD (2011b). How’s Life? Measuring Well-being, OECD Publishing, Paris. Papanek, G. and O. Kyn (1986). “The Effect on Income Distribution of Development, the Growth Rate and Economic Strategy,” Journal of Development Economics, Vol. 23(1), pp. 55–65. Perotti, R. (1996). “Growth, Income Distribution, and Democracy: What the Data Say,” Journal of Economic Growth, Vol. 1(2), pp. 149–187. Pigou, A. (1920). The Economics of Welfare, Transaction Publisher, New Brunswick, New Jersey. Rajan, R.G. (2010). Fault Lines: How Hidden Fractures Still Threaten the World Economy, Princeton University Press, Princeton, New Jersey. Rodrik, D. (1998). “Why Do More Open Economies Have Bigger Governments?,” Journal of Political Economy, Vol. 106, pp. 997–1032. Sen, A. (1976). “Real National Income,” The Review of Economic Studies, Vol. 43(1), pp. 19–39. Sen, A. (1992). Inequality Re-examined, Oxford University Press, Oxford. Wilkinson, R. and K. Pickett (2010). The Spirit Level: Why Equality Is Better for Everyone, Penguin Books Ltd, London.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

2. MAPPING INCOME INEQUALITY ACROSS THE OECD Peter Hoeller, Isabelle Joumard, Mauro Pisu, and Debbie Bloch

Introduction and Main Findings OECD-wide, income inequality has drifted up over the past decades. This chapter assesses how inequality measures change when moving from individual labor earnings (ILEs) to adjusted household disposable income (HDI), by successively taking into account other market income sources, the effects of household formation as well as taxes and transfers. It provides country profiles for four countries that summarize the various inequality dimensions and a cluster analysis that identifies groups of OECD countries that share similar inequality patterns. It also touches on inequality issues in the large emerging economies.

Main findings • Countries differ widely with respect to the level of labor income inequality among individuals of working age. Labor income inequality is shaped by differences in wage rates, hours worked, and inactivity rates. • Individual labor income inequality is the main driver of household market income (HMI) inequality, with family formation and the dispersion of self-employment and capital income playing a smaller role. While the dispersion of self-employment and capital income is much wider than that of labor earnings, their contribution to HMI inequality is muted, because their income shares are small in most, though not all, countries. • Taxes paid by households and cash transfers reduce the market household income dispersion by around a quarter on average across the OECD countries. But given the different starting points in terms of HMI distribution

13

October 21, 2013

14

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

and the differences in redistribution via the tax and transfer system, the distribution of HDI still varies widely. • Country profiles have been assembled in diamonds. These allow comparing 24 inequality dimensions with the OECD average and identifying how these inequality dimensions map into inequality of HDI. The country profiles reveal that inequality of HDI does not have the same origins. In some countries, wage dispersion among those working is an important factor (e.g. for the US), while in others, the non-employment rate (NER) and/or inequality in capital income are driving inequality in HDI (e.g. Italy). The country profiles also show that tax and transfer systems as well as publicly provided services (in particular education and health) have a larger redistributive impact in some countries (e.g. Sweden compared with the UK). • A cluster analysis helps to pin down the origins of inequalities. Four groups of countries sharing similar inequality patterns and drivers of inequality have been identified. For example, one group consists of four English-speaking countries (Australia, Canada, New Zealand, and the UK). In this group, a wide wage dispersion and a high share of part-time employment drive inequality in labor earnings above the OECD average. Public cash transfers are largely means-tested, and household taxes are progressive, thus reducing income inequality, though the dispersion in HDI remains above the OECD average. In another group of countries, comprising four Nordic countries (Denmark, Iceland, Norway, and Sweden) and Switzerland, inequality emerging from the labor market is below the OECD average because wage dispersion is narrow and the employment rate is high. Cash transfers tend to be universal and are thus less progressive, as are household taxes. Still, inequality is considerably below the OECD average in this group. • Five large emerging countries (Brazil, China, India, Indonesia, and South Africa) show considerably higher economic growth and higher inequality than most OECD countries. Income dispersion has increased in three of them since the early 1990s: 24% in China, 16% in India, and 4% in South Africa, while it has remained broadly stable in Indonesia and declined by around 10% in Brazil. While income dispersion trends have diverged across these countries, absolute poverty has declined in all of them, owing to rapid economic growth. From the early 1990s to the late 2000s, China experienced the largest drop in poverty among the five large emerging countries. Although on a declining trend, poverty is still high and often concentrated on children and rural populations. The experience of these

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

15

large emerging countries, where non-contributory benefit schemes were introduced over the 1990s, shows that transfer programs, if well targeted, can be effective in reducing poverty and inequality.

Drivers of Inequality According to the Stiglitz–Fitoussi–Sen Report (Stiglitz et al., 2009), HDI adjusted for publicly provided in-kind services should be the focal point when assessing inequality as it is the most comprehensive income concept. Inequality in adjusted HDI is shaped by various factors — originating from the labor market, tax and transfer systems, etc. — which differ across countries and need to be disentangled. Thus, before moving to adjusted HDI, the analysis assesses cross-country differences in the distribution of five main income concepts following the approach of OECD (2011a) (Fig. 2.1): • Individual labor earnings. The dispersion of ILEs (wages and income from self-employment) for the working-age population reflects both the wage dispersion for working full-time employees and the labor income dispersion of other groups forming the working-age population (part-time workers and the self-employed), while the unemployed and people not looking actively for a job have no earnings.

Family formation and composition

Individual labor earnings

Labor market and education policies, migration & gender policies, etc.

Capital income

Household labor earnings

Family policies (child and elderly care), etc.

Fig. 2.1: From ILEs to Adjusted HDI.

Taxes & cash transfers

Household market income

Tax policies (wealth, capital income)

Individual consumption of public goods

Household disposable income

Cash transfers and tax policies

Household adjusted disposable income

Education & health (access and coverage) policies

October 21, 2013

16

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

• Household labor earnings (HLEs). Working-age families may differ in size and composition, affecting the sharing of labor income in households. • Household market income. Capital income complements HLEs, though in varying proportions across countries and also across households within the same country. As the focus of the first three income concepts is on market income, the population covered is the working-age population. • Household disposable income. HDI covers all household and income sources, after taxes and cash transfers. • Adjusted HDI (HADI). This concept is the most comprehensive as it also takes into account in-kind transfers, such as education and healthcare spending. This sequential approach allows identifying the main policy and nonpolicy drivers of inequalities at each income level, as well as their build-up along the chain. Data issues are discussed in Box 2.1 and different income inequality measures are discussed in Annex 2.1.

Individual Labor Earnings1 Countries differ widely with respect to the level of labor earnings inequality among individuals of working age (Fig. 2.2).2 Differences in labor earnings inequality (inequality among those who earn an income from employment) and labor income inequality (inequality among all people in the workingage population) are shaped by cross-country differences in wage rates, hours worked, and inactivity rates. Among the OECD countries, earnings inequality for full-time employees is highest in Chile, the US, and Portugal, while Switzerland, Belgium, and Denmark are the most equal countries. Inequality is generally higher for all the full-time employed, reflecting the wider dispersion of earnings among the self-employed. Extending the analysis to part-time workers, the unemployed and the inactive raises the Gini index, reflecting the large income differentials for these groups and the group of full-time workers (unemployed individuals and the inactive enter the calculation with zero income as transfers are not taken into account). The increase in the Gini index is particularly large

1 OECD (2011a) provides much more detail on the five main income concepts and also discusses changes over time. 2 The policy and non-policy drivers of labor income inequality are discussed in Chapter 3.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

17

for countries where part-time workers make up a sizable share of total employment and for countries with a high unemployment rate and many inactive people of working age. While the Gini indices of the population subgroups are highly correlated (the correlation coefficients are between 0.8 and 0.9), there are several countries, such as Belgium, Italy, and Estonia, for which the choice of the group matters considerably for the inequality ranking.

Gini index,1 15-to-64-year olds, 2008 Full-time employees

Full-time employed

Full-time and part-time employed

Employed and unemployed

Working age population

0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20

Fig. 2.2: Inequality in Labor Earnings. 1. The Gini index ranges from 0 (perfect equality) to 1 (one individual receives all earning). Note: The group of employed individuals includes both dependent and self-employed individuals. The working-age population includes all persons aged 15 to 64 years except for students and persons above the country’s statutory retirement age. The Gini coefficients take into account labor earnings only; the precise definition of labor earnings differs across countries (see Fournier and Koske (2012), for details). Data refer to 2005 for Israel, 2007 for France, Korea, and the US, and 2009 for Australia and Japan. The values for the OECD are calculated as an unweighted average across all OECD countries for which data are available. Source: Panel Study of Income Dynamics (PSID) for the US, Household Income and Labour Dynamics in Australia Survey (HILDA) for Australia, National Socioeconomic Characterization Survey (CASEN) for Chile, Korean Labor and Income Panel Study (KLIPS) for Korea, Luxembourg Income Study (LIS) for Israel, Japan Household Panel Survey (JHPS) for Japan, Swiss Household Panel (SHP) for Switzerland, and European Union EU-SILC for the other countries.

October 21, 2013

18

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

Box 2.1. Data Sources and their Pros and Cons The discussion of labor incomea inequality in Chapter 3 relies on national surveys, while the OECD Household Income Distribution and Poverty Database is mainly used in the other chapters (Ireland is not included in this database). The data issues are similar for both, but the OECD income distribution data have been harmonized.b They are based on national income surveys, but the dataset does not allow access to the original micro-data which constrains the analysis that can be performed. Tax data can also be used when addressing specific issues — in particular top incomes (Chapter 6) — or for providing a long-term perspective (e.g. the Top Incomes Database by Alvaredo et al., 2011). Both tax and household survey data have limitations which are summarized below. Data from household surveys: main limitations • Household surveys tend to be biased at both ends of the income ladder. The richest often fail to respond and, when they do, they tend to underreport their income. The poorest may be too marginalized to respond. Inequality, thus, tends to be underestimated. • Nonresponse rates and misreporting vary across countries. As an illustration, in the EU Statistics on Income and Living Conditions (EU-SILC), the 2008 nonresponse rate exceeded 30% in Belgium, Denmark, Luxembourg, and Norway and stood below 10% in Portugal and the Slovak Republic (Wolff et al., 2010). The nonresponse rate is also often higher for some income components, such as the income from self-employment and capital income (Verma and Betti, 2010). Moreover, comparisons over time are hampered by changes in survey design or income accounting. • The income data refer to cash income and thus exclude imputed components such as home production and imputed rent from owner occupation — the data on adjusted disposable income, which include imputed public spending on health, education, and social housing, being the main exception. • Social security contributions paid by employers are not included, while social security contributions paid by households are, in principle, included. This makes it difficult to draw cross-country comparisons on the size and effect of the tax system based on household surveys. (Continued )

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

19

Box 2.1. (Continued ) • Some income components are not treated consistently in household surveys, distorting cross-country comparisons. For instance, occupational pensions should be treated as capital income according to the OECD terms of reference. In practice, however, they are treated as cash transfers (e.g. for France). Data from tax returns: main limitations Tax data should in principle cover all high-income recipients, that is, provide “census” rather than sample data. They do not suffer from high nonresponse rates for high-income levels. Although tax evasion and avoidance may distort the data, the bias is generally perceived to be smaller than for household surveys since the income data are audited by the tax authorities. Using tax data, the income share of the top 10% richest households is often considerably higher as compared with household surveys. Still tax data also have limitations, in particular: • (i) Under-declaration of income can be significant and varies across countries, reflecting the penalties imposed for under-declaration of income and the efficiency of tax authorities in fighting tax evasion. (ii) Possibilities for tax planning and tax avoidance will affect the amount of income declared in tax returns. (iii) The tax authorities generally only collect information on income that is taxable. Differences in tax codes can thus lead to differences in the concept of income used in different countries. (iv) The tax unit (joint versus individual filing) varies between countries. • Relying on tax data for assessing the income at the bottom end of the distribution may also be difficult since, in some countries, people are not required to declare their income if this remains below the taxable income threshold, and in some countries, if tax affairs are simple, the tax can be deducted at source.

a. ILEs are defined as gross earnings, thus excluding social security contributions paid by employers. For some countries, other types of (Continued )

October 21, 2013

20

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

Box 2.1. (Continued ) compensation are also included such as overtime pay or special allowances for Christmas and holidays. b. See OECD (2008), Table 1.A1.1, for detailed country information.

Moving from Individual to Household Labor Earnings HLEs include the earnings of all individuals of working-age population, whether they work or not, and take into account economies of scale in consumption, which is done by adjusting household earnings for the number of household members with an “equivalence scale.”3 Extending the coverage of the earnings data step by step from individual workers to all people of working age yields the following insights (OECD, 2008): (i) accounting for partnership formation among working household members narrows the earnings distribution, because of the ensuing economies of scale in consumption; (ii) including dependents (e.g. nonworking spouses) in households of workers widens the earnings distribution, with only minor differences across countries; and (iii) including households with no labor income earnings widens the distribution of household earnings considerably and the effect differs across countries, because of the large cross-country differences in the share of people living in households with no earners.

Moving from HLEs to Total Market Income Total market income includes income from dependent work, selfemployment, and capital income. Overall, capital income is more concentrated than labor earnings. Not surprisingly, the property income distribution is as highly skewed as that of the wealth distribution (Chapter 7). What matters most for the strong concentration of both is that higher income individuals save more and thereby accumulate more wealth than poorer households, while they typically also hold riskier and 3 In OECD (2008), equivalization is achieved via the “square root elasticity,” which implies that the needs of a household composed of four people are twice as large as those of a single.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

21

higher-yielding assets. Lower taxation of wealth and of capital income (Chapter 6) has probably also played a role in fostering a more rapid wealth accumulation at the top.

The Contribution of the Market Income Components to Overall HMI Inequality Wages and salaries are in general the main driver of total market income dispersion (Fig. 2.3).4 This market income component accounts for around Working-age population, in the late 2000s 0.6

Wages and salaries

Self-employment income

Capital income

0.5

0.4

0.3

0.2

0.1

0.0

Fig. 2.3: Contributions to Overall HMI Inequality. Note: Contributions to overall HMI inequality are derived by multiplying the concentration coefficients of each income source by their weight in total market income. The data for Greece, Hungary, Mexico, and Turkey are net of taxes. Source: OECD Household Income Distribution and Poverty Database, December 2012.

4 Inequality

in HMI can be expressed as a weighted average of the concentration ratios of market income sources with weights equal to the share of an income source in total market income (Lerman and Yitzhaki, 1985; Shorrocks, 1982). The analysis here and below uses household data from the OECD Household Income Distribution and Poverty Database. As it does not include individuals but grouped data with households grouped by equivalized disposable income, the Gini index for the different income components cannot be computed, but the concentration coefficient can. The concentration coefficient

October 21, 2013

22

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

75% of the dispersion in market income on average in the OECD. Only in Italy, it is not the main determinant, the contribution of self-employment income being slightly larger and well above the OECD average of 15%. Although capital income is generally more concentrated than wages and salaries, it is not a strong determinant as its share in total market income is modest (around 7% in the OECD on average). Focusing on the entire population yields similar results and country rankings. The only noticeable change regards Turkey, as the concentration coefficient of market income drops from being the sixth highest to slightly below the OECD average when considering the whole population. Shorrocks (1982; 1983) questions the decomposition of the Gini (or concentration coefficient) by income sources on the grounds that results are sensitive to the specific decomposition rule adopted. After imposing additional assumptions, he shows that the decomposition of the squared coefficient of variation (i.e. the variance to squared-mean ratio) is independent of any specific decomposition rule and recommends using this approach. Overall, results using this alternative methodology are similar to those presented in Fig. 2.3.

The Distribution of HDI Taxes paid by households and cash transfers reduce the market household income dispersion by around a quarter on average across the OECD countries. But given the different starting points in terms of HMI distribution and the differences in redistribution via the tax and transfer system (Chapter 4), the distribution of HDI still varies widely. In around 2008, the Gini index ranged from somewhat below 0.25 in Slovenia to nearly 0.5 in Chile (Fig. 2.4). The HDI distribution is fairly narrow in the Nordic countries, some Eastern European countries as well as Austria and Belgium. The Gini index is somewhat higher in a number of continental European countries and higher still in the English-speaking countries as well as Japan, Korea, Italy, and Portugal. Income inequality is highest in the poorer OECD countries (Chile, Mexico, and Turkey). Percentile ratios provide a measure of HDI inequality at specific points of the income distribution and are a more intuitive way to gauge the width of the income distribution. The gap (footnote 4 continued) of any earnings’ component is computed in the same way as the Gini coefficient, with the only difference being that households are not ranked by the value of the earnings but rather by their equivalized disposable income.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

23

Gini index and gap between the 10th and 90th centile 0.6 Late 2000s

Mid-1990s

0.5

Gap between poorest and richest 10% 0.4

1:6 1:4

0.3 1:3 0.2

0.1

0.0

Fig. 2.4: The HDI Distribution: Cross-country Comparisons and Trends. Note: The Gini index ranges from 0 (perfect equality) to 1 (one individual receives all of the income and the others receive none). The OECD-26 includes countries for which data are available for the mid-1990s. Source: OECD Household Income Distribution and Poverty Database, December 2012.

between the upper bound value of the 1st decile and the upper bound value of the 9th decile of households is close to 1:3 for the three countries with the most narrow HDI distribution and above 1:6 for the three countries with the widest. Also the cross-country differences in the share of top income earners (top percentile) in total HDI are very wide, ranging from 5% for Denmark to 17% for the US. Since the mid-1990s, the disposable income distribution has widened in 12 countries, while it has narrowed in 8 countries. Although it has come down in some of the poorer countries (Chile, Greece, Hungary, and Mexico), it has widened considerably in the Nordic countries (except in Norway) as well as in Australia, Canada, and Israel. For the OECD on average, there was little change in income inequality as measured by the Gini index,5 contrary to public perceptions of a sharp rise in income inequality.6 In the countries that experienced greater inequality, the increase was often driven 5 For

the 19 countries for which long data series are available, the Gini index rose by 2.1 points between the mid-1980s and the mid-1990s. 6 A French Institute of Public Opinion (Ifop) poll covering 12 countries found that 83% of the Germans, 80% of the French, and 72% of the Swiss thought that inequality had

October 21, 2013

24

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

by rising inequality in the upper half of the income distribution, largely due to rapidly rising incomes at the very top (Chapter 6).

Adding In-kind Transfers In-kind transfers are an important tool for redistribution in addition to public cash transfers. On average, the size of in-kind services (health, education, and other social services) is more important than that of cash transfers — 15% versus 13% of GDP (Fig. 2.5). The Nordic countries and France stand out with very high in-kind transfers, whereas Chile, Korea, Poland, and Mexico spend little. Health care and education services are by far the most important in-kind transfers in all OECD countries, while other services (child care, public transport, and social housing) play a minor role, except in the Nordic countries.7 The variation in the size of in-kind transfers, as measured by the variance-to-mean ratio is highest for other social spending and lowest for health, suggesting more policy heterogeneity across countries in education and other social policies than in health. OECD (2011a) provides cross-country results on the inequalityreducing effect of in-kind transfers. Whatever the measure used, inequality declines in all OECD countries when in-kind transfers are taken into account.8 Total in-kind transfers reduce the interquintile ratio (Q5/Q1) (footnote 6 continued) increased over the past 10 years, while only 42% of the US citizens, 47% of the Australians, and 50% of the Brazilians thought so (Ifop, 2010). 7 Assessing the effect of in-kind benefits on inequality poses substantial methodological challenges. Government services are valued by the cost of production, and services have to be disaggregated to the household level. Household surveys often provide only limited information on the actual use of different government services. Two disaggregation methods exist: the actual consumption approach and the insurance value approach (OECD, 2011a). The former is used for public education, for which it is relatively easy to identify the beneficiaries, for example, households with children. In the presence of children, the income of a household is increased by the average spending at the relevant educational level. The insurance value approach is used for public health care as information on the number of medical treatments is usually not available. An insurance value is calculated, depending on individual characteristics such as age, sex, and socioeconomic position. This insurance value is equal to what such an individual would have to pay so that a third-party provider would fund the claims. 8 These calculations concern only those 27 OECD countries for which data were available for imputing the value of spending on public services. For some of the remaining countries, evidence from national sources suggest that public services have a significant redistributive impact (see, for instance, L´ opez and Miller (2008) for Chile).

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

25

In percentage of GDP, 2009 Health services

Education services

Other social services

a

Cash transfers

b

25

25

20

20

15

15

10

10

5

5

0

0

Fig. 2.5: Public Spending on In-kind Transfers. Note: Countries are ranked in increasing order of total expenditure on all social services. Data on education services for Greece refer to 2005. Data on health services, other social services, and cash transfers for Switzerland refer to 2008. a Other social services include services to survivors, disabled persons, unemployed, as well as those in respect of housing and social assistance (estimates of social housing are, however, not included). b Cash transfers include cash transfers to the elderly, survivors, disabled persons, families, unemployed, as well as those in respect of social assistance. Private mandatory spending, which accounts for a large share of total social spending in some countries (in particular Chile, Germany and Switzerland), is not included here. Source: OECD Social Expenditure Database, OECD Education Database, February 2012.

by nearly 1.6 points (from 5 to 3.4) on average, while the Gini index for the OECD area is reduced from 0.30 to 0.24. As measured by the Gini index, inequality is reduced by one-fifth, which is less than the reduction achieved by cash transfers (one-third). Despite the large overall decline in inequality, country rankings do not change much (Fig. 2.6). However, some of the countries with a high level of inequality in terms of disposable income (Mexico, Portugal, and the UK) show the largest decline in inequality (above 21%), while it is among the smallest for Slovenia (17%), the country with the lowest inequality in terms of disposable income.

October 21, 2013

26

Gini

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al. Gini of cash disposable income

Gini of disposable income plus in-kind transfers

0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00

Fig. 2.6: The Reduction in Inequality due to Public Spending on In-kind Transfers. Source: OECD (2011a).

Characterizing Cross-country Inequality Patterns What are the inequality dimensions which differentiate OECD countries most? Can groups of countries sharing similar inequality patterns be identified? This section uses two approaches — country profiles and a cluster analysis — to answer these questions. Both approaches rely on a set of inequality indicators, that is, those identified above subject to the constraint that they are available for a large number of OECD countries and reliable from a cross-country perspective. For each of the five income concepts (Fig. 2.1), three to six indicators of income inequality have been selected. Two additional indicators have been added, one on relative poverty and another on regional income disparities. The statistical analysis thus relies on a set of 24 indicators of income inequality for country profiles and on 12 core indicators for the cluster analysis.9

9 For some countries, data for a few indicators were not available. Country profiles show only the available data. For the cluster analysis, only core indicators have been used to avoid giving too much and artificial weight to some income concepts. For instance, the dispersion in HMI is included in the analysis but not the dispersion in capital income or in self-employment income.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

27

Country Profiles Trace the Various Inequality Dimensions Country profiles have been assembled in diamonds. These allow comparing the 24 inequality dimensions for a country with the OECD average and identifying how these inequality dimensions map into inequality of HDI. Box 2.2 Box 2.2. Country Profiles of Inequality Dimensions: The Case of Italy, Sweden, the UK and the US For Sweden, the indicator set reveals that inequality in HDI, whether adjusted or not for in-kind public services, is low in international comparison (Fig. 2.7). Inequality in ILEs for the working-age population is Italy

Gini regional

ILE Gini 18-65 3

Sweden

Gini regional

ILE P9/P1

Poverty

ILE P9/P5

HADI (Q5/Q1)

ILE P5/P1

ILE P5/P1 1

1 Men/Women

0

Education (Q5/Q1)

Part-time

Health (Q5/Q1)

Part-time 2 -2

-2 2 Non-employment

-3

HDI Gini 65+

HLE Gini 18-65

HDI Gini 18-65

Gini head

CC taxes

HDI Gini all

CC transfers

HLE Gini 18-65

HDI Gini 18-65

Gini head

CC taxes

Gini spouse

CC transfers

CC capital

CC capital

HMI Gini all

CC self-employed

Household market income (HMI)

Household market income (HMI)

United Kingdom

Gini regional

ILE Gini 18-65 3

United States

ILE P9/P1

Gini regional

ILE P9/P5 ILE P5/P1

ILE P9/P5

HADI (Q5/Q1)

ILE P5/P1

1

1 Men/Women

0

Education (Q5/Q1)

-1 1 Part-time

Health (Q5/Q1)

Part-time

-2 2

2 -2 Non-employment

-3

HDI Gini 65+

HLE Gini 18-65

HDI Gini 18-65

Gini head

CC taxes

Gini spouse

CC transfers

CC capital CC self-employed HMI Gini 18-65

Household market income (HMI)

Fig. 2.7:

Men/Women

0

-1 1 Health (Q5/Q1)

HMI Gini all

ILE P9/P1

2

HADI (Q5/Q1)

HDI Gini all

ILE Gini 18-65 3

Poverty

2

Education (Q5/Q1)

CC self-employed HMI Gini 18-65

HMI Gini 18-65

Poverty

Non-employment

-3

HDI Gini 65+

Gini spouse

HMI Gini all

Men/Women

0 -1 1

-1 1 Health (Q5/Q1)

HDI Gini all

ILE P9/P1

2

HADI (Q5/Q1)

Education (Q5/Q1)

ILE Gini 18-65 3

Poverty

ILE P9/P5 2

HDI Gini all

Non-employment

-3

HDI Gini 65+

HLE Gini 18-65

HDI Gini 18-65

Gini head

CC taxes

Gini spouse

CC transfers

CC capital

HMI Gini all

CC self-employed HMI Gini 18-65

Household market income (HMI)

Inequality Indicators for Italy, Sweden, the UK, and the US.

(Continued )

October 21, 2013

28

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

Box 2.2. (Continued ) Fig. 2.7 (Continued ) Note: The dotted line represents the OECD average and the solid line represents the country shown. Where the solid line falls inside the OECD average, this implies less inequality than the OECD average. Inversely, where the solid line is outside of the OECD average, inequality is greater. The indicators are presented in standard deviation units. Legend: ILE ILE Gini 18–65 = ILE Gini index for working-age population, including wage earners, self-employed, unemployed, and non-employed ILE P9/P1 = 9th to 1st decile wage earnings for full-time employees ILE P9/P5 = 9th to 5th decile wage earnings for full-time employees ILE P5/P1 = 5th to 1st decile wage earnings for full-time employees Men/Women = Median wage earnings of men to women Part-time = Ratio of part-time workers to total employment NER = Non-employment rate HLE HLE Gini 18–65 = HLE Gini index for working-age population Gini head = Gini index for heads of household Gini spouse = Gini index for spouses HMI CC capital = Concentration coefficient for capital income CC self-employed = Concentration coefficient for self-employment income HMI GINI 18–65 = HMI Gini index for working-age population HMI Gini all = HMI Gini index for total population

HDI CC transfers = Concentration coefficient for cash transfers CC taxes = Concentration coefficient for household taxes HDI Gini 18–65 = HDI Gini index for working-age population HDI Gini 65+ = HDI Gini index for population aged 66 and over HDI Gini all = HDI Gini index for total population Adjustment to HDI for public spending on: Health = Public spending on health, 5th to 1st quintile Education = Public spending on education and early childhood education and care, 5th to 1st quintile HADI = HDI adjusted for in-kind health and education public spending, 5th to 1st quintile Others Poverty rate = Relative poverty rate Gini regional = Gini index for regional GDP

(Continued )

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

29

Box 2.2. (Continued ) low, reflecting both a narrow wage dispersion (in particular at the bottom end, as represented by the ratio of the average wage for the 5th decile to the average wage of the 1st decile), a NER below the OECD average and little concentration in self-employment income. Overall, the Gini index of HMI for the entire population stands well below the OECD average. Taxes and cash transfers bring inequality in HDI significantly below the OECD average. In Italy, wage income is less dispersed than in the OECD on average. However, inequality in HMI is adversely affected by a high NER, and a high concentration of self-employment and capital income. With occupational pensions playing a pivotal role in the Italian welfare system, cash transfers and the tax system do not succeed in bringing HDI inequality and the poverty rate below the OECD average. In the UK, inequality in ILEs is above the OECD average, reflecting both a wide wage dispersion for those working full-time and the large proportion of part-time workers by OECD standards. The concentration in self-employment income further adds to inequality in HMI. Cash transfers benefit low-income households most and taxes are progressive, thus bringing down inequality in HDI to slightly above the OECD average. In the US, the wage dispersion for full-time workers is high by OECD standards, both at the top and bottom of the income ladder. The NER and part-time employment are both below the OECD average, bringing inequality in labor earnings for the working-age population close to the OECD average. Taxes have a large redistributive impact (the concentration coefficient for taxes is one of the highest in the OECD and much higher than the concentration coefficient for HMI). As a share of HDI, cash transfers are smaller than in most other OECD countries and have, thus, little redistributive impact. Overall, inequality in HDI is considerably above the OECD average.

shows, how to interpret them for four countries. The country profiles reveal that inequality of HDI, whether adjusted for in-kind transfers or not, does not have the same origin. In some countries, wage dispersion among those working is an important factor (e.g. the US), while in others, the NER and/or inequality in capital income are driving inequality

October 21, 2013

30

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

in HDI (e.g. Italy). The country profiles also show the extent to which tax and transfer systems as well as publicly provided services (in particular education and health) reduce income inequality. Some countries (e.g. Sweden) are characterized by relatively low inequality in HMI but still redistribute considerably via large tax and cash transfer systems which brings inequality in HDI well below the OECD average. The UK and the US display a similar inequality in HMI, which is clearly above the OECD average. Still, taxes and cash transfers redistribute more in the UK than in the US.

A Cluster Analysis Allows Identifying Groups of Countries Sharing Similar Inequality Patterns Four groups of countries sharing similar inequality patterns have been identified using a cluster analysis (Fig. 2.8), which is based on 12 core inequality indicators. The four groups, ordered by rising HDI inequality, are as follows: • A group — including four Nordic countries plus Switzerland — is characterized by below average inequality originating from the labor market, owing to a narrow wage dispersion, in particular at the upper end, combined with a high employment rate. However, the share of parttime employment is above average in all these countries, except Sweden. Cash transfers are often universal and household taxes tend to be largely proportional to household income. Overall, the dispersion in disposable income and the poverty rate are well below the OECD average. • In a group of 16 European countries (Austria, Belgium, Czech Republic, Estonia, Finland, France, Germany, Greece, Hungary, Italy, Luxembourg, the Netherlands, Poland, Slovak Republic, Slovenia, and Spain) along with Japan and Korea, inequality originating from the labor market is close to the OECD average. The underlying causes vary, however. The wage dispersion is wide at the lower end of the wage distribution in Germany, at the upper end of the income distribution in Greece, Hungary, and Poland, and at both ends in Korea. Inequality in labor earnings is driven by a low employment rate (in particular for Greece, Hungary, Italy, Korea, Luxembourg, Poland, the Slovak Republic, Slovenia, and Spain), while the share of part-time employment is high in Austria, Germany, Japan, and the Netherlands. However, in some of these countries (including Greece and Korea), an important

October 21, 2013 15:59

Lower inequality in HDI

Higher inequality in HDI

Income Inequality in the OECD. . . 9in x 6in

Fig. 2.8: Four Groups of Countries Share Similar Inequality Patterns.

31

b1598-ch02

Mapping Income Inequality Across the OECD

United States

October 21, 2013

32

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

redistribution in labor earnings takes place at the family level.10 Overall, both the dispersion in HDI and the poverty rate are close to the OECD average. • Four English-speaking countries (Australia, Canada, New Zealand, and the UK) are characterized by a large share of part-time employment driving inequality in labor earnings. Pushing in the other direction, the employment rate stands above the OECD average in all these countries. While small in size in all these countries, taxes and transfers have a sizable redistributive impact, as cash transfers tend to be more targeted and taxes more progressive than in other OECD countries. Still, the dispersion in HDI is above the OECD average in all these countries. • Chile, Israel, Mexico, Portugal, Turkey, and the US are characterized by above average inequality originating from the labor market, reflecting a very wide wage dispersion, coupled with a low employment rate (the US and Portugal being exceptions in this regard). Capital and selfemployment income also tend to be highly concentrated. Cash transfers have little redistributive impact because they are small in size and often largely insurance-based. Inequality in HDI and the poverty rate are well above the OECD average.

Inequality and Poverty in Large Emerging Economies Five emerging countries (Brazil, China, India, Indonesia, and South Africa) show both considerably higher economic growth and higher inequality than most OECD countries. Drawing cross-country comparisons is, however, difficult because income distribution data and methods often differ across these and the OECD countries. As an example, income dispersion in India is measured via a consumption survey (Gini coefficient of 0.38), while an income survey is used for South Africa (Gini coefficient of 0.70). Also, poverty in these countries is most often measured in absolute terms, rather than in relative terms as in most OECD countries, with the poverty line varying across countries.

10 In Korea, the number of households composed of only a single adult above 65 years is very low by OECD standards, while in Greece, the number of lone parents with children is low.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

33

Income dispersion has increased in three of the five countries considered here since the early 1990s. Income dispersion has risen by 24% in China, 16% in India, and 4% in South Africa (Arnal and F¨ orster, 2011; OECD, 2010b). It has remained broadly stable in Indonesia and declined by around 10% in Brazil. While income dispersion trends have diverged across these countries, absolute poverty has declined in all of them, owing to rapid economic growth. From the early 1990s to the late 2000s, China experienced both the highest economic growth rate and largest drop in poverty among the countries. Over the same period, the poverty rate in Indonesia and Brazil fell more than in India even though the latter grew faster.11 Poverty also fell in South Africa over the same period but to a lesser extent than in the other countries. Institutional settings and poverty reduction policies play a key role in explaining these differences. Although on a declining trend, poverty is still high and often concentrated on children and rural citizens. Poverty still affects 19% of the population in Indonesia and 17% in South Africa, but less than 4% in Brazil.12 In Brazil, India, and South Africa, children are more at risk of being poor than adults and the elderly (Arnal and F¨ orster, 2011). Whereas in the past two decades old age poverty declined in these countries, only Brazil managed to lower children’s poverty risk. The rapid urbanization and the rural–urban divide typical of emerging economies may also raise specific issues. In China, poverty has fallen more markedly in rural than urban areas since the 1990s reflecting both layoffs by state-owned enterprises, which were concentrated in urban agglomerations, and rural-to-urban area migration flows. In India, poverty also appears to have fallen more in rural than urban areas. Yet, migration does not seem to explain these different trends as migration flows were stable over the past two decades (Kundu and Mohanan, 2009). Reducing poverty and inequality in these countries is challenging for a number of reasons. First, the labor market needs to absorb a fast-growing labor force. And labor earnings account for a larger share of total household income than in developed countries. Second, rampant labor market informality often implies lower income, no social protection, little opportunity to develop human capital, and thus risks perpetuating poverty. Yet, a

11 The

elasticity of the poverty rate with respect to economic growth varies from −4.3% in Brazil, −0.8% in China, and −0.4% in India (Ravallion, 2011). 12 Poverty is defined as the population living with less than $1.25 a day (purchasing power parity [PPP] adjusted) in the mid or late 2000s.

October 21, 2013

34

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

higher degree of informality does not necessarily imply higher inequality.13 Over the past decade, inequality decreased in Brazil, but recent research attributes this decline to better education rather than to the concomitant drop in labor market informality (Arnal and F¨ orster, 2011). Also for China, informality affects income dispersion only slightly with differences in human capital and the rural–urban divide playing a more important role (Cai et al., 2009). Third, the emerging countries have less developed tax and transfer systems, with often low unemployment benefits or social safety nets. The experience of the large emerging economies shows, however, that transfer programs, if well targeted, can be powerful in reducing poverty and inequality. These countries introduced non-contributory benefit schemes over the 1990s, although in markedly different forms: • Ferreira et al. (2010) show for Brazil that from 1994 onwards the fall in the poverty rate is largely attributable to the expansion and reform of social assistance spending, including Bolsa Familia, along with a more stable macroeconomic environment under the Real Plan. • In 1999, China launched its first social welfare scheme, the Minimum Living Standard Scheme (Dibao), as a last resort assistance scheme for the urban poor. And this scheme was progressively extended to rural areas.14 This program has contributed to reduce poverty albeit by only a limited extent because of problems in its design and implementation (Cai et al., 2009; Herd, 2010). Although the number of beneficiaries has increased rapidly, Dibao reached only 3.9% and 6.7%, respectively, of the urban and rural populations in 2008.15 In both urban and rural areas, the benefit level is low at around 10% of average household income and there are considerable targeting problems. 13 A

growing informal sector affects income dispersion in two opposite directions. It increases income dispersion since workers in the informal sector have lower incomes than in the formal sector. At the same time, it reduces income dispersion as some of those employed in the informal sector may, otherwise, have no income. 14 Historically, China’s attempt to narrow the income gap between regions in the fastgrowing East and the laggard West has relied on large capital-intensive projects in the less developed areas. The impact of such projects on the local economy, poverty, and inequality has been below expectations (Herd, 2010). 15 Migrant workers from rural areas without urban registration are not eligible to Dibao benefits. Gao and Riskin (2009) have estimated that if Dibao were extended to such workers, the number of beneficiaries in urban areas would increase by around 65%.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

35

• In India, spending on social welfare is relatively high compared to other countries with a similar income level. However, the system’s fragmentation and poor coverage reduce its effectiveness (OECD, 2011b). The largest program is the Public Distribution Scheme. It aims at providing the neediest with subsidized food and other essential items, but it is beset by poor targeting and major delivery problems. India also relies on the National Rural Employment Guarantee Scheme, which provides 100 days of government-funded work per year at a minimum wage to all rural residents who wish to participate. However, its impact on poverty reduction has been limited as the program does not target the poor and cannot reach children, the elderly, and disabled people. Besides, the urban poor are not covered by this program. • In South Africa, social assistance programs target mostly poor children, through the Child Support Grant, and the poor elderly, through the Old-Age Pension. These schemes have significantly contributed to reduce poverty in addition to improving education and health (Leibbrandt et al., 2011). • In 2005, Indonesia strengthened its social protection program, with the implementation of conditional cash transfers. These have been effective in reducing poverty, especially during crisis periods (OECD, 2010a).

Global Inequality Inequality has also an international dimension. Income inequality among world citizens is affected by two components: inequality among individuals living in the same country (the within-country component) and crosscountry differences in average income (the between-country component). Higher within-country inequality does not necessarily result in higher global inequality, if average income in poor and large countries, such as China and India, grow faster than that in rich and less populous ones. The literature dealing with this issue has used two approaches: • The first one (also labeled international inequality) relies on national accounts data and focuses on the between-component only. In this approach, individuals in a given country are assigned the same level of income, usually GDP per capita, thus assuming perfect equality of income among people inhabiting the same country. Inequality is then computed applying standard inequality indices, such as the Gini and the Theil index, to GDP per capita data.

October 21, 2013

36

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

• The most recent approach (i.e. global inequality) relies on household survey data. This allows researchers to take into account not only differences in average income across countries (the between-country component) but also income inequality among individuals within the same country (the within-country component). International inequality measures (computed by weighing countries by their population share) show a clear downward trend in the Gini index from more than 0.65 in 1950 to around 0.55 in mid-2000s. This suggests that country-average incomes have converged. Starting from early 1980s to until the late 1990s, this development is totally attributable to China, given its large population share and strong growth performance. Excluding China, international inequality increased from the mid-1970s to the late 1990s. Thereafter, the international inequality measure excluding China also declined, mostly because of faster growth in India (Milanovic, 2011). Global inequality, which considers income inequality between and within countries, is higher than international inequality. The global Gini index was around 0.71 in 2005, well above the Gini of the most unequal countries (which is around 0.6) (Milanovic, 2011). From the late 1980s to 2005, global inequality does not show any clear trend. If anything there was a slight upward drift interrupted by a small decline between 1995 and 2000. The between-country component explains around 70% of the Gini, indicating that differences in living standards between countries are the main driver of global inequality. Preliminary figures for 2008 show a decreasing Gini to around 0.67–0.68, driven by the rising standard of living in poor and large economies, such as China, India, and Indonesia, rising faster than that in rich countries (Milanovic, 2011). The large contribution of the between-country component to global inequality is consistent with the historical evidence from early 19th to late 20th centuries (Bourguignon and Morrison, 2002).16 They found that global inequality increased from 1820 to 1950 and leveled off thenceforth. All the increase can be ascribed to the rising difference in country-average incomes. The share of global inequality explained by the between-country component rose from around 12% in the early 19th century to around 60% in 1950 and has remained stable since then. 16 Bourguignon and Morrison (2002) do not use household survey data. They calculate within-country inequality multiplying quantile income shares, obtained from different sources, by GDP per capita in PPP terms. They also divide the world in 33 country groups.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

37

Table 2.1: Decomposition of OECD Total Inequality.a Late 2000s

Total Within (%) Between (%)

Theil

MLD

0.279 68.81 31.19

0.320 66.52 33.48

Note: a MLD is the Mean Logarithmic Deviation index. Computation is conducted using grouped data by deciles. Source: OECD Household Income Distribution and Poverty Database, December 2012.

As regards OECD countries, total inequality of individuals inhabiting them is mostly due to inequality within countries rather than inequality between countries. Computing the Theil index and Mean Logarithmic Deviation (MLD) indices for the whole OECD for the late 2000s reveals that around 70% of total OECD inequality is attributable to the withincountry component and the remaining 30% to the between-country component (Table 2.1).17 Considering changes over time, the between-country share of the Theil and MLD index increased, whereas the within-country share decreased. This suggests that divergence in average income between countries was a more important factor in rising overall inequality than changes in withincountry inequality. The Theil and the MLD indices are not comparable over time as their upper-limit value varies with the number of individuals considered. However, the between- and within-country component share can be compared over time. The Gini index and its decomposition in between, within, and overlapping components corroborate these findings. The overlapping component measures how much countries’ income distributions overlap (Lambert and Aronso, 1993). From the mid-1990s to the late 2000s, the Gini index for the OECD area comprising the 19 countries for which data are available 17 The Theil and MLD indices are two special cases of the generalized entropy index (Theil, 1967). They differ as the Theil index is more sensitive to changes in income at the top end of the income distribution than the MLD.

October 21, 2013

38

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

increased by around 4% (from 0.406 to 0.423). The between-country component made the largest contribution to the increase in the overall Gini, followed by the within-country component. The overlapping component diminished, thus making a negative contribution to the rise in inequality. Overall, this evidence is consistent with that provided by the Theil index as the increase in inequality from the mid-1990s to the late 2000s was mostly due to the increasing dispersion in average income across countries.

References Alvaredo, F. et al. (2011). “The Top Incomes Database.” Available at: http:// topincomes.g-mond.parisschoolofeconomics.eu/. Anand, S. and S.M.R. Kanbur (1993). “Inequality and Development: A Critique,” Journal of Development Economics, Vol. 41(1), pp. 19–43. Arnal, E. and M. F¨ orster (2011). “Growth, employment and inequality in Brazil, China, India and South Africa,” In: Tackling Inequalities in Brazil, China, India and South Africa: The Role of Labour Market and Social Policies, Paris: OECD Publishing. Bourguignon, F. and C. Morrisson (2002). “Inequality among World Citizens: 1820–1992,” American Economic Review, Vol. 92(4), pp. 727–744. Cai, F., D. Yang and M. Wang (2009). “Employment and Inequality Outcomes in China.” Paper presented at the OECD Seminar on “Employment and Inequality Outcomes: New Evidence, Links and Policy Responses in Brazil, China and India,” April. OECD Headquarters, Paris, France. Evans, W.N. et al. (2004). “Assessing the effect of economic inequality,” In: Neckermann, K. (ed.), Social Inequality, New York: Russell Sage Fund, pp. 933–968. Ferreira, F.H., P.G. Leite and M. Ravallion (2010). “Poverty Reduction without Economic Growth? Explaining Brazil’s Poverty Dynamics, 1985–2004,” Journal of Development Economics, Vol. 93(1), pp. 20–36. Fournier, J.M. and I. Koske (2012). “Less Income Inequality and More Growth — Are They Compatible? Part 7. The Drivers of Labour Earnings Inequality — An Analysis Based on Conditional and Unconditional Quantile Regressions,” OECD Economics Department Working Paper No. 930. OECD Publishing, Paris. Gao, Q. and C. Riskin (2009). “Market versus social benefits: explaining China’s changing income inequality,” In: Davis, D. and Wang, F. (eds.), Creating Wealth and Poverty in Contemporary China, Palo Alto, California: Stanford University Press. Herd, R. (2010). “A Pause in the Growth of Inequality in China?” OECD Economics Department Working Paper No. 748.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

39

Ifop (2010). “Perception of Inequalities. Comparison of 12 Countries.” Results of a survey for La Fondation Jean Juar`es carried out by Ifop presented in a Powerpoint presentation. Available at: www.ifop.com/media/poll/1191-2study file.pdf. Kundu, A. and P.C. Mohanan (2009). “Employment and Inequality Outcomes in India.” Paper presented at the OECD Seminar on “Employment and Inequality Outcomes: New Evidence, Links and Policy Responses in Brazil, China and India,” April. OECD Headquarters, Paris, France. Lambert, P.J. and J.R. Aronson (1993). “Inequality Decomposition Analysis and the Gini Coefficient Revisited,” The Economic Journal, Vol. 103(420), pp. 1221–1227. Leibbrandt, M., I. Woolard, H. McEwen and C. Koep (2011). “Better employment to reduce inequality further in South Africa,” In: Tackling Inequalities in Brazil, China, India and South Africa: The Role of Labour Market and Social Policies, Paris: OECD Publishing. Lerman, R.I. and S. Yitzhaki (1985). “Income Inequality Effects by Income Source: A New Approach and Applications to the United States,” The Review of Economics and Statistics, Vol. 67(1), pp. 151–156. L´ opez, R. and S.J. Miller (2008). “Chile: The Unbearable Burden of Inequality,” World Development, Vol. 36(12), pp. 2679–2695. Milanovic, B. (2011). “Global Inequality Recalculated and Updated: The Effect of New PPP Estimates on Global Inequality and 2005 Estimates,” Journal of Economic Inequality, Vol. 10(1), pp. 1–18. OECD (2008). Growing Unequal? Income Distribution and Poverty in OECD Countries, OECD Publishing, Paris. OECD (2010a). OECD Economic Surveys: Indonesia 2010, OECD Publishing, Paris. OECD (2010b). Tackling Inequalities in Brazil, China, India, and South Africa: The Role of the Labour Market and Social Policies, OECD Publishing, Paris. OECD (2011a). Divided We Stand. Why Inequality Keeps Rising, OECD Publishing, Paris. OECD (2011b). OECD Economic Surveys: India 2011, OECD Publishing, Paris. Ravallion, M. (2011). “A Comparative Perspective on Poverty Reduction in Brazil, China and India,” The World Bank Research Observer, Vol. 26(1), pp. 71–104. Shorrocks, A.F. (1982). “Inequality Decomposition by Factor Components,” Econometrica, Vol. 50(1), pp. 193–211. Shorrocks, A.F. (1983). “The Impact of Income Components on the Distribution of Family Incomes,” The Quarterly Journal of Economics, Vol. 98(2), pp. 311–326. Stiglitz, J., A. Sen and J.P. Fitoussi (2009). Report by the Commission on the Measurement of Economic Performance and Social Progress. Theil, H. (1967). Economics and Information Theory, North-Holland Publishing Co., Amsterdam.

October 21, 2013

40

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

Verma, V. and G. Betti (2010). “Data accuracy in EU-SILC,” In: Atkinson, A.B. and Marlier, E. (eds.), Income and Living Conditions in Europe, Luxembourg: Eurostat. Wolff, P., F. Montaigne and G. Rojas Gonzalez (2010). “Investing in statistics: EU SILC,” In: Atkinson, A.B. and Marlier, E. (eds.), Income and Living Conditions in Europe, Luxembourg: Eurostat.

ANNEX Annex 2.1: Measures of Income Inequality Many inequality measures exist. They fall into two categories: one-number summary statistics, such as the Gini, Theil, or Atkinson indices and shares of income or percentile ratios. One-number summary statistics measure the income distribution throughout the distribution and differ somewhat in their sensitivity to changes in the tails versus the middle of the distribution. Also the concentration coefficient, which is a variant of the Gini index, is often used in this chapter and Chapter 4.18 Shares of income and percentile ratios provide a picture of income inequality at specific points in the income distribution. Table A2.1 highlights the properties and main characteristics of those measures that are most commonly used. The main conclusion that Table A2.1: Main Income Inequality Measures and Their Properties.

Measure Gini coefficients

Scale Replication Transfer Decompoinvariance invariance principle sability X

X

X

Main features Simple and robust measure of inequality. Sensitive to tails at the top or bottom of the distribution. It does not permit to locate, where changes in the distribution occur. (Continued)

18 The concentration coefficient is computed in the same way as the Gini coefficient, with the only difference being that individuals are not ranked by the value of the earnings they receive but rather by their equivalized disposable income.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Mapping Income Inequality Across the OECD

41

Table A2.1: (Continued)

Measure

Scale Replication Transfer Decompoinvariance invariance principle sability

Coefficient of variation

X

X

X

X

Percentile ratios and income shares Atkinson’s index

X

X

X

X

X

X

Theil’s index

X

X

X

X

Main features Simple and intuitive. Possible to conduct robust decompositions by income sources. Ignore information about incomes other than the percentiles and shares selected. Ability to gauge movements in different segments of the income distribution. Can be turned into a normative measure by imposing a coefficient to weigh individuals in different part of the distribution differently. Can be expressed as weighted average of inequality within subgroups, plus inequality among those subgroups. Allows the analysis of inequality among population subgroups.

Note: Scale invariance: inequality is unchanged if the income of each individual is multiplied by a given constant (i.e. how income is expressed, e.g. euro, yen, or dollar, is not important). Replication invariance (or population principle): a replication of the population (i.e. adding to the population under consideration n times the same individuals) does not change the inequality index. Transfer principle: the inequality index decreases in the case of a progressive income transfer (i.e. a transfer from richer to poorer individuals) and increases in the case of a regressive transfer (i.e. a transfer from poorer to richer individuals). Decomposability: the inequality index can be expressed as a weighted sum of the inequality values of different and mutually exclusive population subgroups plus the inequality between the subgroups’ means.

October 21, 2013

42

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch02

Hoeller et al.

can be drawn from the table is that looking at several of these measures at the same time might allow overcoming weaknesses of an individual index (e.g. an analysis based on Gini indices can be complemented by using percentile shares for different parts of the income distribution). Evans et al. (2004) show that while Gini, Theil, and Robin Hood indices as well as the share of income of the top quartile are highly correlated, the coefficient of variation and the 90:10 percentile ratio are less well correlated. However, estimates of the change in inequality over time can differ, depending on the income distribution measure.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

3. THE DISTRIBUTION OF LABOR INCOME Isabell Koske, Jean-Marc Fournier and Isabelle Wanner

Introduction and Main Findings Inequality in labor income is the main driver of inequality in household disposable income (HDI) in virtually all OECD countries. As the rise in inequality happened alongside rapid technological progress as well as rapidly rising trade and financial integration, much of the debate has focused on the role of technological change and globalization in shaping the distribution of income. However, these two factors are only part of a more complex story as policy also influences labor income, both directly and indirectly by affecting the linkages with macroeconomic developments. This chapter investigates the determinants of labor income inequality arising from hourly wage dispersion, differences in hours worked, and inactivity rates. It draws on both existing and new empirical evidence to analyze the role of macroeconomic factors — in particular, technological change and globalization — and the role of structural policies — in particular, education policy as well as labor and product market regulation — in shaping the distribution of labor income. While it is ultimately disposable income inequality (ideally adjusted for publicly provided in-kind services) that matters, labor income inequality is one of its major sources and thus warrants a distinct analysis. The focus on labor income implies that important population subgroups (such as retirees) and income sources (such as capital income) are ignored. Moreover, tax and transfer policies are only briefly touched upon in this chapter, as a thorough elaboration of their inequality effects can be found in Chapter 4, where the focus is on disposable rather than labor income inequality. The remainder of this chapter is structured as follows. The second section discusses cross-country differences in both the level and recent 43

October 21, 2013

44

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

evolution of labor income inequality.1 The third section investigates the role of non-policy factors (globalization and technological change) in shaping the distribution of labor income, while the fourth section focuses on policy factors (in particular education, labor, and product market policies). The fourth section also touches upon gender inequality and migration. The fifth section presents country diamonds that can help to identify policy priorities for individual OECD countries.

Main Findings • Both labor earnings inequality (inequality among those who earn an income from employment) and labor income inequality (inequality among all people in the working-age population, whether they work or not) differ widely across the OECD, reflecting cross-country differences in wage rates, hours worked, and inactivity rates. • Technological change and globalization have played a role in shaping inequality patterns, but the marked cross-country differences also reflect differences in policy and institutional settings. • While the link between education and labor income inequality is ambiguous from a theoretical point of view, empirical evidence indicates that policies that increase upper-secondary and tertiary graduation rates help reduce inequality. Policies that promote more equal access to education (e.g. no early tracking or improving the provision of early childhood care) are also likely to reduce income inequality. • A lower minimum wage tends to widen the distribution of earnings, but the implied negative impact on labor income equality may at least partly be offset by higher employment. • Higher union density appears to reduce labor income inequality through a more equal distribution of earnings. Although not investigated in this chapter, legal extensions of collective wage agreements may have similar effects. • The available evidence does not allow drawing firm conclusions about the inequality effects of the overall level of employment protection. 1 The focus on labor income inequality implies a focus on the working-age population, meaning that retirees and children are ignored. Whenever subgroups of the workingage population are referred to (e.g. when discussing specific empirical results), this is explicitly stated in the text.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

•

•

•

•

45

However, reducing labor market dualism by narrowing the gap between the protection of permanent and temporary jobs seems to lower income inequality through a lower wage dispersion and lower unemployment. Several studies suggest that spending on active labor market policies (ALMPs) may reduce income inequality by boosting employment, though the effectiveness of such measures varies widely across different types of programs. The evidence on the link between product market reform and the dispersion of earnings is rather mixed, but several studies indicate that the removal of competition-unfriendly product market regulations reduces labor income inequality by boosting employment. Inequality between men and women can also be important. The empirical analysis suggests that gender differences in working hours and choice of occupation and sector account for a sizable part of the earnings gap. Policies to reduce these differences (e.g. improvements in the access to childcare) could thus lead to more equal labor market outcomes among men and women. A large part of the gender earning gap remains unexplained after accounting for other factors, indicating that policies that reduce gender discrimination may also help. Immigrants underperform relative to natives in terms of employment and wages in most countries, implying that policies that foster the integration of immigrants could reduce inequality.

Cross-country Patterns and Recent Trends in Labor Income Inequality in OECD Countries Countries differ widely with respect to the level of labor earnings inequality among individuals of working age (Fig. 3.1). Earnings inequality for fulltime employees is highest in Chile, the US, and Portugal, while Switzerland, Belgium, and Denmark are the most equal countries. Inequality is generally higher for all the full-time employed, reflecting the wider dispersion of earnings among the self-employed. Extending the analysis to part-time workers, the unemployed, and the inactive raises the Gini index, reflecting the large income differentials for these groups and the group of full-time workers (unemployed individuals and the inactive enter the calculation with zero income as transfers are not taken into account). The increase in the Gini index is particularly large for countries where part-time workers make up a sizable share of total employment (e.g. Australia, Germany,

October 21, 2013

46

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

Gini Indexa 15-to-64-year olds, 2008 Full- me employees Full- me and part- me employed

Full- me employed Employed and unemployed

Working-age popula on

BEL DNK FIN CZE SWE SVK ITA FRA NOR AUS NLD ISL ESP SVN DEU OECD EST IRL AUT JPN HUN LUX GRC KOR GBR POL ISR CAN PRT USA CHL

0.75 0.70 0.65 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25

Fig. 3.1: Labor Earnings Inequality. Note: The group of employed individuals includes both dependent and self-employed individuals. The working-age population includes all persons aged 15 to 64 years except for students and persons above the country’s statutory retirement age. The Gini coefficients take into account labor earnings only; the precise definition of labor earnings differs across countries. 2005 for Israel, 2007 for France, Korea, and the US, 2009 for Chile and Japan. The values for the OECD are calculated as unweighted averages across all OECD countries for which data are available. a

The Gini index ranges from 0 (perfect equality) to 1 (one individual receives all of the income and the others none). Source: PSID for the US; HILDA for Australia; CASEN for Chile; KLIPS for Korea; LIS for Israel; JHPS for Japan; SHP for Switzerland; and EU-SILC for the other countries.

and the Netherlands) and for countries with a high unemployment rate and many inactive people in the working-age population. While the Gini indices of the population subgroups are highly correlated (the correlation coefficients are between 0.8 and 0.9), there are several countries for which the choice of the group matters considerably for the inequality ranking. The evolution of inequality among the full-time employed also differed widely. While many OECD countries saw a marked rise in inequality over the past decade, it remained broadly unchanged or even declined in others (Fig. 3.2).2 In about half of the countries that experienced an increase in labor earnings inequality between the mid-1990s and the mid-2000s, it was

2 Diverging

trends in working hours may account for part of the differences in the evolution of labor earnings inequality, but the size of this effect is difficult to gauge.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

47

Percentage change in 90/10 percentile ratio and percentage point change in employment rate Full-time employees, mid-1990s to mid-2000s 90/10 percen le ra o

Employment rate

3.0 2.5 2.0 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 ESP CHE CAN IRL FRA AUT SWE JPN GBR NLD USA FIN DEU DNK AUS BEL ITA NOR CZE HUN NZL KOR POL

Fig. 3.2: Annual Change in Labor Income Inequality. Note: For the employment rate, the annual change between 1995 and 2005 is shown. For the decile ratios, the mid-1990s refer to 1993 for Belgium, 1994 for Canada, 1996 for Denmark and the Czech Republic, 1997 for Ireland and Norway, and 1995 for all other countries; the mid-2000s refer to 2004 for Poland and 2005 for all other countries. Source: National household and enterprise surveys.

driven by rising inequality in the upper half of the earnings distribution. In the remaining countries, the increase in inequality was more broad-based, affecting all parts of the earnings distribution. Only in Germany the rising inequality was due to the increasing inequality at the bottom half of the earnings distribution. In the majority of countries, the widening of the earnings distribution was accompanied by a rise in employment, the only exceptions being the Czech Republic, Poland, and the US.

The Role of Non-policy Factors in Shaping Labor Income Inequality Skill-biased Technological Change One prominent explanation for the rise in labor income inequality is skillbiased technological change. One theory (Acemoglu and Autor, 2010) assumes that highly skilled workers have an advantage in performing nonroutine abstract tasks, while medium-skilled workers have an advantage in performing routine tasks that are based on precise and well-understood

October 21, 2013

48

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

procedures, whereas low-skilled workers have an advantage in performing non-routine manual tasks.3 To the extent that computers substitute for routine tasks, they reduce the demand for medium-skilled workers, while the demand for the high-skilled and (possibly) low-skilled workers rises — the so-called process of employment polarization. If the demand shifts are not offset by equal shifts in supply, technological progress reduces the earnings or employment of medium-skilled workers relative to both the lowand high-skilled ones. The effect on the earnings of high-skilled relative to low-skilled workers is a priori ambiguous, but it is likely to be positive, because laid-off medium-skilled workers move down in the task distribution and depress the earnings of the low-skilled workers so that the relative earnings of high-skilled workers rise. Developments in wage and employment patterns have provided empirical support for this theory. The past decades have seen a polarization of employment by skill level (e.g. Autor et al., 2006; Goos et al., 2009). In particular, labor demand seems to have shifted from routine to nonroutine tasks, with the shift concentrated in those industries that computerized rapidly (Autor et al., 2003). It is more difficult to find support at the macro level, possibly reflecting problems in identifying the right group of workers and in properly capturing technological change. New OECD empirical work, based on cross-country time series analysis, suggests that technological change has contributed to a rise in earnings inequality (see Annex 3.1, and OECD, 2011a). The choice of specification matters, however, for the significance and size of the effect and also for where in the earnings distribution the impact is most pronounced.

Globalization While globalization has many facets, the debate on the link between globalization and labor income inequality has mostly focused on trade, migration (which is discussed below), and, to a lesser extent, investment flows. Early empirical studies were based on the Stolper–Samuelson theorem, which 3 The

(older) canonical view (see, e.g. Katz and Murphy, 1992; Tinbergen, 1974, 1975) focused instead on high- and low-skilled workers, with technological progress raising the productivity and thus wages more for the high-skilled than the low-skilled. While this view was successful in explaining wage developments in the US between the 1960s and 1980s, it fails to explain more recent developments, in particular the polarization of wages and employment.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

49

predicts that trade increases the real return for the factor with which a country is most abundantly endowed. However, the theorem proved to be inconsistent with the evidence and, in particular, the observed rise in inequality in low-income countries (e.g. Feenstra, 2008; Goldberg and Pavcnik, 2007; Stone and Cavazos, 2011). Several new mechanisms have thus been explored (see Harrison et al., 2010, for an overview). For the most part, they predict that labor income inequality should increase with globalization: • Offshoring: Feenstra and Hanson (1996) explain the simultaneous rise in labor income inequality in both rich and poor countries by the reallocation of tasks from the richer to the poorer ones. The tasks that are reallocated are typically not skill-intensive from the perspective of the skill-rich country, but they are from the perspective of the skill-poor country. As a result, offshoring makes labor demand more skill-intensive in both groups of countries. Empirical studies tend to confirm that offshoring from high- to low-wage countries leads to a rise in inequality in the source country, though there is disagreement about the magnitude of this effect.4 Workers performing routine tasks appear to suffer the most from offshoring (Ebenstein et al., 2009). • Firm heterogeneity: If firms differ in their profitability and the least productive firms are battered by import competition, trade may increase labor income inequality by lowering employment or by reducing the relative earnings of low-income workers if high-income workers work disproportionately in (high-productivity) exporting firms (e.g. Egger and Kreickemeier, 2009; Helpman et al., 2010). Empirical studies support the thrust of these theoretical models, showing, for example, that trade liberalization is associated with an increase in the wages of workers at multinational enterprises and a decline in wages at domestically oriented firms (Amiti and Davis, 2008). • Trade-induced innovation: By spurring innovation, trade may raise labor income inequality both temporarily — since the R&D process is skill-intensive (Dinopoulos and Segerstrom, 1999; Neary, 2003) — and permanently, provided it results in skill-biased technological change (Acemoglu, 2002). There is evidence that firms that begin exporting after

4 Liu

and Trefler (2008), for example, conclude that the effects are at best small, while Ebenstein et al. (2009), who also account for occupational exposure to globalization, find sizable negative effects.

October 21, 2013

50

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

trade liberalization upgrade technology and workers’ skills more quickly (Bustos, 2007). While the positive link between globalization and inequality is supported by a growing body of firm-level studies, it is more difficult to establish a robust link between globalization and inequality at the aggregate level.5 New OECD regressions on cross-country time series suggest that trade increases the dispersion of earnings only if unions are weak or if employment protection is lax, whereas the negative employment effects of trade are more pronounced under the presence of strong unions (see Figs. 3.7 and 3.8 and the discussion below). However, these results should be interpreted with great care given the many limitations of the analysis. The impact of cross-border financial flows that may, for instance, be associated with offshoring appears to depend on their direction.

The Role of Structural Policies in Shaping Labor Income Inequality This section discusses the impact of education, labor, and product market policies on labor income inequality both from a theoretical and empirical perspective. The empirical discussion draws on new cross-country time series regressions (Annex 3.1) as well as existing literature, in particular the study by Fournier and Koske (2012a) which estimates quantile regressions based on household survey data for 32 different countries (Annex 3.2).6

Education Policy Theory suggests that the link between education and labor income inequality is far from straightforward. The impact of a change in the educational composition of the workforce can have two effects (Knight and Sabot, 1983): 5 While some studies confirm that trade raises inequality (e.g. Richardson, 1995; Wood, 1995), others draw the opposite conclusion (e.g. Checchi and Garc´ıa-Pe˜ nalosa, 2010) or are unable to establish a significant link (e.g. Edwards, 1997; OECD, 2011a). 6 The quantile regressions do not allow investigating as many policies as the cross-country time series regressions, but they have the important advantage of relying on large household survey datasets, so that the results are more reliable. The discussion of gender and migration issues draws on the existing literature as well as Oaxaca–Blinder decompositions based on household survey data.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

51

(i) a composition effect, whereby a rise in the share of highly educated (high-wage) workers raises earnings inequality up to a certain point, but will then lower it as fewer low-education (low-wage) workers remain and (ii) a rate-of-return effect, whereby a rise in the share of highly educated workers alters the returns to education.7 Given these theoretical ambiguities, the impact of education — and, hence, education policy — on labor income inequality is an empirical issue. Owing to the lack of a comprehensive dataset on education policy indicators, the empirical literature has focused on the inequality effects of education outcomes rather than policies.8 In general, the better educated are found to benefit from higher employment rates and higher earnings than less educated individuals (Fig. 3.3 and OECD, 2010). Similarly, workplace training appears to boost wages, though there is uncertainty regarding the magnitude of this effect (e.g. Arulampalam and Booth, 2001; Goux and Maurin, 2000; Leuven and Oosterbeek, 2002). The quantile regressions presented in Fournier and Koske (2012a) allow investigating the impact of a general rise in the educational attainment of the workforce on the level of income inequality. They also show that, through pure composition effects: • The dispersion of earnings falls as more individuals go beyond lowersecondary school and acquire upper-secondary degrees — a result that is to be expected as the majority of individuals in the countries considered already have upper-secondary education (Fig. 3.4, Panel A)9 ; and

7 The

direction of the change in (relative) returns depends on many factors such as the substitutability or complementarity between low educated workers, highly educated workers and capital, the interplay between innate ability and schooling (Dur and Teulings, 2004), and the signaling role of education (e.g. Hendel et al., 2005). 8 The main exception is education spending. However, the empirical results are mixed, possibly because spending alone does not take into account the country-specific institutional setup and because it is a poor measure of the quality of the education system. For example, Sylwester (2002) concludes that higher public spending on education is associated with lower earnings inequality, while Checchi (2000) draws the opposite conclusion. Hoxby (2008) looks specifically at redistributive school spending and finds for the US that it reduces earnings inequality among adults, but that the effect is very small. 9 Two notable exceptions are Portugal and Brazil, where upper-secondary education is found to be more profitable for those at the top of the earnings distribution. This could be due to the lower average education level compared with the other countries in the sample. The results for Portugal are in line with existing empirical evidence (e.g. Hartog et al., 2001; Machado and Mata, 2001).

October 21, 2013

52

Income Inequality in the OECD. . .

15:59

9in x 6in

b1598-ch03

Koske, Fournier and Wanner Number of 25–64 year-olds in employment as a percentage of the population aged 25 to 64, 2008 Below upper secondary

Upper secondary and post-secondary non-ter ary

Ter ary

100 90 80 70 60 50 40 AUS AUT BEL BRA CAN CHE CHL CZE DEU DNK ESP EST FIN FRA GBR GRC HUN IRL ISL ISR ITA JPN KOR LUX MEX NLD NOR NZL OECD POL PRT SVK SVN SWE TUR USA

30

Fig. 3.3: Employment Rates of 25-to-64-year-olds by Educational Attainment. Source: OECD (2010), Education at a Glance 2010, Table A6.3a.

Panel A. Effect on log earnings of raising the number of upper-secondary or post-secondary non-tertiary graduates 90th cen le 1.2 1.0

Panel B. Effect on log earnings of raising the number of tertiary graduates 90th cen le 2.5 BRA 2.0

PRT

0.8

CHL

BRA

1.5

0.6 1.0

0.4

POL

GRC 0.5

0.2

GRC

POL 0.0

SWE

0.0 0.0

0.2

0.4

0.6

0.8

1.0 1.2 10th cen le

Wage effect differs significantly

0.0

0.5

1.0

1.5

2.0 2.5 10th cen le

Wage effect is the same at the 10th and 90th cen les

Fig. 3.4: The Impact of Education on the Distribution of Earnings. Note: The horizontal axis shows the impact of a 1% increase in the proportion of workers with respectively upper-secondary or post-secondary non-tertiary (Panel A) and tertiary (Panel B) education on the log earnings of the 10th quantile. The vertical axis shows the impact of the same change on the log earnings of the 90th quantile. A data point below (above) the 45 degree line indicates that the change in the educational composition of the workforce is associated with a fall (rise) in earnings inequality. Source: Fournier and Koske (2012a).

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

53

• A rise in the number of tertiary graduates, by contrast, raises the dispersion of earnings (Fig. 3.4, Panel B). These composition effects may be strengthened or weakened by changes in the returns to education that result from changes in the relative supply of low- and highly-educated workers, which are not taken into account by the quantile regressions, which are static in nature. As regards tertiary education, cross-country time series analysis, which accounts for both the composition and the rate-of-return effects, tentatively indicates that a rise in the number of tertiary graduates may lower the relative returns to tertiary degrees enough to more than offset the composition effect. Overall, a rise in tertiary graduation rates is thus associated with lower earnings inequality (Annex 3.1 and OECD, 2011a).10 A decomposition of cross-country differences in earnings inequality suggests that differences in the educational composition of the workforce play an important role (see Annex 3.3). The importance of acquiring knowledge for labor market outcomes has increased over the last couple of decades. An upper-secondary degree is today typically considered the minimum requirement to find a decent job (OECD, 2010).11 Moreover, while the returns to lower-secondary education have declined over time, those to post-secondary education have surged (e.g. Lemieux, 2006a; Machado and Mata, 2001), most likely reflecting a rise in the demand for skills that has outpaced the rise in supply.12 Among workers with post-secondary education, relative wage gains have been higher for those higher up in the wage distribution.

10 This

implies that the return to tertiary education falls with a rise in the share of individuals who hold a tertiary degree. A simple panel regression analysis that regresses the average rate of return to a tertiary degree on the share of employed and unemployed individuals holding such a degree and country-fixed effects lends support to this negative relationship. For upper-secondary and post-secondary non-tertiary degrees by contrast, the regression does not provide any evidence for a significant link between the returns to such degrees and the share of workers holding them, suggesting that the composition effect estimated with the quantile regressions can be regarded as the total effect. See Fournier and Koske (2012a), for details. 11 Upper-secondary education (ISCED level 3) includes both general and (pre-) vocational/(pre-)technical education. 12 Specifically, while the rise in the supply of workers with post-secondary education is likely to have put downward pressure on the returns to post-secondary degrees, this effect was more than offset by an opposite effect stemming from a rise in demand associated with skill-biased technological change.

October 21, 2013

54

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

Both theoretical and empirical studies suggest that greater equality in educational attainment — measured, for example, by the variance of the number of years of schooling — lowers earnings inequality (e.g. De Gregorio and Lee, 2002). This is supported by the quantile regressions, which show that the returns to education differ across education levels (Fig. 3.4). Policies to promote equity in education by ensuring a basic minimum standard of education for all and by limiting the influence of personal and social circumstances on educational potential should thus also be beneficial for earnings equality.

Labor Market Institutions Labor market institutions affect labor income inequality through their (possibly conflicting) impacts on relative earnings and employment. This section covers the impact of minimum wages, wage bargaining systems, employment protection legislation (EPL), unemployment and social assistance benefits, and active labor market policies (ALMPs).

Minimum wages A minimum wage set above the market-clearing level can raise unemployment and thus inequality by lowering labor demand. At the same time, a rise in the minimum wage may reduce the dispersion of wages among those who remain employed (by raising the wages of those at the bottom of the wage distribution), thereby contributing to a decline in inequality. The latter effect is confirmed by a large number of empirical studies (e.g. Di Nardo et al. (1996); Lee (1999) and Lemieux (2006b), for the US; Dickens et al. (1999), for the UK; Checchi and Garc´ıa-Pe˜ nalosa (2010) and Koeniger et al. (2007), for panels of OECD countries) as well as the cross-country time series analysis conducted in this study (see Annex 3.1). However, the literature is less conclusive regarding employment: while some studies, including the present one, cannot detect a link between minimum wages and overall employment, others find negative employment effects for at least some types of workers such as the very low-skilled (see Neumark and Wascher (2007), for a recent review). Although this issue is unsettled, the few studies that looked at the combined effect conclude that an increase in minimum wages from their current level would generally raise inequality (as measured by the Gini index), as unemployment increases and this dominates the effect of a more compressed wage distribution (e.g. Calder´ on et al., 2005; Checchi and Garc´ıa-Pe˜ nalosa, 2008).

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

55

Wage bargaining The influence of the wage bargaining system on wage inequality is, in theory, ambiguous and depends on: (1) The number of workers who are covered by collective agreements (be it through union membership or through administrative extensions of collective agreements). (2) The strength of unions as manifested in their wage effects — both the wage level and wage dispersion may vary between union and non-union members. (3) The level at which bargaining takes place (e.g. at the firm, industry, or economy-wide level). The new cross-country time series analysis conducted in this study (Annex 3.1) suggests that, on average across countries, a rise in the share of workers affiliated to a trade union is associated with lower wage inequality among full-time workers — consistent with earlier evidence (e.g. Burniaux et al., 2006; Kahn, 2000). The quantile regressions, which are more robust and also account for the strength of unions, confirm this conclusion for some, but not all countries (Fig. 3.5). Higher union membership is found to decrease the dispersion of wages in Australia, Canada, Switzerland, and the US, but to have no effect in Korea and favors middle-income earners in Chile and Japan. Studies that account for legal extensions by looking at union coverage instead of union membership also point to wage equalizing effects (e.g. Kahn, 2000; OECD, 2011a), though legal extensions appear to be detrimental to employment (Murtin et al., 2013). Regarding the level of bargaining, several studies show that highly centralized/coordinated wage bargaining systems deliver better employment outcomes than decentralized/uncoordinated or intermediate systems (e.g. Bassanini and Duval, 2006).13 Moreover, the wage compression effect of unionization appears to be smaller if wage bargaining is more coordinated, possibly because unions take the adverse employment effects of wages that are pushed beyond the market-clearing level better into account (Koeniger et al., 2007).

13 The literature on the relationship between the level of bargaining and employment is, however, far from being conclusive (Flanagan, 1999). Some studies even indicate a humped-shaped relationship so that intermediate systems based on branch-level bargaining yield the worst labor market outcomes (e.g. Calmfors and Driffill, 1988; Elmeskov et al., 1998).

October 21, 2013

56

Income Inequality in the OECD. . .

15:59

9in x 6in

b1598-ch03

Koske, Fournier and Wanner Effect of raising the share of workers affiliated to a union by one percentage point on log earnings Panel A. Australia

0.8

Panel B. Canada

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0.0

0.0

-0.2

-0.2

-0.4

-0.4

-0.6

-0.6 10

20

30

40 50 60 Quan le

70

80

Panel C. Chile

0.8

10

90

20

30

0.6

0.4

0.4

0.2

0.2

0.0

0.0

-0.2

-0.2

-0.4

-0.4

-0.6

70

80

90

70

80

90

70

80

90

Panel D. Japan

0.8

0.6

40 50 60 Quan le

-0.6 10

20

30

40 50 60 Quan le

70

80

90

Panel E. Korea

0.8

10

20

0.6

0.4

0.4

0.2

0.2

0.0

0.0

-0.2

-0.2

-0.4

-0.4

-0.6

40 50 60 Quan le

Panel F. Switzerland

0.8

0.6

30

-0.6 10

20

30

40 50 60 Quan le

70

80

90

10

20

30

40 50 60 Quan le

Panel G. United States

0.8 0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 10

20

30

40 50 60 Quan le

70

80

90

Fig. 3.5: Union Membership — Unconditional Quantile Regression Results. Note: The solid lines show the effect of a one percentage point increase in the share of workers affiliated with a union (evenly spread across the population) and the white areas show the 95% confidence band. Source: Fournier and Koske (2012a).

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

57

EPL: overall stringency and gap between regular and temporary contracts If firms incur costs when dismissing workers, they will be reluctant to hire, thereby lowering labor turnover. The impact of EPL on employment is thus theoretically ambiguous and also the empirical literature is inconclusive (Bassanini and Duval, 2006). Empirical evidence on the impact of EPL on the earnings distribution is scarce. The cross-country time series regressions carried out in OECD (2011a) as well as in the present study (Annex 3.1) tentatively indicate that strict EPL narrows the dispersion of wages among full-time employees,14 thereby confirming the results of Koeniger et al. (2007). The substantial fixed administrative burden that is associated with employment protection may make it relatively more costly to fire low-income workers, so that EPL raises the bargaining power of low-income earners more than that of their high-income counterparts (Boeri et al., 2006). The negative effect of EPL on wage dispersion has been found to be stronger in the presence of minimum wages, which prevent firms from passing the cost of EPL on to low-income workers (Koeniger et al., 2007). Not only the overall stringency of EPL but also the gap between EPL on regular versus temporary contracts may matter for earnings inequality. If EPL is much stricter for regular than for temporary contracts, this may contribute to higher inequality since workers at the margin of the labor market risk getting trapped. The young, for instance, may be on a temporary contract or unemployed but cannot access indefinite contracts, with adverse implications for human capital and career progression (OECD, 2004). This premise is supported by the quantile regressions. Low-income workers on temporary contracts indeed earn less than those on permanent ones, and this effect fades for workers that are better paid (Fig. 3.6). Moreover, the cross-country panel analysis tentatively indicates that a larger gap between EPL on regular and temporary contracts is associated with more dispersed earnings (Annex 3.1). At the same time more stringent EPL for regular workers appears to be associated with higher unemployment, at least for

14 When drawing conclusions from the cross-country time series regressions for variables other than the minimum wage, the focus is on those specifications that exclude the minimum wage since the inclusion of this variable cuts the sample size approximately in half.

October 21, 2013

58

Income Inequality in the OECD. . .

15:59

9in x 6in

b1598-ch03

Koske, Fournier and Wanner Effect on log earnings of raising the share of workers with a temporary work contract by one percentage point.

0.2 0.0 -0.2 -0.4 -0.6 -0.8 -1.0 -1.2 -1.4 -1.6 -1.8 10

20

30

40

50 Quan le

60

70

80

90

Fig. 3.6: Earnings Effect of Having a Temporary Instead of a Permanent Work Contract. Note: The figure shows the estimated effect on the logarithm of gross earnings of increasing the share of workers with a temporary work contract by one percentage point. The thick bars depict the cross-country mean of the estimated effect ±1 standard deviation across countries on the logarithm of gross earnings of full-time and part-time workers, while the thin bars depict the cross-country maximum and minimum of the estimated effect. Source: Fournier and Koske (2012a).

certain categories of workers (e.g. Bassanini et al., 2010; Murtin et al., 2013).

Unemployment and social assistance benefits By providing income support, such benefits have a direct income inequality reducing effect (Chapter 4). At the same time, since unemployment benefits are progressive in most OECD countries due to benefit floors and ceilings that imply relatively higher benefits for people with low income, they may reduce inequality by strengthening the bargaining position of low-income workers more than that of high-income workers (Koeniger et al., 2007). However, they may also increase inequality by lowering employment.15 15 Specifically,

such benefits may reduce the job-search intensity of the unemployed and their willingness to accept job offers (though this effect could be mitigated through more conditionality and activation strategies) and put upward pressure on their wage claims by lowering the economic cost of being unemployed. On the other hand, they may

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

59

The negative employment effects of unemployment benefits are supported both by existing empirical studies (e.g. Bassanini and Duval, 2006; Murtin et al., 2013; Nickell et al., 2005; Nunziata, 2002; OECD, 2011a) and the new cross-country time series regressions presented in Annex 3.1. While Koeniger et al. (2007) and OECD (2011a) find that generous unemployment benefits also compress the wage distribution, the analysis summarized in Annex 3.1 does not find support for this view. It appears difficult to establish a significant relationship with broader measures of inequality such as the Gini index (e.g. Burniaux et al., 2006; Vanhoudt, 1997), possibly because the various effects cancel each other.

Active labor market policies ALMPs can reduce unemployment and thus labor income inequality through improvements in the job matching process and by raising the skills of the inactive workers via training programs. However, the effectiveness of ALMPs in reducing unemployment appears to vary widely across different types of programs, suggesting that program design is key (Kluve and Schmidt, 2002; Martin and Grubb, 2001). In addition, high public spending on ALMPs may contribute to reducing inequality by mitigating the negative employment effects of generous unemployment benefits (Bassanini and Duval, 2006). In line with the positive labor market effects of ALMPs, Vanhoudt (1997) finds that higher spending on ALMPs improves the income share of those at the bottom while reducing the income share of those at the top.

Interactions between labor market policies and globalization Labor market policies may also shape the way in which globalization has affected labor income inequality. New OECD cross-country time series analysis finds, for example, that the effect of trade on labor income inequality may be influenced by institutional features. Trade seems to increase the dispersion of wages among full-time employees if unions are weak but seems to have no effect on them if unions are strong (Fig. 3.7, Panel A). By contrast, the negative employment effects of trade are more pronounced under the presence of strong unions (Fig. 3.8), pointing to a role of unions (footnote 15 continued) allow jobseekers more time to find better matches, thereby lowering the likelihood of subsequent job separations (Bassanini and Duval, 2006).

October 21, 2013

60

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner Percentage point change in 90/10 percentile ration of full-time employees due to a 1% increasein the trade-to-GDP ratio as a function of: Panel A. Union density

Panel B. EPL

0.020

0.020

0.015

0.015

0.010

0.010

0.005

0.005 0.000

0.000 0

25 50 75 Union density (in %)

100

4 0 1 2 3 EPL index (a higher value means stricter EPL)

Fig. 3.7: The Interaction between Labor Market Institutions and Trade — Wage Dispersion. Note: The lines show the point estimates of the effect of trade on the 90/10 decile ratio among full-time employees and the shaded areas indicate the 90% confidence intervals. Source: Based on the coefficient estimates presented in Annex 3.1.

Percentage point change in the employment rate due to a 1% increase in the trade-to-GDP ratio as a function of union density

0.2 0.1 0.0 -0.1 -0.2 -0.3 -0.4 0

15 30 45 Union density (in %)

60

Fig. 3.8: The Interaction between Labor Market Institutions and Trade — Employment. Note: The line shows the point estimate of the effect of trade on the 90/10 decile ratio among full-time employees and the shaded area indicates the 90% confidence interval. Source: Based on the coefficient estimates presented in Annex 3.1.

in maintaining existing wage levels in the face of cost-cutting pressure from foreign competition. Moreover, trade is found to increase the dispersion of wages among full-time employees if employment protection is rather lax, possibly because strict employment protection may delay wage adjustments associated with trade liberalization (Fig. 3.7, Panel B).

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

61

Product Market Regulation Rents associated with lower product market competition may not be fully appropriated by capital owners but could accrue at least partially to workers. Greater competition, including through the removal of anticompetitive product market regulations, could thus affect earnings inequality through several channels (Guadalupe, 2007; Nicoletti et al., 2001): • It tends to reduce rents that can be captured by workers through collective bargaining, potentially leading to a decline in union power or more decentralized bargaining, that may, in turn, result in greater labor income inequality; • It may induce firms to innovate, which, to the extent that technological advances favor the highly skilled, could raise wage dispersion; • Since stringent product market regulations are often associated with wage premia, inter-industry differences in regulation are likely to affect interindustry wage relativities and thus labor income inequality; • By boosting potential output, product market liberalization could potentially reduce inequality through higher employment. There is abundant empirical evidence that product market liberalization raises employment (e.g. Bassanini and Duval, 2006; Fiori et al., 2007; Griffith et al., 2007; Nicoletti and Scarpetta, 2005; Nicoletti et al., 2001). However, its impact on the dispersion of earnings is less clear-cut. While Nicoletti et al. (2001) could not find much evidence that product market reforms have a long-run impact on earnings inequality, the empirical analysis by Guadalupe (2007) shows that greater competition is associated with a wider earnings distribution. The cross-country time series evidence by OECD (2011a) also points in this direction, suggesting that product market liberalization may raise wage dispersion by lowering the relative wages of those at the bottom of the income distribution. However, this result is not robust across different specifications.

Tax Policy The tax system influences inequality by altering the after-tax distribution of income. While this effect is discussed in detail in Chapter 4, tax policy may also influence the distribution of pre-tax earnings through its impact on labor demand and supply. In a perfectly competitive labor market, higher labor taxes should not affect equilibrium unemployment since workers will

October 21, 2013

62

Income Inequality in the OECD. . .

15:59

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

bear the entire tax burden through lower net wages. However, if firms cannot shift the entire tax burden onto workers (e.g. because of minimum wages or strong trade unions), higher taxes will reduce labor demand. Several empirical studies support this view (e.g. Bassanini and Duval, 2006; Belot and van Ours, 2004; Murtin et al., 2013). The impact appears to be particularly strong in countries with strong trade unions and an intermediate (or low) degree of wage bargaining coordination (e.g. Daveri and Tabellini, 2000; Elmeskov et al., 1998; Murtin et al., 2013). To the extent that job losses are more prevalent among low-wage workers, a higher tax wedge may, however, reduce the dispersion of wages (among those who retain their job). New OECD evidence tentatively supports this premise (OECD, 2011a and Annex 3.1).

Other Policy Issues Gender inequality Inequality among women and among men has increased between the mid1990s and mid-2000s in the majority of OECD countries (Fig. 3.9). In many countries, the increase in inequality was more pronounced among men than Difference in logarithm of percentile ratios (full-time employees) 0.04 0.03 0.02 0.01 0.00 -0.01

KOR

POL

HUN

AUS

ITA

NZL

CZE

GBR

NOR

BEL

USA

FIN

NLD

Men

DNK

JPN

DEU

AUT

SWE

IRL

FRA

CAN

ESP

CHE

-0.02

Women

Fig. 3.9: Annual Change in Inequality among Women and Men between the mid-1990s and the mid-2000s. Note: The mid-1990s refer to 1995 for all countries but Canada (1994), Ireland (1997), Denmark (1996), Norway (1997), Belgium (1990), and the Czech Republic (1996). The mid-2000s refer to 1995 for all countries but Poland (2004). Source: National household and enterprise surveys.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

63

among women. At the same time, despite some convergence between men and women, gender differences in labor market performance are still strikingly large in most countries. Women are less likely to be employed than men (76% of men of working age are in employment on average across countries, compared with 67% of women) and those who are working typically earn less than their male counterparts (on average across countries, women earn one-third less than men). Using an Oaxaca–Blinder-type decomposition to break down the gender earnings gap into differences in personal characteristics and differences in the returns to these characteristics the following may be suggested (Fig. 3.10): • Women’s shorter working hours (which are most likely due to the fact that they take on more caring obligations for children and elderly relatives than men; see OECD 2011b) play an important role in explaining differences in earnings between men and women — a finding that is in line with previous evidence (e.g. Ponthieux and Meurs, 2005). • Secondary and tertiary attainment rates of women are equal to, or exceed those of, men in most OECD countries as reflected in a zero or negative

Full-time and part-time workers, difference in log gross earnings, 2008 or latest available yeara

1.2 1 0.8 0.6 0.4 0.2 0 SVN HUN POL PRT DNK SVK ITA FIN ESP SWE FRA BEL CZE NOR EST CAN ISL USA IRL AUT AUS GRC GBR LUX NLD KOR DEU CHE JPN

-0.2

Composi on effects: Educa on Sector Hours worked Occupa on Sum of all rate of return effects plus the contribu on of the constant Total earnings gap between men and women

Age

Fig. 3.10: Decomposition of the Gender Earnings Gap. a 2009

for Australia and Japan; 2007 for France, Korea, and the US. Ratio between men’s and women’s average log earnings. Source: PSID for the US; HILDA for Australia; SLID for Canada; CASEN for Chile; KLIPS for Korea; JHPS for Japan; SHP for Switzerland; and EU-SILC for the other countries.

October 21, 2013

64

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

contribution of the composition of education to the gender earnings gap.16 • In the majority of countries, gender differences in the choice of occupation and/or sector of employment raise the gender wage gap — notable exceptions are Ireland, Poland, and Portugal. • Gender discrimination is also at play. A large part of the gender earnings gap is explained by the constant and rate of return effects (i.e. differences between men and women in the returns to personal characteristics such as age or the level of education). While this might be due to the omission of important variables in the regression,17 it may also capture gender discrimination.

Migration Migration influences labor earnings inequality in the host country because the labor market outcomes of the foreign-born differ from those of natives. Immigrants tend to under-perform natives in terms of employment, earnings, or both, suggesting that immigration raises inequality through this channel. Even though the gap decreases over time as immigrants assimilate, it typically persists even after a long period of stay (Jean et al., 2010). Decomposing the average earnings gap between natives and immigrants into differences in personal characteristics and differences in the returns to these characteristics points to considerable cross-country heterogeneity (Fig. 3.11). This heterogeneity may, for example, be due to differences in the structure of the immigrant population (in terms of country of origin, timing of immigration waves or motivation) or differences in countries’ policy settings. Two factors that contribute positively to the gap in many countries are a lower average level of education of immigrants and the sector composition, with immigrants working in sectors that pay less well. In addition, the contribution of the constant and rate of return effects raise the gap, as migrants earn less than natives, even if they have 16 Flabbi (2011) shows that although women are slightly more likely than men to obtain a tertiary degree, these are more often in the first level of tertiary education (e.g. Bachelor) than in the post-graduate level (e.g. Master or PhD). 17 For example, women might not acquire as much job market experience as men because of career breaks around childbirth (e.g. Buligescu et al., 2008; Dupuy and Fern´ andezKranz, 2011) or might pursue different fields of study which may give rise to different returns to tertiary education.

October 21, 2013

Income Inequality in the OECD. . .

15:59

9in x 6in

b1598-ch03

The Distribution of Labor Income

65

Full-time and part-time workers, difference in log gross earnings, 2003 to latest available year 0.5 0.4 0.3 0.2 0.1 0 -0.1 -0.2 -0.3

Composi on effects:

Occupa on

FIN

LUX

USA

SWE

ITA

Age

DNK

AUT

ESP

NOR

GRC

ISL

Hours worked

BEL

EST

CZE

DEU

SVN

Sector

NLD

FRA

Educa on

CHE

PRT

CAN

GBR

IRL

SVK

AUS

HUN

CHL

POL

-0.4

Gender

Sum of all rate of return effects plus the contribu on of the constant Total earnings gap between na ves and immigrants

Fig. 3.11: Decomposition of the Earnings Gap between Natives and Immigrants. Note: Individuals are defined as immigrants if they are born outside the country or, in case of EU member countries, outside the EU. The earnings gap is defined as the ratio between natives’ and immigrants’ average log earnings. Source: PSID for the US; HILDA for Australia; CASEN for Chile; KLIPS for Korea; JHPS for Japan; SHP for Switzerland; and EU-SILC for the other countries.

exactly the same characteristics (in terms of age, gender, education level, and so on). This may, for example, reflect firms’ difficulties in properly assessing qualifications obtained in a foreign country, the fact that at least some of the immigrants may have acquired their skills in a lower-quality educational environment, immigrants’ lack of work experience in the host country (which is an important channel of integration, but not controlled for in the regression), and also discrimination. Migration could also affect labor earnings inequality by altering the labor outcomes of natives. While there is not much evidence that immigrants have a sizable impact on the employment rate of natives with similar skills or occupations (e.g. Bonin, 2005; Cohen-Goldner and Paserman, 2004; D’Amuri and Peri, 2011; Jean and Jim´enez, 2011), they appear to reduce the wages of workers with similar skills or occupations (e.g. Aydemir and Borjas 2007; Borjas, 2003; Orrenius and Zavodny, 2007; Steinhardt, 2011). At the same time, the wages of natives whose skills are complementary to those of immigrants (Ottaviano and Peri, 2012) may rise in response to the arrival of new immigrants. The impact of immigration on earnings inequality among natives is thus unclear and depends in particular on the skill mix of the immigrant population. Aydemir and Borjas (2007),

October 21, 2013

66

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

who investigated the impact of immigration in countries with very different skill-structures among immigrants, found that it reduces inequality in Canada (characterized by high-skilled immigration), but increases inequality in the US (low-skilled immigration). In many OECD countries the distribution of educational attainment among immigrants is increasingly U-shaped, with an overrepresentation of both highly skilled and unskilled individuals (Jean et al., 2010). Migration may influence earnings inequality also in the source country. Most importantly, emigration alters the skill-composition of the workforce in the source country and, thus, the relative returns to skills. The direction of the impact on earnings inequality is theoretically ambiguous and depends in particular on the skill mix of emigrants and the substitutability between low-skilled workers, high-skilled workers, and capital (Davies and Wooton, 1992). Aydemiar and Borjas (2007) found that, for example, for Mexico, emigration has increased relative wages in the middle of the skill distribution (where emigration rates are highest in the case of Mexico) and lowered the relative wages at the extremes.

Identifying Reform Options to Reduce Labor Income Inequality As discussed in the previous section, structural policies influence labor income inequality through their impact on employment and the dispersion of earnings. To identify reform options for individual OECD countries that could help them to reduce inequality, policy indicators can be gathered into a country diamond enabling to compare its indicators with the OECD average. In addition, the diamond shows a countries’ performance with respect to equality of opportunity. For each policy indicator, a value above the OECD average means that the country is better positioned than the OECD average to reduce inequality and vice versa. To illustrate how the indicators can be used, the case of Canada is examined here. The indicator set (Fig. 3.12) shows that in Canada labor and product market regulations tend to be more employment-friendly than the OECD average (light shaded area), but they have adverse effects on the equality of earnings (dark shaded area). The education system delivers more equitable outcomes than the systems of most other OECD countries,

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

PMR*

2

Min wage*

67

PMR Min wage

1 Union density*

Union density 0

EPL perm / temp

EPL perm / temp

-1 1

Unemp benefits*

Unemp benefits

-2

Public emp* LT Unemp Poverty mobility

PISA soc econ

Income mobility

PISA rural / city Social mobility

Equality of opportunity

Fig. 3.12: Indicators of Policies that Influence Labor Income Inequality — the Example of Canada. PMR = Economy-wide product market regulation Min wage = Ratio of minimum to median wage Union density = Share of workers affiliated to trade union EPL = Employment protection legislation EPL perm/temp = Difference in EPL between permanent and temporary workers Unemp benefits = Average gross unemployment benefit replacement rate Public emp = Employment in general government and public corporations as percentage of total labor force LT Unemp = Long-term unemployed as share of total unemployed PISA soc econ = Average reading score point difference associated with socioeconomic background PISA rural/city = Reading performance, difference between students attending city schools versus rural/village schools Social mobility = Intergenerational earnings elasticity Income mobility = Average share of people staying within HDI quintile (average of all quintiles) Poverty mobility = Share of people experiencing persistent poverty Note: The dotted line represents the OECD average, the solid line and diamond markers represent the country shown. Where the solid line falls outside the OECD average, this implies better results than the OECD average with regard to the three policy goals shown. The series with asterisks have been inversed to take into account the relationship between these policy variables and the three policy goals. The indicators are presented in standard deviation units.

October 21, 2013

68

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

as indicated by a smaller impact of pupils’ socioeconomic background and area of residence (rural versus urban) on Programme for International Student Assessment (PISA) scores (white area). This is also reflected in relatively high intergenerational social mobility: the impact of parental earnings on the earnings of their offspring is relatively small. However, intragenerational mobility is relatively low: both the probability of remaining poor once fallen into poverty (indicator on poverty mobility) and the probability of staying in a given income quintile (indicator on income mobility) are high compared with other OECD countries. This does not seem to be driven by difficulties in reentering the labor market after a spell of unemployment, as the share of long-term unemployed (defined here as those who are unemployed for more than one year) is somewhat lower than in other countries.

References Abiad, A., E. Detragiache and T. Tressel (2010). “A New Database of Financial Reforms,” IMF Staff Papers, Vol. 57, pp. 281–302. Acemoglu, D. (2002). “Technical Change, Inequality, and the Labor Market,” Journal of Economic Literature, American Economic Association, Vol. 40(1), pp. 7–72. Acemoglu, D. and D.H. Autor (2010). “Skills, Tasks and Technologies: Implications for Employment and Earnings,” NBER Working Papers No. 16082, National Bureau of Economic Research. Amiti, M. and D.R. Davis (2008). “Trade, Firms, and Wages: Theory and Evidence,” NBER Working Papers No. 14106, National Bureau of Economic Research. Arulampalam, W. and A.L. Booth (2001). “Learning and Earning: Do Multiple Training Events Pay? A Decade of Evidence from a Cohort of Young British Men,” Economica, Vol. 68(271), pp. 379–400. Autor, D.H., F. Levy and R. Mumane (2003). “The Skill Content of Recent Technological Change: An Empirical Exploration,” Quarterly Journal of Economics, Vol. 118(4), pp. 1279–1333. Autor, D.H., L.F. Katz and M.S. Kearney (2006). “The Polarization of the US Labour Market,” American Economic Review, Vol. 96(2), pp. 189–194. Aydemir, A. and G.J. Borjas (2007). “Cross-Country Variation in the Impact of International Migration: Canada, Mexico, and the United States,” Journal of the European Economic Association, Vol. 5(4), pp. 663–708. Bassanini, A. and R. Duval (2006). “Employment Patterns in OECD Countries: Reassessing the Role of Policies and Institutions,” OECD Economics Department Working Papers No. 486.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

69

Bassanini, A., A. Garnero, P. Marianna and S. Martin (2010). “Institutional Determinants of Worker Flows: A Cross-Country/Cross-Industry Approach,” OECD Social, Employment and Migration Working Papers No. 107. Belot, M. and J. van Ours (2004). “Does the Recent Success of Some OECD Countries in Lowering their Unemployment Rates Lie in the Clever Design of their Labour Market Reform?,” Oxford Economic Papers, Vol. 56(4), pp. 621–642. Boeri, T., J. Ignacio Conde-Ruiz and V. Galasso (2006). “The Political Economy of Flexicurity,” mimeo, Bocconi University. Bonin, H. (2005). “Wage and Employment Effects of Immigration to Germany: Evidence from a Skill Group Approach,” IZA Discussion Papers No. 1875. Borjas, G.J. (2003). “The Labor Demand Curve Is Downward Sloping: Reexamining the Impact of Immigration on the Labor Market,” Quarterly Journal of Economics, Vol. 118(4), pp. 1335–1374. Buligescu, B. et al. (2009). “Panel Estimates for the Wage Penalty for Maternal Leave,” Oxford Economic Papers, Vol. 61, pp. i35–i55. Burniaux, J., F. Padrini and N. Brandt (2006). “Labour Market Performance, Income Inequality and Poverty in OECD Countries,” OECD Economics Department Working Papers No. 500. Bustos, P. (2007). “The Impact of Trade on Technology and Skill Upgrading: Evidence from Argentina,” mimeo, CREI and Universitat Pompeu Fabra. Cald´eron, Chong, A. and R. Vald´es, (2005). “Labor market regulations and income inequality: evidence for a panel of countries,” In: Restrepo, J.E. et al. (eds.), Labor Markets and Institutions, Chapter 7, Central Bank of Chile, Santiago, pp. 221–279. Calmfors, L. and J. Driffill (1988). “Bargaining Structure, Corporatism and Macroeconomic Performance,” Economic Policy (6). Checchi, D. (2000). “Does Educational Achievement Help to Explain Income Inequality?” UNU/WIDER Working Papers No. 208. Checchi, D. and C. Garcia-Pe˜ nalosa (2008). “Labour Market Institutions and Income Inequality,” Economic Policy, Vol. 23(56), pp. 601–649. Checchi, D. and C. Garcia-Pe˜ nalosa (2010). “Labour Market Institutions and the Personal Distribution of Income in the OECD,” Economica, Vol. 77(307), pp. 413–450. Cohen-Goldner, S. and M.D. Paserman (2004). “The Dynamic Impact of Immigration on Natives’ Labor Market Outcomes: Evidence from Israel,” IZA Discussion Papers No. 1315. D’Amuri, F. And G. Peri (2011). “Immigration, Jobs and Employment Protection: Evidence from Europe,” NBER Working Paper No. 17139. Daveri, F. and G. Tabellini (2000). “Unemployment, Growth and Taxation in Industrial Countries,” Economic Policy (30), pp. 47–104.

October 21, 2013

70

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

Davies, J. and I. Wooton (1992). “Income Inequality and International Migration,” Economic Journal, Vol. 102, pp. 789–802. De Gregorio, J. and J. Lee (2002). “Education and Income Inequality: New Evidence from Cross-Country Data,” Review of Income and Wealth, Vol. 48(3), pp. 395–416. Dickens, R., S. Machin and A. Manning (1999). “The Effects of Minimum Wages on Employment: Theory and Evidence from Britain,” Journal of Labor Economics, Vol. 17(1), pp. 1–22. DiNardo, J., N.M. Fortin and T. Lemieux (1996). “Labor Market Institutions and the Distribution of Wages, 1973–1992: A Semiparametric Approach,” Econometrica, Vol. 64(5), pp. 1001–1044. Dinopoulos, E. and P. Segerstrom (1999). “A Schumpeterian Model of Protection and Relative Wages,” American Economic Review, Vol. 89(3), pp. 450–472. Dupuy, A. and D. Fern´ andez-Kranz (2011). “International Differences in the Family Gap in Pay: The Role of Labour Market Institutions,” Applied Economics, Vol. 43, pp. 413–438. Dur, R.A.J. and C.N. Teulings (2004). “Are education subsidies an efficient redistributive device?,” In: Agell, J., Keen, M. and Weichenrieder, A. (eds.), Labor Market, Institutions and Public Regulation, Cambridge: MIT Press. Ebenstein, A. et al. (2009). “Estimating the Impact of Trade and Offshoring on American Workers Using the Current Population Surveys,” NBER Working Papers No. 15107, National Bureau of Economic Research. Edwards, S. (1997). “Trade Policy, Growth, and Income Distribution?,” The American Economic Review, Vol. 87(2), pp. 205–210. Egger, H. and U. Kreickemeier (2009). “Firm Heterogeneity and the Labor Market Effects of Trade Liberalization,” International Economic Review, Vol. 50(1), pp. 187–216. Elmeskov, J., J. Martin and S. Scarpetta (1998). “Key Lessons for Labour Market Reforms: Evidence from OECD Countries’ Experiences,” Swedish Economic Policy Review, Vol. 5(2), pp. 205–252. Feenstra, R.C. (2008). “Offshoring in the Global Economy: Lecture 1: Microeconomic Structure; Lecture 2: Macroeconomic Implications,” The Ohlin Lectures Presented at the Stockholm School of Economics, September 17–18. Feenstra, R.C. and G.H. Hanson (1996). “Foreign Investment, Outsourcing and Relative Wages,” NBER Working Paper No. 5121, National Bureau of Economic Research. Fiori, G., G. Nicoletti, S. Scarpetta and F. Schiantarelli (2007). “Employment Outcomes and the Interaction between Product and Labor Market Deregulation: Are they Substitutes or Complements?” IZA Discussion Papers No. 2770, Institute for the Study of Labor (IZA). Firpo, S., N. Fortin and T. Lemieux (2007a). “Unconditional Quantile Regression,” NBER Technical Working Paper No. 339, July.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

71

Firpo, S., N. Fortin and T. Lemieux (2007b). “Decomposing Wage Distributions Using Recentered Influence Function Regressions,” unpublished manuscript, University of British Columbia, June. Firpo, S., N. Fortin and T. Lemieux (2009). “Unconditional Quantile Regression,” Econometrica, Vol. 77(May), pp. 953–973. Flabbi, L. (2011). “Gender Differentials in Education, Career, Choices and Labour Market Outcomes on a Sample of OECD Countries,” mimeo. Flanagan, R.J. (1999). “Macroeconomic Performance and Collective Bargaining: An International Perspective,” Journal of Economic Literature, Vol. 37(3), pp. 1150–1175. Fortin, N.M., T. Lemieux and S. Firpo (2011). “Decomposition methods,” In: Ashenfelter, O. and Card, D. (eds.), Handbook of Labor Economics, Amsterdam: North-Holland, Elsevier, Vol. 4A, pp. 1–102. Fournier, J.-M. and I. Koske (2012a). “Less Income Inequality and More Growth — Are they Compatible? Part 7: The Drivers of Labour Earnings Inequality — An Analysis Based on Conditional and Unconditional Quantile Regressions,” OECD Economics Department Working Papers No. 930. Fournier, J.-M. and I. Koske (2012b). “The Determinants of Earnings Inequality: Evidence from Quantile Regressions,” OECD Journal: Economic Studies, Vol. 2012/1, pp. 7–36. Goldberg, P.K. and N. Pavcnik (2007). “Distributional Effects of Globalization in Developing Countries,” Journal of Economic Literature, Vol. 45(1), pp. 39–82. Goos, M., A. Manning and A. Salomons (2009). “Job Polarization in Europe,” The American Economic Review, Vol. 99(2). Papers and Proceedings of the One Hundred Twenty-First Meeting of the American Economic Association, American Economic Association, pp. 58–63. Gosling, A. and S. Machin (1995). “Trade Unions and the Dispersion of Earnings in British Establishments, 1980–90,” Oxford Bulletin of Economics and Statistics, Vol. 57(2), pp. 167–184. Goux, D. and E. Maurin (2000). “Returns to Firm-Provided Training: Evidence from French Worker-Firm Matched Data,” Labour Economics, Vol. 7(1), pp. 1–19. Griffith, R., R. Harrison and G. Macartney (2007). “Product Market Reforms, Labour Market Institutions and Unemployment,” The Economic Journal, Vol. 117(519), pp. C142–C166. Guadalupe, M. (2007). “Product Market Competition, Returns to Skill, and Wage Inequality,” Journal of Labor Economics, Vol. 25(3), pp. 439–474. Harrison, A., J. McLaren and M.S. McMillan (2010). “Recent Findings on Trade and Inequality,” NBER Working Papers No. 16425, National Bureau of Economic Research.

October 21, 2013

72

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

Hartog, J., P.T. Pereira and J.A.C. Vieira (2001). “Changing Returns to Education in Portugal During the 1980s and Early 1990s: OLS and Quantile Regression Estimators,” Applied Economics, Vol. 33(8), pp. 1021–1037. Helpman, E., O. Itskhoki and S.J. Redding (2010). “Inequality and Unemployment in a Global Economy,” Econometrica, Vol. 78(4), pp. 1239–1283. Hendel, I., J. Shapiro and P. Willen (2005). “Educational Opportunity and Income Inequality,” Journal of Public Economics, Vol. 89(5–6), pp. 841–870. Hoxby, C.M. (2008). “School Spending, Income, and Inequality: The Efficient Redistribution Hypothesis,” mimeo, Stanford University. Jean, S. and M. Jim´enez (2011). “The Unemployment Impact of Immigration in OECD Countries,” European Journal of Political Economy, Vol. 27(2), pp. 241–256. Jean, S., O. Causa, M. Jim´enez and I. Wanner (2010). “Migration and Labour Market Outcomes in OECD Countries,” OECD Economic Studies, Vol. 2010, OECD Publishing. Kahn, L. (2000). “Wage Inequality, Collective Bargaining and Relative Employment from 1985 to 1994: Evidence from 15 OECD Countries,” Review of Economics and Statistics, Vol. 82(4), pp. 564–579. Katz, L.F. and K.M. Murphy (1992). “Changes in Relative Wages, 1963–1987: Supply and Demand Factors,” Quarterly Journal of Economics, Vol. 107(1), pp. 35–78. Kluve, J. and C.M. Schmidt (2002). “Can Training and Employment Subsidies Combat European Unemployment?,” Economic Policy, Vol. 17(35), pp. 409–448. Knight, J.B. and R.H. Sabot (1983). “Educational Expansion and the Kuznets Effect,” The American Economic Review, Vol. 73(5), pp. 1132–1136. Koencker, R. and G. Basset (1978). “Quantile Regression,” Econometrica, Vol. 46(1), pp. 33–50. Koeniger, W., M. Leonardi and L. Nunziata (2007). “Labor Market Institutions and Wage Inequality,” Industrial and Labor Relations Review, Vol. 60(3), pp. 340–356. Lee, D.S. (1999). “Wage Inequality in the US during the 1980s: Rising Dispersion or Falling Minimum Wage?,” Quarterly Journal of Economics, Vol. 114(3), pp. 977–1023. Lemieux, T. (2006a). “Postsecondary Education and Increasing Wage Inequality,” American Economic Review, Vol. 96(2), pp. 195–199. Lemieux, T. (2006b). “Increasing Residual Wage Inequality: Composition Effects, Noisy Data, or Rising Demand for Skill?,” American Economic Review, Vol. 96(3). Leuven, E. and H. Oosterbeek (2002). “A new Method to Estimate the Returns to Work-related Training,” mimeo, Department of Economics, University of Amsterdam.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

73

Liu, R. and D. Trefler (2008). “Much Ado About Nothing: American Jobs and The Rise of Service Outsourcing to China and India,” NBER Working Papers No. 14061, National Bureau of Economic Research. Machado, J.A.F. and J. Mata (2001). “Earning Functions in Portugal 1982–1994: Evidence from Quantile Regressions,” Empirical Economics, Vol. 26(1), pp. 115–134. Martin, J.P. and D. Grubb (2001). “What Works and for Whom: A Review of OECD Countries’ Experiences with Active Labour Market Policies,” Swedish Economic Policy Review, Vol. 8(2), pp. 9–56. Murtin, F., A. de Serres and A. Hijzen (2013). “The Ins and Outs of Unemployment: The Role of Labour Market Institutions,” OECD Economics Department Working Papers, No. 1020, OECD Publishing, Paris. Neary, J.P. (2003). “Presidential Address: Globalization and Market Structure,” Journal of the European Economic Association, Vol. 1(2–3), pp. 245–271. Neumark, D. and W. Wascher (2007). “Minimum Wages and Employment,” IZA Discussion Papers, No. 2570. Nickell, S., L. Nunziata and W. Ochel (2005). “Unemployment in the OECD since the 1960s: What Do We Know?” Economic Journal, Vol. 115(500), pp. 1–27. Nicoletti, G. et al. (2001). “Product and Labour Markets Interactions in OECD Countries,” OECD Economics Department Working Papers No. 312. Nicoletti, G. and S. Scarpetta (2005). “Product Market Reforms and Employment in OECD Countries,” OECD Economics Department Working Papers No. 472. Nunziata, L. (2002). “Unemployment, Labour Market Institutions and Shocks,” Nuffield College Working Papers in Economics 2002-W16. OECD (2004). OECD Employment Outlook, OECD Publishing, Paris. OECD (2010). Education at a Glance 2010, OECD Publishing, Paris. OECD (2011a). Divided We Stand: Why Inequality Keeps Rising, OECD Publishing, Paris. OECD (2011b). Report on the Gender Initiative: Gender Equality in Education, Employment and Entrepreneurship, Meeting of the OECD Council at Ministerial Level, May 25–26, 2011. Orrenius, P.M. and M. Zavodny (2007). “Does Immigration Affect Wages? A Look at the Occupation-Level Evidence,” Labour Economics, Vol. 14(5), pp. 757–773. Ottaviano, G. I. P. and Peri, G. (2012). Rethinking the Effects of Immigration on Wages,” Journal of the European Economic Association, Vol. 10, pp. 152–197. Ponthieux, S. and D. Meurs (2005). “The Gender Wage Gap in Europe: Women, Men and the Public Sector,” INSEE, Direction des Statistiques D´emographiques et Sociales Working Paper No. F0502.

October 21, 2013

74

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

Richardson, J.D. (1995). “Income Inequality and Trade: How to Think, What to Conclude,” Journal of Economic Perspectives, Vol. 9, pp. 33–55. Sala-i-Martin, X., G. Doppelhofer and R.I. Miller (2004). “Determinants of Long-Term Growth: A Bayesian Averaging of Classical Estimates (BACE) Approach,” American Economic Review, Vol. 94, pp. 813–835. Steinhardt, M.F. (2011). “The Wage Impact of Immigration in Germany — New Evidence for Skill Groups and Occupations,” The B.E. Journal of Economic Analysis and Policy, Vol. 11(1), Article 31. Stone, S. and R. Cavazos (2011). “Wage Implications of Trade Liberalisation: Evidence for Effective Policy Formation,” OECD Trade Policy Working Papers No. 122. Sylwester, K. (2002). “Can Education Expenditures Reduce Income Inequality?” Economics of Education Review, Vol. 21(1), pp. 43–52. Tinbergen, J. (1974). “Substitution of Graduate by Other Labour,” Kyklos, Vol. 27(2), pp. 43–52. Tinbergen, J. (1975). “Substitution of Academically Trained by Other Manpower,” Review of World Economics, Vol. 111, pp. 217–226. Vanhoudt, P. (1997). “Do Labor Market Policies and Growth Fundamentals Matter for Income Inequality in OECD Countries?,” IMF Staff Papers, Vol. 44(3), pp. 356–373. Wood, A. (1995). “How Trade Hurt Unskilled Workers,” Journal of Economic Perspectives, Vol. 9(3), pp. 15–32.

ANNEXES Annex 3.1: Bayesian model averaging analysis: additional estimation results and details on the dataset used. Building on recent OECD work (OECD, 2011a), the determinants of labor income inequality are explored by estimating a series of reduced-form equations for 22 OECD countries18 that relate different measures of income inequality Y to measures of globalization GLOB, technological change TECH, human capital HC, policy indicators P and a set of control variables X: Yit = αGLOBit + βTECHit + γHCit + δPit + φXit + µi + λt + εit ,

(1)

18 Australia, Austria, Belgium, Canada, Czech Republic, Denmark, Finland, France, Germany, Hungary, Ireland, Italy, Japan, Netherlands, New Zealand, Norway, Poland, Spain, Sweden, Switzerland, the UK, and the US.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

75

where i and t denote country and time. All equations include country and period fixed effects to control for unobserved cross-country heterogeneity and unobserved shocks that affect all countries simultaneously. Equation (1) is estimated for several dependent variables: (i) the 90/10, 90/50, and 50/10 percentile ratios for full-time employees to capture the dispersion of wages among the employed; (ii) the employment rate; and (iii) the Gini index to capture the combined impact. As time-series data for the Gini index for labor income are not available for a wide range of countries, the Gini indices are based on disposable income and are calculated over the whole population (not just the working-age population). The estimated coefficients in the Gini regressions thus capture more than just the impact of the explanatory variables on labor income inequality (e.g. the coefficient on the unemployment benefit replacement rate captures both the indirect effects, whereby a higher replacement rate may reduce employment as well as the dispersion of wages, and the direct effect, whereby it raises the income of the unemployed). Three different measures of globalization are used: total trade (exports plus imports) and foreign assets and liabilities (all expressed as a ratio to GDP). Technical change is measured by the number of patents (per one million inhabitants). The stock of human capital is captured by the share of the population with post-secondary education. The policy indicators span four different policy areas: labor market policy (measured by the overall level of employment protection and by the level of protection on regular and temporary work, the unemployment benefit replacement rate, and the ratio of the minimum to the median wage), product market regulation (measured by an index of regulation in seven non-manufacturing industries), financial market regulation (measured by a financial reform index), and taxation (measured by the labor tax wedge). Moreover, union density is included to investigate the role of wage setting mechanisms. The set of control variables includes the output gap and, except for the employment rate specifications, the female employment share and employment in agriculture and industry (both as a share of total employment). Details on the dataset are provided in Table A3.1. Taking a macroeconomic perspective allows exploiting a broad range of reform experiences simultaneously and capturing the full general equilibrium effects of the explanatory variables. However, such cross-country time series regressions are typically less robust to variations in, for example, samples, econometric methods or the set of variables included in the regression

October 21, 2013

76

Income Inequality in the OECD. . .

15:59

9in x 6in

b1598-ch03

Koske, Fournier and Wanner Table A3.1: Details on the Dataset used in the BMA Analysis.

Title

Definition

Sources

Dependent variables Earnings decile ratios Gini

Wage refers to gross weekly or monthly earnings of full-time workers Ratio of the area between the 45 degree line and the Lorenz curve over the total area under the 45 degree line

Trade (as share of GDP)

Macro variables Trade exposure (weighted average of import penetration and export intensity)

Foreign assets (as share of GDP)

Total assets = FDI assets + portfolio equity assets + debt assets + derivatives assets + FX reserves as a share of GDP

Foreign liabilities (as share of GDP)

Total liabilities = FDI liabilities + portfolio equity liabilities + debt liabilities + derivatives liabilities as a share of GDP

Patents (per 1 million inhabitants)

Patents (total patent applications to both European patent office and the US Patent and Trademark Office), per million population.

OECD Earnings Database UNU-WIDER World Income Inequality Database, Version 2.0c, May 2008 United Nations Conference on trade and development (UNCTAD) Updated and extended version of the External Wealth of Nations Mark II Database developed by Lane and Milesi-Ferretti (2007) Updated and extended version of the External Wealth of Nations Mark II Database developed by Lane and Milesi-Ferretti (2007) OECD Science and Technology Indicators and OECD Patents Database

Policy variables Share of population with post-secondary education Employment protection legislation Union density (in %)

Percentage of population that has post-secondary educationa

OECD Education at a Glance and Barro and Lee (2000) dataset

Index scale of 0–6 from least to more restrictive.

OECD, Employment Outlook Database

Percentage of employees who are members of a trade union.

OECD, Employment Outlook Database (Continued)

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

77

Table A3.1: (Continued) Title Unemployment benefit replacement rate Minimum wage (as ratio to median wage) Tax wedge (in %)

Product market regulation indicator

Financial reform index

Sectoral employment shares Female employment share Output gap

Definition

Sources

Average gross unemployment benefit replacement rate

OECD, Benefits and Wages Database

Minimum wage relative to median of full-time worker

OECD, Employment Outlook Database

Tax wedges are calculated by expressing the sum of personal income tax as a percentage of labor costs. The reference rates are for single persons without children at 100% of the average level. Indicators of regulation in seven non-manufacturing sectors (telecoms, electricity, gas, post, rail, air passenger transport, and road freight). Index scale of 0–6 from least to more restrictive. Financial Reform Index, 0 to 21, sum of seven components; higher number means more liberal/developed financial system.

OECD, Taxing Wages

Control variables Percentage of employment in agriculture and industry as a share of total employment. Percentage of female employment as a share of total employment. Deviations of actual GDP from potential GDP as a percentage of potential GDP

OECD, Product Market Regulation Database

Abiad et al. (2010)

OECD, Employment Database OECD, Employment Database OECD, Economic Outlook Database

Note: a Data for 1980, 1985, 1990, 1995, and 2000 are drawn from the Barro and Lee (2000) dataset, and for the years 2001–2008 are from OECD Education at a Glance. For the years between 1985 and 2000, data are interpolated linearly. Source: Lane, P.R. and G.M. Milesi-Ferretti (2007). “The External Wealth of Nations Mark II,” Journal of International Economics Vol. 73, pp. 223–250; Barro, R.J. and J.-W. Lee (2000). “International Data on Educational Attainment: Updates and Implications,” CID Working Paper No. 42; and Abiad et al. (2010).

October 21, 2013

78

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

than analysis based on micro data. In order to deal with the model uncertainty problem, the Bayesian model averaging (BMA) technique proposed by Sala-i-Martin et al. (2004) — which the authors call Bayesian Averaging of Classical Estimates — is used. The method essentially constructs estimates as a weighted average of ordinary least square estimates for every possible combination of the included variables, where the weights applied to individual regressions are justified on Bayesian grounds. The results of the exercise are summarized in Tables A3.2 and A3.3. Since institutions may influence inequality not only directly but also indirectly by shaping the impact of non-institutional drivers, additional equations are estimated that interact the policy variables with the measures of globalization and technological change. The equations are estimated using a standard fixed-effects approach. The interactions are specified as multiplicative terms, which take the form of products of deviations of policies and macroeconomic variables from their sample means: Yit = αGLOBit + βTECHit + γHCit + δPit + φXit + µi + λt + ω(GLOBit − GLOB∗ )(Pit − P ∗ ) + τ (TECHit − TECH∗ )(Pit − P ∗ ) + εit ,

(2)

where GLOB∗ , TECH∗ and P ∗ are the sample means (across countries and over time) of GLOB, TECH and P, respectively, and the other variables are denoted as before. Since no robust evidence could be found for an interaction between policies and, respectively, technological change and financial globalization, the final results are based on a specification that only interacts the trade-to-GDP ratio with the policy indicators. The regression results are summarized in Table A3.3. Annex 3.2: Assessing the determinants of labor earnings inequality based on quantile regressions. Fournier and Koske (2012a) investigate the determinants of labor earnings inequality using household survey data from 32 countries.19 The authors consider all individuals aged between 15 and 64 years who are working 19 Austria, Belgium, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Luxembourg, the Netherlands, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, and the UK (EU-SILC); Australia (HILD); Canada (Survey of Labour and Income Dynamics); Chile (CASEN); Korea (KLIPS); Japan (JHPS); Switzerland (SHP); the US (PSID); Brazil and Israel (LIS).

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

79

either part-time or full-time and earn a positive income during the reference year.20 For each country, they estimate several equations that relate the logarithm of an individual’s labor income to different sets of explanatory variables. In the baseline model, the explanatory variables include the logarithm of the individual’s number of working hours, the gender, the age and age squared, and dummy variables for the education level.21 Several additional variables are of interest but are excluded from the authors’ baseline equation because they exist only for a subset of countries or pose potential endogeneity problems. These are dummy variables for the sector of work and the occupation, the number of years of work experience, and dummy variables for having a temporary as opposed to a permanent work contract, for being self-employed, for being member of a union, for working in the public sector, for having foreign citizenship, for being born in a foreign country, and for having obtained a PhD. These variables are added to the baseline equation in a series of alternative specifications. While standard least squares techniques could be used to study the relationship between the explanatory variables and labor earnings, they would only provide an estimate of the effect on the mean. To draw conclusions about the impact on earnings inequality, it is necessary to go beyond the mean and estimate the effect of the explanatory variables also on other parts of the earnings distribution. Fournier and Koske (2012a) do this using the unconditional quantile regression technique proposed by Firpo et al. (2007a, 2009).22 The method allows investigating the marginal impact of certain characteristics (such as the level of education) on different parts of the earnings distribution such as the median or the 10th or 90th centile. 20 Individuals

with zero or negative income are excluded from the analysis (although rare in the datasets, negative incomes may occur if self-employed individuals make a loss). 21 The level of education is captured by two dummy variables, the first one being equal to unity for all individuals who have at least finished upper-secondary education, and hence also takes value one for those who have a tertiary degree. The second one being equal to unity for all individuals who have finished tertiary education. Hence, the coefficient on the first dummy gives the impact of an upper-secondary or post-secondary nontertiary education relative to lower-secondary education or less and the coefficient on the second dummy gives the impact of tertiary education relative to upper-secondary or post-secondary non-tertiary education. 22 Since this technique is still fairly new, the more established approach of Koencker and Bassett (1978) is also used by Fournier and Koske (2012a). However, because this approach does not allow estimating the impact on overall earnings inequality of changes in the explanatory variables, the focus of the chapter is on the method by Firpo et al. (2007a, 2009).

October 21, 2013

80

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

Table A3.2: Results of the BMA Analysis. Impact of a one-standard-deviation change in the explanatory variable on the dependent variable (in ppt) Dependent variable Specification

90/10 decile ratio (1)

(2)

Trade (as share of 0.6 1.1 GDP) Foreign assets (as −13.0a −19.0a share of GDP) Foreign liabilities 5.0 6.6a (as share of GDP) Patents (per 1 6.6a 4.9a million inhabitants) Share of population −12.6a −5.1a with post-secondary education EPL −9.8a 0.0 EPL on regular work EPL on temporary work Union density (in %) Unemployment benefit replacement rate Minimum wage (as ratio to median wage) Tax wedge (in %) Product market regulation indicator Financial reform index Number of observations Prior mean model size Posterior mean model size

90/50 decile ratio

(3)

(4)

0.1

0.0

0.0

4.2

−0.1

0.6

6.7a

50/10 decile ratio (7) 0.7

−0.1

0.0

0.0

0.0

−2.8

−1.1

−4.2

0.5

(8)

−7.7a

−2.0a

−6.7a

3.1a

0.0

0.1

0.8

0.4

−1.2

0.1

6.2a

5.4a

−5.1a

−5.1a

0.0

−6.9a

0.1

0.0

1.5

−8.5a

−0.4

0.0

a

−0.3

15.8a

−8.2a

0.2

−7.2a

−2.7 4.6a

−7.6a 0.1

−18.9a −23.4a 0.0 0.2

0.0

(9)

3.6a

−0.1

−0.1

−1.2

0.1

0.2

1.4

−0.3a

0.0

(6) 4.5a

16.0

−15.1a

(5)

−9.8a 0.9

−0.1 0.1

−1.8a 0.2

−1.6a 0.2

−8.4a

0.0

0.0

0.2 2.5a

−0.1

−3.6a

−0.2

−2.0a −1.3

−0.1a

0.0

−1.7 8.8a

−1.2

−6.4a −11.2a −1.3a −0.1

0.0

0.0

−4.3a 0.3

−3.1a

311

311

172

311

311

172

311

311

172

8

8

8

8

8

8

8

8

8

9

8

9

3

3

6

9

8

11

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

81

Table A3.2: (Continued ) Dependent variable Specification Trade (as share of GDP) Foreign assets (as share of GDP) Foreign liabilities (as share of GDP) Patents (per 1 million inhabitants) Share of population with post-secondary education EPL EPL on regular work EPL on temporary work Union density (in %) Unemployment benefit replacement rate Minimum wage (as ratio to median wage) Tax wedge (in %) Product market regulation indicator Financial reform index Number of observations Prior mean model size Posterior mean model size

Gini coefficient

Employment rate

(10)

(11)

(12)

(13)

(14)

(15)

0.0

−0.1

−2.6a

−0.5

−2.1a

−3.1

−2.8a

−1.5a

8.1a

0.0

0.4

−3.6

1.5a

1.3a

4.4 −4.3

1.3a

0.3

−8.4a

0.0

0.0

0.1

−4.0a

−3.7a

0.3

0.0

0.1

0.0

−0.2

−1.0a

0.0 −2.3a

0.0

0.0

0.0 −2.3a

−0.2

0.1

−1.4

0.0

−1.5a

−2.5a

−1.9a

−0.1

−2.3a

1.6

−2.0a

−1.6a

0.1

−2.5a

−2.4a

−2.4

0.0a

0.0

0.1 0.8a

0.1 0.2

0.3 1.6a

−1.5a 0.0

−1.0a 0.0

−1.5 −0.6

0.8a

0.7a

0.7a

0.5a

0.3a

−1.1

356

356

177

385

385

203

8

8

8

6

7

7

10

8

10

7

9

12

Note: The one-standard-deviation change in the explanatory variables is calculated across all countries that are included in the regression and across all years between 1981 and 2007 for which data are available. a Denotes that the inclusion probability of the variable is above 50%.

October 21, 2013

82

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

Table A3.3: Results of the Analysis of Interaction Terms for Selected Inequality Measures. Dependent variable

90/10 decile ratio

90/10 decile ratio

Employment rate

Trade (as share of GDP)

0.575∗∗∗ (0.209)

0.659∗∗∗ (0.188)

−12.880∗∗∗ (2.622)

Foreign assets (as share of GDP)

−0.195∗∗∗ (0.044)

−0.149∗∗∗ (0.042)

−0.022 (0.543)

Foreign liabilities (as share of GDP) Patents (per 1 million inhabitants)

0.179∗∗∗ (0.043) −0.002 (0.015)

0.135∗∗∗ (0.042) 0.037∗∗ (0.015)

Share of population with post-secondary education

−1.846∗∗∗ (0.343)

−1.180∗∗∗ (0.326)

10.569b (4.868)

Union density (in %)

−0.821∗∗∗ (0.173)

−3.396 (2.560)

Union density∗∗∗ trade

−0.013∗∗∗ (0.005)

−0.488∗∗∗ (0.070)

EPL EPL∗ trade

Number of cross sections Number of observations Adjusted R2

1.056∗ (0.561) −0.925∗∗∗ (0.201)

−10.845∗∗∗ (3.016) −0.354∗∗∗ (0.130)

22 377 0.97

22 420 0.96

22 516 0.92

Note: All models include country and period fixed effects. Moreover, the models of the 90/10 decile ratio include the female employment share, the employment shares of agriculture and industry, and the output gap as control variables and the model of the employment rate includes the output gap as a control variable. ∗ , ∗∗ , and ∗∗∗ denote significance at the 10%, 5% and 1% significance level, respectively.

The procedure is implemented by simultaneously estimating nine quantile regressions, one for each decile in the range 0.1 to 0.9. For each year and country, the regression thus does not yield a single coefficient, but nine different coefficients, one for each decile. By comparing the different coefficients, it is possible to draw conclusions about the impact of a change in a certain variable on earnings inequality. For example, if a rise in the education level of the workforce has a positive impact on the earnings of all individuals, but the impact is larger for those at the bottom of the earnings distribution (i.e. for lower deciles)

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch03

The Distribution of Labor Income

83

than for those at the top (i.e. for higher deciles), this means that a rise in the education level of the workforce is associated with a fall in earnings inequality. It is important to note that the method is static in nature, thus abstracting from general equilibrium effects. This implies in particular that it accounts only for changes in the composition of the workforce in terms of certain characteristics, ignoring that such changes may alter the relative returns to these characteristics (e.g. a rise in the share of upper-secondary graduates among the working-age population reduces the relative returns to upper-secondary degrees). Annex 3.3: Explaining cross-country differences in labor earnings inequality — a decomposition based on unconditional quantile regressions. There are many ways to decompose cross-country differences in labor earnings inequality (see, e.g. the survey by Fortin et al., 2011). Building on unconditional quantile regressions, Fournier and Koske (2012b) use a methodology close to the one proposed by Firpo et al. (2007b) to decompose cross-country differences into: (i) differences in the composition of the workforce (e.g. inequality should be higher in countries with a more unequal distribution of education endowment) and (ii) differences in rates of return (e.g. inequality should be higher in countries with a larger wage gap between high- and low-educated workers). The measure of labor earnings inequality used for the decomposition is the 90/10 percentile ratio calculated over full-time and part-time workers with positive earnings during the reference year. The US is used as reference country so that each country’s 90/10 percentile ratio is compared with the 90/10 percentile ratio of the US.23 For each explanatory variable, the composition effect is constructed from a comparison of the variable’s mean in the two countries.24 The rate-of-return effect is computed from two separate unconditional quantile regressions, one on the US and another one on the country of interest, and then comparing the coefficients for the 10th and 90th percentiles obtained from the two regressions.25 23 Although

the results depend somewhat on the choice of the reference country, the general conclusions are fairly robust to this choice. 24 While focusing on the mean is a strong simplification for continuous variables such as hours worked, the mean contains all information in the case of dummy variables. 25 The precise magnitude of the rate of return effect needs to be interpreted with caution as it also reflects the uncertainty that is attached to the unconditional quantile regression estimates.

October 21, 2013

84

Income Inequality in the OECD. . .

15:59

9in x 6in

b1598-ch03

Koske, Fournier and Wanner

The US is used as the reference country, 2007a 1.0 0.5 0.0 -0.5 -1.0

Composi on effects: Hours worked Gender Age Upper-sec. or post sec. non-tert. educa on Sum of all rate of return effects Total difference in the 90/10 percen le ra ob

JPN

CAN

DEU

GRC

AUS

GBR

IRL

KOR

CHL

CHE

PRT

AUT

LUX

NLD

POL

ISL

EST

ITA

ESP

FIN

FRA

SVN

NOR

BEL

HUN

CZE

SVK

DNK

-1.5

Ter ary educa on

Fig. A3.1: Decomposition of Cross-country Differences in the Logarithm of the 90/10 Percentile Ratio. Note: a 2008 for Canada; 2009 for Japan. b 90/10

percentile ratio of the country is shown on the horizontal axis minus 90/10 percentile ratio of the US. Source: Fournier and Koske (2012b).

The following findings emerge from this decomposition (Fig. A3.1 and Fournier and Koske, 2012b): • Differences in the rates of return to personal characteristics such as the level of education, the age, or hours worked account for a large part of the gap in the 90/10 percentile ratio between OECD countries. • Cross-country differences in the composition of the workforce in terms of hours worked and the education level also play a role. • Factors such as the share of public sector employment, the share of financial sector employment, or the age structure of the workforce do not play a big role in shaping differences in earnings inequality within the OECD.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

4. INCOME REDISTRIBUTION VIA TAXES AND TRANSFERS Isabelle Joumard, Mauro Pisu and Debbie Bloch

Introduction and Summary The recent financial and fiscal crisis has pushed governments to contain public spending, including cash transfers, while both fiscal consolidation and equity considerations have led some government to raise taxes, in particular on top incomes. Against this background, this chapter assesses different tax and transfer policy approaches and their effectiveness in reducing income inequality. The second section of this chapter reviews the redistributive impact of taxes and transfers. The third section summarizes tax and transfers policy indicators in country profiles and provides an illustration on how these profiles can be used to identify reform options for two OECD countries (Australia and Germany). It also provides a typology of tax and transfer systems, identifying five groups of countries sharing broadly similar features.

Main Findings The redistributive impact of taxes and transfers • Taxes and transfers have a significant redistributive impact. Inequality in income after taxes and transfers, as measured by the Gini index, was about 25% lower than for income before taxes and transfers on average in the OECD area in the late 2000s. For the same period, poverty measured after taxes and transfers was 55% lower than before taxes and transfers for the OECD average.

85

October 21, 2013

86

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

• Countries with a more unequal distribution of market income tend to redistribute more. • Cash transfers reduce income dispersion more than taxes in most OECD countries. On average, three quarters of the reduction in inequality as between market and disposable income are due to transfers, the rest to taxes. • The redistributive impact of cash transfers varies widely across countries. Countries with a similar dispersion of household market income (HMI) opt for distinct redistributive strategies. In some, cash transfers account for a large share of household disposable income (HDI), redistributing income mainly over the lifecycle rather than across individuals — oldage pensions often fall into this category — and their progressivity is low in many countries. Other countries with smaller cash transfers tend to rely more on targeted benefits. And some make little use of cash transfers to reduce income inequality. • Family and housing benefits are, in most countries, the most progressive cash transfers, though their redistributive impact is limited as they are often small in size. Disability and unemployment benefits reduce income inequality although their degree of progressivity depends to a large extent on their design. • The cross-country variation in the redistributive impact of household taxes is more limited than that of transfers, despite large differences in tax-to-GDP ratios. High-tax countries tend to have less progressive household taxes. • The progressivity of labor taxes (including social security contributions) has increased in the majority of OECD countries. Although personal income tax rate schedules have often become flatter reflecting the steep decline in top marginal tax rates, social security contributions for lowincome earners have been cut or tax reliefs made more generous in some countries so as to reduce the cost of labor of groups at high unemployment risk. Furthermore, earned income tax relief has been raised to make work more attractive for low-income earners, raising the progressivity of labor income taxes. • The personal income tax is the most progressive tax, though there are significant cross-country variations. Social security contributions, consumption taxes, and real estate taxes tend to be regressive in most countries.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Income Redistribution via Taxes and Transfers

87

• Tax expenditures pertaining to the personal income tax tend to benefit the well-off, a main exception being in-work tax credits. • The taxation of capital income, wealth, and inheritance has been reduced in many countries, which has reduced the redistributive impact of tax systems.

Indicators of tax and transfer policies help in identifying reform options and different country models • The redistributive impact of taxes and transfers depends on their size, mix, and the progressivity of each component. A set of policy indicators has been compiled which breaks down the redistributive impact of both taxes and transfers into these three dimensions. Individual country profiles facilitate the comparison of each country to the OECD average and thus help in identifying reform options. • Five groups of countries sharing broadly comparable tax and transfer systems have been identified empirically, based on the set of policy indicators: — A “Nordic model” characterized by large and mostly universal cash transfers, a high level of spending on in-kind services and a tax mix which promotes redistribution (all Nordic countries and also Belgium, the Netherlands and Spain are in this group). — A “Continental European model” characterized by large cash transfers with the lion’s share for old-age pensions — that is, redistributing income mostly over the lifecycle instead of across individuals — and a tax mix which does not promote redistribution across individuals reflecting a small role for the personal income tax (Austria, France, and Germany are representative). — An “Anglo-Saxon model”, characterized by small cash transfers and a tax mix which promotes income redistribution. — A lower-income group, where the welfare system is not well developed. Spending on transfers and the level of taxation are considerably below the OECD average, with a heavy reliance on consumption taxes (Chile and Turkey are in this group). — Another group of countries which has little in common, except low public spending on in-kind transfers, low progressivity of cash transfers, and a small role given to personal income taxes.

October 21, 2013

88

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

The Redistributive Impact of Taxes and Transfers Welfare systems, and the taxes and transfers they rely on, are key policy levers to influence distributional outcomes. In most OECD countries, welfare systems have three main objectives: • Redistributing income over the lifecycle (from working age to retirement age), with public old-age pensions largely financed out of social security contributions or general taxation. • Providing income maintenance or insurance to cope with adverse risks, such as unemployment, disability, and sickness. • Avoiding poverty or a too wide dispersion in living standards, with benefits financed mostly out of general taxation. These benefits can be either universal or means-tested. There are large differences in the priorities and organization of welfare systems across OECD countries. Bismarckian-type welfare states rely on social insurance, with benefits financed out of social security contributions. They give priority to the first two objectives. Beveridgean-type welfare states give priority to the third objective with social benefits targeted on those in need and financed by tax revenues. A third welfare model, as recognized by Esping-Andersen (1990), is implemented in the Nordic countries. It involves universal benefits — with a “de-familialization” of welfare responsibilities through a complete coverage both for child and elderly care — and thus a high level of taxation. In practice, welfare systems involve a mix of redistribution between the rich and the poor, risk insurance and lifetime redistribution. They also often provide both means-tested and universal benefits. These features, however, differ between countries. In a comparative study, St˚ ahlberg (2007) estimated that in Australia 38% of lifetime benefits received by individuals were financed through taxes they paid at another stage of their lifecycle, and the remaining 62% involved redistribution between the rich and the poor. In Sweden, 18% of lifetime benefits involved redistribution between individuals and 82% involved redistribution over different phases of the lifecycle of an individual. Sweden, as the other Nordic countries, relies heavily on universal benefits, while Australia relies more on targeted and means-tested transfers. Cross-country differences in the public/private nature of insurance mechanisms and in the taxation of social benefits are also important. Some countries rely mostly on private pension funds to ensure income redistribution across the lifecycle and, often to a lesser extent, on private insurance

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Income Redistribution via Taxes and Transfers

89

companies to provide insurance against health risks. Because contributions to, and benefits received from, private funds are not considered as part of the redistributive system, the size of the welfare system is smaller than in the countries which rely mostly on public coverage. At the same time, the countries that rely more on private schemes may display more progressive public schemes since these do not include pensions and other benefits governed by insurance mechanisms, which often benefit lower-income households less. The taxation of social benefits also affects cross-country comparisons. In some OECD countries, transfers are subject to broadly the same tax treatment as wage income (e.g. Nordic countries), while in others (e.g. Japan), they are largely untaxed. Adema and Ladaique (2009) provide estimates for net public social expenditure, that is, adjusting for the impact of the taxation of social benefits and tax breaks with a social purpose as well as for indirect taxes. This leads to a reassessment of the size of welfare states and to a greater similarity in social expenditure-to-GDP ratios across countries. Although there are various approaches to assess the redistributive impact of taxes and transfers (Box 4.1), all suggest that taxes and transfers

Box 4.1. Assessing the Redistributive Impact and Progressivity of Cash Transfers and Taxes Various measures can be used when assessing the impact of taxes and transfers on income inequality. For both taxes and transfers, three concepts are extensively used in this paper: the redistributive impact, the concentration coefficient, and a progressivity measure. The redistributive impact of taxes and transfers can be defined as the difference in income concentration before transfers and taxes (i.e. HMI) and after taxes and transfers (i.e. HDI). The time horizon matters, however. As an illustration, the rich may pay more social security contributions when they work but also receive more pensions when they retire. This should be adjusted when assessing the redistributive impact of taxes and transfers. If time series on household income before and after taxes and transfers based on micro-data are available, one can disentangle redistribution over the lifecycle from redistribution from the rich to the poor. Unfortunately, these data are scarcely available for international comparisons. Most of the results presented here, thus, rely on (Continued )

October 21, 2013

90

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

Box 4.1. (Continued ) cross-section data, making it difficult to differentiate redistribution over the lifecycle from redistribution from the rich to the poor. How to measure the concentration of market income is also an issue. Two approaches can be used, that is, by ranking households by their market income or by their disposable income. Throughout this chapter, households have been ranked by disposable income. The redistributive impact of taxes and transfers is stronger when calculating the concentration of market income based on ranking households by market income. The difference is most pronounced for countries, such as France, Italy, Austria, and Belgium, which are characterized by a large share of pensions paid to the working-age population due to a low effective retirement age. Pensions redistribute more over the lifecycle of individuals than they do among individuals. Measuring redistribution by the difference between the Gini index of market income, with households ranked by market income, and disposable income may, thus, overestimate the redistributive impact of taxes and transfers. The concentration coefficient for cash transfers provides an indication of who receives the transfers. It is zero when everyone receives the same transfer. The concentration coefficient is negative when the poor receive more in absolute terms than the rich — for example, when benefits are means-tested — and becomes positive when the rich receive more in absolute terms (e.g. old-age pensions). Even in the latter case, there may be redistribution, if market income is distributed more unevenly. Likewise, the concentration coefficient for taxes provides an indication of who pays the taxes across the income distribution. If the richest pay most of the taxes, the concentration coefficient will be high. Still, the concentration coefficient for taxes reflects not only the progressivity of the tax system but also the dispersion of pre-tax income. A more unequal country will raise more tax revenues from the wealthy and less from the poor. It will, therefore, show a higher concentration coefficient even if it features the same tax system as the less unequal country. The redistributive impact of cash transfers and taxes is best measured by the difference in the concentration coefficients for income before and after cash transfers and taxes (Musgrave and Thin, 1948; Norregaard, 1990). It depends on both the size of cash transfers and taxes and their progressivity (Kakwani, 1977, 1979). (Continued )

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Income Redistribution via Taxes and Transfers

91

Box 4.1. (Continued ) The redistributive impact of cash transfers can be expressed as: Concentration coefficient of market income plus transfers − concentration coefficient of market income It is also equal to: Size of cash transfers × Progressivity of cash transfers (Kakwani index), where the size of cash transfers is measured as their share in market income plus transfers and the Kakwani index is defined as (concentration coefficient of cash transfers − concentration coefficient of market income, that is, before taxes and transfers). The redistributive impact of taxes can be expressed in the same way: Concentration coefficient of disposable income − concentration coefficient of market income plus transfers And it is also equal to: Size of taxes × Progressivity of taxes (Kakwani index), where the size of taxes is measured as their share in HDI and the Kakwani index is defined as (concentration coefficient of taxes − concentration coefficient of market income plus transfers). A tax is considered to be progressive when high-income groups face a higher average tax rate than low-income groups (relative progressivity). In some cases, for example as regard consumption taxes, high-income groups pay a higher amount of taxes than low-income groups (absolute progressivity) but still face a lower average tax rate. Taxes are then considered as regressive and the Kakwani index becomes negative. Similarly, cash transfers are considered to be progressive when they account for a larger share of the low-income groups’ income. This definition implies that flat cash transfers (e.g. a minimum pension for all) are considered to be progressive. And cash transfer programs which benefit the rich most in absolute terms (e.g. subsidies for tertiary studies) are still considered as progressive as long as the share of these transfers in household income is lower for high-income than for low-income groups. (Continued )

October 21, 2013

92

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

Box 4.1. (Continued ) In measuring the redistributive impact of taxes and cash transfers, it is assumed that cash transfers are received first and taxes paid afterwards, an approach consistent with OECD (2008a). Indeed, benefits are taxable in many countries. Thus, assessing the redistributive impact of taxes by comparing the concentration coefficient of market income and the concentration coefficient of market income minus taxes would distort the picture — the tax system would appear more regressive than it is. However, in some countries benefits are largely set on the basis of after-tax market income. In this case, comparing the concentration coefficients for market income and for market income plus transfers results in a biased measure of the redistributive impact of benefits. In a study on 14 OECD countries (Immervoll and Richardson, 2011), the measurement approach is chosen to reflect, as far as possible, the actual legal sequence implicit in each country’s tax and benefit system. In particular, the redistributive impact of taxes is determined by comparing Gini indices of market income and net-of-tax income for Australia, the Czech Republic, Germany, Israel, and the US. This option only partly solves the problem since some of the benefits are taxable in these countries (e.g. pensions and unemployment benefits in the US). Overall, however, a simulation carried out on two countries (France and the US) suggests that a different sequencing does not significantly alter the estimated redistributive impact of taxes (or transfers).

are key policy levers to influence distributional outcomes. Information based on household surveys suggests that OECD-wide, taxes and cash transfers reduced the market income dispersion — as measured by the concentration coefficient — by about 25% and relative poverty by about 55% in the late 2000s (Chapter 5).1 Their redistributive impact tends

1 The

OECD Income Distribution and Poverty Database relies on household income surveys carried out by national experts who apply common conventions and definitions, thus enhancing cross-country comparability. In the OECD database, however, income data for Greece, Hungary, Mexico, and Turkey are presented net of taxes and data on household taxes are not available. Data from household surveys face other limitations, discussed in Box 2.1 in Chapter 2.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Income Redistribution via Taxes and Transfers

93

In the late 2000s Redistributive impact of household taxes and transfers 0.20

y = 0.2976x - 0.0136 R² = 0.142

FIN DNK

0.15

CZE BEL

DEU GBR

SWE AUS

NOR SVN

EST AUT

SVK

0.10

FRA NLD OECD-29

LUX POL

NZL ISR

ITA USA

CAN PRT

ESP

JPN

ISL CHE

0.05

CHL KOR

0.00 0.30

0.35

0.40

0.45

0.50

0.55

Inequality in income before taxes and transfers

Fig. 4.1: Taxes and Transfers Reduce Income Dispersion, and More so in “Unequal” Countries. Note: Inequality in income before taxes and transfers is measured by the concentration coefficient. The redistributive impact of taxes and transfers is defined as the difference in the concentration coefficients for income before cash transfers and taxes (i.e. HMI) and after cash transfers and taxes (i.e. HDI). Two approaches can be used to measure the concentration of market income, that is, by ranking households by their market income or by their disposable income. In this illustration, and throughout the paper, households have been ranked by disposable income due to data limitations (see Box 4.1, for more discussion). The trend line shown above has been calculated excluding Chile. Source: OECD Household Income Distribution and Poverty Database, December 2012.

to be high in the Nordic countries (Iceland being an exception) and eastern European countries (Fig. 4.1). It is low in Iceland, Korea, and Switzerland, all characterized by little market income dispersion, and in Chile. The data suggest that there is a positive link between the redistributive impact of taxes and transfers and the level of market income

October 21, 2013

94

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch Point reduction in the concentration coefficients, in the late 2000s

Redistributive impact of household taxes 0.06

0.05 ITA SVN USA 0.04

DEU

AUS

ISR

DNK FIN

CAN PRT

NOR

CZE

FRA

NLD

0.03

GBR

AUT LUX

NZL

OECD-29

BEL

EST

SWE

0.02 SVK ESP POL 0.01

CHL

ISL JPN

KOR CHE a 0.00 0.00

0.02

0.04

0.06

0.08

0.10

0.12

Redistributive impact of public cash transfers

Fig. 4.2: Cash Transfers Reduce Income Dispersion More than Taxes. Note: The redistributive impact of public cash transfers is measured as the difference between the concentration coefficient of market income and that of income after transfers. The redistributive impact of household taxes is measured as the difference between the concentration coefficient of post-transfer income and that of disposable income (i.e. post-tax and transfers). a The redistributive impact of household taxes for Switzerland is slightly negative (−0.006), but has been set to zero. Source: OECD Household Income Distribution and Poverty Database, December 2012.

inequality. Focusing on the working-age population yields a similar pattern.2 Cash transfers reduce income dispersion more than taxes (Fig. 4.2).3 The US, however, is an outlier with virtually the same redistribution achieved through taxes as cash transfers. It relies heavily on the tax code to provide support to low-income groups — the Earned Income Tax Credit 2 When assessing the redistributive impact of taxes and transfers, most analyses (including this paper) assume that taxes and transfers do not affect economic behavior. Incidence issues are discussed at the end of this section. 3 Looking at the working-age population broadly yields the same message.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Income Redistribution via Taxes and Transfers

95

(EITC) is one of the largest US social programs — while other countries rely more on cash transfers.4 The redistributive impact of taxes does not vary widely across countries, despite large cross-country differences in the size of the tax take and progressivity of the tax system. Some countries have opted for a high tax take but, as they are constrained to tax virtually all citizens, with little progressivity. Others have a smaller but more progressive tax system. By contrast, the redistributive impact of transfers displays large cross-country differences. In Denmark, Finland, and Sweden, it is more than five times higher than in Korea and about three times higher than in the US. However, the annual income distribution data may significantly overstate the degree of redistribution across individuals as social security schemes have a — sometimes large — component that provides redistribution over the lifetime rather than redistributing across individuals. Countries that spend the most on cash transfers tend to concentrate more on redistribution across the lifecycle (in particular through old-age pensions). In contrast, those countries that focus more on redistribution between the rich and the poor, through extensive use of targeting, spend less.

The Redistributive Impact of Cash Transfers: Cross-country Differences and Driving Forces The redistributive impact of cash transfers is large but varies a lot across countries The main features of the size and redistributive impact of cash transfers are as follows: • On average across the OECD, cash transfers amounted to 13% of GDP and 24% of HDI in 2009 (Fig. 4.3). They reduced income inequality as measured by the fall in concentration of market income before and after transfers by about 19% in the late 2000s. Those countries that spend the most are not always those where the redistributive impact is strongest. Cash transfers ranged from under 3% of GDP in Mexico to over 20% in

4 In

the OECD Income Distribution and Poverty Database, the Earned Income Tax Credit is booked as a transfer.

October 21, 2013

96

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch 2009 Old age

Incapacity (b)

Family

Unemployment

Other social policy areas

%GDP 21

Total (right scale)

% Household disposable income 40

18 15

35 30 25

12 20 9 15 6

10

3

5

0

0

Fig. 4.3: Public Cash Transfers to Households: Level and Composition.a Note: a The data shown here exclude private mandatory spending which accounts for an important share of total social spending in some countries (in particular Chile, Germany, and Switzerland). In addition, public cash transfers shown here may not fully account for those programs and services provided, or co-financed, by local governments. Measurement gaps may be high, notably in federal countries such as Canada. b Incapacity-related spending covers expenditure on disability pensions and sick leave schemes (occupational injury and other sickness daily allowances). c Data for Switzerland refer to 2008. Source: OECD Social Expenditure Database, February 2013.

Austria and Italy. In Austria, however, the redistributive impact is close to the OECD average and in Italy it is below the OECD average. • There is no clear link between the degree of market income inequality and the redistributive impact of transfers — the most unequal countries do not redistribute more (Fig. 4.4, Panel A). The redistributive impact is highest in the Czech Republic, Finland, Sweden, and Denmark, all characterized by a dispersion of market income close to the OECD average. The redistributive impact of cash transfers is especially low in Korea and Chile, followed by Iceland, the US, and Portugal. Moreover, countries with a similar dispersion in HMI (e.g. Finland and Canada) can opt

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Income Redistribution via Taxes and Transfers

97

In the late 2000s (A) Unequal countries do not redistribute more through cash transfers Redistributive impact of transfers 0.15

0.12

FIN

SWE DNK

0.09

SVK

CZE BEL EST

NOR AUT

SVN

NLD

CHE

ESP

GBR NZL AUS

ITA

JPN

LUX

0.06

DEU

POL FRA OECD-29

ISR

CAN

USA

PRT

ISL 0.03

CHL

KOR

0.00 0.30

0.35

0.40

0.45

0.50

0.55

Inequality in household pre-transfer income

(B) Countries with large cash transfers tend to have less progressive systems Progressivity index of cash transfers 1.0 y = -0.01x + 0.72 R² = 0.30

AUS NZL 0.8

CHL

DNK GBR NLD

ISR

0.6

USA

CZE

CAN

NOR

OECD-29 JPN CHE

0.4

DEU

SVK

FIN BEL

POL SVN

ESP

ISL

KOR

SWE

EST

PRT

FRA

AUT

ITA LUX

0.2

0.0 0

5

10

15

20

25

30

35

40

Cash transfers as a share of household disposable income

Fig. 4.4: The Redistributive Impact of Cash Transfers. Note: Inequality in household pre-transfer income is measured by the concentration coefficient for HMI. The redistributive impact of cash transfers is measured as the difference between the concentration coefficient of market income and that of market income after transfers but before taxes. The progressivity index of cash transfers is the Kakwani index — defined as the concentration coefficient for market income less the concentration coefficient for transfers (see Box 4.1). Source: OECD Household Income Distribution and Poverty Database, December 2012.

October 21, 2013

98

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

for distinct redistributive strategies — the redistributive impact of cash transfers in Finland is about twice as large as in Canada. • The cross-country variation in the redistributive impact of cash transfers reflects differences in the size and progressivity of these transfers. Countries obtain a similar redistributive impact through drastically different size and progressivity combinations (Fig. 4.4, Panel B). For instance, in Portugal and the US, transfers attain about the same reduction in inequality but for different reasons. In the US, the limited reduction in inequality is due to the smaller size of transfers compared with the OECD average, whereas in Portugal it is mainly due to their lower progressivity. • From the mid-1990s to the late 2000s, the redistributive impact of cash transfers slightly weakened on average for the 19 countries for which data are available. This decline is due to a lower size, partly due to a reduction in unemployment, whereas progressivity increased. The lack of, or incomplete, indexation of cash transfers has impinged negatively on their generosity. The examination of 10 countries suggests that incomplete indexation of benefits resulted in recipients losing ground in a majority of these countries (OECD, 2011a). The choice of indexation — to prices versus wages — affects the income loss, which is often more pronounced at lower income levels.5 • The progressivity of transfers significantly varies across benefit schemes. It tends to be lower for old-age and disability pensions as well as unemployment benefits than for family and housing benefits. The transfer mix, thus, plays a role in explaining cross-country differences.

The redistributive impact of pension systems depends on their design Old-age pensions account for the largest share of total cash transfers — 54% in 2009 for the OECD on average. Unsurprisingly, household surveys suggest that for people above 65 years, old-age pensions accounted for the bulk (more than 90%) of total transfers received in the mid-2000s, but for working-age people, their share was also large: 37% for the OECD average 5 As an example, social transfers targeted at low-income groups in France — the minimum income (RMI/RSA) and minimum pensions — are only adjusted for price inflation. As the average wage has grown more rapidly than prices over the past decade, the relative income of social transfer recipients has declined substantially, undoing the discretionary measures to improve their generosity (Conseil des Pr´el` evements Obligatoires, 2011).

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Income Redistribution via Taxes and Transfers

99

but around 80% in Italy and 60% in Poland.6 Old-age pensions can be subdivided into three tiers (OECD, 2011c). The first two tiers are mandatory, whereas the third is voluntary. The first tier has a redistributive function and is publicly provided. The second tier mostly plays an insurance or income-replacement role, ensuring a living standard comparable to that prior to retirement. The third tier reflects voluntary pension arrangements, and is thus likely to have little or no redistributive impact. The OECD has produced a progressivity index for mandatory oldage pensions. For flat basic pensions, the index reaches 100 (OECD, 2011a; Whitehouse, 2006). New Zealand and Ireland are the only OECD countries where public pensions are flat and universal payments without second-tier schemes and, thus, are considered as the most progressive pension system (Fig. 4.5). Other countries with highly progressive pension systems are Canada, the UK, Israel, Korea, and the Czech Republic. On the other hand, Sweden has a regressive mandatory pension scheme, due to the U-shaped profile of the replacement rate.7 Some of the southern and eastern European countries, as well as Finland and the Netherlands, also show little progressivity. Among these countries, Italy and Poland now have defined-contribution second-tier pensions, which have been specifically designed to forge a strong link between contributions and benefits — that is, redistribution over the lifecycle instead of across individuals. Differences in mortality rates across individuals and the design of tax systems — two issues not covered by the progressivity index above — often reduce the progressivity of pensions. Low-income earners tend to die at a younger age than high-income earners (e.g. Christia, 2007; Marmot and Shipley, 1996; Waldron, 2007). The shorter average lifespan

6 The

high share of pensions received by working-age individuals in some countries is attributable to the low effective retirement age. There is a negative correlation between this share and the standard or effective retirement age. The correlation coefficient between them is 0.61 (based on data for the mid-2000s) and is statistically significant at the 1% level. 7 In Sweden, the replacement rates of mandatory pensions are progressive for a large part of the income distribution. The replacement rate declines with income up to a certain income level beyond which it starts to increase. It is also worth emphasizing that pension systems with high payments can be characterized by low overall progressivity and vice versa, as progressivity basically depends on the difference in replacement rates between low- and high-income earners and not on the level of replacement rates.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

100

2008, pensions before tax 100 Pension Gini

Progressivity index 80

60

40

20

0

Ireland

New Zealand

Canada

Israel

United Kingdom

Korea

Czech Republic

Belgium

Australia

Denmark

Switzerland

Japan

Mexico

Iceland

Norway

OECD-34

United States

Chili

France

Austria

Estonia

Slovenia

Spain

Germany

Turkey

Luxembourg

Finland

Portugal

Greece

Netherlands

Italy

Poland

Slovak Republic

Sweden

Hungary

-20

Fig. 4.5: Progressivity Index of the Pension System.a Note: a The progressivity index is calculated considering only the mandatory part of the pension system plus the quasi-mandatory parts with broad coverage. For instance, in Denmark and Sweden there are quasi-mandatory, occupational defined-contribution schemes with broad coverage that are included in the index. The index shown here is not a Kakwani index. It is based on pension systems’ parameters and is computed as 100 minus 100 times the ratio of the Gini of pension payments to the Gini of personal gross earnings in 2008. The Gini indices are calculated using the OECD average earnings distribution. Pension entitlements are computed using the OECD pension model and refer to workers entering the labor market in 2008. The calculations are based on the rules applying in 2008. They include the effects of pension reforms legislated by 2008 but to be phased in later. The pension Gini is the Gini index for pension payments multiplied by 100. Source: OECD pension models and OECD (2011c).

of low-income individuals particularly reduces the progressivity of contributive insurance-type pension systems, as part of low-income earners’ contributions ultimately finance pension payments of high-income earners.8 Taxes and social security contributions can also substantially affect the

8 Empirical

evidence corroborates this (Garrett (1995) and Goda et al. (2009) for the US, and Hachon (2009) for France). Hachon (2009) shows that, for sufficiently large differences in mortality rates, insurance-based pension systems can become regressive.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

101

progressivity of pensions, and therefore the degree of redistribution they achieve (Keenay and Whitehouse, 2003). Pension income is in general taxed at lower rates than work-related income because: (i) the personal income tax is progressive and gross replacement rates are generally below 100% and (ii) pensioners are often exempted from certain types of social security contributions or pay them at a reduced rate. Over the past two decades, pension reforms have slightly reduced pension progressivity on average for the 20 countries for which data are available (Whitehouse et al., 2009), though there was again wide cross-country heterogeneity. The introduction of defined-contribution schemes, which provide a stronger link between pensions and earnings, has resulted in a drop in the progressivity of pensions in countries such as Hungary, Poland, and the Slovak Republic. On the other hand, pension reforms which increased the number of years required to calculate the earnings basis for pension payments (in addition to raising the effective retirement age) have tended to make pension systems more progressive, because higher income individuals typically have a steeper wage profile over their working life.

Disability benefits are redistributive but risk creating poverty traps Cash transfers associated with disability and sickness benefits are also large. They amounted to 2.8% of GDP on average in the OECD in 2009 and reached more than 4% in the Nordic countries, the Netherlands, and Switzerland (Fig. 4.6). In 2007, around 6% of the working-age population received such benefits on average across the OECD, and in some countries this share was well above the unemployment rate (OECD, 2010d). During the past 10–15 years, the number of recipients increased in around half of the OECD countries partly reflecting the use of disability programs to soften the impact of downturns (Ben´ıtez-Silva et al., 2010) and, to a lesser extent, population ageing. The increase also reflects a shift away from early retirement schemes and unemployment benefits as governments have tightened the eligibility criteria of these programs. Although disability benefits reduce income inequality at a given point in time, as net replacements rates for low-wage earners are higher than for high-wage earners (OECD, 2010d), they may increase lifetime income inequality by reducing income mobility and by creating poverty traps. Across the OECD, disability benefits are often the only means of sustenance for people with a disability but often do not suffice to escape

October 21, 2013

102

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

% GDP 7

6

2009

2000

1990

5

4

3

2

1

0

Fig. 4.6: Spending on Incapacity-related Benefits. Note: Data include public and private disability pensions and paid sick leave schemes (occupational injury and other sickness daily allowances). Data for Canada do not include spending on provincial social assistance payments with a disability designation (which would roughly double the spending figure), nor spending on voluntary private long-term disability plans. Data for Switzerland refer to 2005 instead of 2009. Source: OECD Social Expenditure Database, February 2013.

poverty.9 People receiving disability benefits, indeed, have lower employment and higher unemployment rates than people with no disability. Many countries have thus started to reform disability benefits to reach a better, and more sustainable, balance between income security and labor market (re-)integration of disabled people.

Unemployment benefits are mostly insurance-based and thus not very progressive The redistributive effect of unemployment benefits depends on the system’s parameters including eligibility criteria, replacement rates for low- and 9 On

average in the OECD, 22% of households with a disabled person receiving benefits live in poverty, compared with around 14% for other households. This gap is especially high in Australia, Ireland, Korea, the UK, and the US (OECD, 2010d).

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

103

high-income earners and benefit duration. These benefits are conditional on past contributions and are earnings-related in most countries. Greece, Iceland, Ireland, Poland, and the UK are exceptions since they provide flat though relatively low benefits, enhancing the redistributive impact of unemployment benefits. Many countries impose a ceiling on unemployment benefits, thus also introducing an element of progressivity, but it may be high, whereas in a few others, there is no upper limit. The progressivity of unemployment benefits can be gauged by comparing the net replacement rates of low- and high-income earners. Figure 4.7 shows that unemployment benefits for the initial phase of unemployment tend to be progressive, with a wide cross-country variation.10 Unemployment benefits have become slightly more progressive over the past decade, in particular in those countries where they were initially less progressive (Fig. 4.8). In contrast, progressivity has declined in countries where the systems were the most progressive. Most OECD countries also operate minimum-income programs as a last resort safety net (Immervoll, 2010). The benefit level is unrelated to previous income and has, thus, a considerable redistributive effect. The overall redistributive impact is, however, difficult to gauge as it depends on how minimum income benefits are combined with other social assistance schemes. For instance, in continental Europe, minimum income programs often complement other benefits delivering important first-tier safety nets, whereas in Australia and New Zealand, it represents the main benefit for individuals without income.

Family cash benefits are targeted toward low-income groups Family cash benefits have a rather strong redistributive impact. Although they account for a rather small share of total cash transfers in most countries, they tend to be more progressive than other transfers.11 Data from

10 Progressivity

of unemployment benefits also depends on the type of family. For instance, unemployment benefits for a family with children are slightly less progressive than transfers to those without. Long-term unemployment benefits and social assistance for people who have been unemployed for 5 years or more are on average as progressive as those granted during the initial phase of unemployment spells. But the most progressive countries are not the same. 11 Family cash benefits do not include public spending on services for families with children (e.g. direct financing and subsidizing of childcare providers and early education facilities) nor financial support for families provided through the tax system.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

104

a

Measured by the difference in net replacement rates between low and high income earners 2009, initial phase of unemployment 50

Single person with no children

Single person with two children

40

30

20

10

0

-10

Fig. 4.7: Progressivity of Unemployment Benefits Net of Taxes. Note: a Progressivity is defined as the difference in the net replacement rate for low and high earners having earnings equal to 67% and 150% of the average wage. The larger the difference, the more progressive are unemployment benefits. Countries are ordered from the smallest to the largest values in progressivity for a single person with no children. These replacement rates refer to all unemployment benefits and not only to those insurance-based but no social assistance “top-ups” are included in either the in-work or out-of-work situation. One limitation of this indicator is that it does not take into account the duration of the unemployment spell. In addition, it is based on net replacement rates and, thus, also reflects the progressivity of personal income taxes (any income taxes payable on unemployment benefits are determined in relation to monthly values, multiplied by 12, even if the maximum benefit duration is shorter than 12 months). Children are aged 4 and 6 years and neither childcare benefits nor childcare costs are considered. Source: OECD, “Tax-benefit Models.” Available at: www.oecd.org/els/social/workincentives, September 2011.

the mid-2000s household surveys suggested that the redistributive impact of family cash benefits was the largest in Ireland, followed by the Netherlands, Australia, and Austria, but well below the OECD average in the US, Switzerland, Portugal, Denmark, and Norway. As with other cash benefits, cross-country variations in the redistributive impact reflect differences in the size and progressivity of such benefits. While the majority of OECD

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Income Redistribution via Taxes and Transfers

105

2001 to 2009 Change 25

ITA

20 KOR 15 CZE JPN 10 PRT

ESP HUN BEL CAN AUT NOR DNK OECD-28 GRC GBR CHE USA SWE AUS NZL LUX POL FRA FIN ISL IRL

5

NLD

DEU 0 -5 -10

SVK -35 -20

-10

0

10

20

30

40 50 Level in 2001

Fig. 4.8: Change in the Progressivity of Initial Unemployment Benefits for a Single Person. Source: OECD, “Tax-benefit Models.” Available at: www.oecd.org/els/social/workincentives, September 2011.

countries implement universal family cash benefits, some rely on incometested schemes. In addition, child benefits based on the number of children in the household can benefit low-income groups more as these households often have more children, at least in some countries including the UK (ONS, 2010).

The Redistributive Impact of Taxes: Cross-country Differences and Driving Forces The overall redistributive impact of taxes depends on the amount of taxes collected (size), the tax mix, and the progressivity of each tax. Household surveys are useful as they provide actual tax payments including tax

October 21, 2013

106

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

relief. But they exclude social security contributions paid by employers and consumption taxes. In addition, household taxes covered by household surveys vary across countries. This section thus starts with information derived from household surveys to assess the redistributive impact of taxes and then completes the picture by using additional information.

Redistribution via household taxes as gauged by household surveys The redistributive impact of taxes can be gauged by the difference in the concentration coefficients for income before and after taxes, as derived from household surveys (Box 4.1). In the late 2000s, the redistributive impact was the highest in Australia, Denmark, Germany, Israel, Italy, Slovenia, and the US (Fig. 4.9, Panel A). It was by far the lowest in Switzerland, followed by Korea, Chile, Iceland, and Japan. Some of the countries with the highest inequality in market income tend to redistribute more through household taxes than less unequal countries. Australia, Israel, Italy, the UK, and the US are examples. Chile, however, clearly stands out, having a wide market income dispersion combined with little redistribution via the tax system. The redistributive impact of household taxes depends on both their share in disposable income (i.e. their size) and their progressivity. It varies little across countries, despite large cross-country differences in the size of taxes. As an illustration, household taxes absorbed more than 35% of HDI in Austria, Denmark, and Sweden in the late 2000s, but their redistributive impact was lower than in Australia, Israel, and the US, all characterized by a much lower tax-to-income ratio. In many high tax countries, taxes have a relatively low redistributive impact because they embody little progressivity (Fig. 4.9, Panel B) — this is particularly the case in Denmark, Iceland, and the Netherlands. And household taxes are more progressive in the US than in most EU countries.12 However, some countries (including Chile, Korea, and Japan) combine a relatively low tax burden with very little progressivity. 12 Various studies have compared the progressivity of tax systems of European countries with that of the US (see, for instance, Joumard, 2001; Piketty and Saez, 2007; Prasad and Deng, 2009). Though they use different definitions, methods, and databases, they reach the same conclusion: the US tax system is more progressive than that of the continental European countries.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Income Redistribution via Taxes and Transfers

107

In the late 2000s Panel A. Unequal countries tend to redistribute more through household taxes Redistributive impact of taxes 0.05 0.05 ITA

DEU SVN DNK

0.04 0.04

FIN

y = 0.1228x - 0.0105 R² = 0.1465

ISR USA

AUS CAN PRT

GBR

NOR NLD CZE

0.03 0.03

BEL AUT

SWE

LUX

NZL

OECD-28 EST

0.02 0.02 SVK ESP POL JPN

ISL

0.01 0.01

CHL

KOR

0.00 0.00

CHE -0.01 -0.01 0.25 0.25

0.30 0.30

0.35 0.35

0.40 0.40

0.45 0.45

0.50 0.50

Gini of household market income plus transfers

Panel B. High tax countries tend to have less progressive household taxes Progressivity index of household taxes 0.25

AUS

y = -0.0019x + 0.1594 R² = 0.1202

ISR

0.20 USA CZE CAN PRT GBR

0.15 EST

KOR

SVN

NOR

OECD-28

ESP

LUX NZL

BEL

AUT

DNK

SWE

CHL

0.05

DEU

FIN

SVK

0.10

ITA

NLD POL

JPN

ISL

0.00 CHE

-0.05 0

10

20

30

40

50

60

Household taxes as a share of household disposable income

Fig. 4.9: The Redistributive Impact, Size, and Progressivity of Household Taxes. Note: The redistributive impact of household taxes is measured as the difference between the concentration coefficient of income after transfers but before taxes and that of disposable income (i.e. after taxes and transfers). The progressivity index of household taxes is the Kakwani index computed as the concentration coefficient for taxes less the concentration coefficient for income after transfers and before taxes (see Box 4.1). In Panel A, the trend line excludes Chile. Source: OECD Household Income Distribution and Poverty Database, December 2012.

October 21, 2013

108

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

Going beyond household surveys when assessing the redistributive impact of taxes Assessing the redistributive impact of tax systems based on household surveys has serious limitations. The data are available for only a few years, which make it difficult to assess the impact of tax reforms. Furthermore, most household surveys focus on the personal income tax, social security contributions paid by employees, and, sometimes, property taxes.13 They do not, therefore, take into account consumption taxes, employers’ social security contributions, and corporate income taxes, thus leaving aside more than 50% of total tax revenues on average across the OECD. This omission creates serious biases since the tax mix varies widely both across countries and over time (Fig. 4.10). In particular, consumption taxes — which are often seen as being regressive (see below) — have declined as a share of total tax revenue in most OECD countries.14 In 2010, this share ranged from about 15% in Japan and the US to over 40% in Chile, Hungary, Mexico, and Turkey. The rest of this section will go beyond household surveys to assess the redistributive impact of taxes, by relying on statutory tax schedules and actual tax revenues as well as a literature review.

Labor income taxes: progressivity indicators based on statutory tax schedules The progressivity of tax systems can be measured by statutory tax schedules. Compared with those derived from household surveys, such measures have the main advantage of being unaffected by cross-country differences in the definition of household taxes and by differences in responding to household surveys. The OECD has built synthetic indicators of labor income tax progressivity, based on statutory tax schedules, over a wide income range.15 Focusing on a single wage earner, Fig. 4.11 (Panel A) shows that 13 In

OECD (2008a), the data are drawn from household surveys and household taxes should in principle cover personal income tax, social security contributions paid by employees (but not by employers), and property taxes. In practice, however, the coverage varies from one country to another. Furthermore, the database does not allow a breakdown of these taxes. 14 Between 1975 and 2010, the share of consumption taxes in total tax revenues dropped from 26% to 23% in Canada, from 32% to 24% in France and from 17% to 15% in the US. 15 Taxes on labor income consist of personal income taxes, social security contributions paid by employees, as well as payroll taxes when relevant. The calculation and coverage

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Income Redistribution via Taxes and Transfers OECDa

109

United States

Personal income tax

Consumption taxes

Property taxes

Personal income tax

Consumption taxes

Property taxes

Corporate income tax

SS & payroll taxes (b)

Other

Corporate income tax

SS & payroll taxes (b)

Other

% GDP

% GDP

40

40

40

40

35

35

35

35

30

30

30

30

25

25

25

25

20

20

20

20

15

15

15

15

10

10

10

10

5

5

5

5

0

0

0

0

European Unionc

Japan Personal income tax

Consumption taxes

Property taxes

Personal income tax

Consumption taxes

Property taxes

Corporate income tax

SS & payroll taxes (b)

Other

Corporate income tax

SS & payroll taxes (b)

Other

% GDP

% GDP

40

40

40

40

35

35

35

35

30

30

30

30

25

25

25

25

20

20

20

20

15

15

15

15

10

10

10

10

5

5

5

5

0

0

0

0

Fig. 4.10: Tax Revenues: Level, Composition, and Change Over Time. Note: a Unweighted average excluding Chile, the Czech Republic, Estonia, Hungary, Israel, Mexico, Poland, the Slovak Republic, and Slovenia. b Social security contributions paid by employees and employers. c Unweighted average excluding the Czech Republic, Estonia, Hungary, Poland, the Slovak Republic, and Slovenia. Source: OECD Revenue Statistics Database, February 2013.

(footnote 15 continued) of progressivity indices based on tax schedules differ from the Kakwani index presented above. In particular, they include standard cash (mostly family related) transfers and do not account for non-standard tax reliefs, such as those associated with mortgage interest payments. Drawing international comparisons for single taxpayers without children allows leaving aside most cash transfers to focus mostly on taxes. The correlation between the progressivity index based on tax schedules (for single taxpayers and under the assumption of a similar distribution of income across countries) and the Kakwani progressivity index based on household surveys is 0.58, significant at 5%.

October 21, 2013

110

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch Based on statutory tax schedules for single tax payers without children Net personal tax progressivity, synthetic indicator Progressivity at the lower end of the income distribution, right scale Progressivity at the higher end of the income distribution, right scale

15

0.25

(A) Level, 2009 0.20

12

0.15

9

0.10

6

0.05

3

0.00

0

(B) Change over the period 2000–2009 0.08

8

0.06

6

0.04

4

0.02

2

0.00

0

-0.02

-2

-0.04

-4

-0.06

-6

-0.08

-8

Fig. 4.11: Progressivity of Statutory Personal Income Tax and Employee Social Security Contribution Schedules. Note: Net personal income tax is defined as the sum of personal income tax and employee social security contributions net of standard cash transfers. Standard tax relief measures — including those linked to marital and family status and income level — are accounted for. Non-standard tax relief measures, that is, those determined by reference to actual expenses incurred (such as the amount of interest paid on loans), are not

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

111

there is wide cross-country variation in the degree of progressivity of statutory schedules for personal income taxes and employees’ social security contributions. Also interesting are the differences in the progressivity structure along the income ladder. Progressivity at the higher end of the income distribution — that is, income equal to 167% of the average wage — is strongest in Ireland, Sweden, and Denmark. At the lower end of the income distribution — that is, income equal to 67% of the average wage — tax schedule progressivity is highest in Hungary, Luxembourg, and Belgium. The US tax schedule has slightly more progressivity at the upper end but its overall progressivity is below the OECD average — at least for single wage earners — while Japan and Korea have very little progressivity embedded in their tax schedule for labor income.16

Despite cuts in top marginal rates, labor taxes have often become more progressive Statutory tax schedules can also be used to assess changes in progressivity over time. Personal income tax schedules have generally become flatter over the past decades (Sabirianova Peter et al., 2008; Piketty and Saez, 2007). Top marginal rates have declined in the vast majority of the OECD countries since 2000, by more than 10 percentage points in Belgium, the Czech Republic, France, Mexico, and the Slovak Republic (Table 4.1). In addition, the income thresholds from which these top marginal rates apply ←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− Fig. 4.11: (Continued) included. The synthetic indicator for net personal tax progressivity presented here is not a Kakwani index. It is calculated as the difference between the average net personal tax rate at two income levels based on the assumption of a similar income dispersion across OECD countries. This difference is then divided by the change in income level. Progressivity at the lower end (respectively higher end) of the income distribution is computed as the difference in personal income tax rates (personal income tax and employee social security contributions expressed as a percentage of gross wage earnings) between the average wage and 67% of the average wage (respectively between 167% of the average wage and the average wage). Source: OECD (2008b). 16 Due to the various tax reliefs for low-income earners with children, the US personal income tax is much more progressive for families with children than for single tax payers.

October 21, 2013

15:59

112

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch Table 4.1: Top Personal Income Tax Rates and Thresholds. Top statutory income tax rate (%)a

Threshold (multiple of the average wage)b

2000

2009

Change 2000 to 2009

2000

2009

Change 2000 to 2009

Australia

48.5

46.5

−2.0

1.2

2.8

1.6

Austria Belgium

50.0 63.9

50.0 53.7

0.0 −10.2

2.3 1.2

2.1 1.1

−0.2 −0.1

Canada Czech Republic

46.4 32.0

46.4 15.0

0.0 −17.0

1.7 2.4

2.9 0.4

1.2 −2.0

Denmark

59.7

51.6

−8.1

1.0

1.0

0.0

Finland France

55.2 58.3

49.1 47.8

−6.1 −10.5

2.1 2.9

1.8 2.8

−0.3 −0.1

Germany

53.8

47.5

−6.3

1.7

6.2

4.5

Greece Hungary

45.0 40.0

40.0 36.0

−5.0 −4.0

3.8 0.9

3.6 0.8

−0.2 −0.1

Iceland Ireland

45.4 44.0

37.2 41.0

−8.2 −3.0

1.5 1.0

0.3 0.9

−1.2 −0.1

Italy

46.4

44.9

−1.5

3.9

3.2

−0.7

Japan Korea

50.0 44.0

50.0 38.5

0.0 −5.5

4.5 5.5

4.6 3.2

0.1 −2.3

Luxembourg

47.2

38.9

−8.3

2.1

1.0

−1.1

Mexicoc Netherlands

40.0 60.0

28.0 52.0

−12.0 −8.0

49.3 1.6

4.7 1.2

−44.6 −0.4

New Zealand

39.0

38.0

−1.0

1.7

1.5

−0.2

Norway Poland

47.5 40.0

40.0 32.0

−7.5 −8.0

2.6 3.3

1.6 2.8

−1.0 −0.5

Portugal Slovak Republic

40.0 35.0

42.0 19.0

2.0 −16.0

3.4 3.2

4.3 0.5

0.9 −2.7

Spain

48.0

43.0

−5.0

4.4

2.4

−2.0

Sweden Switzerland

55.4 43.2

56.5 41.7

1.1 −1.6

1.5 4.0

1.5 3.6

0.0 −0.4

Turkey

35.6

35.6

0.0

8.1

3.0

−5.1

United Kingdom United States

40.0 46.7

40.0 41.9

0.0 −4.8

1.4 8.9

1.3 9.6

−0.1 0.7 (Continued )

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Income Redistribution via Taxes and Transfers

113

Table 4.1: (Continued ) Top statutory income tax rate (%)a

OECD average Standard deviation

2000

2009

Change 2000 to 2009

46.7

41.5

−5.2

7.9

9.3

1.5

Threshold (multiple of the average wage)b

2009

Change 2000 to 2009

2.9

2.5

−0.4

2.0

2.0

0.0

2000

Note: a These are the top statutory tax rates (combined central and sub-central) that apply from the threshold levels reported in the fourth and fifth columns. b These columns report the level of gross wage earnings (expressed as a multiple of the average wage) at which the top personal income tax rate starts to apply. The average and dispersion exclude Mexico. c The threshold figure for Mexico in 2000 reflects a tax schedule with two supplementary brackets designed to tax very high earners more heavily. These supplementary brackets were removed in 2002, resulting in the threshold of the upper bracket coming down sharply as a proportion of average earnings. Source: OECD Tax Database, 2009.

have been raised in some countries including Australia, Canada, Germany, and the US. Finally, cross-country variation in top rate thresholds is wide: the top rate applies to those earnings equal to the average wage in Denmark while in the US, it applies only to those earnings at about 10 times the average wage. Despite cuts in top rates, tax schedule progressivity has increased in a majority of OECD countries since 2000, largely driven by changes at the lower end of the income distribution (Fig. 4.11, Panel B). To make work more attractive for spouses and low-paid workers, many countries (including Belgium, Canada, Finland, France, the Netherlands, the Slovak Republic, Sweden, the UK, and the US) have introduced or strengthened in-work benefits (IWB) targeted at low-income groups, thereby incidentally increasing the progressivity of the personal income tax. Several countries (including Austria, Finland, Germany, Italy, and the Slovak Republic) have also raised the tax-free allowance17 or made social security contributions 17 The Slovak Republic replaced the progressive personal income tax system by a flat rate system in 2004. However, the basic tax allowance, which declines as income grows, and the refundable tax credit targeted at low-income earners introduced in 2009 significantly reinforce tax progressivity at the lower end of the income distribution.

October 21, 2013

114

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

less regressive by removing or raising social security contribution caps or floors (e.g. France and the US).18 In parallel, partial or total exemptions for social security contributions below a given income threshold have been introduced or made more generous in some European countries (including Austria, Belgium, France, Spain, and the UK) to reduce the cost of labor for low-paid workers.

The progressivity of the personal income tax is often hollowed out by tax expenditures The use of tax expenditures has grown in many OECD countries (OECD, 2010e) and their value tends to increase with income. With the main exception of earned-income tax credits targeted at low-income groups, the value of tax reliefs often increases for higher tax brackets, because the income or transaction targeted is most commonly used by higher-income individuals.19 Tax breaks for health and child care, education, owner-occupied housing, and retirement savings often fall into this category. Regarding the latter, Antolin et al. (2004) confirm that in Canada, the UK, and the US voluntary tax-favored retirement schemes disproportionately benefit upper income individuals. Overall, tax expenditure can, thus, significantly reduce the progressivity of the personal income tax. Landais et al. (2011) further show that, in France, tax expenditures result in a decline of the effective personal income tax rate beyond an income threshold. Only the US provides much information on who benefits from tax expenditures. There, tax expenditures pertaining to the personal income tax clearly raise after-tax incomes more for higher-income than lowerincome taxpayers (Burman et al., 2008). More than 90% of the savings 18 In

Denmark, Hungary, Spain, and Switzerland, minimum amounts of social security contributions — floors — still have to be paid by employees and/or employers. Caps are in place in many OECD countries. 19 Tax reliefs take different forms, including: (i) tax allowances and exemptions (amounts are deducted/excluded from the tax base); (ii) rate relief (a reduced rate of tax applied to a group of taxpayers or transactions); (iii) tax deferral (a delay in paying the tax); and (iv) tax credits (amounts deducted from tax liabilities). Standard tax reliefs — including those linked to marital and family status and income — are accounted for in the data on progressivity of personal income taxes and employee social security contributions from Taxing Wages. Thus, EITC-type tax reliefs are included. Non-standard tax reliefs, that is, reliefs determined by actual expenses incurred (such as the amount of interest paid on loans), are not included.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

115

from preferential tax rates on long-term capital gains and qualified dividends go to taxpayers in the top quintile of the income distribution, and nearly half of the benefits go to people in the top 0.1% (Williams, 2011). Concerning healthcare-related tax expenditure (1.3% of GDP), Toder et al. (2009) estimated that more than 40% of the implicit subsidy accrues to the 20% richest households. Likewise, almost 70% of the implicit subsidy associated with the deductibility of mortgage interest on owner-occupied homes benefits the top income quintile.20

Taxes on capital income have been reduced and are often lower than taxes on labor income Various savings schemes have long been granted a preferential tax treatment in most OECD countries. Since capital income tends to be concentrated in upper income brackets, such tax relief implies less progressivity of the income tax. Governments promote private pensions by means of tax incentives in most OECD countries. In the most common regime, private pension savings can be deducted from the income tax base and accrued return on investment is exempt from taxation, but pension benefits arising from these savings are taxed (Yoo and de Serres, 2004). Housing investment also often benefits from a favorable tax treatment with imputed income from owner-occupied dwellings and capital gains on the sale of a principal residence taxed less than other capital income, or not taxed at all, while interest payments on debt-financed investment in owner-occupied housing are sometimes deductible from taxable personal income. These tax-favored schemes, however, tend to affect the mix, rather than the volume, of private savings. They may, thus, divert saving and investment away from other activities that may be more conducive to growth. The low taxation, if any, of capital gains on shares also has important distributional consequences, though its impact on growth is subject to debate. It is often argued that a lower tax rate on capital gains encourages risktaking and entrepreneurship, thus promoting growth, while high capital gains taxes create an inefficient “lock-in” effect. Burman and Moynihan (2011) show 20 Matsaganis and Flevotomou (2007) used the tax-benefit model EUROMOD to quantify the distributional impact of mortgage interest tax relief in the Netherlands, Sweden, Finland, Italy, and Greece. They show that higher-income groups capture a disproportionate share of mortgage interest tax relief in all these countries. The effect is most regressive in the Netherlands and least regressive in Sweden.

October 21, 2013

116

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

that 94% of the value of capital gains tax breaks in the US benefit the top quintile. They consider that, since capital losses are often supported by the government (in the form of reduced taxes), capital gains should be taxed. They also argue that the lock-in effect is small. Because it is often difficult to distinguish between labor and capital income, in particular for the selfemployed, low taxation of capital income further creates opportunities for income-shifting and tax planning (Diamond and Saez, 2011). And it may ultimately favor top-income groups most (top executives, finance professionals, and entrepreneurs), who can benefit from carried interest arrangements and the low taxation of stock options (OECD, 2011a).21 Some countries have moved further toward a non-progressive and reduced taxation of most capital income, in particular, in Europe. The main objective of such reforms is to reduce tax distortions across savings instruments and incentives for capital exports. Precursors were the Nordic countries which adopted a dual income tax system in the late 1980s or early 1990s. Under such a system, a unique flat tax rate applies to net capital income (interest income, dividends, and capital gains), while labor income is subject to a progressive tax schedule. Many countries have not adopted a “pure” dual income tax system but they increasingly tax interest income at flat rates, usually lower than the marginal rates which apply on labor income (Joumard, 2001). In contrast, several countries have continued to tax most types of capital income as labor income, that is, at progressive rates, including Canada, Ireland, Korea, Luxembourg, Switzerland, Turkey, the UK, and the US (OECD, 2006).

Property taxes play a minor role in many OECD countries Raising property taxes is often presented as one option to increase the redistributive impact of tax systems. Property taxes amount to more than 10% of total tax receipts in several OECD countries and about 2% of GDP on average across the OECD. As a share of GDP, they are highest in the UK, Canada, France and the US and lowest in several continental European countries and Mexico (Fig. 4.12). Recurrent taxes on immovable property account for the bulk in most countries though taxes on financial and capital

21 Individuals may benefit from “carried interest” arrangements when they have a relatively small equity stake in a business. If successful, rewards are taxed as capital gains, hence at a rate that is generally below their marginal personal income tax rate.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

117

2010, percentage of GDP Recurrent taxes on immovable property Estate, inheritance and gift taxes % GDP 4.5

Net wealth taxes Taxes on financial and capital transactions

Other

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Fig. 4.12: The Property Tax Take Varies Significantly across OECD Countries. Source: OECD Revenue Statistics Database, February 2013.

transactions play a dominant role in Belgium, Greece, Korea, Italy, and Turkey.

Real estate taxes are regressive in some countries While high-income households pay more recurrent taxes on immovable property in absolute terms, real estate taxes often absorb a larger share of the income of the poorer households. In the UK, for instance, the Council Tax amounts to 6% of income for those in the bottom decile, but to less than 2% for those in the top decile (Fig. 4.13). A similar pattern is found in Canada (Chawla and Wannel, 2003) and the US (Davies et al., 2009). One explanation is that recurrent taxes on immovable property are often a subnational government tax whose amount should reflect the benefit of local public services (waste collection, etc.) which do not increase much with income. In some countries, real estate taxes are also paid by renters,

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

118

% Gross income

United Kingdom

7 6 5 4 3 2 1 0 1st

2nd

3rd

4th

% Household income 6

5th

6th

7th

8th

9th

10th Decile

New York

5 4 3 2 1 0 Lowest

Second

Middle

Fourth

20%

20%

20%

20%

% Household income 6

Next 15%

Next 4%

Top 1%

To p 2 0 %

Texas

5 4 3 2 1 0 Lowest

Second

Middle

Fourth

20%

20%

20%

20%

Next 15%

Next 4%

Top 1%

To p 2 0 %

Fig. 4.13: Real Estate Taxes Tend to be Regressive in the UK and the US. Note: Data for the UK refer to the Council tax and Northern Ireland rates, less discounts, Council tax benefits, and rate rebates, for non-retired households for the fiscal year 2009/2010. They are expressed as a percentage of gross income (market income plus transfers). The data for New York and Texas cover property taxes — property

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

119

who often have low income. Part of the regressive nature of real estate taxes may also reflect the fact that many pensioners own expensive houses but receive relatively little income. Still, Palameta and Macredie (2005) found that this is only part of the story in Canada, since non-seniors make up the majority of lower-income homeowners. Some countries have introduced tax allowances, income-conditional exemptions, or progressive tax rates, to reduce real estate tax payments by low-income groups. For instance, regressivity is mitigated in some US states by a flat dollar amount exemption — for example, homestead exemptions — or a tax credit designed to assist low-income taxpayers. Similar tax relief is provided by some Canadian provinces and municipalities. And in France, generous income- and family-related tax relief has succeeded in making the largest recurrent tax on immovable property (Taxe d’habitation) slightly progressive since 2000 (Conseil des Pr´el`evements Obligatoires, 2011; Marical, 2009) at least for the first part of the income distribution.22

←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− Fig. 4.13: (Continued) primarily includes homes, but may include property other than real estate such as cars — paid by non-elderly households in 2007. The states of New York and Texas were chosen as examples for the US to represent states with above-average (New York) and belowaverage (Texas) property values. In New York State, there is no general homestead exemption for all homeowners, but homeowners with income below $500,000 are entitled to a partial school tax exemption. In Texas, there is a basic homestead exemption available to all homeowners. Source: ONS (2011). “The Effects of Taxes and Benefits on Household Income, 2009/10.” Available at: http://www.ons.gov.uk/ons/rel/household-income/the-effectsof-taxes-and-benefits-on-household-income/2009-2010/the-effects-of-taxes-and-benefitson-household-income-2009-10.pdf. Accessed December 2012; ITEP (2009). “Who Pays? A Distributional Analysis of the Tax Systems in All 50 States.” Available at: http://www. itepnet.org/whopays3.pdf.

22 The

Conseil des Pr´ el` evements Obligatoires (2011) also notes that existing taxes on real estate transactions — Droits de mutation a ` titre on´ ereux — are regressive as they are calculated as the sum of a fixed amount and a percentage of the value of the transaction. The effective tax rate on small transactions is de facto much higher and often paid by low-income families.

October 21, 2013

120

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

Wealth, inheritance, and gift taxes have been reduced in many countries Taxing household wealth, either annually or at the time of transfer (gift and inheritance), is attractive for various reasons. First, the tax base is large and had grown briskly prior to the crisis. As an illustration, households’ net worth in France is five times higher than GDP and it has grown by over 7% on average per year and in real terms between 1997 and 2007, compared to a 1.7% increase for GDP (Conseil des Pr´el`evements Obligatoires, 2011). Taxing wealth at a rather low rate should thus generate large and rising tax revenues. Second, real estate accounts for a large share of household net worth (Chapter 7) and the tax can thus hardly be avoided. Third, inheritance and gift taxes, not only on immovable property but on all net assets, could offer an alternative to the taxation of lifelong saving. It can be considered as a way of taxing, for example, income or capital gains that were tax-exempt during a person’s lifetime. Inheritance taxes have the advantage of generating less distortions than annual wealth taxes because inheritance is difficult to plan. Several countries, including the US, have made inheritance and gift taxes highly progressive by providing tax-free allowances and by applying progressive rates.23 Fourth, wealth is more concentrated than income and is becoming more unequally distributed. In the seven OECD countries covered by the Luxembourg Wealth Study (LWS), the 10% richest households hold between 40% (Italy) and 70% (the US) of total wealth. And the value of inheritances and gifts as a share of GDP has grown rapidly in some countries — it has risen threefold since 1950 to reach almost 15% in 2008 in France (Piketty, 2010) — perpetuating income inequality. In practice, however, the use of wealth and inheritance taxes has declined. Wealth taxes have been abolished in about one-third of the OECD countries since the mid-1990s and several countries (including the UK) never had this type of tax.24 In 2010, only four OECD countries (Canada, France, Norway, and Switzerland) still imposed wealth taxes (Price and Dang, 2011).

23 Roach (2003) shows that the US estate and gift taxes were the most progressive element of federal taxation because they applied only to large estates/gifts and because rates were progressive. 24 Among the countries that reduced or abolished wealth taxes since the mid-1990s, some have increased the taxation of top incomes (Germany) or capital income (Luxembourg and the Netherlands). Spain introduced a wealth tax in September 2011.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

121

Inheritance and gift taxes are applied rather widely but several countries have reduced or abolished them since the mid-1990s (including Austria, France, New Zealand, Portugal, Slovak Republic, and Sweden). The risk of wealth/capital flight to low-tax countries, as well as administrative and collection costs, have often been cited as the main reasons for the limited use of wealth taxes.25 It could also be argued that wealth and inheritance taxes entail a double taxation of immovable property, since real estate taxes may already be high, while financial wealth is too mobile to be taxed. However, real estate taxes often finance local services that benefit local populations and businesses. The double-taxation argument has, thus, a weak basis.

Consumption taxes tend to be regressive Consumption taxes account for a significant revenue share in all OECD countries (about 31% on average in 2010) and tend to decline as a share of HDI since lower-income households consume a larger share of their income (Prasad and Deng, 2009; Roach, 2003; Warren, 2008). As an illustration, in the UK, indirect taxes amounted to 13% of household gross income in 2008 (defined as market income plus cash benefits) but to over 25% for the lowest quintile and less than 10% for the top quintile (ONS, 2010).26 Likewise, in the US, sales and excise taxes levied by States are found to be highly regressive. Poor families pay almost eight times more as a share of their income than the best-off families (Davies et al., 2009). It should be noted, however, that calculations based on annual income data may overstate the regressive nature of consumption taxes since consumption largely depends on lifetime income which is less variable than annual household income (Poterba, 1989). In particular, pensioners with low annual income may consume out of their previous (accumulated) earnings. A lack of data makes it difficult to investigate the redistributive impact of consumption taxes in a cross-country setting. Implicit consumption tax rates can, however, be used to derive estimates (OECD, 2008b). They suggest that the regressive impact of consumption taxes — as measured by the 25 In France, about 5000 individuals paying the wealth tax (ISF) left the country between 1996 and 2005 (Conseil des Pr´el` evements Obligatoires, 2011). 26 In cash terms, the top fifth of households pay three times as much indirect tax as the bottom fifth. This simply reflects higher expenditure by higher-income households. The only indirect taxes where average payments in absolute terms do not vary much across the income distribution are those on tobacco, television licences, and the tax element of the National Lottery.

October 21, 2013

122

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch Rate level, 100% of average earnings

Difference in rates at 167 and 67% of average earnings

16 12 8

4 0 -4 -8 United States

Australia

Mexico

Belgium

United Kingdom

Spain

France

Austria

Ireland

Slovak Denmark Republic

Fig. 4.14: Average Consumption Tax Rate at Different Income Levels. Note: Average consumption tax rates are estimated by using micro data on consumption patterns available from household budget surveys and the corresponding tax rates (VAT, sales taxes and excise duties), in order to calculate the tax payments for each individual/family by income level. Estimated tax payments are then divided by net income (i.e. gross earnings minus personal income tax and employees’ social security contributions plus family benefits). Source: OECD (2008b).

difference in the implicit consumption tax rate for those at 167% of average earnings and those at 67% — is higher in the European countries than in the other OECD countries, due to higher consumption tax rates (Fig. 4.14).27 Warren (2008) also found a higher regressivity of consumption taxes — measured as the contribution to the Gini coefficient — in Denmark, Finland, Hungary, Norway, and Sweden and the lowest regressivity in Japan and the 27 Reflecting differences in the level of consumption taxes (standard rates at 5% in Japan and about 7% in the US, but up to 25% for the standard VAT rate in the Nordic countries), Garfinkel et al. (2006) noted that the same hypothetical USD1000 cash transfer to low-income households buys more goods in the US and Japan than in the Nordic countries — more than USD930 compared with USD750. Adema and Ladaique (2009) also recognize that consumption taxes reduce the real value of consumption which can be financed out of a given level of benefits. Furthermore, they note that, in some countries, policy explicitly took into account the impact of indirect taxation on the financial position of low-income households. For example, when the Goods and Services Tax was introduced in Australia in July 2000 at a rate of 10% (with food being exempt), a compensation package for social protection benefit recipients was introduced at the same time.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

123

US. Thus, omitting consumption taxes affects estimates of redistribution achieved through the tax and transfer system and how they differ across countries and evolve over time. To mitigate the regressive impact of consumption taxes, many OECD countries apply reduced rates and exemptions for goods and services deemed to account for a large share of poorer households’ consumption basket. For instance, food, water supply, medical care, and public transport are often granted reduced rates or exemptions.28 This approach typically implies a considerable dead-weight loss and people at higher income levels often benefit more in absolute terms since they consume more. In the case of Mexico, the total implicit subsidy due to the zero-rating of food was estimated at some 1.8% of GDP in the mid-1990s (Dalsgaard, 2000). The distribution by income decile showed that the highest decile captured nearly 30% of this amount, while the lowest three deciles together received only 12% of the value of the subsidy. Likewise in the Czech Republic, the lower VAT rate covers about 41% of the consumption of goods and services subject to VAT, and there is only very little variation in the share of such goods in the consumption baskets of households across the income deciles. Reduced VAT rates benefit the average individual in the top income decile about 2.5 times as much as the average consumer in the bottom decile (OECD, 2010b). Moving toward better targeted aid through in-kind benefits, cash transfers, and vouchers, thus, appears as a more effective redistribution tool (European Commission, 2007; van den Noord and Heady, 2001).

A Limitation of the Analysis: The Incidence of Taxes and Transfers When assessing the impact of the tax and transfer system on income distribution, most analyses (including this chapter), assume that taxes and transfers do not affect economic behavior. In practice, taxes and transfers affect incentives to work and save. The assumption of “no behavioral response” is likely to overstate the amount of redistribution achieved by cash transfers as these also typically weaken the work incentives of recipients (Moffit, 2011). The ultimate incidence of taxes and transfers crucially depends on how individuals and firms respond to a change in relative prices. 28 For

details, see OECD (2011b). Reduced VAT rates and exemptions are taken into account in the OECD estimates of the consumption tax rate at different income levels.

October 21, 2013

124

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

The greater the responsiveness, that is, the higher the price elasticities of supply and demand, the more likely it is that someone else will bear the tax burden or somebody else will benefit from a transfer. For instance, if labor demand is more sensitive to wages than labor supply, then payroll taxes end up being mainly borne by the employee in the form of a lower wage. Conversely, if labor supply is relatively inelastic and labor demand is relatively elastic, then the enterprise will bear more of the tax burden. Changes in tax progressivity may have a different impact on the labor supply of low- versus high-income earners. For low-income groups, more progressivity through EITC schemes increases work incentives but the resulting increase in labor supply may reduce wages — the EITC may increase the dispersion of before-tax income.29 For high-income groups, more progressive taxes may dampen work incentives and lower working hours. This could narrow the earnings dispersion. Taxes may also affect labor demand, with potential job losses more likely to affect low-skilled workers.30 The incidence and ultimate income inequality effect of property taxes, housing transfers, and consumption taxes may also differ from the firstround effect. While most property taxes are paid by owners, they may largely be passed onto renters in the form of higher rents as the supply of housing is relatively inelastic, at least in the short term. Similarly, housing cash transfers targeted on low-income groups may be reflected in higher rents, in which case they benefit the (higher-income) owners.31 Consumption taxes will be paid by consumers in the case of strong competition. However, the degree of competition could differ for different goods and services. As an example, recent cuts in French consumption taxes on 29 Rothstein (2008) found for the US that the EITC increases labor supply and studied the effect on wages. Low-skilled single mothers keep only $0.70 of every dollar they receive. Employers of low-skilled labor capture $0.72 (of which about 40% from single mothers and the rest from ineligible workers whose after-tax incomes fall). The net transfer to low-skilled workers is about $0.28 per dollar spent. 30 In a perfectly competitive labor market, higher labor taxes should not affect equilibrium unemployment since workers bear the entire tax burden through lower net wages. However, if firms cannot shift the entire tax burden onto workers (e.g. because of minimum wages or strong trade unions), higher taxes will reduce labor demand. Several empirical studies support this view (e.g. Bassanini and Duval, 2006; Belot and van Ours, 2004). 31 On the final incidence of property taxes, see Fullerton and Metcalf (2011). On the impact of housing cash benefits on rents paid by low-income groups, see Conseil des Pr´ el` evements Obligatoires (2011) and Facks (2005).

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

125

restaurants have not been fully passed on to consumers. They have partly financed higher wages and employment as well as raised profit margins. These examples suggest that the overall distributional impact of taxes may partly depend on behavioral responses.

Tax and Cash Transfer Policy Indicators Help Identify Reform Options and Types of Welfare Systems Countries rely on various tax and transfers instruments which differ in their design and impact on the income distribution. As discussed above, there are wide variations in the size, mix, and progressivity of both taxes levied on and cash transfers paid to households. Indicators on tax and transfer policies have been assembled and brought together in country profiles. The country profiles put the tax and transfer policy framework into an international perspective and allow the identification of reform options to address income inequality. Indicators on tax and transfers policies have also been used to identify groups of countries sharing broadly comparable welfare systems.

A Set of Policy Indicators on Taxes and Cash Transfers The redistributive impact of taxes depends on: (i) a size effect (the overall tax level); (ii) the tax mix (some taxes are more progressive than others); and (iii) the progressivity of each tax. The same logic applies to cash transfers. For both taxes and transfers, policy indicators have been assembled for each of the three dimensions using the OECD Income Distribution and Poverty Database and various other OECD Databases. To identify the tax and transfer policy framework, the country profiles show the value of each indicator compared with the OECD average. As an illustration, the main features that can be gleaned from the set of policy indicators for Australia and Germany are presented below (Fig. 4.15). The redistributive impact of taxes and transfers is higher than the OECD average in the two countries but the size, mix, and progressive nature of both taxes and transfers significantly differ.

October 21, 2013

126

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

Australia Redistributive impact of taxes and cash transfers Redistributive impact of Redistributive impact of 3 cash transfers household taxes 2 Taxes/GDP Cash transfers/GDP 1 Consumption tax

Old age pensions

0 -1 1

Personal income tax

Invalidity pensions

-2 2 --3 Property tax

Unemployment benefits

Household taxes

Family benefits

PIT & SSCs

Household cash transfers

PIT & SSCs, upper end PIT & SSCs, lower end

Old age benefits Unemployment benefits

Germany Redistributive impact of taxes and cash transfers Redistributive impact of Redistributive impact of 3 household taxes cash transfers 2 Taxes/GDP Cash transfers/GDP 1 Consumption tax

Personal income tax

0 -1 1

Old age pensions

Invalidity pensions

-2 2 --3 Property tax

Household taxes

PIT & SSCs PIT & SSCs, upper end PIT & SSCs, lower end

Unemployment benefits

Family benefits

Household cash transfers Old age benefits Unemployment benefits

Fig. 4.15: Tax and Cash Transfer Policy Indicators for Australia and Germany. Note: The dotted line represents the OECD average, the solid line represents the country shown. Where the solid line falls inside the OECD average, the variable considered stands below the OECD average. For instance, the tax/GDP ratio in Australia is lower than in the OECD area. Inversely, where the solid line is outside the OECD average, the variable is above the OECD average (the cash transfer/GDP ratio in Germany is

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

127

In Australia, taxes and transfers are smaller but more progressive than the OECD average The redistributive impact of taxes and cash transfers is above the OECD average in Australia. This mainly reflects the higher redistributive impact of taxes, even though the size of the Australian tax system, as measured by the tax-to-GDP ratio, is considerably below the OECD average (26% and 34%, respectively in 2010). Both the tax mix and the progressive nature

←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− Fig. 4.15: (Continued) higher than the OECD average, for example). The indicators are presented in standard deviationunits. Legend: Size and mix of taxes Taxes/GDP = Total tax revenue, percentage of GDP Consumption tax = Taxes on goods and services, percentage of total tax revenue Personal income tax = Income taxes on individuals or households, percentage of total tax revenue Property tax = Taxes on property, percentage of total tax revenue Progressivity of taxes Household taxes = Progressivity of total household taxes (Kakwani index, based on household surveys) PIT & SSCs = Net personal income tax progressivity, synthetic indicator, based on income tax plus employee contribution schedules net of standard cash transfers as a percentage of gross wage earnings, single person without children PIT & SSCs, upper end = As above, gap in tax rate between those earning 167% of the average wage and those at the average wage. PIT & SSCs, lower end = As above, gap in tax rate between those earning the average wage and those at 67% of the average wage

Size and mix of cash transfers Cash transfers/GDP = Total cash transfers, public and mandatory private sources, percentage of GDP Old age pensions = Old age and survivors’ pensions, percentage of total cash transfers Invalidity pensions = Incapacity related cash transfers, percentage of total cash transfers Unemployment benefits = Unemployment cash benefits, percentage of total cash transfers Family benefits = Family cash benefits, percentage of total cash transfers Progressivity of cash transfers Household cash transfers = Progressivity of total household cash transfers (Kakwani index, based on household surveys) Old age benefits = Progressivity of pensions Unemployment benefits = Progressivity of unemployment benefits, net of taxes for a single person

October 21, 2013

128

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

of taxes raise the redistributive impact of taxes. Personal income taxes account for a larger share of total taxes, while consumption taxes — which, as discussed above, tend to be regressive — play a less important role than in many other OECD countries. The data contained in the OECD Income Distribution and Poverty Database further reveal that household taxes are more progressive — as measured by a Kakwani index — than in the OECD on average. So are the personal income tax and social security contributions paid by employees (based on statutory tax schedules from the OECD Tax Database), in particular at the lower end of the income distribution. On the spending side, household cash transfers as a share of GDP are below the OECD average — 8.4% versus 13.2% in 2009 according the OECD Social Expenditure Database. However, the transfer mix favors redistribution across individuals, with family benefits accounting for a relatively high share of total transfers. The progressivity of total cash transfers to households (as measured by the Kakwani index derived from the OECD Income Distribution and Poverty Database) is also above the OECD average, reflecting the heavy reliance on means-tested, or even assets-tested benefit schemes, in Australia. Public old-age pensions are also more progressive (based on the OECD indicator), while the progressivity indicator for the unemployment benefit system is close to the OECD average.

In Germany, taxes and transfers are larger but less progressive than the OECD average The redistributive impact of taxes and cash transfers in Germany is above the OECD average. On the tax side, the total tax-to-GDP ratio stands above the OECD average (36% and 34%, respectively, in 2010), while the progressivity of household taxes is close to the OECD average. However, the progressivity of the personal income tax and social security contributions are steeper at the lower end than at the upper end of the income distribution. Total cash transfers to households, at almost 17% of GDP in 2009, are higher than in most OECD countries. Old-age pensions and unemployment benefits account for a larger share in total transfers, while family benefits are less important. The redistributive impact of household transfers is small, because of the low progressivity of the two main transfer components — old-age pensions and unemployment — reflecting their largely insurance-based nature.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

129

Five Tax and Transfer Systems Can Be Identified A cluster analysis of 19 policy indicators of the size, mix, and progressivity of both taxes and transfers has been performed.32 It allows identifying five groups of countries with similar tax and transfer systems: • A group of five Nordic countries (Denmark, Finland, Iceland, Norway, and Sweden) plus Belgium, the Netherlands, and Spain, is characterized by large cash transfers to households, which are not dominated by old-age pensions and tend to be more progressive than the OECD average. The tax-to-GDP ratio is high (Spain being an exception in this regard), with a tax mix which promotes redistribution — a rather large role for personal income taxes, while consumption taxes account for a small share of total taxes. Labor income taxes are rather progressive, in particular at the upper end. Overall, the redistributive impact of taxes and transfers — as measured by the decline in the concentration coefficient before and after taxes and transfers — is above the OECD average in all of these countries but Iceland and Spain. • A group of 10 continental European countries (Austria, Czech Republic, Estonia, France, Germany, Greece, Italy, Portugal, Slovak Republic, and Slovenia) features large cash transfers to households and a high taxto-GDP ratio. However, old-age pensions dominate cash transfers whose overall progressivity tends to be below the OECD average — the welfare system is dominated by transfers that redistribute over the lifecycle rather than across individuals. On the tax side, the personal income and property taxes often play a small role in total taxes. Overall, however, the tax and transfer system tends to redistribute more in most of these countries than in the OECD on average. • A group of seven countries (Australia, Canada, Japan, New Zealand, Switzerland, the UK, and the US) have in common relatively small cash transfers and taxes, combined with a large role of property and personal 32 The following indicators have been used for the cluster analysis: (i) for the size — the tax, transfer, and in-kind spending to GDP ratio; (ii) for the mix — the shares of old-age pensions, disability, family and unemployment in total transfers, the shares of consumption, personal income and property taxes in total taxes, and the share of education and health in total in-kind transfers; and (iii) for the progressivity — a Kakwani progressivity indicator for total household taxes and total cash transfers and specific progressivity indicators for old-age pensions, unemployment, personal income tax, and social security contributions.

October 21, 2013

130

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

income taxes. Cash transfers to households are more progressive than in the OECD on average since means-testing is often used. • A group of four countries — Chile, Korea, Poland, and Turkey — is characterized by a small transfer system largely dominated by old-age pensions. Two additional features further limit the redistributive impact of the welfare system: public spending on in-kind services (mostly education and health) as a share of GDP is low and consumption taxes play a dominant role in total taxes. • Another group of four countries — Hungary, Israel, Luxembourg, and Mexico — have little in common except very low in-kind spending, low progressivity of cash transfers, a rather small share of personal income taxes in total government revenue but progressive labor income taxes. It is interesting to note that there is a close, but not perfect, correspondence between country clusters based on tax and transfer policy indicators and country clusters based on inequality outcomes. As an illustration, the Nordic countries and Switzerland are all characterized by relatively low inequality emerging from the labor market and they appear in the same inequality outcome group (Chapter 2). Looking at the tax and transfer side, the Nordic countries again fit within the same group because their tax and transfer policies are broadly similar. However, Switzerland is in a different policy group because it implements much less redistribution via the tax and transfer system than the Nordic countries.

References Adema, W. and M. Ladaique (2009). “How Expensive is the Welfare State? Gross and Net Indicators in the OECD Social Expenditure Database (SOCX),” OECD Social, Employment and Migration Working Papers No. 92, OECD Publishing. Antolin P., A. de Serres and C. de la Maisonneuve (2004). “Long-Term Budgetary Implications of Tax-favoured Retirement Plans,” OECD Economics Department Working Paper No. 393. Bassanini, A. and R. Duval (2006). “Employment Patterns in OECD Countries: Reassessing the Role of Policies and Institutions,” OECD Economics Department Working Paper No. 486. Belot, M. and J. van Ours (2004). “Does the Recent Success of Some OECD Countries in Lowering their Unemployment Rates Lie in the Clever Design of their Labour Market Reform?” Oxford Economic Papers, Vol. 56(4), pp. 621–642.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

131

Ben´ıtez-Silva, H., R. Disney and S. Jimenez-Martin (2010). “Disability, Capacity for Work and the Business Cycle,” Economic Policy, Vol. 63(25), pp. 483–536. Burman, L.E. and D.P. Moynihan (2011). “Tax Reform Options: Marginal Rates on High-income Taxpayers, Capital Gains, and Dividends,” Statement before the Senate Committee on Finance, September 2011. Available at: http://www.urban.org / uploadedpdf / 901447-Burman-Tax-Reform-Options. pdf. Burman, L.E., E. Toder and C. Geissler (2008). “How Big are Total Individual Income Tax Expenditure, and Who Benefits from Them?” Tax Policy Center Discussion Paper, No. 16, the Urban Institute, Washington DC. Chawla, R.K. and T. Wannell (2003). “Property Taxes.” Perspectives, Statistics Canada, Vol. 4(7) (July). Available at: http://www.statcan.gc.ca/pub/75001-x/75-001-x2003007-eng.pdf. Christia, J.P. (2007).“The Empirical Relationship between Lifetime Earnings and Mortality,” CBO Working Paper Series No. 20071. Conseil des Pr´el`evements Obligatoires (2011). “Pr´el`evements obligatoires sur les m´enages: progressivit´e et effets redistributifs,” May. Available at: http:// www.ccomptes.fr/fr/CPO/documents/divers/Rapport conseil prelevements obligatoires prelevements obligatoires sur les menages 11052011.pdf. Dalsgaard, T. (2000). “The Tax system in Mexico — A Need for Strengthening the Revenue-Raising Capacity,” OECD Economics Department Working Paper No. 233. Davies, J.B., S. Sandstr¨ om, A. Shorrocks and E. Wolff (2009). “The Level and Distribution of Global Household Wealth,” The Economic Journal, Vol. 121, pp. 223–254. Diamond P. and E. Saez (2011). “The Case for a Progressive Tax: From Basic Research to Policy Recommendations,” Journal of Economic Perspectives, Vol. 25(4), pp. 165–190. Esping-Andersen, G. (1990). The Three Worlds of Welfare Capitalism, Princeton University Press, Princeton, N.J. European Commission (2007). “Study on Reduced VAT Applied to Goods and Services in the Member States of the European Union.” Available at: http:// ec.europa.eu/taxation customs/resources/documents/taxation/vat/how vat works/rates/study reduced vat.pdf. Facks, G. (2005). “Pourquoi les M´enages Paient-ils des Loyers de Plus en ´ es? L’incidence des Aides au Logement en France (1973–2001),” Plus Elev´ Economie et Statistiques, (381–382), Vol. 381, pp. 17–40. Fullerton, D. and G.E. Metcalf (2011). “Tax incidence,” In: Handbook of Public Economics, Vol. 4. Available at: http://www.sciencedirect.com/science/ handbooks/15734420.

October 21, 2013

132

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

Garfinkel, I., L. Rainwater and T.M. Smeeding (2006). “A Re-examination of Welfare States and Inequality in Rich Nations: How In-kind Transfers and Indirect Taxes Change the Story,” Journal of Policy Analysis and Management, Vol. 25(4), pp. 897–919. Garrett, D.M. (1995). “The Effect of Differential Mortality Rates on the Progressivity of Social Security,” Economic Inquiry, Vol. 33(3), pp. 457–475. Goda, S.G., J.B. Shoven and S.N. Slavov (2009). “Differential Mortality by Income and Social Security Progressivity,” SIEPR Discussion Paper No. 08-6. Hachon, C. (2009). “Who Really Benefits from Pension Systems? When Life Expectancy Matters,” Revue d’´economie politique, Vol. 119(4), pp. 613–632. Immervoll, H. (2010). “Minimum Income Benefits in OECD Countries: Policy Design, Effectiveness and Changes,” OECD Social, Employment and Migration Working Papers No. 100. Immervoll, H. and L. Richardson (2011). “Redistribution Policy and Inequality Reduction in OECD Countries: What Has Changed in Two Decades?” OECD Social, Employment and Migration Working Papers, No. 122, OECD Publishing. Joumard, I. (2001). “Tax Systems in European Union Countries,” OECD Economics Department Working Paper No. 301. Kakwani, N. (1977). “Measurement of Tax Progressivity: An International Comparison,” The Economic Journal, Vol. 87 (March), pp. 71–80. Kakwani, N. (1979). “Measurement of Tax Progressivity: A Reply,” The Economic Journal, Vol. 89(355), pp. 653–657. Keenay, G. and E. Whitehouse (2003). “The Role of the Personal Tax System in Old-Age Support: A Survey of 15 Countries,” Fiscal Studies, Vol. 24(1), pp. 1–21. Landais, C., T. Piketty and E. Saez (2011). Pour une R´evolution Fiscale, Editions Seuil, Paris. Marical, F. (2009). “Les M´ecanismes de R´eduction des In´egalit´es de Revenus ´ en 2008,” France, Portrait Social — Edition 2009, INSEE. Available at: http://www.insee.fr/fr/ffc/docs ffc/ref/FPORSOC09H.PDF, Paris. Marmot, M.G. and M.J. Shipley (1996). “Do Socioeconomic Differences in Mortality Persist after Retirement? 25 Year Follow-up of Civil Servants from the First Whitehall Study,” BMJ, Vol. 313(7066), pp. 1177–1180. Matsaganis, M. and M. Flevotomou (2007). “The Impact of Mortgage Interest Tax Relief in the Netherlands, Sweden, Finland, Italy and Greece,” EUROMOD Working Paper No. EM2/07. Moffit, R.A. (2011). “Welfare programs and labour supply,” In: Handbook of Public Economics, Vol. 4. Available at: http://www.sciencedirect.com/science/ handbooks/15734420.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

Income Redistribution via Taxes and Transfers

b1598-ch04

133

Musgrave, R.A. and T. Thin (1948). “Income Tax Progression 1929–48,” Journal of Political Economy, Vol. 56(6), pp. 498–514. Norregaard, J. (1990). “Progressivity of Income Tax Systems,” OECD Economic Studies (15) (Autumn), OECD Publishing. OECD (2006). OECD Tax Policy Studies: Fundamental Reform of Personal Income Tax, OECD Publishing, Paris, France. OECD (2008a). Growing Unequal? Income Distribution and Poverty in OECD Countries, OECD Publishing, Paris. OECD (2008b). Taxing Wages 2008–2009, OECD Publishing, Paris. OECD (2010a). Education at a Glance 2010, OECD Publishing, Paris. OECD (2010b). OECD Economic Surveys: Czech Republic 2010, OECD Publishing, Paris. OECD (2010c). Revenue Statistics, OECD Publishing, Paris. OECD (2010d). Sickness, Disability and Work: Breaking the Barriers, OECD Publishing, Paris. OECD (2010e). Tax Expenditures in OECD Countries, OECD Publishing, Paris. OECD (2011a). Divided We Stand. Why Inequality Keeps Rising, OECD Publishing, Paris. OECD (2011b). Consumption Tax Trends 2010: VAT/GST and Excise Rates — Trends and Administration Issues, OECD Publishing, Paris. OECD (2011c). Pensions at a Glance 2011, OECD Publishing, Paris. ONS (2010). “Income Inequality Remains Stable.” Available at: http://www. statistics.gov.uk/pdfdir/taxbhinr0610.pdf. Palameta, B. and I. Macredie (2005). “Property Taxes Relative to Income,” Statistics Canada, Perspectives, March. Available at: http://www.statcan.gc. ca/pub/75-001-x/10305/7796-eng.pdf. Piketty, T. (2010). “On the Long-Run Evolution of Inheritance: France 1820– 2050,” Paris School of Economics Working Paper, May. Piketty, T. and E. Saez (2007). “How Progressive is the US Federal Tax System? A Historical and International Perspective,” Journal of Economic Perspectives, Vol. 21(1), pp. 3–24. Poterba, J.M. (1989). “Lifetime Incidence and the Distributional Burden of Excise Taxes,” American Economic Review, Vol. 79(2), pp. 325–330. Prasad, M. and Y. Deng (2009). “Taxation and the Worlds of Welfare,” Luxembourg Income Study, Working Paper Series No. 480. Price, R. and T.T. Dang (2011). “Adjusting Fiscal Balances for Asset Price Cycles,” OECD Economics Department Paper No. 868. Roach, B. (2003). “Progressive and Regressive Taxation in the United States: Who’s Really Paying (and Not Paying) their Fair Share?” Global Development and Environment Institute Working Paper No. 03-10. Rothstein, J. (2008). “The Unintended Consequences of Encouraging Work: Tax Incidence and the EITC,” CEPS Working Paper No. 165.

October 21, 2013

134

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch04

Joumard, Pisu and Bloch

Sabirianova Peter, K., S. Buttrick and D. Duncan (2008). “Global Reform of Personal Income Taxation, 1981–2005: Evidence from 189 Countries,” IZA Discussion Paper No. 4228. St˚ ahlberg, A. (2007). “Redistribution across the life course in social protection systems,” In: Modernising Social Policy for the New Life Course, OECD Publishing, Paris. pp. 201–221. Toder, E.J., B.H. Harris and K. Lim (2009). “Distributional Effects of Tax Expenditures,” Tax Policy Center Discussion Paper, the Urban-Brookings Tax Policy Institute. Available at: http://www.taxpolicycenter.org/UploadedPDF/411922 expenditures.pdf. van den Noord, P. and C. Heady (2001). “Surveillance of Tax Policies: A Synthesis of Findings in Economic Surveys,” OECD Economics Department Working Paper No. 303. Waldron, H. (2007). “Trends in Mortality Differentials and Life Expectancy for Male Social Security-Covered Workers, by Socioeconomic Status,” Social Security Bulletin, Vol. 67(1), pp. 1–29. Warren, N. (2008). “A Review of Studies on the Distributional Impact of Consumption Taxes in OECD Countries,” OECD Social, Employment and Migration Working Paper No. 64. Whitehouse, E. (2006). “New Indicators of 30 OECD Countries’ Pension Systems,” PEF, 5(3), Cambridge University Press, November. Whitehouse, E., A. D’Addio, R. Chomik and A. Reilly (2009). “Two Decades of Pension Reform: What has been Achieved and What Remains to be Done?” The Geneva Papers, 34, The International Association for the Study of Insurance Economics. Williams, R. (2011), “Who Benefits from Tax Expenditures?” The Tax Policy Center, Urban Institute and Brookings Institution. Available at: http://www. taxpolicycenter. org/UploadedPDF/ 1001537 - Who- Benefits- From-Expenditures.pdf Yoo, K.-Y. and A. de Serres (2004). “Tax Treatment of Private Pension Savings in OECD Countries and the Net Tax Cost Per Unit of Contribution to TaxFavoured Schemes,” OECD Economics Department Working Paper No. 406.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

5. POVERTY Mauro Pisu

Introduction and Main Findings Very few people in OECD countries now fall below the absolute poverty line of $2 a day defined by the World Bank. Still, relative poverty remains a pressing policy issue. In the late 2000s, the average relative poverty rate in OECD countries stood at around 11%, up by around one percentage point since the mid-1990s.1 The recent crisis and the fiscal consolidation efforts risk increasing poverty further, making it an even more pressing policy issue. In developed countries, attention has progressively shifted to relative poverty as it has become clear that improving absolute living standards does not guarantee that all people can afford the goods and services that modern life has made necessities. Also, the concept of relative poverty is more closely aligned to that of capabilities’ deprivation (Sen, 1983) as people’s capabilities (i.e. people’s opportunity and ability to lead the kind of life they have reason to value) are partly determined by their position in the income distribution. The remainder of this chapter is structured as follows. The second section provides an overview of poverty rates and trends across countries. The third section compares the poverty rates of different age cohorts, namely children, working-age and retirement-age individuals. The fourth section focuses on poverty-rate differences by gender, whereas the fifth section distinguishes between households with and without children. In-work poverty is analyzed in the sixth section. The following sections focus on the policies

1 The

poverty rate is defined as the share of the population with an income lower than or equal to the national poverty line, which is set at 50% of the median equivalized household disposable income (HDI). The percentage increase in poverty is calculated for a subset of OECD countries (29 countries) for which data are available.

135

October 21, 2013

136

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu

issues. The seventh section compares the effects of taxes and transfers across countries, the eighth section reviews work strategies, while policies to combat child poverty are discussed in the ninth section. The tenth section provides a brief review of the effect of in-kind transfers on poverty.

Main Findings • In the late 2000s, the poverty rate in OECD countries varied greatly, ranging from 6% in the Czech Republic, Denmark, Hungary, and Iceland to above 15% in Turkey, the US, Chile, Israel, and Mexico. On average, working-age people recorded the lowest poverty incidence (10% of them are poor) followed by children (13%) and people of retirement age (16%).2 Over the past decades, poverty shifted away from retirement-age people towards children and working-age people as governments in many OECD countries raised retirement income. • Poverty is more widespread among women than men because of their longer life expectancy and shorter working life. Women are more likely to survive their spouse and thus more likely to live alone in the late part of their lives. In addition, they are less likely than men to have gained full pension rights, because of their shorter employment histories, which increase women’s poverty risk, noticeably during old age. • Households with children suffer from higher poverty rates than those without. On average, in the late 2000s, the poverty rate among households with a head of working age and with children was 11% against 9% for the childless ones. The US, Israel, Estonia, and Mexico are the countries where the difference in poverty rates between households with children and without is largest, whereas it is lowest in the Nordic countries, Poland, Chile, and Korea. • Although work is the best antidote against poverty, it alone is not enough to guarantee a life without it. On average, in the late 2000s, the poverty rate of households with a head of working age and with no workers was about 42% against 15% and 3% for those with one and two or more workers, respectively. In-work poverty is especially pronounced in Mexico, Israel, Chile, and the US where more than 20% of people in households 2 The poverty rate for population subgroups, such as children, retirement-age people, and women, are calculated as the percentage of the individuals in each subgroup with an equivalized HDI lower than or equal to the national poverty line.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

•

•

•

•

137

with one worker are poor. It is lowest in Germany, Norway, and the Czech Republic. Redistribution through taxes and transfers reduces poverty but differently across countries. For instance, the US and Sweden have similar pre-tax and transfer poverty rates, but the latter has a post-tax and transfer poverty rate significantly lower than the former. Those countries achieving a greater reduction in poverty via taxes and transfers tend to redistribute more toward people at the bottom of the income distribution. Net transfers benefit some types of households more than others. On average, net transfers are less effective in reducing poverty among households with children than among households without. Targeting cash transfers to the people most in need enhances the effectiveness of these transfers in reducing poverty, while limiting their size. Cash transfers are more strongly targeted toward the poor in Australia, New Zealand, and Denmark where the population in the lowest income quintile receives more than 35% of all cash transfers. Conversely, they are less well targeted in Italy, Luxembourg, Portugal, and Spain where the share is less than 15%. As regards household taxes, the share paid by the individuals in the lowest quintile is lowest in some English-speaking countries (Australia and the US) where it is below 2%, and highest in Switzerland, at around 10%. The degree of targeting and the size of gross public cash transfers explain much of the cross-country variation in poverty reduction among OECD countries, while taxation plays a minor role. Policies that facilitate paid work and employment-conditional cash transfers to top-up the income of low-wage workers are effective in reducing in-work poverty. Today, more than half of all OECD countries provide some form of IWBs or employment-conditional tax credits. Although some studies suggest that the per -job costs of IWBs are high, when considering the effect these policies have on inequality and poverty, IWBs appear to cost little. Minimum wages are most effective in reducing poverty, if set at a judicious level and when complemented by IWB schemes. Setting a wage floor precludes employers from appropriating the value of IWBs by reducing salaries. Successful poverty-reducing work strategies cannot be based on the minimum wage alone. Minimum wages suffer from poor targeting as they cannot account for specific family characteristics associated with poverty, such as the presence of children or adults working part time. If set too high, minimum wages are likely to result in higher unemployment.

October 21, 2013

138

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu

• Countries with a low child-poverty rate combine policies facilitating employment among parents along with redistribution programs targeted towards children. These two sets of policies are complementary and relying exclusively on only one is likely to be insufficient to lower child poverty. Pro-employment policies for parents include ALMPs providing education and training, provision of subsidized child care and out-ofschool care. Public spending on early childhood education is also crucial as there is evidence of a causal relationship running from family income to educational and cognitive outcomes, which is stronger early in life. • Cash income may not be the best measure to assess poverty as governments also provide in-kind transfers. However, measuring their impact on poverty is challenging as it is difficult to quantify how much such transfers benefit people in different parts of the income distribution. The available cross-country comparable evidence suggests that in-kind transfers are not pro-poor as they appear to be nearly uniformly distributed across the income distribution.

International Comparisons and Poverty Trends At around 11% on average in the late 2000s, the relative poverty rate in OECD countries is not negligible. However, this figure masks a wide variation across countries (Fig. 5.1). Poverty rates range from around 6% in the Czech Republic, Denmark, Hungary, and Iceland to above 15% in Turkey, the US, Chile, Israel, and Mexico. This ranking is robust to alternative measures of the relative poverty threshold, set at 40% or 60% of median income (OECD, 2008). Several OECD countries also compute poverty rates using absolute measures based on the cost of a basket of goods and services deemed to guarantee a minimum standard of living. According to this measure, absolute poverty has trended down in most OECD countries in the past 20 years, although relative poverty has generally increased (OECD, 2008). Comparisons based on absolute poverty rates are problematic since what is considered a minimum standard of living can vary greatly across countries (Box 5.1). Furthermore, in higher-income countries, the concept of relative poverty is closely connected to that of social exclusion and capability deprivation (Sen, 1983). In this chapter, only relative poverty measures will be used. Low poverty rates are generally accompanied by low overall inequality and vice versa (Atkinson and Marlier, 2010). Indeed, poverty and inequality

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

Percent

139

In the late 2000s

25

20

15

10

5

0

Fig. 5.1: Relative Poverty Rates. Note: The relative poverty rate is defined as the share of individuals with equivalized HDI less than 50% of the median income of the whole population. Source: OECD Household Income Distribution and Poverty Database, December 2012 and OECD calculations.

were highly correlated across countries in the late 2000s (Fig. 5.2).3 However, within countries, poverty and inequality does not always change over time in the same direction. This suggests that policy and macroeconomic conditions do not necessarily affect poverty and inequality in the same way and poverty-reducing policies may be effective even in a context of rising overall inequality. Rising inequality driven by a sharp increase in top earners’ incomes can take place at the same time as poverty is falling. Although no country pursues explicitly low poverty/high inequality policies, some of them may tolerate rising inequality if accompanied by declining poverty rates. This was epitomized by the focus of the UK Labour Party government in the late 1990s and 2000s on reducing child poverty accompanied by an “intensely relaxed” attitude towards rising overall inequality.4 Feldstein (1998) maintains that the real policy issue is poverty and not inequality. He argues that focusing on inequality and not just on poverty is inappropriate: it violates the Pareto principle and is equivalent to using a welfare function putting negative weights on increments in income of better-off individuals. 3 The 4 See

correlation coefficient is 0.86 and is statistically significant at 1%. the article “The Rich and the Rest” in The Economist edition of January 20, 2011.

October 21, 2013

140

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu

Box 5.1. Absolute and Relative Poverty Should measures of poverty focus on absolute or relative poverty? Absolute poverty rates are estimated using a cut-off income line below which individuals are not supposed to be able to afford a bundle of predefined basic goods, such as food and clothes. Typically, the poverty line is calculated in an initial year and is then adjusted to allow for price changes. Relative poverty, on the contrary, is defined with respect to the current standard of living in a country, with the relative poverty line defined as a percentage (usually 50 or 60%) of median income. These two concepts of poverty have been at the center of a lively research and policy debate for the past 50 years. The first attempts to rigorously measure poverty relied on absolute poverty lines (Bowley and Burnett-Hurst, 1915; Bowley and Hogg, 1925). In the early 1960s, this approach started to be challenged (Townsend, 1962), as it became clear that the share of people who could afford basic necessities such as food, clothes, and shelter had increased enormously since poverty measurement had started, but many people could still not afford goods or services that modern life had made necessities. Since then, a consensus has slowly emerged favoring the use of relative poverty measures at least in the context of developed countries (Sen, 1983). The US is the most notable exception in this regard, as its official poverty statistics are based on an absolute poverty line developed in the 1960s. More recently, the Census Bureau has started calculating and publishing supplemental poverty measures based on the recommendation of the 1995 National Academy of Science’s report Measuring Poverty (Citro and Michaels, 1995), which relies on alternative poverty lines but falls short of adopting the relative poverty approach. One of the purported advantages of the relativistic approach when compared to the absolutist one is that it relates more closely to the concept of capability deprivation advanced by Sen (1992). Already Adam Smith in the Wealth of Nations linked pauperism to capabilities when he defined the absence of poverty as the ability to appear in public (Continued )

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

141

Box. 5.1. (Continued ) without shame.a Townsend (1962) has made a similar point when defining poverty as the lack of resources needed to participate in the activities of the community, which in turn echoes the concept of social exclusion. Sen (1983) argues that the right approach for assessing the standard of living is to focus not on basic commodities or utility but on capabilities or functionings, which can be loosely interpreted as what people can choose to do. Possession of material goods or services may not say much about what people can choose to do with them. For instance, in societies with limited public transport, car ownership enables the capability of travelling to work whereas in others, car ownership may be irrelevant for this type of functioning. Following this notion, poverty can be thought of as an economic and social state with a level of capabilities that is unacceptably low from the point of view of the members of a society. There is uncertainty of what constitutes those basic capabilities whose lack can be associated with being poor, but they can be reasonably thought of as involving the ability of participating in the activities of the community without shame (Townsend, 1962) and of having selfrespect (Rawls, 1971). It is clear that the resource requirements needed to have these capabilities increase as countries develop. In the poorest societies, the necessary resources to escape poverty (i.e. participating in the activities of the community) are likely to involve meeting nutritional needs plus some others, such as being clothed and free from disease. As societies become more affluent, these needs will be more easily met, but others will arise that may depend on the relative position of individuals in the income distribution. These arguments lend support to the use of relative poverty measures, although mapping relative poverty to capabilities remains challenging. a In the Wealth of Nations Adam Smith defined necessities as “not only those commodities which are indispensably necessary for the support of life, but whatever the custom of the country renders it indecent for creditable people, even in the lowest order, to be without.” Taking the example of leather shoes in 18th century England, he remarked that “[t]he poorest creditable person of either sex would be ashamed to appear in public without them,” although the same was not true in France or Scotland.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu

142

In the late 2000s Inequality CHL

0.5

MEX

0.5

TUR 0.4 ISR USA PRT 0.4

GBR

AUS

ITA NZL

CAN

OECD ISL

NLD

0.3

GRC LUX

FRA

HUN

DEU

JPN EST

CHE

ESP

KOR

POL

AUT SWE

SVK

FIN

CZE

0.3

BEL

NOR DNK SVN 0.2 4

6

8

10

12

14

16

18

20

22

Poverty

Fig. 5.2: Poverty and Household Income Inequality. Note: The relative poverty rate is defined as the percentage of individuals with equivalized HDI less than 50% of the median income of the whole population. Inequality is measured by the Gini index of equivalized HDI inequality. Source: OECD Household Income Distribution and Poverty Database, December 2012 and OECD calculations.

Children and the Elderly are More Likely to be Poor than the Working-age Population Within countries, poverty rates vary considerably across age groups. On average, across the OECD, working-age people record the lowest poverty incidence (10% of them are poor) followed by children (13%) and people

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

143

of retirement age (16%).5 The majority of OECD countries have a higher poverty rate among children than among working-age people, the exception being Denmark, Sweden, Finland, Norway, Korea, Germany, and Iceland (Fig. 5.3). Chile, Israel, Mexico, Turkey, and the US, on the other hand, feature high poverty among children and working-age people. Across the OECD, the poverty rate of retired people exhibits the largest degree of variation followed by that of children and working-age people.6 This pattern suggests that across countries policies to combat poverty among the most vulnerable age groups (i.e. children and people aged over 65 years) are more heterogeneous than those targeted at working-age people. For instance, one of the main determinants of the vastly different retirement-age poverty rates is the level of old-age safety-net benefits. In Australia, the old-age public pensions in 2005 were lower than the respective poverty threshold.7 For Korea, the markedly higher poverty rate among the elderly can be attributed to the public pension scheme that was implemented in 1988, leaving retirees in the late 2000s with little or no entitlements.

Women are More Likely to be Poor than Men Poverty rates also differ by gender, with women suffering from a poverty risk that is one percentage point higher than that for men (OECD, 2008). The gender difference in poverty rates is partly related to the different age profile and working life of men and women. Women are more likely to survive their spouse/partner and consequently more likely to live alone in old age (Box 5.2). In addition, they are less likely than men to have gained pension rights, because of their shorter employment histories, which increase women’s poverty risk during old age (Fig. 5.4). Countries having large gender differences in old-age poverty rates tend not to be same as those

5 The

poverty rate does not reflect house ownership and associated imputed rents, which may be high for pensioners, and in-kind transfers (healthcare services or public transport), which are sometimes provided at a lower price or for free for pensioners. The poverty rate of older people may thus be overestimated in some countries. 6 The standard deviation-to-mean ratio for the three groups is 0.6, 0.5, and 0.3, respectively. 7 In 2005, in Australia, the full public pension was equivalent to 86% of the poverty line (OECD, 2009b).

October 21, 2013

144

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu In the late 2000s Poverty rates among people of working age, percent

18 MEX CHL ISR USA 14 JPN ESP

KOR GBR EST

10

POL PRT

AUS ITA GRC OECD BEL NZL NLD AUT LUX FRA HUN SVK SVN CHE

DEU SWE

NOR FIN ISL

6

TUR

CAN

DNK CZE

2 2

6

10

14

18

22

26

30

Poverty rates among children, per cent Poverty rates among children, percent 35

30 ISR 25

MEX

TUR CHL USA

20 ESP

PRT CAN

15

POL

AUS

ITA

LUX GBR NLD

10

FRA

GRC BEL

SVK DEU

CZE HUN

EST

NZL KOR

AUT

CHE SVN

SWE ISL

5

JPN

OECD

FIN

NOR DNK 0 0

5

10

15

20

25

30

35

40

45

Poverty rates among people of retirement age, percent

Fig. 5.3: Poverty Rates among People of Working Age, Retirement Age, and Children.a

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

145

Box 5.2. The Gender Poverty Gap Women are often underpaid (as they tend to earn less than males for the same job) and overqualified (as they have jobs not corresponding to their ability level, while bearing a disproportionate share of household duties). Promoting gender equality will not only have direct effects on poverty and inequality but also result in higher growth. Despite the vast progress made in the last decades, women are still more likely to be poor than men. Gender equality does not necessarily mean complete equality of income for men and women as unequal outcomes may result from personal and rational choices. Rather it refers more broadly to equality of opportunities and the ability to make informed choices in the labor market and on family issues. Although nowadays in OECD countries young women have broadly the same educational attainment as men, women still suffer from discrimination and lower opportunities with respect to men in the labor market. This tends to perpetuate poverty among women. Demographics will also affect the income gap between men and women. Buvinic et al. (2009) argue that the rapid population ageing that OECD countries will experience in the future due to persistently low fertility rates and declining mortality rates will strain public and private resources to support the elderly, of whom a disproportionate number are women as they typically live longer than men. The share of women living in poverty is thus likely to rise by more than that of men in the future. In addition, improving employment opportunities for women, furthering their access to entrepreneurship and self-employment is key to enhance women’s contribution to society. For instance, some recent studies report that in Italy and Eastern Europe, women are at a (Continued ) ←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−− Fig. 5.3: (Continued) Note: a Poverty is measured as the share of individuals with equivalized HDI less than 50% of the median income of the whole population. Children are individuals being 17 years old or younger, working-age people are those between 18 and 65 years old, and retired people are aged 66 years and above. Source: OECD Household Income Distribution and Poverty Database, December 2012.

October 21, 2013

146

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu

Box. 5.2. (Continued ) disadvantage in obtaining finance from commercial banks (Alesina et al., 2008; Sabarwal and Terrell, 2008). Women are also much more likely to exit the labor market when they have children. In this case, policies to promote maternity and paternity leave and child care can help to ensure a stronger attachment of women to the labor market.

In the mid-2000s, difference in percentage points Working age

Pension age

Germany

Ireland

United States

Finland

Australia

Norway

Italy

United States

Ireland

Italy Austria

Canada

Slovak Republic

Mexico

Japan

Korea

Germany

Luxembourg

Korea

Greece

United Kingdom

Spain

Canada

Japan

Hungary

Netherlands

Spain

Belgium

Australia

Portugal

OECD-30

OECD-30

Greece

Turkey

Switzerland

Switzerland

France

Czech Republic

Denmark

United Kingdom

Sweden

Iceland

Poland

Slovak Republic

Czech Republic

France

Turkey

Austria

Portugal

New Zealand

Mexico

Sweden

Netherlands

Denmark

Belgium

Finland

New Zealand

Hungary

Iceland

Poland

Luxembourg

Norway

-2

0

2

4

6

8

10

12

-2

0

2

4

6

8

Fig. 5.4: Difference in Poverty Rates by Gender.a Note: a Poverty is measured as the share of individuals with equivalized HDI less than 50% of the median income of the whole population. Source: OECD (2008).

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

147

having a large gender gap for working-age adults.8 There is also evidence that the gender gap in poverty rates rises with age even among retirementage persons (OECD, 2009b). Different factors can explain this phenomenon. One has to do with the increasing labor market attachment of younger women and the consequently higher pensions they have gained. Another concerns the higher life expectancy of women with respect to men and that many women are married to older men. This implies that within the retirement-age group widowed women living on survivors’ benefits, which are limited in many countries, are overrepresented with respect to men.

Households with Children are Poorer than Those Without The presence of children in a household is generally associated with a higher poverty risk. Across the OECD, the poverty rate among households with children and with a head of working age was 11% in the late 2000s against 9% for their childless counterparts (Fig. 5.5). The cross-country dispersion in poverty rates among households with children (and with a head of working age) is larger than those without. In addition, the correlation between the two series is not strong, being 0.30, thus suggesting that the factors affecting the poverty rates of these two types of households differ. The larger cross-country variation in the poverty rate among households with offsprings is likely to depend on several factors, such as the effectiveness of social programs aiming at reducing the financial burden of having children. Countries where having children is linked to large increases in the poverty risk are the US (13 percentage points), Israel (12), Estonia (11), and Mexico (10). On the other hand, households with children have a markedly lower poverty incidence than those without in the Nordic countries, Poland, Chile, and Korea and to a lesser extent in Japan and Germany. The labor market participation of parents has a considerable effect on child poverty (Fig. 5.6). Overall, poverty rate differentials between households with and without children are largest for non-working couples and non-working single adults. In other words, the presence of children increases the poverty rate for these two types of households the most. In all OECD countries, non-working couples with children have higher poverty rates than the same type of households without children. Also, single parents with 8 The

correlation coefficient between these gender poverty gaps is indeed very low (0.05).

October 21, 2013

148

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu In the late 2000s

Households without children 20 CHL 18 16 SWE JPN 14 POL

KOR CAN

12

MEX

GBR AUS 10

DEU

FIN

8

DNK

EST

NLD BEL AUT FRA HUN

ISR

ESP

OECD ITA GRC TUR LUX

USA

6 SVN

4

NZL

CHE CZE

2 0

0

5

10

15

20

25

Households with children Fig. 5.5: Poverty Rates of Households with a Head of Working Age with and without Children. Note: Poverty is measured as the share of individuals with equivalized HDI less than 50% of the median income of the whole population. Source: OECD Household Income Distribution and Poverty Database, December 2012.

no job have in general higher poverty rates than non-working single adults. The difference in poverty rates between these two types of households is highest for some Mediterranean countries, namely Italy (57 percentage point difference), the Czech Republic and Luxemburg (52), and Greece (39). Yet, in some countries the reverse is true. In Korea, Norway, Estonia, Chile, and Switzerland the poverty rate for single parents with no job is more than 10 percentage points lower than for non-working single adults. Overall, having children does not seem to significantly raise the poverty risk of couples with two or more workers.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

149

In the late 2000s 80 70 60 50 40 30 20 10 0 -10 -20 -30 -40 -50

Couple, two or more workers

Couple, one worker

Single adult working

Single adult not working

Couple, no workers

Fig. 5.6: Difference in Poverty Rates between Households with Children and without. Note: The graph shows the Box- and-Whisker plot of the differences in relative poverty rates between each household type with and without children. Differences are expressed in percentage points. The lowest mark corresponds to the minimum value, whereas the highest to the maximum. The lower and upper sides of the rectangle represent the 25th and 75th percentiles, respectively. All OECD countries are included except for Estonia, Israel, Slovenia, and Switzerland. Source: OECD Household Income Distribution and Poverty Database, December 2012.

Paid Work Reduces but Does not Eliminate Poverty Although work is the best antidote against poverty, it is not a panacea. OECD-wide, jobless households are six times as likely as working households to be poor. Still, 60% of the poor live in households with at least one worker (OECD, 2008). On average, in the late 2000s, the poverty rate of households with a head of working age and one worker was around 15% and 3% for those with two workers (Fig. 5.7). The corresponding figure for households with no worker was 42%. In-work poverty is especially pronounced in Mexico, Israel, Chile, and the US where more than 25% of people in households with one worker are poor. It is lowest in Germany, Norway, and the Czech Republic. The in-work poverty-reducing effect of an additional worker is substantial in all countries, with a few exceptions. Labor market participation appears to reduce the likelihood of poverty especially

October 21, 2013

150

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu In the late 2000s

35

Households of working age with 30

One worker

Two workers

25 20 15 10 5 0

Fig. 5.7: In-work Poverty. Note: Figures refer to households with a head of working age only. Data refer to the late 2000s. Source: OECD Household Income Distribution and Poverty Database, December 2012.

in some English-speaking countries, to wit New Zealand, Australia, Canada, and the US plus Israel, Chile, and Estonia.

Redistribution Strategies Taxes and transfers are effective policy instruments in reducing poverty. Countries starting with the same proportion of people living in poverty before taxes and transfers are taken into account may end up with markedly different after-tax and transfers poverty rates. For instance, Fig. 5.8 shows that the US has a pre-tax and transfer poverty rate comparable to that of Sweden, but the latter reduced it by more than twice than the former through taxes and transfers. Also, although Australia has a market income poverty rate that is 10 percentage points higher than that of Korea, they have a similar share of people living in poverty after taxes and transfers. Countries that achieve a greater reduction in poverty through taxes and transfers redistribute more toward people at the bottom of the income distribution (Fig. 5.9). Sweden, Denmark, Finland, the Czech Republic, and

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty Before taxes and transfers

80%

40

40%

65%

GBR

151

GRC LUX

35

ITA

SWE

FRA DEU

BEL

FIN 30 HUN CZE

25

AUT POL SVN OECD

SVK

JPN

PRT ESP EST

MEX

CAN

NOR NLD

20

DNK ISL

ISR

USA AUS TUR

CHL

NZL CHE

KOR

15

10

5

0 0

5

10

15

20

25

After taxes and transfers

Fig. 5.8: Poverty Rates Before and After Taxes and Transfers. Note: The after-tax and transfer poverty rate is measured as the share of the population with equivalized HDI less than or equal to the national poverty line, which is set at 50% of the median equivalized HDI. The before-tax and transfers poverty rate is computed as the share of the population with equivalized household market income (HMI) less than or equal to the national poverty line. The dotted lines show different percentage reductions in poverty rates due to taxes and transfers. Data refer to the late 2000s. Greece, Hungary, and Mexico report income net of taxes, so that their before-tax and transfer poverty rate is before transfers only. Source: OECD Household Income Distribution and Poverty Database, December 2012.

Belgium are the countries that redistribute most toward the poor, whereas the opposite is true for Chile and Korea. However, net transfers benefit some individuals more than others. Overall the reduction in poverty is larger for retirement-age individuals than for working-age people and children. After taxes and transfers, the poverty rate for people over 65 years drops by 77%, whereas that for working-age individuals and children falls by only 50% and 46%, respectively (OECD, 2008). This partly reflects the large role public pensions play in most OECD countries.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu

152

Percentage reduction in poverty 90

R² = 0.4538 CZE

FRA

80 NLD

LUX

70

ITA

DEU

AUT SVN

NOR

ISL

FIN

DNK

SVK BEL

SWE

GBR 60

POL

PRT OECD

CHE

50

CAN

EST

ESP

NZL AUS

JPN 40 USA 30 ISR 20 CHL

KOR

10

0 0

1

2

3

4

5

6

7

Net transfers to lowest quintile

Fig. 5.9: Net Transfers to the Lowest Quintile and Reduction in the Poverty Rate. Note: Net transfers are gross public cash benefits minus household taxes as a percentage of equivalized HDI. The lowest quintile refers to the distribution of equivalized HDI. The reduction in the poverty rate is computed as the percentage difference of income before and after taxes and transfers. Source: OECD Household Income Distribution and Poverty Database, December 2012.

Taxes and transfers are less effective in reducing poverty among households with children or with one or more workers than among those with neither of them (Fig. 5.10). On average in the OECD, for households with a head of working age, net transfers reduce the poverty rate of such households with offspring by about 36% and by around 51% of those without. This pattern holds true for the majority of OECD countries. Exceptions are Australia, Canada, Poland, the UK, and to a lesser extent Germany, New Zealand, Denmark, and Finland. Poor households with children are especially disadvantaged as compared to childless ones in Switzerland, Spain, and Japan. As regard poor households without workers, net transfers reduce

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

153

Late 2000s, percentage reduction

Households with children 90 80

FIN DNK

70 POL

60 50

CHL

SVN LUX

OECD USA

EST

20

CZE

BEL

SWE

30

TUR

NLD

CAN

40

FRA AUT

DEU

AUS

HUN

NZL

GBR

GRC

ISR

ITA

10 MEX

KOR

0 -10

ESP

JPN

CHE

-20 -30 0

10

20

30

40

50

60

70

80

90

Households without children Household with workers 90 80 70

NZL FIN

AUS

60

POL

50

CAN

GBR

TUR

DEU

AUT DNK BEL SVN

NLD

SWE

HUN FRA

CZE

GRC

40 ISR

OECD

30

ITA

EST

20

CHL KOR

10

CHE

ESP

MEX

JPN

0

LUX

USA

-10 0

10

20

30

40

50

60

70

80

90

Household without workers

Fig. 5.10: Reduction in the Poverty Rate among Different Types of Households due to Net Transfers. Note: Net transfers are gross public cash benefits minus household taxes as a share of total equivalized HDI. Households with a head of working age are depicted, only. For Greece, Hungary, and Mexico, the figures refer to the reduction in the poverty rate due to gross transfers and not net transfers. Source: OECD Household Income Distribution and Poverty Database, December 2012.

October 21, 2013

154

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu

their poverty rate by about 50% on average as against 40% for those households comprising one or more workers. Only in Australia, Poland, Canada, and Israel, poor households with workers benefit markedly more from redistribution than those without them. Targeting cash transfers to the people most in need enhances their effectiveness, while limiting their size. There is a great degree of variation across OECD countries in how effectively they target the poor (Fig. 5.11). In the late 2000s 50 45

Share of total transfers paid to the lowest quintile Share of total transfers in total HDI

40 35 30 25 20 15 10 5 0

60

12 Share of total taxes paid by lowest quintile (left scale) 10

Share of total taxes in total HDI (right scale)

50

8

40

6

30

4

20

2

10

0

0

Fig. 5.11: Size and Targeting of Cash Transfers and Household Taxes. Note: Household taxes include employee social security contributions. The lowest quintile refers to the distribution of HDI. Source: OECD Household Income Distribution and Poverty Database, December 2012.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

155

Cash transfers are more strongly targeted towards the poor in Australia, New Zealand, and Denmark where the population in the lowest income quintile receives more than 35% of all cash transfers. Conversely, they are less well targeted in Italy, Luxembourg, Portugal, and Spain, where the share is less than 15%. Overall, there is a negative relationship between the degree of targeting of cash transfers and their size (as a share of total HDI).9 As regard household taxes, the share paid by the individuals in the lowest quintile is lowest in some English-speaking countries (Australia and the US), where it is below 1.5%, and highest in Switzerland, at about 10%. In general, countries with the most targeted cash transfer programs are not the same as those collecting the lowest share of taxes from the poorest individuals. The degree of targeting and the size of gross public cash transfers explain much of the cross-country variation in poverty reduction among OECD countries, with household taxes having no effect. Simple regression analysis reveals that countries that are more successful in reducing poverty have larger and better targeted public transfer programs (Table 5.1). The point estimates suggest that a one percentage point rise in the percentage of gross public cash transfers in total household income is associated with an additional two percentage point reduction in poverty rates, while a one percentage point increase in the degree of targeting (share of total gross public transfers accruing to the poorest 20%) is related to 0.7 percentage point reduction in poverty.10 Although net social transfer programs can be effective in reducing poverty, they can undermine work incentives and introduce economic distortions. Ideally, governments should aim at developing and introducing poverty-reducing strategies that limit distortions in economic activity so as to minimize deadweight losses. Such measures are likely to include IWB programs as employment-conditional cash transfers to low-paid workers varying according to family characteristics, lower pay-roll taxes on employers that hire or retain low-paid personnel, flexible work arrangements permitting to reconcile working and parenting duties, and up-skilling policies allowing low-paid workers to move to better jobs.

9 The

correlation coefficient is −0.53 and is statistically significant at the 1% level. as dependent variable, the percentage difference in the poverty gap, rather than the head-count poverty, before and after taxes yields qualitatively and quantitatively similar results. 10 Using

October 21, 2013

156

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu Table 5.1: The Effectiveness of the Size and Progressivity of Taxes and Transfers in Reducing Poverty.a Dependent variable Public gross cash transfers Size Targeting Household taxes Size Targeting Constant R2 Observations

Reduction in the poverty rate

2.02b [0.40] 0.71c [0.31] 0.22 [0.30] −0.32 [0.79] −6.07 [9.71] 0.77 29

Note: a The reduction in the poverty rate is the percentage difference between pre- and post-tax and transfer poverty rates. The size of gross public transfers (household taxes) is the ratio of gross public transfers (households’ tax payments) to total household income times 100. The targeting of gross public transfers (household taxation) is the share of total gross public transfers paid to (household taxes collected from) the poor. The poor are people in the lowest income quintile of equivalized HDI. The data refer to the late 2000s. The countries in the regression are the same as in Fig. 5.11; b denotes statistical significance at 1% level; c denotes statistical significance at 5% level. Robust standard errors in brackets. Source: OECD Household Income Distribution and Poverty Database, December 2012 and OECD calculations.

Work Strategies Helping people to move from benefits to work is the central aim of ALMPs. These policies are based on the premise that governments should not just remedy the consequences of poverty through redistribution but also prevent poverty through facilitating access to paid work. Introducing welfareto-work policies has featured prominently in reforming social programs in several OECD countries and their outcomes in terms of moving people from welfare rolls to employment have sometimes exceeded expectations, though there are also programs that are ineffective (OECD, 2009a).

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

157

Policies aiming at facilitating paid work along with employmentconditional cash transfers to top-up the income of low-wage workers can offer effective ways to combat poverty. A large number of countries has introduced IWBs or employment-conditional tax credits as a core element of “make-work-pay” programs and now more than half of all OECD countries provide some form of IWBs (Immervoll and Pearson, 2009). Although narrow assessments of their per-job-created costs suggest that IWBs are rather expensive, evaluations considering also the effect these policies have on inequality and poverty indicate that IWBs cost little per dollar transferred. Immervoll et al. (2007) found that among 15 European countries, the cost for taxpayers of redistributing one euro through an IWB could be small, because they raise employment. In general, IWBs are more likely to be an efficient redistributive measure in countries characterized by a wide income dispersion at the bottom of the income distribution and a high tax burden. However, some studies caution about the effectiveness of IWBs as an income redistribution tool. The labor supply increase and the ensuing decline in wages generated by IWBs can markedly reduce the actual transfer received by beneficiaries of such programs, with a large part of them accruing to employers. Overall, the larger the share of transfers that employers are able to capture, the more elastic the supply of labor and the less elastic the demand for labor are, and the lower is the redistributive effect of IWBs. Rothstein (2008), studying the expansion of the US’ EITC in the mid-1990s, found that one dollar spent on the EITC generated a transfer of USD 0.70 for the recipient with the remaining part captured by employers. In addition to this, employers also benefit from an additional transfer of USD 0.42 coming from the lower wages of non-eligible workers, who compete in the labor market with EITC beneficiaries. If set at an appropriate level, the minimum wage can be a valuable policy instrument to alleviate in-work poverty, especially for working-age adults having full-time jobs. Several OECD countries have a minimum wage, while many others have collective agreements providing wage minima. Yet, the relationship between poverty and minimum wages is ambiguous (Freeman, 1996). The “labor economist’s view” maintains that a high minimum wage leads necessarily to an increase in poverty because of higher unemployment, whereas the “trade unionist’s view” reaches the opposite conclusion because of the income sharing through family and social links. Yet, both views are incomplete as the impact of the minimum wage on poverty depends on several economic and social factors, which vary across countries. Based on a theoretical model, Fields and Kanbur (2007) argued

October 21, 2013

158

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu

that these are the level of the minimum wage with respect to the poverty line, the elasticity of demand for labor, the degree of intra-family income sharing between the employed and unemployed, and the degree of societal poverty aversion. Minimum wages have the drawback that they do not effectively target the individuals and households most at risk of poverty as they cannot account for specific family characteristics associated with higher poverty rates, such as the presence of children or adults working part time. Also, if set too high, the minimum wages are likely to lead to less employment opportunities. As a result, effective policies to alleviate inwork poverty cannot be based on the minimum wage alone (OECD, 2009a). Minimum wages are most effective in reducing poverty if set at a judicious level and complemented with IWB schemes (Freeman, 1996; Immervoll and Pearson, 2009). Establishing a wage floor precludes employers from appropriating the value of IWBs by reducing salaries. Yet, a too high minimum wage renders the implementation of IWB programs that do not involve a steep benefit phase-out challenging and makes them more expensive.

Policies to Reduce Poverty among Children Child poverty is especially important not only because children are the most vulnerable members of society but also because of the pernicious long-term consequences it entails. Higher child poverty is linked to higher intergenerational inequality as poor children are more likely to be poor when adult (D’Addio, 2007; OECD, 2009c). The classic family investment model by Becker and Tomes (1986) points to the role of financial constraints, whereby poor parents are unable to borrow for investing in their children’s human capital. Also, low family income might adversely affect long-term income prospects through heightened parental stress, which in turn can adversely impinge on child well-being, affecting educational and cognitive outcomes negatively (Duncan, 2006; Mayer, 1997). Various redistribution instruments exist. These include birth grants, child benefits, and tax credits for working parents. Broadly speaking, they can be divided into “benefit strategies,” involving cash transfers or tax credits for poor families so as to ensure a minimum income level to non-working parents or to supplement the income of those working and “work strategies,” aimed at increasing the employment rate of parents. Overall, the countries with very low levels of child poverty combine low levels of joblessness with

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

159

effective redistribution policies towards children, suggesting that these two policy approaches are complementary and that relying exclusivelyon only one of them is insufficient to reduce child poverty (Whiteford and Adema, 2006). Increasing the income of poor families with children through the tax and benefit system is the most direct way of reducing child poverty. Such spending can overcome the financial constraints that poor parents face in investing in their children because of credit-market imperfections. As there is evidence of a causal relationship running from family income to educational and cognitive outcomes that is stronger early in life (Heckman 1999, 2007), governments should focus on spending on early childhood education and care. One unresolved issue is how much of transfers should be spent on children. Some argue that the effect of such policies on the well-being of children is limited as poor and dysfunctional families will spend the money they receive on consumer goods that do not benefit children. Yet, there is evidence that if mothers receive the transfer, a large share of them will be spent on children (Lundberg et al., 1997). Evidence suggests that parental employment is a first step on the route out of poverty (Grogger and Karoly, 2007; Morris et al., 2004). Pro-employment policies for parents include ALMPs providing education and training, provision of subsidized child care and out-of-school care. Evidence also suggests such work-based strategies are effective in lowering child poverty, although their effects on other important aspects of the well-being of children and their future income are less clear cut.

In-kind Transfers and Poverty When considering in-kind transfers, it is important to distinguish between pro-poorness and progressivity. Transfers are progressive when their share of individual income disproportionately goes to the lower end of the distribution. A special case of this is when in absolute terms they are higher at the bottom of the distribution, in which case transfers are said to be pro-poor. OECD (2008) suggests that in-kind transfers are for the most part progressive, but not pro-poor. The value of most in-kind benefits households receive appear not to vary with income, though they still narrow the income differences between individuals because they constitute a higher addition to the cash income of the poor relative to the rich. However, the low progressivity of in-kind transfers may also result from the methodological difficulties in identifying the households who actually benefit from them and allocating

October 21, 2013

160

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Pisu

the value of such transfers accordingly. Such calculations may thus underestimate the role of in-kind transfers in reducing poverty in those countries where such benefits are effectively targeted to poor households. Recent calculations suggest that on average in the OECD, in-kind transfers reduce poverty rates by about 50%. In Belgium, Ireland, and the UK, poverty rates drop by close to 60%, and by around 27% in Estonia and Sweden (OECD, 2011).

References Alesina, A.F., F. Lotti and P.E. Mistrulli (2008). “Do Women Pay More for Credit? Evidence from Italy,” Harvard Institute of Economic Research Discussion Paper No. 2159, Cambridge, Massachusetts. Atkinson, A.B. and E. Marlier (2010). “Living conditions in Europe and the Europe 2020 agenda,” In: Atkinson, A.B. and Marlier, E. (eds.), Income and Living Conditions in Europe, Luxembourg: Office for Official Publications of the European Communities (OPOCE), pp. 21–35. Becker, G. and N. Tomes (1986). “Human Capital and the Rise and Fall of Families,” Journal of Labor Economics, Vol. 4(3), pp. S1–S39. Bowley, A.L. and A.R. Burnett-Hurst (1915). Livelihood and Poverty, Bell, London. Bowley, A.L. and M.H. Hogg (1925). Has Poverty Diminished? P.S. King, London. Buvinic, M., M. Das Gupta and U. Casabonne (2009). “Gender, Poverty and Demography: An Overview,” The World Bank Economic Review, Vol. 23(3), pp. 347–369. Citro, C. and R. Michaels (1995). Measuring Poverty: A New Approach, The National Academies Press, Washington, D.C. D’Addio, A.C. (2007). “Inter-generational Transmission of Disadvantage: Mobility or Immobility across Generations?” OECD Social, Employment and Migration Working Paper No. 52. Duncan, G. (2006). “Income and Child Well-Being,” 2005 Geary Lecture, The Economic and Social Research Institute, Dublin. The Economist (2011). “The Rich and the Rest,” January 20, 2011. Feldstein, M. (1998). “Income Inequality and Poverty,” NBER Working Paper No. 6770. Fields, G.S. and R. Kanbur (2007). “Minimum Wages and Poverty with IncomeSharing,” Journal of Economic Inequality, Vol. 5(2), pp. 135–147. Freeman, R.B. (1996). “The Minimum Wage as a Redistributive Tool,” Economic Journal, Vol. 106(436), pp. 639–649. Grogger, J. and L. Karoly (2007). “The Effects of Work-Conditioned Transfers on Marriage and Child Well-being: A Review,” NBER Working Paper No. 13485.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

b1598-ch05

Poverty

161

Heckman, J. (1999). “Policies to Foster Human Capital,” NBER Working Paper No. 7288. Heckman, J. (2007). “The Economics, Technology and Neuroscience of Human Capability Formation,” IZA Discussion Paper No. 2875. Immervoll, H. and M. Pearson (2009). “A Good Time for Making Work Pay? Taking Stock of In-work Benefits and Related Measures across the OECD,” OECD Social, Employment and Migration Working Paper No. 81. Immervoll, H., H. J. Kleven, C.T. Kreiner and E. Saez (2007). “Welfare Reform in European Countries: A Microsimulation Analysis,” The Economic Journal, Vol. 117(516), pp. 1–44. Lundberg, S., R. Pollak and T. Wales (1997). “Do Husbands and Wives Pool Resources? Evidence from the UK Child Benefit,” Journal of Human Resources, Vol. 32(3), pp. 463–480. Mayer, S. (1997). What Money Can’t Buy: Family Income and Children’s Life Chances, Harvard University Press, Cambridge, MA. Morris, P., G. Duncan and C. Rodrigues (2004). “Does Money Really Matter? Estimating Impacts of Family Income on Children’s Achievement with Data from Random-Assignment Experiments.” Available at: www.cerforum.org/ conferences/200406/papers/MorrisDuncanRodrigues.pdf,4. Accessed March 2011. OECD (2008). Growing Unequal? Income Distribution and Poverty in OECD Countries, OECD Publishing, Paris. OECD (2009a). OECD Employment Outlook, OECD Publishing, Paris. OECD (2009b). Pensions at a Glance 2009: Retirement-Income Systems in OECD Countries, OECD Publishing, Paris. OECD (2009c). Doing Better for Children, OECD Publishing, Paris. OECD (2011). Divided We Stand, OECD Publishing, Paris. Rawls, J. (1971). A Theory of Justice, Harvard University Press, Cambridge, MA. Rothstein, J. (2008). “The Unintended Consequences of Encouraging Work: Tax Incidence and the EITC,” Center for Economic Policy Studies Working Paper No. 1049, Princeton University. Sabarwal, S. and K. Terrell (2008). “Does Gender Matter for Firm Performance? Evidence from Eastern Europe and Central Asia,” IZA Discussion Paper No. 3758, Institute for the Study of Labor (IZA). Sen, A. (1983). “Poor, Relatively Speaking,” Oxford Economic Papers, Vol. 35(2), pp. 153–169. Sen, A. (1992). Inequality Reexamined, Clarendon Press, Oxford. Townsend, P. (1962). “The Meaning of Poverty,” The British Journal of Sociology, Vol. 13(3), pp. 210–227. Whiteford, P. and W. Adema (2006). “Combating Child Poverty in OECD Countries: Is Work the Answer?” European Journal of Social Security, Vol. 8(3), pp. 235–256.

October 21, 2013

15:59

Income Inequality in the OECD. . .

9in x 6in

This page intentionally left blank

b1598-ch05

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch06

6. TOP INCOMES Peter Hoeller

Introduction and Main Findings Many OECD countries have experienced a rise in income inequality over the past decades. The widening dispersion in income is often shaped by the increasing concentration of income at the top end of the income distribution. In the US, for example, the top 1% of the population received 18% of pretax income in 2008, up from 8% in 1980. While the income share of the top income recipients also rose in most other OECD countries, there is great variation across countries with respect to both the extent of this increase and the time when it started. Despite a growing interest in the rise in top incomes, there is still substantial disagreement about the causes and their relative importance. The second section of this chapter reviews top income developments across 19 OECD countries, while the third section assesses the forces shaping the rise in top incomes, focusing on taxation, globalization, technical progress, and remuneration issues.

Main Findings Top income developments • Excluding capital gains, the share of the top 1% income recipients in total income ranged from 5% in Denmark to 17% in the US in 2009. • After declining for many decades, top income shares started to rise in the mid-1970s in the English-speaking countries and later in the continental European countries and Japan. The rise was especially steep in

163

October 21, 2013

164

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch06

Hoeller

the English-speaking countries, Finland, and Portugal with the US top income share being as high in recent years as it was in 1913. • Top income recipients tend to be found among executives in the nonfinancial sector and people working in financial professions and to a lesser extent among the “stars” of the sports and entertainment industries.

Forces shaping top income developments Changes in taxation • Top income tax rates have come down considerably over time which may have boosted the income declared by top earners to the tax authorities. Studies estimating the elasticity of taxable income at the top with respect to the marginal tax rate typically put it at around −1 for a top marginal tax rate of 50%. • Tax regimes may influence the mix of compensation, tilting it toward lower taxed forms of compensation, and thereby boost disposable income particularly at the top. For example, capital gains are often taxed at a lower rate than other income and, in a few countries, they are not taxed at all. Stock options also benefit from preferential tax treatment in many OECD countries and the same holds for carried interest arrangements. Globalization, technological change, and remuneration issues • New information technologies together with globalization have widened the market for “stars” and thus boosted top incomes in the sports and entertainment industries. • The skill requirements and responsibilities of top managers have become more complex, owing not least to stronger competition associated with deregulation and globalization. Moreover, the stability of top management positions has declined, while the outside options of top managers have improved, raising their bargaining power. To the extent that these outside options include jobs in foreign countries, the latter may explain why the top income shares of some countries influence those of others. For example, the top income share in the US is found to have a considerable influence on the one in Canada, while that in the UK and Australia influence the one in New Zealand. • Globalization has also led to a sharp increase in the market capitalization of large multinational companies, with the rise in executive pay closely following the rise in company size.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

Top Incomes

b1598-ch06

165

• Collective bargaining is unlikely to affect top earners, but the growing use of performance-related pay and changes in pay norms, has boosted their earnings. Moreover, the winner-takes-all pay-off structure has spread beyond fields such as sport and entertainment, with rewards of CEOs or financial market traders depending heavily on performance relative to others. These developments seem to be more prevalent in the Englishspeaking countries than in the continental European countries and Japan.

Top Income Developments The share of income going to the top earners has risen sharply in many countries, most notably in the US. Top income earners comprise by definition only a small share of the population, but they can earn a very large share of total income. In the US, for instance, the top 1% of the population captured 18% of total income in 2008, up from 8% in 1980. Changes in top incomes can have a considerable effect on the overall income distribution. Atkinson et al. (2011), for instance, calculated that the change in the share of the top 1% more than explains the increase in the US inequality between 1976 and 2006. For those OECD countries for which data are available, the cross-country correlation between the change in household market income (HMI’s) Gini indices and the change in the top 10% income share from the mid-1980s to the mid-2000s is 0.46.1 This is in line with the finding in Chapter 3 that in about 50% of the countries that experienced an increase in labor income inequality, the increase was driven by rising inequality in the upper half of the earnings distribution. Top incomes can also have a sizeable influence on measures of income growth of different population groups (Atkinson et al., 2011). In the US, for instance, real household income rose by 1.2% on average per year in the 10 years to 2007. Excluding the top 1%, the average income of the bottom 99% grew by only 0.6%, which implies that the top 1% captured 58% of the real income gain. The skewing effect of top incomes can also be gauged by the difference in the development of the mean and median incomes. According to data from the US Census Bureau, average real household 1 Data

for top incomes come from the World Top Incomes Database and Matthews (2011). HMI inequality is sourced from the OECD Household Income Distribution and Poverty Database.

October 21, 2013

166

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch06

Hoeller

income grew by 34% from 1980 to 2007, but median real household income increased by just 18% during the same period. Also international income growth rankings can change. Real household income growth was faster in the US (at 32%) between 1975 and 2006 than in France (at 27%). However, excluding the top percentile, the US growth rate was only 18%, while the French growth rate was virtually the same. Different measures of income growth can thus lead to different conclusions about growth performance (Atkinson et al., 2011). Atkinson and Piketty (2007; 2010) and their research associates have assembled data on top incomes for many countries and over a very long time span. The data on top incomes are based on income tax returns, except for Finland. Figure 6.1 shows the top income share for the G7 countries from about the start of the previous century. A pronounced downward trend can be detected for all countries for many decades, but there was a substantial increase for the English-speaking countries starting in the late 1970s or early 1980s, with the US top income share being as high in 2009 as it was in 1913. While there was an up-tick in top income shares in Germany, Japan, and France that started later than in the US, it was much less pronounced. Looking at developments since 1980 and including also many other smaller European countries as well as Australia and New Zealand, confirms this picture (Fig. 6.1). Inequality among the top income earners has also risen. The increase in the income share of the 1% and 0.01% of the top earners relative to the average of the top 10% was especially steep in the US, but occurred also in all other countries with the exception of Japan. While capital income provided the largest income source for top income earners early on in the previous century, its share has shrunk, and that of labor and self-employment income has risen considerably since the early 1980s. In the US, for instance, labor and self-employment income accounted for about 50% of the total income of the top 0.1% earners in 2006. The contribution of capital gains to top incomes is very volatile, but provided a very large contribution in the US prior to the Great Depression and prior to the recent economic and financial crisis. A sharp increase prior to the recent recession can also be observed for Canada, Spain, and Finland, but not for Australia and Norway.2

2 Capital

gains data are only available for seven countries.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch06

Top Incomes

Percent 23

167

Top 1% income share United States

United Kingdom

Canada

21 19 17 15 13 11 9 7 5 1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Percent 23

Japan

Germany

France

Italy

21 19 17 15 13 11 9 7 5 1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010

Fig. 6.1: Top Income Shares Over the Long Run. Note: Income is defined as market income excluding realized capital gains plus transfers. Source: The World Top Incomes Database, http://topincomes.g-mond.parisschoolof economics.eu/, February 2013.

Top income recipients tend to be concentrated in certain professions: In the US, they are mostly found among executives in the nonfinancial sector (41%) and people working in financial professions (18%) (Bakija et al., 2010).3 In the UK, 21% of the top 1% income recipients work in financial markets, though they make up only 3% of the tax payers. In 3 For

the US, Kaplan and Rauh (2007) estimate that among the top 0.1% earners, the most important groups are Wall Street-related professionals, nonfinancial public company CEOs, and lawyers. These groups comprise between 15% and 26.5% of the individuals at and above the 0.1% top earners.

October 21, 2013

168

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch06

Hoeller

France, where a finer split exists, 40% of the salaried top 1% earners are managers in headquarters of companies, while 15% work in the financial industry (INSEE, 2010). About 4% are sportsmen or work in the media and entertainment sector; 87% of the top earners are men. Among the selfemployed, doctors and lawyers are the largest group, with a share of 27% and 12%, respectively.

Forces Shaping Top Income Developments The significant decline in top income shares until the 1980s has been shaped by a combination of factors (Atkinson et al., 2011). The major ones include: (i) the destruction of capital during the two World Wars and war-time inflation during the First World War or hyperinflation afterwards in some countries; (ii) wealth losses due to the financial crisis of the 1930s; (iii) a narrowing of the labor income distribution during the Second World War, partly as a result of wage controls; (iv) nationalizations after the Second World War; and (v) income taxation, that became more progressive and taxation of inheritances and wealth, the latter making it more difficult for the top earners to accumulate wealth. The more recent increase in top income shares has differed considerably across countries as has the timing when the increase started (Fig. 6.2). Though it is clear that a sharp rise in capital income is not a major culprit, there is substantial disagreement about the causes of the rise in top incomes and their relative importance. Only a start has been made on testing different explanations and the impact of policy (Atkinson et al., 2011).

Taxation Top income tax rates have come down considerably and this may have contributed to boosting the income declared by top earners to the tax authorities. Top marginal tax rates stood at or above 70% in the early 1980s in Italy, Japan, and the US and were above 50% in the other G7 countries, while they are currently between 40% and 50% in all G7 countries. Lower personal income tax rates can have the following effects: (i) they reduce the benefits of tax evasion and aggressive tax avoidance and can lead to tax planning that can imply more income taking the form of personal,

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

Top Incomes

b1598-ch06

169

Share of top 1 % in selected years 18 1980

2009 or latest available year

16 14 12 10 8 6 4 2 0

Fig. 6.2: Share of Top 1% Income Recipients in Total Earnings. Note: Income data exclude capital gains, except for Australia and Finland. Source: The World Top Incomes Database, http://g-nd.parisschoolofeconomics.eu/ topincomes, February 2013 and Matthews (2011).

rather than corporate income4 ; (ii) they raise incentives to work and foster entrepreneurial spirits5 ; and (iii) they increase the ability of high earners to accumulate wealth. Concerning work incentives, estimated elasticities tend to be small (Gruber and Saez, 2002; Meghir and Phillips, 2010). These elasticities are based on a broad income concept, which also includes income items that are untaxed. However, the elasticity of taxable income is substantially higher, 4 In

the US, the use of “pass-through” entities has risen considerably since the 1986 tax reform, which brought the personal below the corporate income tax rate (CBO, 2011). 5 Andrews et al. (2010) investigated the link between rising top incomes and growth. They found evidence that from 1960 to 2000 a rise in top income shares was associated with a rise in growth rates during the following year, but the effect is fairly small. They also found that as long as the increase in top income shares persists, the rise in the growth rate also persists. Initially, increasing the share of income going to the top decile lowers the income of the bottom nine deciles. But if the positive effect of greater inequality on growth persists indefinitely, the resulting rise in total income will eventually exceed the loss entailed by lowering the share of personal income going to the bottom nine deciles, but this would take 13 years to materialize.

October 21, 2013

170

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch06

Hoeller

at 0.57 in the US, for high-income individuals, largely because they engage in more tax planning, and tax planning responds to marginal tax rates. In a review of the literature, Saez et al. (2009) report a similarly sized taxable income elasticity for Canada (after controlling for top incomes in the US), Sweden, and the UK. These estimates are semi-elasticities. At a marginal tax rate of 50%, a rise in the tax rate to 51% leads to a decline in tax receipts of 1%. For Canada, Department of Finance Canada (2010) reported that the taxable income elasticity is rising sharply among top income earners: the elasticity is 0.2 for the top 10%, but larger than 0.5 for the top 1%. In a study that estimates elasticities by income source for Canada, a very high elasticity was found for self-employment income and a very low one for labor income. A similar difference was found in a Swedish study. There is, thus, considerable evidence that the elasticity of taxable income is higher for high-income individuals who have access to avoidance opportunities, but that it also depends on the income source and the broadness and extent of enforcement of the tax base. Tax avoidance incentives are also shaped by the different treatment of income sources. Capital gains on shareholdings escape the income tax rate schedule in many countries. Capital gains are often taxed at a lower, flat rate, while a few countries do not tax capital gains at all (Table 6.1). And in a qualitative assessment of the taxation of stock options, OECD (2005) found that the tax treatment was preferential in many, though not all OECD countries. The same probably holds for carried interest arrangements,6 though such schemes are not well documented. The differential tax treatment favors mainly executives, finance professionals, and entrepreneurs, and there is no apparent market failure that would justify a preferential treatment (Matthews, 2011). Apart from the decline in income tax rates, the taxation of property has come down as a percentage of total taxation, despite the sharp rise in property values in most countries. Dwindling taxation of property, wealth, and inheritances makes it easier to accumulate wealth. Table 6.1 shows that many countries do not tax wealth, or abolished wealth taxation in recent years and tax revenues tend to be small. While most countries tax inheritances and gifts, tax receipts are small in this case in most countries. For the US, Burman et al. (2008) show that the federal estate tax is the 6 Carried

interest is a share of the profits of a partnership that is paid to the manager of the partnership (a private equity fund or hedge fund) as a form of compensation.

October 21, 2013

Table 6.1: The Tax Treatment of Capital Gains on Shares, Net Wealth and Inheritances.

16:9

2010, resident taxpayers. Estate/ Inheritance

Net wealth

Australia Taxable. Indexation or halving of capital gain allowed for disposals 1 year+ Austria Not taxable except speculative gains

Canada

Taxable

Marginal income tax rate

N

0.0

N

Normally zero/half of effective income tax rate on speculative gains Normally zero; 33% on speculative transactions; 16.5% on short-term immovable property gains 50% of realized gains taxed at income tax rate

N since 1994

0.3

N since 2008

0.1

Y

N after 2 years

0.9

N

0.0

N

Y

0.0

N (from Aug 08)

0.0

0.1

b1598-ch06

171

(Continued )

9in x 6in

Not taxable except speculative transactions

N (since 1979)

Weight in total tax revenues %a

Top Incomes

Belgium

Rate and regime

Weight in total tax revenues %a Tax imposed (Y) or not (N)

Income Inequality in the OECD. . .

Shares

Tax imposed (Y) or not (N)

Gift

Taxable

28/43/45% above ceiling Flat rate tax of 28% Taxed at flat 18%

0.0

Y

Y

0.1

N since 1997

0.1

Y

Y

0.5

N since 1/1/06

0.1

Y

Y

0.6

Y

0.4

Y

Y

1.0

N since 1998

0.5

Y

Y

0.4

N

0.1

Y

Y

0.6

N N since 2006

0.0 1.2

Y Y

Y N

0.1 0.2

N

0.0

Y

0.5

(Continued )

b1598-ch06

Normally zero; speculative gains taxed at income tax rates Now flat rate of 10% on shares Flat rate of 25% 18% in 2010; 10% prior to June 2009 Flat rate of 20% until 2008; 25% since April 2009

N

9in x 6in

Ireland

Not taxed

Income Inequality in the OECD. . .

Czech Not taxable Republic Denmark Taxed as income from shares Finland Taxable as income from capital France Taxable as professional income Germany Not taxable until 2010, except speculative gains: 1 year for shares Greece Exempt prior to 1/1/2010 Hungary Taxable Iceland Taxable

Rate and regime

Gift Weight in Weight in total total tax tax revenues %a Tax imposed (Y) or not (N) revenues %a

16:9

Shares

Tax imposed (Y) or not (N)

Hoeller

Estate/ Inheritance

Net wealth

October 21, 2013

172

Table 6.1: (Continued )

October 21, 2013

Table 6.1: (Continued )

Shares

Japan

49.72% subject to tax

0.2

N

0.0

Y

Y

1.5

Not taxed Maximum rate of 19.5%

N N since 2006

0.0 5.6

Y Y

Y Y

0.9 0.3

Subject to income tax rates

N

0.0

N

N among families

0.0

N since 2001 N

0.3 0.0

Y N (since 1992)

Y Y

0.7 0.1

Not taxed Not taxed

Y (since 1/1/07)

0.1

b1598-ch06

173

(Continued )

9in x 6in

N

Top Incomes

Withholding tax of 1.05% or tax assessment on 26% of proceeds Korea Exempt Luxembourg Taxable as miscellaneous income; speculative gains as ordinary income Mexico Exempt when securities are classified as available to the general public Netherlands Not taxable New Not taxable Zealand

Subject to progressive income tax rates 23–43% Taxed at 10%

Gift Weight in Weight in total total tax tax revenues %a Tax imposed (Y) or not (N) revenues %a

Income Inequality in the OECD. . .

Italy

Rate and regime

Tax imposed (Y) or not (N)

16:9

Estate/ Inheritance

Net wealth

Rate and regime

Taxable

Included in taxable income

Poland Portugal

Taxable Taxable

Flat rate of 19% 10% unless the taxpayer opts for its inclusion in his or her taxable income Flat rate of 19%

Switzerland Exempt except for professional share dealing

Income tax at rates of 24–43% Flat rate tax on capital income: 30% Cantons set their own tax rates

Y: up to 0.7% municipal and0.4% national N N

1.3

Y

Y

0.2

0.0 0.0

Y Y N since 1/1/04

0.1 0.2

N

0.0

N since 1/1/04

0.0

N since 1/1/08

0.5

Y

Y

0.6

N since 2007

0.4

N

N

0.0

Y at cantonal level

4.5

Y

Yc

0.9

b1598-ch06

(Continued )

9in x 6in

Slovak Rep Exempt up to a ceiling Spain Treated as ordinary income Sweden Included in income from capital

Weight in Weight in total total tax tax revenues %a Tax imposed (Y) or not (N) revenues %a

Income Inequality in the OECD. . .

Norway

Gift

16:9

Shares

Tax imposed (Y) or not (N)

Estate/ Inheritance

Hoeller

Net wealth

October 21, 2013

174

Table 6.1: (Continued )

October 21, 2013

Table 6.1: (Continued )

Shares

Gift

Weight in Weight in total total tax tax revenues %a Tax imposed (Y) or not (N) revenues %a

N

0.0

Y

Y

0.1

N

0.0

Y

Y

0.6

N

0.0

Yb

Yb

1.0

Top Incomes

b1598-ch06

175

Note: a The period average from 1985 to 2008, except for the countries, where only a shorter sample is available. b Estate tax and generation-skipping transfer tax were gradually phased out from 2002 to 2010 but have been partially reinstated in 2011 to 2012. c No tax was levied at the federal level, but most cantons levy such taxes. Source: Price and Dang (2011).

9in x 6in

Taxable but Income tax rates: exemption for 15–35% with shares in resident inflation companies held for adjustment 3–12 months United Taxable Marginal income tax Kingdom rate before 2008/2009; now at a flat rate of 18% United Taxable Short-term gains States taxed at income tax rate; long-term gains (more than a year) 5–15% until 2010

Estate/ Inheritance

Income Inequality in the OECD. . .

Turkey

Rate and regime

Tax imposed (Y) or not (N)

16:9

Net wealth

October 21, 2013

176

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch06

Hoeller

most progressive tax, applying to only the wealthiest 2% of descendants. Finally, it is shown in Chapter 4 that residential property taxation can be progressive or regressive, depending on the features of the property taxation system. While the high elasticities of taxable income imply that taxing the top earners more heavily may not raise much additional government revenue, reducing tax avoidance incentives and raising property taxation would reduce the after-tax income of top earners and thus inequality.

Technical Progress and Globalization Skill-biased technical progress could have played some role in raising the top part of the income distribution, but it cannot explain the rising income concentration among the top 10% income earners, as they do not include unskilled individuals. On the other hand, globalization in tandem with new information technologies has widened the market for “stars” for some professions (Rosen, 1981), which is likely to be one reason for the increasing inequality among the top earners. Globalization has also led to a sharp increase in the market capitalization of large, multinational companies, with the rise in pay of the US top managers closely following the rise in company size (Gabaix and Landier, 2008). Migration is also likely to have played a role. Saez and Veall (2005), for instance, find that the US top income share has a considerable influence on the Canadian one, while Atkinson and Leigh (2008) finds a similar effect of the UK and Australian income shares on the top incomes in New Zealand. Migration could, thus, provide a strong link between top incomes that is likely to be stronger for the English-speaking countries.

Remuneration Issues While most labor market policy instruments or institutions, such as minimum wages or collective bargaining, are unlikely to affect top earners, the growing use of performance-related pay and changes in pay norms, including in the remuneration of top executives, certainly have. For the US, for instance, Lemieux et al. (2007) found that nearly all of the change in the wage dispersion of the top 20% earners is due to the spread of performancerelated pay. Moreover, Gordon and Dew-Becker (2008) have argued that the winner-takes-all pay-off structure has spread beyond fields such as sport and entertainment, with rewards of CEOs or financial market traders depending

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

Top Incomes

b1598-ch06

177

heavily on performance relative to others. Conyon et al. (2010) estimated that executive pay in the US is 23% higher than in the UK and 55% higher than in continental Europe. The link between corporate governance and executive compensation is likely to differ between the English-speaking and other countries. In the English-speaking countries, companies claim to maximize shareholder value and compensation practices seek to align the interests of the executives with those of the shareholders. Hostile takeovers are a device to discipline managers, and performance-related pay aligns the interest of management with that of the dispersed shareholders. In Europe and Japan, this principal– agent dilemma applies less directly, as other stakeholders are more powerful. Banks are key players in financing firms and there is often an attendant close relationship between banks and companies. Moreover, businesses and labor unions often tend to cooperate and exchange information. Consultations with worker representatives before important decisions, including on pay, are made are likely to have at least some influence on the pay of top executives and how performance-related pay is structured (McCall and Percheski, 2010; Sj¨ oberg, 2009). The economic and financial crisis also brought remuneration issues in financial markets to the fore. In financial markets, managerial compensation schemes were skewed toward the short term and fostered excessive risk taking. Moreover, managers often have an influence on the terms of performance-based pay and pay was often not closely linked to performance, especially in the case of losses. The Financial Stability Board (FSB) (2009) suggested that compensation practices should be harmonized for large, systemically relevant banks and other financial institutions. Compensation packages should include deferral, vesting and claw-back clauses and bonus payments should be limited (as well as dividend pay-outs), when regulatory capital is close to the minimum.7

7 The

UK government slapped a bank payroll tax on banks that awarded discretionary bonuses above £25,000 between December 2009 and April 2010. The banks awarding these bonuses had to pay tax of 50% on bonuses above £25,000. As the tax did not reduce bonus payments, but was paid by the employers, the government introduced a permanent bank levy. Also, the non-taxation of overseas income and capital gains of the non-domiciled, many of whom work in the financial sector was abandoned. They now pay a flat tax of £30,000, when staying in the UK for 7 years and £50,000 after 12 years. Also, the low taxation of carried interest (at only 10%) aroused quite some discussion, and it was raised to 18% in 2008.

October 21, 2013

178

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch06

Hoeller

References Alvaredo, F., A.B. Atkinson, T. Piketty and E. Saez (2011). “The Top Incomes Database.” Available at: http://topincomes.g-mond.parisschoolof economics.eu/. Accessed February 2013. Andrews, D., C. Jencks and A. Leigh (2011). “Do Rising Top Incomes Lift All Boats?” The B.E. Journal of Economic Analysis and Policy, Vol. 11(1), pp. 1–43. Atkinson, A.B. and A. Leigh (2008). “Top Incomes in New Zealand 1921–2005: Understanding the Effects of Marginal Tax Rates, Migration Threat, and the Macroeconomy,” Review of Income and Wealth, Vol. 54(2), pp. 149–165. Atkinson, A.B. and T. Piketty (2007). Top Incomes over the Twentieth Century: A Contrast between Continental European and English-Speaking Countries, Vol. 1, Oxford University Press, Oxford. Atkinson, A.B. and T. Piketty (2010). Top Incomes over the Twentieth Century: A Global Perspective, Vol. 2, Oxford University Press, Oxford. Atkinson, A.B., T. Piketty and E. Saez (2011). “Top Incomes in the Long Run of History,” Journal of Economic Literature, Vol. 49(1), pp. 3–71. Bakija, J., A. Cole and B. Heim (2010). “Jobs and Income Growth of Top Earners and the Causes of Changing Income Inequality: Evidence from U.S. Tax Return Data.” Available at: http://web.williams.edu/Economics/wp/Bakija ColeHeimJobsIncomeGrowthTopEarners.pdf. Accessed November 2012. Burman, L.E., K. Lim and J. Rohaly (2008). Back from the Grave: Revenue and Distributional Effects of Reforming the Federal Estate Tax, Tax Policy Centre, Urban Institute and Brookings Institution, Washington. Congressional Budget Office (CBO) (2011). “Trends in the Distribution of Household Income between 1979 and 2007,” The Congress of the United States. Conyon et al. (2010). “Are US CEOs Paid More than UK CEOs? Inferences from Risk-adjusted Pay.” Available at: http://ssrn.com/abstract=907469. Accessed March 2011. Department of Finance Canada (2010). “The response of individuals to changes in marginal income tax rates,” In: Tax Expenditures and Evaluations, Ottawa. Financial Stability Board (FSB) (2009). “FSF Principles for Sound Compensation Practices,” Basel. Gabaix, G. and A. Landier (2008). “Why has CEO Pay Increased so much?” Quarterly Journal of Economics, Vol. 123(1), pp. 49–100. Gordon, R.J. and I. Dew-Becker (2008). “Controversies about the Rise in American Inequality: A Survey,” NBER Working Paper No. 13982. Gruber, J. and E. Saez (2002). “The Elasticity of Taxable Income: Evidence and Implications,” Journal of Public Economics, Vol. 84(1), pp. 1–32. INSEE (2010). “Les tr`es hauts salaires du secteur priv´e,” INSEE Premi`ere No. 1288.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

Top Incomes

b1598-ch06

179

Kaplan, S.N. and J. Rauh (2007). “Wall Street and Main Street: What Contributes to the Rise in the Highest Incomes?” NBER Working Paper No. 13270. Lemieux, T., W.B. MacLeod and D. Parent (2007). “Performance Pay and Wage Inequality,” NBER Working Paper No. 13128. Matthews, S. (2011). “Trends in Top Incomes and their Tax Policy Implications,” OECD Taxation Working Paper Series No. 4. McCall, L. and C. Percheski (2010). “Income Inequality: New Trends and Research Directions,” Annual Review of Sociology 2010, Vol. 36, pp. 329–347. Meghir, C. and D. Philips (2010). “Labour supply and taxes.” In: Dimensions of Tax Design, Vol. 1 of the Mirrlees Report of the Commission on Reforming the Tax System for the 21st Century, Oxford University Press, Oxford. OECD (2005). The Taxation of Employee Stock Options, OECD Publishing. Price, W.R. and T.T. Dang (2011). “Adjusting Fiscal Balances for the Asset Price Cycle,” OECD Economics Department Working Paper No. 868. Rosen, S. (1981). “The Economics of Superstars,” American Economic Review, Vol. 71(5), pp. 845–858. Saez, E. and M.R. Veall (2005). “The Evolution of High Incomes in Northern America: Lessons from Canadian Experience,” American Economic Review, Vol. 95(3), pp. 831–849. Saez, E., J.B. Slemrod and S. Giertz (2009). “The Elasticity of Taxable Income with Respect to Marginal Tax Rates: A Critical Review,” NBER Working Paper No. 15012. Sj¨ oberg, O. (2009). “Corporate Governance and Earnings Inequality in the OECD Countries 1979–2000,” European Sociological Review, Vol. 25(5), pp. 519–533. US Census Bureau (2011). “Current Population Survey, Annual Social and Economic Supplements.” Available at: www.census.gov/hhes/www/income/ data/historical/household/index.html, accessed March 2011.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

This page intentionally left blank

b1598-ch06

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

7. THE DISTRIBUTION OF WEALTH Kaja Bonesmo Fredriksen

Introduction and Main Findings The wealth distribution is much more concentrated than the income distribution. Of the few countries for which comparable wealth distribution data are available, wealth inequality measured by the Gini index is highest in Sweden and the US and relatively low in Italy. The lack of harmonized wealth data has hampered research on wealth distribution issues. This has particularly been the case for cross-country studies. The situation has recently improved with the launch of the Luxembourg Wealth Study (LWS), a harmonized household survey, which provides data on the size and composition of wealth and also includes a broad range of household characteristics for 11 OECD countries. Trends in wealth inequality can only be assessed using non-harmonized national wealth surveys. They suggest that the wealth concentration came down considerably since the beginning of the 20th century until the 1970s, but over the past 30 years, wealth inequality has increased again. The first section of this chapter reviews measurement issues. The second section examines the inequality of net worth, whereas the third section looks at the role of various assets and liabilities. The fourth section reviews trends in wealth inequality over time. The fifth section analyses the determinants of wealth inequality.

Main Findings • The concentration of wealth is very high. The lowest 50% of households in the wealth distribution hold only a tiny fraction of wealth, while the top 10% hold between 40% (Italy) and 70% (the US) of total wealth.

181

October 21, 2013

182

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

Fredriksen

• The countries with the most unequal income distribution are not necessarily those with the most unequal wealth distribution. Among the countries covered in this study, wealth inequality is particularly high in Sweden and the US and is the lowest in Italy. • In all countries, assets are a more important contributor to overall wealth inequality than debt. Nonfinancial assets contribute more than financial assets, even though financial assets are more unequally distributed in all countries. This is explained by the higher share of nonfinancial assets in total assets. • Wealth inequality has trended down during the 20th century until the mid-1970s to early 1980s, from which point wealth inequality has been rising. The rise in inequality reflects (i) soaring financial markets in the aftermath of financial market deregulation that started in the 1970s; (ii) lower marginal tax rates on top incomes and lower capital gains and wealth taxation which have made the accumulation of wealth easier for the rich; and (iii) at least in France, inheritances and inter vivos gifts have risen again in importance over the past 30 years and stood at nearly 15% of national income in 2008, nearly as high as a century ago.

Measuring the Distribution of Wealth Wealth can be measured in different ways. Marketable wealth is the narrowest concept and includes only assets that generate capital income. Another possibility is to consider household disposable wealth which is the market value of assets minus liabilities that are directly tradable, with housing considered liquid enough to qualify. Augmented wealth is the most encompassing concept both in the time dimension, as it includes the present value of all discounted future income, and in the number of assets considered as both pension rights and human capital are also included. However, given data limitations, studies generally define wealth as household net worth, the sum of household financial and non-final assets minus debt, which is the concept used here. This implies that the value of pension rights and life insurance are excluded from the analysis. However, from a life-time perspective, assets that underpin future consumption should also be included. And not only is an important part of wealth left excluded, the cross-country comparability of the data is also reduced since the omission will bias results less in countries where individuals save more for their retirement outside the public pension scheme.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

The Distribution of Wealth

b1598-ch07

183

Even confined to net worth, measuring wealth holdings and wealth inequality is not easy. As is always the case in distributional studies, one must rely on household survey data or tax data. Such micro data have several shortcomings most notably a tendency to underreport at both tails of the distribution, thereby underestimating inequality. In the case of wealth, this problem is likely to be even greater due to the highly skewed distribution. An additional problem for wealth studies has been the lack of available data to perform cross-country analyses which require harmonized definitions and a common methodology. The LWS launched in 2007 partly remedies this. It is a cross-country database that provides wealth data where a high degree of comparability between countries has been achieved for four financial assets (deposit accounts, bonds, stocks, and mutual funds), nonfinancial assets (principal residence and investment in real estate) and debt. However, because of differences in the breakdown of aggregate wealth in the national surveys, there is a trade-off between the comparability of the cross-country dataset and its completeness. Currently the LWS covers only between 50% of Canadian and the US household wealth and about 80% of the household wealth of the European countries. According to Jantti et al. (2008) the LWS figures provide a better coverage of nonfinancial assets, while coverage is lower for liabilities and lowest for financial assets.1 The country coverage of the LWS database is limited. It contains comparable data for only 11 OECD countries: Austria, Canada, Finland, Germany, Japan, Italy, Luxembourg, Norway, Sweden, the UK and the US. Furthermore, since there are no data for nonfinancial assets for Austria, no debt data for Luxembourg, and because property values for Norway are tax-assessed rather than market based, comparable data on net worth are not available for these three countries. In the remaining countries, the comparable net worth aggregate is only available in the cross-section. The LWS, therefore, cannot be used for analyzing wealth holdings over time.

Inequality of Net Worth Although the percentage of households holding some assets is quite similar across countries, the value of mean and median net worth is quite different

1 In

addition to pension wealth, business equity is also excluded from the data.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

Fredriksen

184

In 2005 PPP USD 300000

Mean

Median

250000

200000

150000

100000

50000

0 Sweden

Finland

Canada

United Kingdom

Japan

Italy

United States

Fig. 7.1: LWS Country Rankings by Mean and Median of Net Worth. Source: LWS.

(Fig. 7.1). Median net worth in Italy and Japan is higher than in the other countries. The UK, Finland, and the US are middling, whereas Sweden is the wealth-poorest country. The large cross-country differences in the ratio of mean and median net worth indicate important cross-country differences in wealth inequality which is confirmed when looking at the share of net worth held at different points in the distribution, as well as the Gini index that measures aggregate wealth inequality (Fig. 7.2). Wealth inequality is particularly high in Sweden (Box 7.1) and the US, whereas it is the lowest in Italy. The high value for the US could also in part reflect efforts to better capture the upper tale of the wealth distribution in the US household survey. Wealth is much more unequally distributed than income in all countries and the countries with the most unequal income distribution are not those with the most unequal wealth distribution (Fig. 7.3). On the other hand, both the property income concentration and that of net worth are very high in all countries. At the upper end of the wealth distribution, wealth in the US is much more unequally distributed than in the other countries and particularly so among the very rich. The 1% wealthiest Americans hold 31% of total wealth, as compared to between 10% and 20% for the other countries (Table 7.1).

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

The Distribution of Wealth

185

In the late 1990s, early 2000s 99 to 100

95 to 99

90 to 95

50 to 90

0 to 50

100 90 80 70 60 50 40 30 20 10 0 Canada (0.75)

Finland (0.68)

Germany

Italy

Japan

Sweden

United Kingdom

United States

(0.80)

(0.60)

(0.71)

(0.89)

(0.67)

(0.84)

Fig. 7.2: The Share of Net Worth Held by Households at Different Points in the Wealth Distribution. Note: For Germany, the first wealth proportion refers to wealth held up to the 90th population decile. For the US, the Survey of Consumer Finance (SCF) is used and the data refer to 2006. The Gini index for wealth is shown in brackets below the country name. Source: LWS and OECD (2008).

Box 7.1. Why Is Wealth Inequality so High in Sweden? Sweden is a particularly interesting case as inequality is very high for net worth, but very low for household disposable income (HDI). There are several potential explanations for this. Whereas in the US the wealth distribution is very skewed at the top, Sweden is characterized by a large fraction of households with negative wealth, with the bottom 30% having negative wealth. The high incidence of debt is likely to be underpinned by interest deductibility for tax purposes, while student debt is also important. Domeij and Klein (2002) found that the generous public pension scheme with a common benefit payable to each senior citizen combined with an upper limit to benefits from the earnings-related pension scheme is a main driver of wealth inequality in Sweden as it (Continued )

October 21, 2013

186

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

Fredriksen

Box 7.1. (Continued ) discourages private asset accumulation at the lower end of the distribution. However, although Sweden has a relatively generous public pension scheme, it is certainly not the only country for which this is the case. Finland also has a generous welfare system, but has also relatively low wealth inequality. Finland also has a lower incidence of negative net worth and a higher home ownership rate (Davies, 2009). Much of the difference in wealth inequality between Sweden and the other countries is caused by a more uneven distribution of nonfinancial wealth. First, the overall home ownership rate at 57% is low compared with most other countries. But the most striking difference with other countries is not the lower incidence, but the higher concentration of nonfinancial wealth. The ratio between the nonfinancial wealth holdings of the 75th and the 25th percentile is double that of the US, and almost treble if one looks only at the primary residence. All other countries have a lower concentration of nonfinancial wealth, which is at least partly due to the structure of the housing market. The social rental market in Sweden is very large at 21% of the total housing stock and is almost the same size as the private rental market. The social rental market is, in addition, not that different from the private rental market. Both are quite heavily regulated with regard to prices and provide secure tenure. This makes renting more lucrative than in other countries. The generous pension system, the large rental sector and incentives to take on debt are likely to be the major reasons for the high wealth inequality in Sweden.

The Role of Various Assets and Liabilities Though there are strong similarities in the relative importance and concentration of specific assets and liabilities, there are also certain notable cross-country differences. As mentioned above, these data do not take into account business equity, life insurance, and pension rights. Nonfinancial assets are more important than financial assets in all countries. Their share in total assets ranges from 87% in Germany to 71% in the US (Table 7.2). Nonfinancial assets consist mainly of the primary residence. Although financial assets have a lower share in total assets, in terms of asset participation

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

The Distribution of Wealth

187

A. Gini for disposable income and wealth Disposable income 0.40 USA ITA 0.35 GBR

CAN

JPN 0.30

DEU FIN 0.25 SWE 0.20 0.5

0.6

0.7

0.8

0.9

1.0 Wealth

B. Gini for property income and wealth

Property incomeb 1.00

FIN 0.90 ITA

SWE

DEU GBR USA

CAN 0.80

0.70

0.60 0.6

0.7

0.8

0.9

1.0 Wealth

Fig. 7.3: Inequality of Income, Wealth, and Property Income.a Note: a The income inequality data are for the mid-2000s. For wealth, the numbers refer to the following years: Canada (1999), Finland (1998), Japan (2003), Italy (2004), Sweden (2002), the UK (2000), and the US (2006). b The property income Gini index is based on LIS data. It ranks individuals by capital income. The Gini index for wealth is based on LWS data. Source: Luxembourg Income Study, LWS, and OECD (2008).

October 21, 2013

188

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

Fredriksen Table 7.1: Share of Net Worth of the Upper Percentiles.

Canada Finland Germany Italy Japan Sweden United Kingdom United Statesa

Top 1 pct.

Top 5 pct.

Top 10 pct.

15 13 16 11 12 18 10 31

37 31 38 29 31 41 30 58

53 46 55 42 47 58 45 70

Note: a The figures for the US are based on the SCF (2006) household survey. Source: LWS and OECD (2008).

Table 7.2: Household Portfolio Composition. Percentage share in total assets Wealth variable Nonfinancial assets Financial assets Total assets Debt Net worth

United United Canada Finland Germanya Italy Sweden Japan Kingdom Statesb 78

84

87

85

72

80

83

71

22

16

13

15

28

20

17

29

100 26 74

100 16 84

100 23 77

100 4 96

100 35 65

100 18 81

100 21 79

100 25 74

Note: a Most of financial assets and non-housing debt are recorded only for values exceeding EUR 2,500. b SCF (2006). Source: LWS and OECD (2008).

they are more important; in almost all countries, 75% of the population or more hold some form of financial asset (Table 7.3). This is mainly because most people hold some wealth in a deposit account. American households hold more financial assets relative to the households in the other countries. Furthermore, American, Swedish, and Finnish households have the greatest preference for holding stocks. As regards debt holdings, there is substantial variation and the lowest level of household debt is found in Italy and the highest in Sweden.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

The Distribution of Wealth

189

Table 7.3: Household Asset Participation. In percentage Wealth variable

United United Canada Finland Germanya Italy Sweden Japan Kingdom Statesb

Nonfinancial assets Financial assets Debt

64

68

43

72

57

77

70

71

90

92

49

81

79

75

80

91

68

52

32

22

70

41

59

77

Nonfinancial assets Financial assets Total debt

64

68

43

72

—

77

70

71

48

53

49

70

—

71

58

64

58

45

32

17

—

40

49

70

Note: a Most of financial assets and non-housing debt are recorded only for values exceeding EUR 2,500. b SCF (2006). Source: LWS and OECD (2008). Table 7.4: Gini Index of Household Net Worth and Its Components.

Gini coefficient Canada Finland Italy Japan Sweden United Kingdom United States

Net worth

Total assets

Financial assets

Nonfinancial assets

Debt

0.75 0.68 0.60 0.71 0.89 0.67 0.84

0.63 0.58 0.58 0.59 0.67 0.58 0.74

0.86 0.80 0.73 0.72 0.78 0.80 0.91

0.62 0.57 0.60 0.63 0.70 0.57 0.70

0.72 0.75 0.91 0.82 0.73 0.78 0.72

Source: LWS.

In all countries, financial assets are substantially more unequally distributed across the population than nonfinancial assets (Table 7.4). There is, however, some cross-country variation worth noting. The US and Canada stand out with a particularly uneven distribution of financial assets. On the other hand, the distribution of financial assets in Germany is quite even and not much more unequal than the distribution of nonfinancial assets. Debt is somewhat more unequally distributed than total assets in all countries except in the US.

October 21, 2013

190

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

Fredriksen

The contribution of specific assets and liabilities to the inequality of net worth depends not only on their relative importance and concentration but also on how their distribution correlates with the distribution of net worth. Table 7.5 shows a decomposition of the contribution of the specific assets and liabilities to overall net worth inequality using the same method as Lerman and Yitzhaki (1985).2 In all countries, assets are a more important contributor to overall wealth inequality than debt. For most of the countries, the distribution of

Table 7.5: Contributions to Overall Net Worth Inequality.

Canada

Finland

Italy

Japan

Sweden

United Kingdom

Contribution of asset- and debt inequality to net worth inequality Gini assets (1a) 0.63 0.59 0.58 0.59 0.67 0.58 Gini debt (2a) 0.72 0.78 0.91 0.82 0.73 0.75 Share assets (1b) 1.36 1.19 1.04 1.23 1.54 1.28 Share debt (2b) −0.36 −0.19 −0.04 −0.22 −0.54 −0.25 Correlation assets/ 0.92 0.94 1.00 0.92 0.88 0.93 net worth (3a) Correlation debt/ 0.14 −0.04 0.11 −0.22 0.07 0.17 net worth (3b) Gini net worth 0.75 0.67 0.60 0.70 0.87 0.65 (1a*2a*3a + 1b*2b*3b) Asset contribution 104.84 99.22 100.63 94.26 102.93 104.77 in % (4a = 1a* 2a*3a/Gini net worth) Debt contribution −4.84 0.78 −0.63 5.74 −2.93 −4.77 in % (4b = 1b* 2b*3b/Gini net worth)

United States

0.74 0.72 1.35 −0.35 0.95 0.39 0.85

111.42

−11.42

(Continued )

2 The

Gini index of net worth (or any aggregate) can be computed as Σi Gi × Ri × Si , where Gi is the Gini of component i, Ri is the Gini correlation between component i and net worth, which varies between −1 and 1, and Si is the share of component i in net worth. Thus, the contribution of component i to the Gini index of net worth can be calculated as the ratio between (Gi × Ri × Si ) and the Gini index of net worth. These contributions sum to 1.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

The Distribution of Wealth

191

Table 7.5: (Continued )

Canada

Finland

Italy

Japan

Sweden

United Kingdom

United States

Contribution of financial and nonfinancial inequality to total asset inequality Gini financial assets (1a) Gini nonfinancial assets (2a) Share financial assets (1b) Share nonfinancial assets (2b) Correlation financial-/total assets (3a) Correlation nonfinancial-/ total assets (3b) Gini total assets (1a*2a*3a + 1b*2b*3b) Financial asset contribution in % (4a = 1a*2a*3a/ Gini total assets) Nonfinancial asset contribution (4b = 1b*2b*3b/ Gini total assets)

0.86

0.79

0.73

0.72

0.78

0.80

0.91

0.62

0.59

0.60

0.63

0.70

0.57

0.70

0.22

0.16

0.11

0.20

0.28

0.16

0.29

0.78

0.84

0.89

0.80

0.72

0.84

0.71

0.85

0.83

0.71

0.79

0.81

0.81

0.91

0.95

0.99

0.99

0.94

0.97

0.96

0.95

0.62

0.60

0.58

0.59

0.67

0.57

0.71

26.12

17.89

9.92

19.76

26.66

18.70

33.63

73.88

82.11

90.08

80.24

73.34

81.30

66.37

Contribution to financial asset inequality Gini deposit accounts (1a) Gini bonds (2a) Gini stocks (3a) Gini mutual funds (4a) Share deposit accounts (1b) Share bonds (2b) Share stocks (3c)

0.82

0.76

0.67

—

0.81

—

0.84

0.97 0.98 0.96

0.99 0.94 0.99

0.94 0.97 0.96

— — —

0.95 0.95 0.83

— — —

0.99 0.98 0.98

0.42

0.59

0.54

—

0.40

—

0.29

0.06 0.30

0.03 0.34

0.18 0.10

— —

0.07 0.21

— —

0.09 0.34 (Continued )

October 21, 2013

192

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

Fredriksen Table 7.5: (Continued )

Canada Finland

Italy

United United Japan Sweden Kingdom States

Contribution to financial asset inequality (Continued) Share mutual funds (3d) Correlation deposit accounts/financial assets (3a) Correlation bonds/financial assets (3b) Correlation stocks/financial assets (3c) Correlation mutual funds/financial assets (3d) Gini financial assets (1a*2a*3a + 1b *2b*3b+1c*2c*3c + 1d *2d*3d) Deposit account contribution in % (4a = 1a*2a*3a/Gini financial assets) Bonds contribution in % (4b = 1b*2b *3b/Gini financial assets) Stocks contribution in % (4c = 1c*2c *3c/Gini financial assets) Mutual funds contribution in % (4d = 1d*2d*3d/Gini financial assets)

0.21

0.04

0.17

—

0.31

—

0.28

0.93

0.95

0.91

—

0.91

—

0.88

0.87

1.00

0.89

—

0.87

—

1.04

0.98

0.95

0.92

—

0.95

—

1.01

0.93

1.04

0.90

—

0.90

—

0.99

0.86

0.80

0.73

—

0.78

—

0.92

37.65

52.81

45.53

—

37.94

—

23.51

6.04

3.74

21.00

—

7.59

—

9.99

34.30

38.37

12.55

—

24.56

—

37.11

22.01

5.09

20.92

—

29.91

—

29.39

(Continued )

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

The Distribution of Wealth

193

Table 7.5: (Continued )

Canada

Finland

Italy

Japan

Sweden

United Kingdom

United States

Contribution to nonfinancial asset inequality Gini primary residence (1a) Gini investment real estate (2a) Share primary residence (1b) Share investment real estate (2b) Correlation primary residence/ nonfinancial assets (3a) Correlation investment real estate/nonfinancial asset (3b) Gini nonfinancial assets (1a*2a*3a + 1b*2b*3b) Primary residence contribution in % (4a = 1a*2a*3a/ Gini net worth) Investment real estate contribution in % (4b = 1b*2b*3b/ Gini net worth)

0.60

0.57

0.57

0.61

0.71

0.56

0.65

0.93

0.87

0.93

0.98

0.95

0.97

0.95

0.83

0.77

0.80

0.88

0.85

0.90

0.73

0.17

0.23

0.20

0.11

0.15

0.10

0.27

0.96

0.94

0.95

0.98

0.98

0.97

0.94

0.85

0.93

0.89

0.95

0.85

0.80

0.95

0.62

0.60

0.60

0.63

0.71

0.57

0.69

77.95

68.80

72.01

83.39

82.47

85.86

65.05

22.05

31.20

27.99

16.61

17.53

14.14

34.95

Note: A detailed asset-decomposition is not available for Japan and the UK. Source: Calculations based on the LWS dataset.

liabilities reduces the concentration of wealth as the correlation between debt and net worth is positive, which indicates that those with high net wealth not only have more assets but also have more debt. Finland and Japan are two exceptions where the correlation between debt and net worth is negative, though in the case of Finland the correlation is small.

October 21, 2013

194

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

Fredriksen

Nonfinancial assets contribute more to asset inequality than financial assets, despite a lower Gini index. The higher contribution of nonfinancial assets is explained by the higher share of nonfinancial wealth than financial wealth in all countries and the strong positive correlation between the concentration of nonfinancial wealth and total assets. The share of nonfinancial wealth is particularly high in Italy, whereas Americans and Swedes have the highest preference for financial assets in relative terms. Deposit accounts are the biggest contributor to financial asset inequality in all countries except for the US, despite the fact that they are the least unequally distributed financial asset. This is explained by the high share of deposit accounts. The contribution of stocks and mutual funds is also high in many countries, whereas bonds have a marginal impact on overall financial asset inequality. Principal residences contribute more to overall inequality in nonfinancial assets than other real estate investment in all countries, although there is quite substantial variation in the size of the relative contributions.

Trends in Wealth Inequality Over Time The LWS dataset is mainly cross-sectional and cannot, therefore, be used to assess the evolution of wealth inequality over time. Tracing developments in wealth inequality for individual countries is, however, possible by relying on national surveys that have not been harmonized. National studies that rely on tax data over a long time span focus on the upper end of the distribution of wealth. They find that, with the exception of Switzerland, wealth inequality decreased from a very high level at the beginning of the 20th century until the mid-1970s to early 1980s. Piketty et al. (2006) conclude that the decline in wealth inequality in France was mostly due to adverse shocks to the portfolios of the wealthiest in the first half of the century with wealth destruction concentrated around the First World War, the stock market crash in 1929, and the Second World War. The enlargement of the welfare state financed by more progressive tax systems prevented a renewed sharp rise in wealth inequality in the aftermath of the Second World War. A similar pattern was found for the US (Kopczuk and Saez, 2004), Denmark, and Sweden (Ohlsson et al., 2007). In the UK (Atkinson et al., 1989) and Norway (Ohlsson et al., 2007), the decline concerned only the top percentile. Switzerland is an outlier, as the top wealth shares remained largely flat between 1915 and 1969, though there was some decline afterwards. Dell et al. (2005) argue that this is

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

The Distribution of Wealth

b1598-ch07

195

because Switzerland has never implemented a highly progressive income tax and wealth taxation was always light, while Switzerland stayed out of both World Wars. Studies based on household surveys indicate a turning point from the mid-1970s to mid-1980s, and the evidence that wealth inequality was rising at the turn of the century appears solid (Table 7.6). Unfortunately, recent data on wealth inequality are sparse. Wolff (2007; 2010) has updated his 1983 study. He found a continued, though small, increase in total wealth inequality in the US and a much larger increase in non-housing wealth inequality. Somewhat surprisingly, the increase was not interrupted during and in the immediate years following the burst of the dot.com bubble. This is explained by the rising indebtedness of the American middle-class, as asset inequality actually fell in the early 2000s. Little is known about the effect of the recent economic crisis on the wealth distribution, but household survey data until 2010 exist for the US. They show that the mean and median net worth fell between 2007 and 2010, but the decline in the median was greater than that of the mean. This suggests that the recession and slow recovery had a more adverse effect on the households in the bottom half of the wealth distribution than those further up the distribution (Levine, 2012).

Determinants of Wealth Inequality Apart from the distribution of market income, differences in the value of wealth holdings and wealth inequality are likely to be influenced by sociodemographic characteristics of households, inheritance patterns, the composition of wealth portfolios, and the evolution of relative returns on assets.

Socioeconomic Characteristics of the Population Age, education, and family type are all variables that one would expect to be correlated with the income and wealth distributions. According to the life-cycle theory, wealth holdings by age should follow a hump-shaped profile where net worth increases until an age close to retirement, and decreases thereafter (in its strictest form, down to zero). Using LWS data, one, indeed, finds evidence of steadily increasing net worth until the age of 65 years. However, dissaving during retirement is limited in most countries and does not occur in the US and Canada (Fig. 7.4). A hump-shaped profile is even

Canada

Household surveys 1984 and 1999

Overall trend in wealth inequality Increasing inequality

Results

Population ageing has decreased wealth inequality. Other socioeconomic factors such as permanent income and other family attributes are not important explanatory factors. Most of the rise in wealth inequality can be explained by growing inheritances, inter vivos transfers, and the evolution of relative returns on savings.

b1598-ch07

(Continued )

9in x 6in

Only those located in the two upper deciles have increased their share in total wealth. Median wealth fell in p0– p30, and rose 27% or more in p70–p100. There was an increase in the wealth Gini index of 4–8% and an increase in the coefficient of variation of 35%. Inequality among non-elderly couples with children increased the most. Older households have become relatively richer. There was a growing proportion of young couples with children with 0 or negative wealth.

Explanations

Income Inequality in the OECD. . .

Morisette et al. (2003), “The Evolution of Wealth Inequality in Canada, 1984–1999”

Time span

16:9

Country

Fredriksen

Study

October 21, 2013

196

Table 7.6: Overview of National Wealth Studies.

October 21, 2013

Table 7.6: (Continued )

Household surveys in 1988, 1994, and 1998

Hauser and Holger (2003), “Inequality of the Distribution of Personal Wealth in Germany 1973–1998”

Germany

Household surveys 1973–1998

Increasing inequality

Gini net worth was 55.1, 60.4, and 61.5 in 1987, 1994, and 1998, respectively. The dispersion of assets occurred mostly in the late 1990s, whereas that of debt occurred earlier.

Sociodemographic factors have only minor explanatory power. Increases in income and wealth inequality seem to stem from same phenomenon. One possible explanation is increased inequality of property and/or property income.

Slight decrease in inequality in West-Germany from 1973 to 1993 and an increase from 1993 to 1998

There was a decreasing inequality of net housing wealth from 1983 to 1993, then a rising trend. Inequality of net financial wealth fluctuated over the period.

It appears that inheritance and inter vivo gifts play an important role and increasing wealth inequality is expected in the future because of the following: retrenched welfare state, reduced social transfers, and larger inheritances. (Continued )

b1598-ch07

Explanations

9in x 6in

Finland

Results

Income Inequality in the OECD. . .

Jantti (2006), “Trends in the Distribution of Income and Wealth: Finland, 1987–98”

Overall trend in wealth inequality

16:9

Time span

197

Country

The Distribution of Wealth

Study

Italy

Household surveys for 1989, 1991, 1993, 1995, 1998, and 2000

Sweden

Overall trend in wealth inequality Slight decrease in inequality from 1989 to 1991, followed by an increase in inequality

Household surveys Increasing for 1984, 1986,1993, inequality 1996, and 1998

Explanations

The Gini index was 0.55, 0.54, and 0.61 in 1989, 1991, and 2000, respectively. The richest 5% had increased their share of the total by 9.1 pp. in 2000. Also older households increased their share of total wealth.

There was an increased concentration in financial wealth. Changes in demographical characteristics contributed only marginally to higher inequality. Increase in the proportion of holders of risky assets amplified the increasing trend in inequality.

There was a small increase in the net worth of the 10th percentile, a larger increase in median net worth, and a largest increase in net worth in the 90th percentile.

There was a peak of the age-earnings profile in the baby-boom generation, an increase in distrust in the social security system, an increase in private savings to compensate expected cuts in public pensions, and increase in the stock to house price ratio.

b1598-ch07

(Continued )

9in x 6in

Results

Income Inequality in the OECD. . .

Klevmarken (2004), “On the Wealth Dynamics of Swedish Families 1984–1998”

Time span

16:9

Brandolini et al. (2004), “Household Wealth Distribution in Italy in the 1990s”

Country

Fredriksen

Study

October 21, 2013

198

Table 7.6: (Continued )

October 21, 2013

Table 7.6: (Continued )

D’Ambrosio and Wolff (2001), “Is Wealth Becoming More Polarized in the US?”

United States

Kennickell (2006), “A Rolling Tide: Changes in the Distribution of

United States

Increasing inequality

During the 1980s and 1990s, real median wealth increased by ∼30%. The increase for the top decile was 50%, whereas there was no increase for the left tail of the distribution.

Household surveys (SCF) for 1983, 1989, 1992, 1995, and 1998

Steeply increasing inequality from 1983 to 1989, then increase but at a much slower pace from 1989

Household surveys (SCF) and Forbes data on the 400 wealthiest

Strongest growth in wealth at the top and bottom of the distribution. Fairly

Period 1983–1989: Wealth Gini increased from 0.80 to 0.83. Share of top percentile increased by 3.6 pp. Period 1989–1998: Wealth Gini decreased from 0.83 to 0.82. Share of top percentile increased by 0.7 pp. Forbes data: Overall mean wealth of the group was fairly flat. There was substantial growth in the

Concerns about the viability of the public pension system have increased private life-cycle savings; deregulation of financial markets; tax reforms, and demographical changes. Most of the increased wealth inequality is due to increased within-group inequality. Changes in sociodemographic factors do not explain much.

(Continued )

b1598-ch07

1990s

Explanations

9in x 6in

Sweden

Results

Income Inequality in the OECD. . .

Klevmarken (2006), “On Household Wealth Trends in Sweden over the 1990s”

Overall trend in wealth inequality

16:9

Time span

199

Country

The Distribution of Wealth

Study

Wealth in the US, 1989–2001”

Americans over the period 1989–2001

United States

Household survey (SCF) 1983, 1989, 1992, 1995, 1998, 2001, 2004, and 2007

Overall trend in wealth inequality even growth in the middle, though simple concentration measures fail to show consist patterns

wealth of the top 50 persons and substantial amount of churning. Household survey: Share of the bottom half of the distribution was approximately unchanged, and there was a decrease in the share of p50–p90 and quite large increases in the share of p90–p100. Gini wealth was 0.80, 0.83, 0.829, and 0.834 in 1983, 1989, 2004, and 2007, respectively. Share of top percentile increased from 33.8% to 34.6%, but peaked in 1998. Share of the top quintile increased over the entire period from 81.3% to 85%. There was a shift in wealth from younger to older households.

Explanations

Relative asset prices (surge in stock prices over the period). Increasing indebtedness of the middle class for the last period (2000–2004) during which there was actually a decline in asset inequality.

9in x 6in

Sharply increasing inequality from 1983 to 1989. Very modest increase from 1989 to 1998. Modest increase in inequality in the 2000s.

Results

Income Inequality in the OECD. . . b1598-ch07

Wolff (2007), “Recent Trends in Household Wealth in the US: Rising Debt and the Middle-Class Squeeze” and Wolff (2010), “Recent Trends in Household Wealth in the US: Rising Debt and the Middle-Class Squeeze — an Update to 2007”

Time span

16:9

Country

Fredriksen

Study

October 21, 2013

200

Table 7.6: (Continued )

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

The Distribution of Wealth

201

2002 US dollars Canada

Finland

Germanya

Sweden

United Kingdom

United States (SCF)

Italy

Net worth 200,000

150,000

100,000

50,000

,0 under 24

25-34

35-44

45-54

55-64

65-75

75 and over

under 24

25-34

35-44

45-54

55-64

65-75

75 and over

under 24

25-34

35-44

45-54

55-64

65-75

75 and over

Financial assets 25,000

20,000

15,000

10,000

5,000

,0

Debt 80,000

60,000

40,000

20,000

,0

Fig. 7.4: Median Wealth Holdings by Age of Household’s Head. Note: Data based on household weights. a In Germany, financial assets and non-housing debt are recorded for values of EUR 2500 or more. Source: Jantti et al. (2008).

October 21, 2013

202

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

Fredriksen

less visible for financial assets, with a continuing strong increase in financial wealth holdings in the US and Canada until a late age. On the other hand, a clear hump-shaped age profile can be seen for indebtedness, a result confirmed by national surveys. A notable exception from these general patterns is Australia. According to Bradbury (2008), Australian income and home ownership patterns in retirement are very different from other countries as income decreases sharply, whereas housing wealth remains high. He argues that at least part of the explanation for this phenomenon is that public pensions are meanstested against income and assets excluding housing. Some of the national studies find evidence that sociodemographic factors have played a role in shaping wealth inequality over time (Table 7.6) but this has mainly been due to an age effect. Changes in other sociodemographic characteristics seem to have mattered little. Almaas and Mogstad (2010) who developed a new age-adjusted Gini indicator found no difference in country rankings compared to the ordinary Gini index, suggesting that age is not an important factor behind the observed country differences in wealth inequality.

Other Determinants of Wealth Inequality Overall, the studies tend to favor macroeconomic factors to explain increasing wealth inequality over the past 30 years. Both returns on assets and the tax treatment vary across assets, and higher yielding financial assets are a much more important part of wealth holdings at the upper end of the distribution. In contrast, the share of deposit accounts, principal residences, and related mortgages are more important in the middle and lower parts of the distribution. Soaring financial markets in the aftermath of financial deregulation during the 1970s and particularly the stock market boom in the late 1990s is consistent with greater wealth concentration at the top. Wolff (2007) finds that wealth inequality is positively related to the ratio of stock prices to house prices. For Canada, Morisette et al. (2006) conclude that the rates of return on assets are an important factor in explaining wealth inequality over time. Changes in tax and transfer systems are also likely to have affected wealth accumulation and distribution. Lower top marginal tax rates and lighter capital gains taxation in many countries make the accumulation

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

The Distribution of Wealth

203

of wealth easier for the rich. Klevmarken (2006), for instance, notes that the Swedish tax reform at the beginning of the 1990s that increased the return on financial assets contributed to increased wealth inequality. However, there has also been increased targeting of transfers, especially in the English-speaking countries, which are clawed back by tax systems as income rises, while some countries also apply asset tests. Taken in isolation, this should have improved the income position of the poor, making it easier to accumulate wealth, though asset-testing public transfers will also have a direct discouraging effect on savings (Davies, 2009). An interesting feature of the German national wealth study is that it compares wealth inequality between East and West Germany and looks specifically at the effect of reunification. Thus, a “natural experiment” is exploited to look at the effect on the wealth distribution of passing from a socialist system to a market economy. At the time of reunification, the average West German household was much wealthier than the average East German household. However, surprisingly, the authors find that although income was more unequally distributed in West Germany, the reverse was true for wealth. Moreover, after reunification, wealth inequality has evolved differently, increasing in West Germany, while decreasing in East Germany. Finally, wealth accumulation and distribution depends on inheritance patterns. On average, over the period between 1989 and 2007, 20.9% of American households received a wealth transfer and all wealth transfers received made up 23.2% of net worth (Wolff and Gittleman, 2011). These results are close to a number of other studies for the US and also an estimated share of inherited wealth in Sweden in 1998 of 19% (Klevmarken, 2006).3 Inheritance appears to play a bigger role in wealth accumulation in France where inherited wealth was estimated at 35% of total wealth in 1975 (Kessler and Masson, 1989). Using French tax data assembled since the beginning of the 18th century, Piketty (2011) found that annual inheritance flows in France were around 15% of national income in 2010. Wealth transfers are very concentrated, and, at least in the case of the US, even more so than total wealth. In 1998, the Gini index for wealth transfers was 0.96 (Wolff and Gittleman, 2011). Using the French Mutations a titre gratuit surveys conducted in 1987 and 1994, Arrondel and Laferrere ` (2001) found that 42% of bequests were made by the top decile and 13% by the top centile. Adding inter vivos gifts made until 10 years before death increases the concentration of inheritance even further. 3 The

same study found that 35% of households received a wealth transfer in 1998.

October 21, 2013

204

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

Fredriksen

Comparing the socioeconomic characteristics of heirs versus non-heirs confirms that not only are wealth transfers concentrated on a few lucky recipients but those who benefit are individuals already favored in life. For the US, Wolff and Gittleman (2011) found an increasing share of recipients as one moves up both in the income and wealth distribution. The likelihood of receiving a bequest also increases with education and for those with a white ethnic origin.

References Almaas, I. and M. Mongstad (2010). “Older or Wealthier? The Impact of AgeAdjustment on Cross-Sectional Inequality Measures,” Norwegian School of Economics and Business Administration Department of Economics Discussion Paper Series. Arrondel, L. and A. Laferrere (2001). “Taxation and Wealth Transmission in France,” Journal of Public Economics, Vol. 79(1), pp. 3–33. Atkinson, A.B., J.P.F. Gordon and A. Harrison (1989). “Trends in the Shares of Top Wealthholders in Britain: 1923–1989,” Oxford Bulletin of Economics and Statistics, Vol. 51(3), pp. 315–332. Bradbury, B. (2008). “Housing Wealth as Retirement Saving: Does the Australian Model Lead to Over-Consumption of Housing?” LWS Working Paper Series No. 7. Brandolini, A., L. Cannari, G. D’Alessio and I. Faiella (2004). “Household Wealth Distribution in Italy in the 1990s,” Levy Economics Institute Working Paper No. 414. D’Ambrosio, C. and E. Wolff (2001). “Is Wealth Becoming More Polarized in the United States?” Levy Institute Working Paper No. 330. Davies, J.B. (2009). “Wealth and inequality,” In: Salverda, W., Nolan, B. and Smeeding, T. (eds.), The Oxford Handbook of Economic Inequality, Oxford: Oxford University Press. Dell, F., T. Piketty and E. Saez (2005). “Income and Wealth Concentration in Switzerland over the 20th Century,” CEPR Discussion Paper No. 5090. Domeij, D. and P. Klein (2002). “Public Pensions: To What Extent Do They Account for Swedish Wealth Inequality?” Review of Economic Dynamics, Vol. 4(1), pp. 503–534. Hauser, R. and S. Holger (2003). “Inequality of the Distribution of Personal Wealth in Germany 1973–1998,” Levy Economics Institute Working Paper No. 398. Jantti, M. (2006). “Trends in the distribution of income and wealth: Finland, 1987–98,” In: Wolff, E. (ed.), International Perspectives on Household Wealth, Cheltenham, UK: Edward Elgar Publishing.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

The Distribution of Wealth

b1598-ch07

205

Jantti, M., E. Sierminska and T. Smeeding (2008). “The Joint Distribution of Household Income and Wealth: Evidence from the Wealth Study,” OECD Social, Employment and Migration Working Paper No. 65, OECD Publishing. Kennickell, A. (2006). “A Rolling Tide: Changes in the Distribution of Wealth in the US, 1989–2001,” In: Wolff, E. (ed.), International Perspectives on Household Wealth, Cheltenham, UK: Edward Elgar Publishing. Kessler, D. and A. Masson (1989). “Bequest and Wealth Accumulation: Are some Pieces in the Puzzle Missing?” Journal of Economic Perspectives, Vol. 3(3), pp. 141–152. Klevmarken, N.A. (2006). “On Household Wealth Trends in Sweden over the 1990s,” Department of Economics, In: Wolft, E. (ed.), International Perspectives on Household Wealth, Cheltenham, UK: Edward Elgar Publishing. Klevmarken, N.A. (2004). “On the Wealth Dynamics of Swedish Families 1984– 1998,” Review of Income and Wealth, Vol. 50(4), pp. 469–491. Kopczuk, W. and E. Saez (2004). “Top Wealth Shares in the United States, 1916– 2000: Evidence from Estate Tax Returns,” National Tax Journal, Vol. 57(2), pp. 445–487. Lerman, R.I. and S. Yitzhaki (1985). “Income Inequality Effects by Income Source: A New Approach and Applications to the United States,” The Review of Economics and Statistics, Vol. 67(1), pp. 51–56. Levine, L. (2012). “An Analysis of the Distribution of Wealth across Households,” Congressional Research Service, 7-5700. Morisette, R., X. Zhang and M. Drolet (2003). “The Evolution of Wealth Inequality in Canada, 1984–1999,” Levy Economics Institute Working Paper No. 369. OECD (2008). Growing Unequal? Income Distribution and Poverty in OECD Countries, OECD Publishing, Paris. Ohlsson, H., J. Roine and D. Waldenstr¨ om (2007). “Long-run Changes in the Concentration of Wealth: An Overview of Recent Findings,” IFN Working Paper No. 699. Piketty, T. (2011). “On the Long-Run Evolution of Inheritance: France 1820– 2050,” Quarterly Journal of Economics, Vol. 126(3), pp. 1071–1131. Piketty, T., G. Postel-Vinay and J-L. Rosenthal (2006). “Wealth Concentration in a Developing Economy: Paris and France, 1807–1994,” American Economic Review, Vol. 96(1), pp. 236–256. Wolff, E. (2007). “Recent Trends in Household Wealth in the United States: Rising Debt and the Middle-Class Squeeze,” Levy Economics Institute Working Paper No. 502. Wolff, E. (2010). “Recent Trends in Household Wealth in the United States: Rising Debt and the Middle-Class Squeeze — An Update to 2007,” Levy Economics Institute Working Paper No. 589.

October 21, 2013

206

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch07

Fredriksen

Wolff, E. (2011). “Pensions in the 2000s: The Lost Decade?” NBER Working Paper No. 16991. Wolff, E. and M. Gittleman (2011). “Inheritances and the Distribution of Wealth or Whatever Happened to the Great Inheritance Boom?” ECB Working Paper Series No. 1300.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch08

8. CONCLUSION: GROWTH-ENHANCING POLICIES AND INEQUALITY Isabelle Joumard and Isabell Koske

Introduction and Main Findings The first chapter argued that it is difficult to find a clear link between growth and inequality. However, a link between growth-enhancing policies and inequality can be established. The OECD analyzes the link between structural policies and GDP per capita and translates the results into concrete policy recommendations (see in particular the annual OECD publication Going for Growth). At the same time, most structural reforms also influence income inequality. Against this background, this chapter assesses possible policy trade-offs and complementarities between growth and income distribution objectives. In doing so, growth-enhancing policy reforms are classified into three categories: (i) those that are likely to reduce income inequality; (ii) those that are likely to raise it; and (iii) those that seem to have an ambiguous effect. Caution is needed in interpreting the results. First, they may not reflect the overall (general equilibrium) effects of policy changes. This may, for example, be the case whenever the conclusions are based on the quantile regression framework. Moreover, the question of how policy reforms are financed is ignored, meaning that the effects shown here are only partial. Second, the uncertainties that surround the estimated effects are not taken into account. Third, no comparison is made of the magnitudes of the effects on GDP per capita and income inequality. Fourth, it abstracts from possible additional direct interactions between economic growth and inequality. While these interactions have attracted great attention in the

207

October 21, 2013

208

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch08

Joumard and Koske

economic debate in recent years, the theoretical and the empirical literature remain by and large inconclusive regarding both causality and the direction of influence (see Ehrhart (2009) and Fields (2001), for surveys of the vast literature). The second section of this chapter reviews the impact of growthenhancing structural policies on labor income inequality. The third section looks at the effect of different taxes on inequality and growth, while arguing that the impact of transfer policies on growth depends on their design.

Main Findings • Many policies entail a double dividend as they reduce income inequality, while at the same time boosting long-run GDP per capita. Examples include facilitating the accumulation of human capital, making educational potential less dependent on personal and social circumstances, reducing labor market dualism or promoting the integration of immigrants and fostering female labor market participation. Concerning taxation, reducing tax expenditures which mostly benefit the well-off, for instance, for investing in housing, contributes to equity objectives, while also allowing a growth-friendly cut in marginal tax rates. • By contrast, several policies may entail a trade-off between reducing income inequality and raising GDP per capita. For instance, administrative extensions of collective wage agreements may reduce the wage earnings dispersion among workers, but if they set labor costs at toohigh levels for some employers they may reduce employment. Shifting the tax mix to less-distorting taxes — in particular away from labor and corporate income taxes toward consumption and real estate taxes — would improve incentives to work, save, and invest, but could undermine equity. Cash transfers targeted to lower incomes can be used to ease this trade-off. • Finally, some policies aimed at boosting GDP per capita have an uncertain impact on income inequality. For instance, avoiding too-high and long-lasting unemployment benefits may raise employment over the long run but also widen the distribution of income among the working-age population, with an ambiguous net effect on inequality. The same holds as regards keeping minimum wages at moderate levels.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch08

Conclusion: Growth-Enhancing Policies and Inequality

209

Reducing Labor Income Inequality and Boosting GDP per Capita: Policy Trade-offs and Complementarities As discussed in Chapter 3, structural policies in the areas of education, labor, and product markets influence labor income inequality by affecting the employment rate and the dispersion of earnings among those that have a job. At the same time, these policies can have an impact on GDP per capita by affecting labor utilization, capital accumulation, and productivity. While some reforms boost GDP per capita and lower the labor income inequality, others have conflicting effects on the two policy goals. Table 8.1 summarizes the main findings and shows which reforms are win–win strategies and which ones imply trade-offs.

Table 8.1: The Impact of Structural Policies on Labor Income Equality.

Employment rate

Earnings equality

Total labor income equality

GDP per capita

Initiatives to increase the tertiary graduation rate Initiatives to increase the upper-secondary graduation rate

∼

+

(+)

+

∼

+

(+)

+

Initiatives to promote equity in education

∼

+

(+)

+

The minimum wage (as share of the median wage)

0/−

+

∼

(0/−)

Unionization Legal extensions of collective wage agreements The overall level of EPL

∼ −

+ ∼

+ (−)

(∼) (−)

0/−

+

∼

−

−

−

(−)

−

−

+

∼

−

0/+

∼

+

+

A rise in:

The gap between EPL on regular versus temporary work Replacement rate and duration of unemployment benefits Spending on ALMPs

(Continued)

October 21, 2013

210

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch08

Joumard and Koske Table 8.1: (Continued)

Employment rate

Earnings equality

Total labor income equality

GDP per capita

Anticompetitive product market regulation Initiatives to foster the integration of immigrants

−

0/+

∼

−

+

+

(+)

(+)

Initiatives to combat discrimination in the labor market Initiatives to raise female labor force participation Initiatives to avoid gender stereotyping in education

+

+

(+)

(+)

+

+

(+)

(+)

∼

+

(+)

(∼)

A rise in:

Note: The term “earnings inequality” refers to inequality among the working population and the term “total labor income inequality” refers to inequality among the workingage population, thus accounting for both employment and earnings inequality effects. +, −, and 0 denote, respectively, a significant rise, a significant fall, and no impact on the variable of interest. In cases where some studies find a significant effect, while other do not, this is indicated by combining the symbols; for example, 0/+ means that some studies cannot find a significant effect while others find a positive effect. A tilde (∼) means that the sign of the effect is unknown because the empirical literature is inconclusive or because studies on the link are not available. When the sign of the total labor income inequality effect is unknown but can be deducted from the signs of the employment and earnings equality effects, the results are reported in brackets. The GDP per capita effects are based on the findings of previous OECD and other studies (reported, e.g. in Barnes et al. (2011) and Bouis and Duval (2011)) or deducted from the employment rate effect (in which case the GDP per capita effect is written in brackets).

Growth-enhancing Policy Reforms that are Likely to Reduce Income Inequality Improving the quality and quantity of education Reforms that facilitate the accumulation of human capital are among the most important for improving long-run living standards. Examples of reforms include inter alia enhanced accountability and autonomy for both secondary and tertiary education institutions and better teacher recruitment and training. Although it can take up to a generation until all the GDP

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

Conclusion: Growth-Enhancing Policies and Inequality

b1598-ch08

211

per capita gains from such reforms have been realized, small improvements in the skills of a nation’s labor force can have a large effect on future GDP per capita (OECD, 2010). Many of the reforms are likely to have additional benefits in terms of a more equitable distribution of labor income and at least some of them could be achieved without raising government spending.1 Since an upper secondary degree is today generally a prerequisite for being successful in the labor market in most OECD countries, reducing the number of pupils that leave school without a lower-secondary degree (e.g. by providing support to pupils at risk to reduce drop outs) should help reduce income inequality. Similar benefits can be expected from reforms that encourage more students to pursue tertiary studies. Although a rise in the number of tertiary graduates raises inequality via a pure composition effect according to the quantile regressions (at least until highly educated workers reach a certain share of the workforce), there is some tentative evidence from the cross-country time-series regressions that this effect is more than offset by a decline in the relative returns to tertiary education. Still, raising tuition fees that make students share at least part of the cost of tertiary education could lower inequality as the current financing of education is regressive, provided the poor are not excluded from tertiary education — for example, by combining tuition fees with student loans whose repayment is contingent on income.2

Promoting equity in education Making educational outcomes less dependent on personal and social circumstances should boost GDP per capita by enhancing entrepreneurship, work incentives, the overall quality and allocation of human capital, and, ultimately, productivity. At the same time, a more equitable distribution of educational opportunities has been shown to entail a more equitable distribution of labor income. Examples of policy initiatives that have been shown to raise equity in education include postponing early tracking (e.g. Bauer and Riphahn, 2005; Hanushek and W¨ oßmann, 2005) and strengthening links between school and home to help disadvantaged children to learn

1 Sutherland

et al. (2007) show that public spending efficiency in primary and secondary educations can be raised considerably in many OECD countries. 2 Empirical evidence suggests that any negative effect of tuition fees on participation rates can be fully offset by raising the financial support for low-income students (Heller, 1999; Santiago et al., 2008).

October 21, 2013

212

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch08

Joumard and Koske

(e.g. Mullis et al., 2003). Providing early childhood care and basic schooling for all is also key as it may yield large positive returns over the entire lifetime, particularly for the most disadvantaged (Chetty et al., 2011; Leseman, 2002; McCabe and Smyth, 2000; OECD, 2006a). Similarly, carefully managing school choice to prevent that pupils with weaker parental support are at a disadvantage, channeling resources to students and regions with the greatest needs, removing dead ends in upper secondary education, and offering a second chance to those who fail at school, would also promote equity in education. (See OECD, 2007, for detailed policy recommendations on how to promote equity in education.)

Reducing the gap between employment protection on temporary and permanent work A smaller gap in job protection between temporary and permanent contracts lowers income inequality by improving the labor market prospects of those at the margin of the labor market (e.g. the youth) both in terms of employment and wages. It is also likely to reduce the income gap between immigrants and natives, as immigrants also suffer disproportionately from labor market dualism (Causa and Jean, 2007). The associated human capital accumulation may spur productivity improvements which, in turn, lead to higher GDP per capita.

Increasing the spending on ALMPs As ALMPs are often accompanied by a strong emphasis on activation, they may limit the adverse effects of high social benefits on work incentives and employment, thereby contributing to higher GDP per capita and, abstracting from their financing, lower labor income inequality. The same effects may arise from ALMP-driven improvements in job matching — to the extent that they benefit disproportionately marginal labor market groups — and skills.

Promoting the integration of immigrants Targeted policies such as language courses and transparent systems of recognizing foreign qualifications should help reduce labor income inequality by closing the gap between immigrants and natives’ labor market performance. Better labor market integration of immigrants should also raise GDP per

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

Conclusion: Growth-Enhancing Policies and Inequality

b1598-ch08

213

capita as it raises labor force participation and improves human capital (and, thereby, productivity).

Fostering female labor market participation Since women tend to take on more caring responsibilities than men, they work fewer hours and thus take home less pay. Ideally, the higher labor supply elasticity of women should be dealt with by taxing them at a lower rate than men. Policies to improve the availability of formal care for children and the elderly can serve as a second best solution. Such policies should help to reduce gender differences in working hours and pay, and at the same time improve long-run living standards through higher participation rates. Such policies might further contribute to higher earnings equality by influencing women’s career choices.3

Avoiding gender stereotyping in education To the extent that differences in occupational choice do not reflect only on personal preferences but also gender stereotyping, this need to be addressed. This should reduce inequality between men and women with a positive effect on GDP per capita through better allocation of resources.

Fighting discrimination The earnings gap between men and women and between immigrants and natives is partly due to discrimination as shown in Chapter 3. Discrimination is likely to entail negative consequences for long-run living standards as it reduces work incentives and leads to a suboptimal allocation of human capital.4 To combat discrimination, legal rules can be made more effective, for example, by empowering well-resourced specialized bodies to investigate employers even in the absence of individual complaints and to take legal action against those who engage in discriminatory practices (OECD, 2011).

3 As

women traditionally take career breaks for child rearing reasons, they may choose occupations with flatter earnings profiles (Polachek, 1981), or trade lower earnings for job characteristics that improve their family life such as shorter commuting or flexible hours (Killingsworth, 1987). 4 Livanos et al. (2009) provide some evidence that discrimination contributes to lower employment rates of women relative to those of men.

October 21, 2013

214

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch08

Joumard and Koske

Growth-enhancing Policy Reforms that are Likely to Raise Income Inequality Increasing the flexibility of wage determination Administrative extensions of collective wage agreements to firms that are not party to the original settlement may set labor costs at too-high levels for some employers, thus hampering productivity through lower competitive pressures from the entry of new firms and possibly reducing employment. At the same time, the less dispersed wages among union members indicate that such extensions may be associated with lower earnings inequality.

Growth-enhancing Policy Reforms that have an Uncertain Impact on Income Inequality Avoiding too high and long-lasting unemployment benefits If unemployment benefits are too high or long-lasting, they risk reducing job-search incentives and raising wages above market clearing levels, thereby lowering employment with negative effects on GDP per capita (OECD, 2006b). These labor market effects have implications for income inequality: the lower employment rate is likely to be associated with higher income inequality, whereas the more compressed wage distribution — to the extent that unemployment benefits are progressive or lower-income workers are more likely to receive them — has the opposite effect. While in the short run, these effects on inequality are likely to be dominated by the direct inequality-reducing effect that stems from the income support for the unemployed (Chapter 4), the direction of the overall effect is less clear-cut in the long run.

Liberalizing product markets A wide range of industry and country-level studies illustrate the large beneficial effects of product market liberalization on the pace of productivity convergence to technologically advanced countries (e.g. Bourl`es et al., 2010; Conway et al., 2006). At the same time, the impact of product market liberalization on labor income inequality is uncertain: while higher

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

Conclusion: Growth-Enhancing Policies and Inequality

b1598-ch08

215

employment should reduce inequality, ceteris paribus, this may potentially be offset by a wider earnings dispersion.

Lowering minimum labor costs If minimum wages are set too high, they risk limiting the job market opportunities of young workers and the low skilled workers. Under such circumstances, lowering relative labor costs may boost employment of marginal groups in the labor market. Higher employment raises GDP per capita and reduces labor income inequality, but a wider wage distribution raises it, ceteris paribus, resulting in an ambiguous effect. The employment effect is likely to be the smaller the lower the initial level of minimum labor costs, in which case the likelihood that cuts result in higher labor income inequality will be greater.

The Inequality and Growth Nexus Associated with Taxes and Transfers Taxes and public spending affect both GDP per capita and the distribution of income. The redistributive implications of various taxes have been discussed in Chapter 4. However, the tax system — in particular the tax mix and the design of taxes — can also have an impact on GDP per capita by affecting labor utilization and labor productivity or both (Johansson et al., 2008). Similarly, public spending programs also affect income levels. Some tax reform options appear to be win–win options — improving growth prospects and narrowing the distribution of income. Most, however, may imply trade-offs between these objectives. Table 8.2 summarizes the main findings.

Growth-enhancing Tax Policy Reforms that are Likely to Reduce Income Inequality Re-assessing those tax expenditures that benefit mainly high-income groups Cutting back tax expenditures (e.g. tax reliefs on mortgage interest) is likely to be beneficial both for long-term GDP per capita, allowing a reduction

October 21, 2013

216

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch08

Joumard and Koske

Table 8.2: The Inequality and Growth Nexus Associated with Taxes. Tax policies Increasing total tax revenues

Income equality ∼, + The impact of taxes on income distribution depends on the level of taxation, the tax mix, and the use of tax revenues. If tax systems are progressive overall, equality is enhanced.

GDP per capita − Taxes dampen incentives to work, save, and invest and are thus detrimental to growth. But some taxes have a less adverse effect than others.

Changing the tax mix while keeping total tax revenues constant Moving from the personal income tax to consumption taxes

−

The personal income tax tends to be progressive, while the consumption tax is regressive. Moving from social security contributions to consumption taxes

− (∼ in the long-run)

Those not employed (e.g. pensioners and unemployed) would suffer most from a decline in their living standards.

+

The personal income tax reduces work and saving incentives. A shift from direct to indirect taxes would raise output. +

The cost of labor would decline and/or incentives to work would increase, raising the growth potential. In the long run, this may help to reduce income dispersion. (Continued)

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch08

Conclusion: Growth-Enhancing Policies and Inequality

217

Table 8.2: (Continued) Tax policies Moving from labor income to property and capital taxes ∗ to wealth, inheritance, and capital income taxes, such as capital gains taxes

Income equality ∼

+

Wealth and inheritance taxes tend to be progressive. The distributive impact depends on the relative progressivity of income versus wealth and inheritance taxes. −

∗ to

GDP per capita

real estate taxes

Real estate taxes are often less progressive than the personal income tax and can even be regressive.

Cutting tax expenditures and marginal rates

+ in most cases (− for in-work tax credits) Most tax expenditures benefit high-income groups (in-work tax credits, such as EITC schemes and other tax expenditures targeted on low-income groups, are the exception). Cutting tax expenditure would narrow the distribution of disposable income.

+ Property taxes are among the least harmful for growth. Moving from income to property taxes tends to improve incentives to work and invest, and thus raise output, at least in the short and medium term. + Cutting marginal rates improves incentives to work, save, and invest, and thus lifts GDP per capita.

Increasing the progressivity of taxes (revenue-neutral) Increasing the progressivity of the personal income tax

+

−

(Continued)

October 21, 2013

218

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch08

Joumard and Koske Table 8.2: (Continued)

Tax policies ∗ ∗

increase in top rates, combined with: expansion of in-work tax credits or tax free allowances

Income equality

GDP per capita

8 >

=

> :

+

+

In-work tax credits narrow the income distribution.

> ;

In-work tax credits raise incentives to work, while higher top rates may reduce working hours, as well as productivity by undermining incentives to learn, invest, and innovate. The GDP per capita impact is thus indeterminate.

Note: Concerning equality effects, a plus sign indicates more equality and a minus sign indicates less equality (i.e. more inequality). For both the equality and growth effect, a tilde indicates an ambiguous effect.

in marginal tax rates, and for a more equitable distribution of income as tax expenditures disproportionally benefit high-income earners. Cutting tax expenditure would also reduce the complexity of the tax system, and thus tax compliance and collection costs.

Reducing distortions in taxing capital income Tax reliefs granted to specific savings instruments — such as reduced taxation of capital gains resulting from the sale of a principal or secondary residence — should be reconsidered since they often distort resource allocation, without boosting total savings and growth, and benefit mainly high-income groups. Specific tax relief may also provide tax avoidance instruments for the top-income earners. In particular, there is little justification for tax breaks for stock options and carried interest. Raising such taxes would cater to equity objectives and allow a cut in marginal labor income tax rates.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

Conclusion: Growth-Enhancing Policies and Inequality

b1598-ch08

219

Growth-enhancing Tax Policy Reforms that are Likely to Raise Income Inequality Shifting the tax mix from personal and corporate income taxes toward real estate and consumption taxes Empirical evidence suggests that recurrent taxes on immovable property are among the least distortive taxes, followed by consumption taxes (Johansson et al., 2008). Personal and corporate income taxes as well as social security contributions are the most distortive as they have sizable adverse effects on labor utilization, productivity, and capital accumulation. Shifting the tax mix away from such taxes should thus raise living standards. However, there is likely to be an important trade-off with the income distribution objective since personal income taxes are progressive, while real estate and consumption taxes tend to be regressive. Targeted transfers can counteract the resulting increase in the regressivity of the tax system and reduce the severity of the trade-off.

Growth-enhancing Tax Policy Reforms that have an Uncertain Impact on Income Inequality Moving from income to wealth or inheritance tax Personal income taxes, wealth and inheritance taxes all tend to be progressive. The distributional impact of a move from income to wealth or inheritance taxes would thus depend on the relative progressivity of each tax but may be broadly neutral. The impact on growth should, however, be positive since such property taxes are among the least distortive taxes.

Moving from social security contributions (or flat labor income taxes) to consumption taxes Lowering the tax burden on labor could lead to increases in both labor supply and demand, boosting economic growth. The increase in consumption taxes would fall on wage earners but also on benefit and capital income recipients. If benefit indexation is incomplete, benefit recipients may suffer

October 21, 2013

220

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch08

Joumard and Koske

a loss of real income and the rise in consumption taxes may have undesirable distributional effects. Incentives to work would increase, however, thus reducing unemployment traps and the dispersion of labor income. Wages would also likely reflect, at least partly, the increase in consumption taxes, thus affecting employment.

The Impact of Transfer Policies on Growth Depends Very Much on Their Design In most OECD countries, household transfers play the most important redistributive role (Chapter 4). Transfers are usually progressive, although their degree of progressivity depends on their design features, for example, on the relative portion of flat versus income-related benefits. Effects on GDP per capita also depend on the design of transfer systems as well as flanking measures. While many transfers may undermine work incentives with adverse effects on hours worked and income levels, this need not be the case if they are properly designed or accompanied by offsetting measures. For instance, high, but degressive, unemployment benefits may have only limited adverse effects on work incentives when a coherent activation strategy is in place. Likewise, high old-age pension benefit replacement rates may not affect labor force participation of older workers much, if the system is actuarially neutral, as may disability benefit schemes that involve strict eligibility criteria and regular reassessments of work capacity. Targeted transfers are likely to have adverse incentive effects during the withdrawal phase, but these can be mitigated by avoiding threshold effects and the associated spikes in marginal effective tax rates. Universal benefits are likely to have comparatively lower incentive effects, but a higher tax take — which itself entails economic distortions — is needed to finance them.

References Barnes, S. et al. (2011). “The GDP Impact of Structural Reform: A Simple Simulation Framework,” OECD Economics Department Working Paper No. 834, OECD Publishing. Bauer, P. and R. Riphahn (2005). “Timing of School Tracking as a Determinant of Intergenerational Transmission of Education,” Economics Letters, Vol. 91, pp. 90–97. Bouis, R. and R. Duval (2011). “Raising Potential Growth After the Crisis: A Quantitative Assessment of the Potential Gains from Various Structural

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

Conclusion: Growth-Enhancing Policies and Inequality

b1598-ch08

221

Reforms in the OECD Area and Beyond,” OECD Economics Department Working Paper No. 835, OECD Publishing. Bourl`es, R. et al. (2010). “Do Product Market Regulations in Upstream Sectors Curb Productivity Growth: Panel Data Evidence for OECD Countries,” OECD Economics Department Working Paper No. 791, OECD Publishing. Causa, O. and S. Jean (2007). “Integration of Immigrants in OECD Countries: Do Policies Matter?” OECD Economics Department Working Paper No. 564, OECD Publishing. Chetty, R. et al. (2011). “How Does Your Kindergarten Classroom Affect Your Earnings? Evidence from Project Star,” Quarterly Journal of Economics, Vol. 126(4), pp. 1593–1660. Conway, P. et al. (2006). “Regulation, Competition and Productivity Convergence,” OECD Economics Department Working Papers No. 509. Ehrhart, C. (2009). “The Effects of Inequality on Growth: A Survey of the Theoretical and Empirical Literature,” ECINEQ Working Paper Series No. 107, February. Fields G. (2001). Distribution and Development, A New Look at the Developing World, Russel Sage Foundation, MIT Press, New York. Hanushek, E.A. and L. W¨ oßmann (2005). “Does Educational Tracking Affect Performance and Inequality? Differences-In-Differences Evidence Across Countries,” NBER Working Papers No. 11124, National Bureau of Economic Research. Heller, D.E. (1999). “The Effects of Tuition and State Financial Aid on Public College Enrolment,” Review of Higher Education, Vol. 23(1), pp. 65–89. Johansson, ˚ A. et al. (2008). “Taxation and Economic Growth,” OECD Economics Department Working Paper No. 620, OECD Publishing. Killingsworth, M. (1987). “Heterogeneous Preferences, Compensating Wage Differentials and Comparable Worth,” Quarterly Journal of Economics, Vol. 102, pp. 727–742. Leseman, P.P.M. (2002). “Early Childhood Education and Care for Children from Low-Income or Minority Backgrounds.” Paper for discussion at the OECD Oslo Workshop, June 6–7, 2002, Oslo. Livanos, I., C ¸ . Yalkin and I. Nu˜ nez, (2009). “Gender Employment Discrimination: Greece and the United Kingdom,” International Journal of Manpower, Vol. 30, pp. 815–834. McCabe, B. and E. Smyth (2000). “The educational situation of disadvantaged children,” In: Nicaise, I. (ed.), The Right to Learn: Educational Strategies for Socially Excluded Youth in Europe, The Policy Press. Mullis, I. et al. (2003). “PIRLS 2001 International Report: IEA’s Study of Reading Literacy Achievement in Primary Schools in 35 Countries,” International Study Center, Lynch School of Education, Boston College. OECD (2006a). Starting Strong II: Early Childhood Education and Care, OECD Publishing, Paris.

October 21, 2013

222

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-ch08

Joumard and Koske

OECD (2006b). OECD Employment Outlook 2006: Boosting Jobs and Incomes, OECD Publishing, Paris. OECD (2007). No More Failures: Ten Steps to Equity in Education, OECD Publishing, Paris. OECD (2010). The High Cost of Low Educational Performance: The Long-run Economic Impact of Improving PISA Outcomes, OECD Publishing, Paris. OECD (2011). Report on the Gender Initiative: Gender Equality in Education, Employment and Entrepreneurship, meeting of the OECD Council at Ministerial Level, May 25–26, 2011, Paris. Polachek, S. (1981). “Occupational Self-selection: A Human Capital Approach to Sex Differences in Occupational Structure,” Review of Economics and Statistics, Vol. 63(1), pp. 60–69. Santiago, P. et al. (2008). Tertiary Education for the Knowledge Society. Vol. 1: Special Features: Governance, Funding, Quality, OECD Publishing, Paris. Sutherland, D., R. Price, I. Joumard and C. Nicq (2007). “Performance Indicators for Public Spending Efficiency in Primary and Secondary Education,” OECD Economics Department Working Paper No. 546, OECD Publishing.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-abt

ABOUT THE AUTHORS

Peter Hoeller is Head of the Public Economics Division in the Economics Department of the OECD. In this position, he was supervising and contributing to the OECD project on income distribution and growth-enhancing policies. Since joining the OECD in 1982, he has contributed to the surveillance of many OECD countries as an economist and later as a Head of Division. His recent research focus has been on health-care systems, counter-cyclical economic policies and fiscal consolidation. Mr. Hoeller studied economics and holds a PhD from the University of Linz and was, for several years, a lecturer and researcher at this university. Prior to joining the OECD, he worked for two years at a commercial bank in Vienna.

Isabelle Joumard is a Senior Economist in the OECD Economics Department. She graduated from the University of Paris I, la Sorbonne. She has contributed to the 2011–2012 OECD project on income distribution and growth. She has also conducted various OECD studies in the following fields: public spending reforms; public spending efficiency in the primary and lower secondary education sector; fiscal decentralization; tax reform; indicators of structural fiscal balances; and fiscal rules. She also led the OECD project on health-care policies and spending efficiency and was the main author of several working papers and of the monograph Health Care Systems — Efficiency and Policy Settings, published in 2010. She has worked on many OECD countries, including Finland, Greece, Japan, 223

October 21, 2013

224

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-abt

About the Authors

Mexico, Norway, Spain and Switzerland, and she is now in charge of the OECD Economic Surveys for Colombia, India and Slovenia. Isabell Koske is a Senior Economist in the Structural Surveillance Division of the OECD Economics Department. In this position, she contributed to the 2011–2012 OECD study on income distribution and growth. Since she joined the OECD in 2006, she has also worked on different research projects in the areas of globalization, inflation, potential output and current accounts and has contributed to the economic surveillance of Germany, Slovakia and Slovenia. Moreover, she is responsible for the construction of structural policy indicators feeding into the OECD’s Going for Growth publication, in particular the Product Market Regulation indicators. Ms. Koske studied Business Administration and Applied Mathematics and holds a PhD in Economics from WHU — Otto Beisheim School of Management in Germany.

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-index

INDEX

Country groups and characteristics

Inequality

Education policy, 50 Female labor market participation, 213 Fighting discrimination, 213 Flexibility of wage determination, 214 Gap in protection between temporary and permanent labor contracts, 57, 212 Gender inequality, 62 Globalization, 48 Household labor earnings, 16, 20 Household market income, 16 Individual labor earnings, 15, 16 Integration of immigrants, 212 Intergenerational social mobility, 68 Labor income inequality, 45 Labor market institutions, 54 Labor market policies and globalization, 59 Migration, 64, 176 Minimum wages, 54, 158 Offshoring, 49 PISA, 68 Product market regulation, 61 Skill-biased technological change, 47 Quantile regressions, 51 Quantity and quality of education, 210 Tax policy, 61 Trade-induced innovation, 49 Wage bargaining, 55

And growth, 2, 215 And welfare, 5

Labor market policy

Labor income inequality

Active labor market policy, 59, 212 Disability benefits, 101 Earned income tax credit, 157

Cluster analysis, 26, 30, 129 Country groups sharing similar inequality patterns, 30 Country profiles of inequality patterns, 26 Inequality and poverty in large emerging economies, 32 Policy indicators for labor income inequality, 66 Policy indicators for tax and cash transfer systems, 125 Gender issues Gender inequality, 62 Gender poverty gap, 62 Gender stereotyping, 213 Female labor market participation, 213 Fighting discrimination, 213 Education Distribution of educational opportunities, 211 Education policy, 50 Gender stereotyping, 213 PISA, 68 Quality and quantity of education, 210

Active labor market policy, 59, 212 225

October 21, 2013

226

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-index

Index

Female labor market participation, 213 Flexibility of wage determination, 214 Gap in protection between temporary and permanent labor contracts, 212 In-work benefits, 157 Labor market institutions, 54 Labor market policies and globalization, 59 Minimum labor costs, 215 Minimum wages, 54, 158 Policy indicators for labor income inequality, 66 Unemployment benefits, 58, 214 Measuring inequality Adjusted household disposable income, 16 Concentration coefficient for cash transfers and taxes, 90 Concentration of wealth, 181 Data sources, 18 Distribution of household disposable income, 22 Distribution of wealth, 181, 182 Global inequality, 35 Household disposable income, 16 Household market income, 16 Individual labor earnings, 15, 16 International comparisons and poverty trends, 138 Kakwani index, 91 Policy indicators for labor income inequality, 66 Policy indicators for tax and cash transfer systems, 125 Quantile regressions, 51 Total market income, 20 Policy Policy indicators for labor income inequality, 66 Policy indicators for tax and cash transfer systems, 125 Trade-offs and complementarities, 207

Product market policy Product market regulation, 61, 214 Poverty Absolute and relative poverty, 140 Among children, 143 Among households with children, 147 Among jobless households, 149 Among the retired, 143 Among working-age people, 142 Gender poverty gap, 62 In-kind transfers, 159 International comparisons and poverty trends, 138 Minimum wages, 158 In-work benefits, 157 Policies to reduce poverty among children, 158 Poverty in large emerging economies, 32 Targeting of cashtransfers, 154 Taxes and transfers, 150 Work strategies, 156 Taxes and transfers Beveridgean-type welfare states, 88 Bismarckian-type welfare states, 88 Concentration coefficient for cash transfers and taxes, 90 Consumption taxes, 121, 219 Disability benefits, 101 Distortions in taxing capital income, 218 EITC, 157 Family cash benefits, 103 Incidence of taxes and transfers, 123 In-kind transfers, 24, 159 In-work benefits, 157 Labor income taxes, 108 Progressivity of cash transfers and taxes, 89 Property taxes, 116 Real estate taxes, 117 Redistributive impact of cash transfers, 95

October 21, 2013

16:9

Income Inequality in the OECD. . .

9in x 6in

b1598-index

Index Redistributive impact of pension systems, 98 Redistributive impact of taxes, 95, 105 Policy indicators for tax and cash transfer systems, 125 Tax avoidance, 170 Tax burden on labor, 219 Tax expenditures, 114, 215 Taxes and transfers, 150, 215 Tax mix, 219 Taxation of property, wealth, and inheritance, 120, 170, 219 Taxes on capital income, 115 Tax policy, 61 Transfer policies and growth, 220 Unemployment benefits, 58, 102, 214 Top incomes Capital gains on shareholdings, 170 Carried interest arrangements, 170 Corporate governance and executive compensation, 177 Distortions in taxing capital income, 218 Migration, 176 Recent trends in top income shares, 168 Remuneration issues, 176 Technical progress and globalization, 176

227

Tax avoidance, 170 Taxation of property, wealth, and inheritance, 170 Top income developments, 163 Top income recipients, 167 Top income tax rates, 164, 168 Top part of the income distribution, 176 Wealth Augmented wealth, 182 Concentration of assets and liabilities, 186 Concentration of wealth, 181 Determinants of wealth inequality, 195 Distribution of wealth, 181 Financial assets, 186 Household asset participation, 189 Household debt, 188 Household net worth, 182 Inequality of net worth, 183 Luxembourg Wealth Study, 181 Marketable wealth, 182 Measuring the distribution of wealth, 182 Non-financial assets, 186 Property taxes, 116 Real estate taxes, 117 Socio-economic characteristics, 195 Trends in wealth inequality, 194