EUROPE AND CENTRAL ASIA STUDIES TOWARD A NEW SOCIAL CONTRACT Taking On Distributional Tensions in Europe and Central Asia Maurizio Bussolo María E. Dávalos Vito Peragine Ramya Sundaram Toward a New Social Contract Toward a New Social Contract Taking On Distributional Tensions in Europe and Central Asia Maurizio Bussolo, María E. Dávalos, Vito Peragine, and Ramya Sundaram © 2018 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www .worldbank .org Some rights reserved 1 2 3 4 21 20 19 18 This work is a product of the staff of The World Bank with external contributions . The findings, interpreta- tions, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent . The World Bank does not guarantee the accuracy of the data included in this work . The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries . Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved . Rights and Permissions This work is available under the Creative Commons Attribution 3 .0 IGO license (CC BY 3 .0 IGO) http://creative commons .org/licenses/by/3 .0/igo . Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: Bussolo, Maurizio, María E . Dávalos, Vito Peragine, and Ramya Sundaram . 2018 . Toward a New Social Contract: Taking On Distributional Tensions in Europe and Central Asia . Europe and Central Asia Studies . Washington, DC: World Bank . doi:10 .1596/978-1-4648-1353-5 . License: Creative Commons Attribution CC BY 3 .0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation . The World Bank shall not be liable for any content or error in this translation . Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank . Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank . Third-party content—The World Bank does not necessarily own each component of the content contained within the work . The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties . The risk of claims resulting from such infringement rests solely with you . If you wish to reuse a component of the work, it is your responsibility to determine whether permission is needed for that reuse and to obtain permission from the copyright owner . Examples of components can include, but are not limited to, tables, figures, or images . All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank .org . ISBN (paper): 978-1-4648-1353-5 ISBN (electronic): 978-1-4648-1354-2 DOI: 10 .1596/978-1-4648-1353-5 Cover photo: © Carlos Reyes / World Bank . Further permission required for reuse . Cover design: Debra Naylor, Naylor Design, Washington, DC Library of Congress Cataloging-in-Publication Data has been requested. Contents Foreword. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii About the Authors and Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix Regional Classifications Used in This Report . . . . . . . . . . . . . . . . . . . . . . . . . . xxi Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Distributional Tensions and the Need to Rethink the Social Contract. . . . . . 2 Equity: A Key Aspiration in the Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Balancing Markets, Policies, and Preferences . . . . . . . . . . . . . . . . . . . . . . . . . 3 The Market-Generated Distribution of Incomes. . . . . . . . . . . . . . . . . . . . . . . 5 Public Policy Responses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Preferences for Equity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Fissures in the Social Contract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Looking Ahead: Public Policies for a Stable Social Contract . . . . . . . . . . . . 15 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Emerging Distributional Tensions in Europe and Central Asia. . . . . . . . . . . 21 The Potential Implications for the Social Contract. . . . . . . . . . . . . . . . . . . . 22 Is a Rethinking of the Social Contract in the Region Warranted?. . . . . . . . . 24 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Are Distributional Tensions Brewing in Europe and 2  Central Asia? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Inequality Across Individuals in Europe and Central Asia . . . . . . . . . . . . . . . 30 Labor Market Polarization and the Shifting Demand for Skills. . . . . . . . . . . 38 An Increasing Generational Divide, and the Young Are Losing Ground . . . 50 Persistent Spatial Disparities across the Region. . . . . . . . . . . . . . . . . . . . . . 65 Rising Inequality of Opportunity, Particularly in the East. . . . . . . . . . . . . . . 77 Distributional Tensions and the Path to a Middle-Class Society . . . . . . . . . . 87 Annex 2A. Statistics Tables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 v References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 vi  ●   Toward a New Social Contract Are Public Policies Equipped to Respond to 3  Distributional Tensions? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 Labor Markets Are Changing, and Policy Is Not Ensuring Equal Protection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 The Impact of Tax and Transfer Systems on Income Redistribution. . . . . . 128 Limited Labor Mobility Affects the Opportunities in High-Productivity Areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 Annex 3A. Decomposition Analysis: Drivers of Change in Redistribution. . . 152 Annex 3B. Policy Changes that Have Contributed to Redistribution. . . . . 157 Annex 3C. The Impact of Taxes and Transfers on Redistribution. . . . . . . . 158 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 The Social Contract: Do Distributional 4  Tensions Matter?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 The Third Component of the Social Contract: The Preference for Equity. . . . 177 There Are Fissures in the Social Contract in the Region . . . . . . . . . . . . . . . 189 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198 How Can the Stability of the Social Contract 5  Be Restored? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 Promoting Growth and Protecting People. . . . . . . . . . . . . . . . . . . . . . . . . 202 Extending Social Protection to Everyone. . . . . . . . . . . . . . . . . . . . . . . . . . 205 More Progressive Taxation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217 Reducing Inequality of Opportunity through Improved Services. . . . . . . . 220 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 Boxes 2.1 Horizontal Inequality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.2 Construction of Occupational Categories. . . . . . . . . . . . . . . . . . . . 40 2.3 Decomposing the Change in Wages: The Role of Occupational Change. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 2.4 Teachers and Drivers: Low Wages in High-Skill Occupations in the former Soviet Union Economies . . . . . . . . . . . . . . . . . . . . . . 48 2.5 The Changing Education and Task Profile of Nonstandard Employment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 2.6 A Closer Look at Spatial Disparities in the Russian Federation . . . . 75 2.7 Calculating Measures of Intergenerational Mobility . . . . . . . . . . . . 85 2.8 Defining the Middle Class. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 2.9 Defining the Absolute Middle-Class Threshold, a Vulnerability Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 Contents ●  vii 3.1 Labor Market Institutions Pick the Winners, France vs. the United States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 3.2 Italy: Toward One Type of Employment Contract . . . . . . . . . . . . . 123 3.3 Housing and Labor Mobility Constraints in Kazakhstan. . . . . . . . . 150 4.1 Preferences for Equity and Demand for Redistribution, a Brief Digression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 5.1 The Perils of Excessive Employment Protection . . . . . . . . . . . . . . 203 5.2 Helping Displaced Workers through Active Labor Market Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 5.3 Progressive Universalism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210 5.4 Types of Social Assistance Cash Transfers. . . . . . . . . . . . . . . . . . . 210 5.5 Distributional and Fiscal Effects of a UBI, Selected EU Countries . . . . 215 5.6 Should Taxes Be Higher on Capital Income or on Wealth?. . . . . . 219 Figures 1 O.­ The social contract as a dynamic equilibrium . . . . . . . . . . . . . . . . . . 4 O.2 Distributional tensions along four dimensions are explored . . . . . . . 5 O.3 Income inequality is much higher among cohorts born in the 1980s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 O.4 The employment share of routine task-intensive occupations has fallen in Europe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 O.5 The share of employment, by occupational category, early 2000s to mid-2010s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 O.6 Between-region spatial inequalities within countries have increased in the European Union . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 O.7 The middle class in the European Union has become more vulnerable. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 O.8 Measured changes in inequality explain little of the demand for redistribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 O.9 Perceived inequality correlates strongly with the demand for redistribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 O.10 At any decile of consumption, Individuals more likely feel poor when they are not in full-time employment . . . . . . . . . . . . . . 14 1.1 Income inequality is lower in Europe and Central Asia than in most of the rest of the world . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 1.2 The social contract as a dynamic equilibrium . . . . . . . . . . . . . . . . . 23 1.3 Distributional tensions along four dimensions are explored . . . . . . 24 2.1 Trends in income inequality, European Union, 1988–2015 . . . . . . . 31 2.2 Trends in consumption inequality, former Soviet Union economies, Turkey, and Western Balkans, 1988–2013 . . . . . . . . . . 32 2.3 Gini index adjusted for the top incomes, 2011. . . . . . . . . . . . . . . . 32 2.4 The number of billionaires and their net worth have increased. . . . 35 2.5 The declining share of labor income, particularly in transition economies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 2.6 The employment share in routine task-intensive occupations has fallen in Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 viii  ●   Toward a New Social Contract 2.7 The share of employment, by occupational category, early 2000s to mid-2010s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 2.8 Changes in wages, Germany, Poland, and Spain, 1990s to 2013 . . . . 44 2.9 Wage changes, Georgia, Kyrgyz Republic, Russian Federation, and Turkey, 1990s to 2010s . . . . . . . . . . . . . . . . . . . . . 47 B2.4.1 Distribution of teaching professionals, drivers, and mobile plant operators, initial year. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 2.10 Nonstandard employment (NSE) has expanded in most of Europe and Central Asia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 2.11 The composition of nonstandard employment differs in countries and regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 B2.5.1 Changes in the education profile of workers, by employment type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 B2.5.2 Changes in task content, by employment type. . . . . . . . . . . . . . . . 55 2.12 Rising nonstandard employment (NSE), Southern and Western Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 2.13 Rising nonstandard employment (NSE), Central and Northern Europe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 2.14 Average job tenure has been mostly stable in Europe and Central Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 2.15 Tenure is decreasing among the young, but less among the middle and older age-groups. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 2.16 Household income, by age of household head, Western, Northern, and Southern Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 2.17 Household income, by age of household head, Central Europe, Baltic States, Russian Federation, and Turkey. . . . . . . . . . . . . . . . . 62 2.18 Average annual earnings, 30–34 age-group, Southern Europe, 2004–14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 2.19 Average annual earnings, 30–34 age-group, Western Europe, 2004–14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 2.20 Average annual earnings, 30–34 age-group, Central Europe, 2004–14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 2.21 Average annual earnings, 30–34 age-group, Northern Europe, 2004–14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 2.22 Income inequality is much higher among cohorts born in the 1980s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 2.23 Spatial disparities in welfare are not uncommon in the region . . . . 67 2.24 Gaps between urban and rural areas are largest in Georgia and Tajikistan and are negative only in Greece . . . . . . . . . . . . . . . . . . . 68 2.25 Between-region inequality has widened in some countries . . . . . . . 69 2.26 Inequality between urban and rural areas has increased in some countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 2.27 Gaps in mean consumption, circa 2003–13 . . . . . . . . . . . . . . . . . . . 70 2.28 Between-region spatial inequalities within countries have increased in the European Union . . . . . . . . . . . . . . . . . . . . . . . . . . 71 2.29 Regional disparities in disposable income rose, were unchanged, or declined . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 Contents ●  ix 2.30 The spatial dispersion of poverty rates has increased. . . . . . . . . . . 72 2.31 Differences in characteristics and in returns to characteristics help explain welfare gaps across geographical areas, circa 2013. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 2.32 Gaps in PISA reading scores: often equivalent to a year of schooling, urban and rural areas. . . . . . . . . . . . . . . . . . . . . . . . . . . 75 2.33 Moldova: indicators of service quality, by region, 2013. . . . . . . . . .77 2.34 Income inequality, Europe, 2005 and 2011. . . . . . . . . . . . . . . . . . . 78 2.35 Trends in inequality of opportunity: France, Germany, Italy, United Kingdom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 2.36 Decomposition of inequality of opportunity in age and cohort effects, France, Germany, Italy, United Kingdom . . . . . . . . . . . . . . . 80 2.37 Decomposition of inequality of opportunity. . . . . . . . . . . . . . . . . . 80 2.38 Income inequality and inequality of opportunity in obtaining income . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 2.39 Inequality of opportunity in tertiary education . . . . . . . . . . . . . . . . 83 2.40 Intergenera­ tional persistence in education, Europe and Central Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 2.41 Trends in the relative size of the middle class, Europe and Central Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 2.42 Income classes, subregions of Europe and Central Asia, excluding the EU15. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 2.43 Age-groups along the income distribution. . . . . . . . . . . . . . . . . . . 93 2.44 Cumulative change in the share of people living in single-adult households, by country. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 2.45 Change in the share of people living in single-adult households, by income, France, Italy, Poland. . . . . . . . . . . . . . . . . 95 2.46 The decline in single-breadwinner households across the region. . . . 95 2.47 The middle class in the European Union has become more vulnerable. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 B2.9.1 The vulnerability-income function: identifying the ­middle-class threshold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 2.48 The profile of those vulnerable to poverty now looks like the middle class of yesterday . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 3.1 Union membership . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 3.2 Employment protection and job quality, Europe and Central Asia vs. the rest of the world . . . . . . . . . . . . . . . . . . . . . . 117 3.3 Employment protection differs within the region and has shifted. . . . 119 3.4 Protections governing contracts, Central Asia and OECD Europe, 1990–2009. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 3.5 Spending on labor market interventions varies across the region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 3.6 Employment protection, by contract type, Eastern Europe and Central Asia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 3.7 Employment structure, selected countries of Eastern Europe and Central Asia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 3.8 Gini index, market vs. disposable income, non-EU countries . . . . 129 x  ●   Toward a New Social Contract 3.9 Gini index, various income concepts, EU28 . . . . . . . . . . . . . . . . . 129 3.10 Nordic countries spend the most on social protection . . . . . . . . . 130 3.11 Structure of taxation, European Union, 2016 . . . . . . . . . . . . . . . . 131 3.12 Most EU15 countries achieve near universality in social protection coverage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 3.13 The Baltic States, Central Europe, and the southern euro area: low benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 3.14 The Baltic States: lowest means testing of nonpension benefits in the EU. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 3.15 Top personal income tax rates declined, Western Europe, 1981–95 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 3.16 . . . and in 1995–2008, but stability returned after the 2008–09 recession . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 3.17 Corporate tax rates declined in Western European, 1995–2008 and 2008–16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 3.18 Changes in redistribution, 2007–14. . . . . . . . . . . . . . . . . . . . . . . .136 3.19 Decomposition of changes in redistribution, Western and Eastern Europe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 3.20 Incidence of transfers on gross income, by decile, 2007 and 2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 3.21 Impact of taxes and social insurance contributions on gross income, by decile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 3.22 Tax changes: progressive in Southern Europe, regressive in Eastern Europe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 3.23 Differences in the reactions of tax systems to job polarization . . . 143 3.24 The limited role of policy changes in transfer systems across occupations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 3.25 Three examples of changes in average tax rates by household type. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 3.26 Labor mobility is low in Europe and Central Asia . . . . . . . . . . . . . 148 3.27 High housing costs in urban areas inhibit internal migration . . . . . 149 B3.3.1 Share of owner-occupied housing. . . . . . . . . . . . . . . . . . . . . . . . . 151 3C.1 Incidence of tax and social security contributions, by income decile. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 3C.2 Incidence of transfers, by income decile. . . . . . . . . . . . . . . . . . . . 160 3C.3 Incidence of taxes and social security contributions, by age-group. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162 3C.4 Incidence of transfers, by age-group . . . . . . . . . . . . . . . . . . . . . . . 163 3C.5 Incidence of tax and social security contributions, by occupational category . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 3C.6 Incidence of transfers, by occupational category . . . . . . . . . . . . . 166 3C.7 Incidence of tax and social security contributions, by household type. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 3C.8 Incidence of transfers, by household type . . . . . . . . . . . . . . . . . . . 168 4.1 Perceptions of inequality differ systematically from objective measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 4.2 Measured changes in inequality explain little of the demand for redistribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 Contents ●  xi 4.3 Perceived inequality correlates strongly with the demand for redistribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 4.4 Individuals more likely feel poor when they are not in full-time employment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 4.5 A large share of people in the region prefer a public sector job . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 4.6 . . . Linked to the value placed on job stability and economic security. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 4.7 Their position in a reference group affects whether individuals feel poor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 4.8 Perceptions of inequality have increased throughout Europe . . . 187 4.9 The value placed on connections in obtaining a job is rising . . . . 188 4.10 Voting for extreme parties has increased in recent years . . . . . . . 189 4.11 Workers facing less demand for their skills tend to vote for extreme parties. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190 4.12 Voter turnout has declined among the young, but not among the old . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 4.13 Distrust in institutions has increased across Europe and Central Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 4.14 Greater inequality is associated with a rise in distrust of government . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 4.15 Distrust of institutions has risen among the losers of occupational change. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 5.1 Social protection policy package . . . . . . . . . . . . . . . . . . . . . . . . . 207 B5.5.1 The costs of a UBI and the various effects of a UBI on income distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 Maps B3.3.1 Housing price index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 4.1 The expansion in voting for extremist parties. . . . . . . . . . . . . . . . 192 4.2 The at-risk-of-poverty measure, 2011 . . . . . . . . . . . . . . . . . . . . . . 193 Tables 2.1 Top 1 Percent Income Shares Vary Across the Region, but Have Risen in Many Countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 2.2 Income Associated with an 8 Percent Probability of Falling into Poverty. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 2A.1 Summary statistics on occupational categories. . . . . . . . . . . . . . . 100 2A.2 Size of income class, by country . . . . . . . . . . . . . . . . . . . . . . . . . . 100 2A.3 Working-age population and old-age dependency, by income class and country. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .101 3.1 A Typology of Labor Market Policy Instruments . . . . . . . . . . . . . . 115 5.1 Main Differences in Social Protection Pillars . . . . . . . . . . . . . . . . . 207 5.2 Comparing Cash Transfer Instruments, by Key Design Features . . . . 212 Foreword Rising inequality is among the most serious problems of our times. Economic progress has been remarkable in the last few decades, but not everyone has enjoyed the same gains, or even the same opportunities. The aggregate indica- tors, such as GDP growth or employment rate, paint a positive picture. Indeed, the setback of the global financial crisis of 2008 has been overcome, and most coun- tries around the world have seen their income and employment not only return to the levels before the crisis but, in most cases, go beyond those and reach new heights. However, a different picture comes into focus when one goes beyond the aggregates. Technological change, globalization, and policy reforms have influenced indus- tries, regions, and ultimately people in very different ways. Entire sectors have lost importance and many occupations are under threat of disappearing. In many countries, the labor share of total income has been declining, and given the accu- mulation of capital wealth in the hands of a minority, incomes are concentrating at the top of the distribution. These changes have created opportunities, but the challenges cannot be over- looked. Services such as education and health care—key inputs to the accumula- tion of productive human capital—are becoming more expensive, and equal access to good-quality services is becoming an issue. Risk-sharing arrangements via targeted assistance or more general insurance have limitations. This uneven playing field generates inequality traps: without mobility and flexibility, ­technology- and globalization-driven opportunities become elusive, some groups are left behind, and distributional tensions arise. In contrast to what populist proposals are promising, there is no quick fix. Curbing the trends—stopping trade or rejecting technologies—as well as pas- sively compensating the losers have not worked in the past, and they will not work in the future. But inaction is not an option. The way societies adjust to distribu- tional tensions and maintain social cohesion can make a big difference, not just in terms of equity but also in terms of future prosperity. Given the long and varied experience with social welfare institutions, one would perhaps think that countries in Europe and Central Asia are well equipped to deal with distributional tensions. But, in fact, these institutions were designed for a very different economic environment. A key difference, even if not the only difference, is a rapidly transforming labor market where long-term wage employment is no longer the norm, especially not for the younger generations. xiii xiv  ●   Toward a New Social Contract Instead of a quick fix, a long-term productive and stable solution requires (1) understanding better how inequality is evolving, and whether the growth process is or is not inclusive, and (2) rethinking the social contract—the shared principles used to regulate markets, define responsibilities and benefits, and redis- tribute incomes. This report aims to offer contributions to these two requirements. Inequality among individuals (or households), usually captured by inequality indexes such as the Gini, has shown a mixed pattern for the Europe and Central Asia region. Compared with the levels at the time of the fall of the Berlin Wall, this vertical inequality is, by the late 2010s, at higher levels. Also, it has been shown that using tax data, the concentration of incomes at the top has increased. However, this report demonstrates that it is persistent unfairness and growing inequality between groups—rather than individuals—that are insidiously corrod- ing social cohesion. Tensions between workers, between generations, and between regions have been increasing. Insecurity, unfairness, and growing tensions among groups have also led to perceptions of increases in overall inequality and influence demands for corrective actions. Fissures in the social contract are becoming more evident. Losers from the distributional tensions—young cohorts, routine task-intensive and low-wage work- ers, inhabitants of lagging regions—choose to voice their discontent by support- ing extreme political movements and parties or choose to exit the social and political dialogue altogether. In terms of rethinking the social contract, rather than prescribing or even iden- tifying a specific set of policies, the report proposes a set of three policy principles that, considered jointly, could help level the playing field and redesign a stable social contract. The principles consist of (1) moving toward equal protection of all workers, no matter their type of employment, while promoting labor markets’ flex- ibility; (2) seeking universality in the provision of social assistance, social insurance, and basic quality services; and (3) supporting progressivity in a broad tax base that complements labor income taxation with the taxation of capital. With its concerns for distribution and fairness and their implications for political stability and sustainable economic growth, this report continues the World Bank work in support of paving the way toward shared prosperity in Europe and Central Asia.1 Cyril Muller Vice President, Europe and Central Asia Region The World Bank Note 1. This paraphrases the title of an earlier study in the same series: Bussolo, Maurizio, and Luis F. López-Calva. 2014. Shared Prosperity: Paving the Way in Europe and Central Asia. Washington, DC: World Bank. About the Authors and Contributors Maurizio Bussolo, Lead Economist in the Chief Economist Office for Europe and Central Asia, has been working on quantitative analyses of economic policy and development with research interests spanning both micro and ­ ­macroeconomic topics. He previously worked at the Organisation for Economic Co-operation and Development (OECD), the Overseas Development Institute in London, and Fedesarrollo and the University of Los Andes in Colombia. He has extensively published in peer-reviewed journals on trade, growth, pov- erty, and income distribution. He holds a PhD in economics from the University of Warwick. María E. Dávalos is Senior Economist in the Poverty and Equity Global Practice. She joined the World Bank in 2010 and has worked in the Latin America and the Caribbean region, as well as the Europe and Central Asia region, on poverty, inequality, economic mobility, migration, and gender. She obtained a master’s degree in economic policy management from the Centre for Studies and Research on International Development (France) and a PhD in economics from Fordham University. Vito Peragine is Full Professor of Economics at the University of Bari. Previously, he was Lecturer in Economics at the University Carlos III of Madrid. He has pub- lished widely in the fields of inequality, poverty, and normative economics. He serves on the editorial boards of the Journal of Economic Inequality and the Review of Income and Wealth. He holds a PhD in economics from the University of York (U.K.). Ramya Sundaram is Senior Economist in the Social Protection and Jobs Global Practice. She has extensive experience in advising governments on improving the effectiveness and coverage of social protection systems, on labor market and acti- vation systems, on measurement and alleviation of poverty, and on inequality and inclusion. Ramya has a PhD in economics from the University of Pennsylvania and was Assistant Professor of Economics at the University of Arizona prior to joining the World Bank. Aylin Isik-Dikmelik is Senior Economist in the Social Protection and Jobs Global Practice at the World Bank. Her work focuses on a wide range of topics on the social protection and labor spectrum, from designing and implementing effective xv xvi  ●   Toward a New Social Contract social protection systems to improving labor markets and employability for inclu- sive growth. She holds a PhD in economics from Johns Hopkins University. Jonathan George Karver is a research analyst in the Poverty and Equity Global Practice for Europe and Central Asia, where he contributes to analytical work on poverty and inequality in the European Union. He has provided leadership and supporting roles for projects related to welfare measurement and fiscal policies, among others. He holds a master’s degree in economics from the Instituto Tecnológico Autónomo de México. Xinxin Lyu worked as a research analyst at the Poverty and Equity Global Practice for Europe and Central Asia, from 2016 to 2018. At the World Bank, she contributed to various pieces of analytical work done in the South Caucasus and Western Balkans countries, focusing on the measurement and analysis of poverty and the estimation of the impact of policies on poverty. She holds a bachelor’s degree from University of International Relations, China, and a master of science in economics from Tufts University. She is currently a PhD student in economics at Purdue University. Mattia Makovec is an economist in the Social Protection and Jobs Global Practice at the World Bank. He works on various topics related to jobs and social protection in Europe and Central Asia, including minimum wages, migration, the integration of refugees, and the effectiveness of social protection systems. Previously, he held positions at Essex University, at the World Bank Country Office in Indonesia, at the University of Chile, and at the Ministry of Labor in Chile. Mattia holds a PhD in economics from Bocconi University and a master’s in economics from University College London. Iván Torre is an economist in the Office of the Chief Economist for Europe and Central Asia at the World Bank. His work focuses on inequality, income distribu- tion, and the political economy of development. He previously worked as a con- sultant for the Inter-American Development Bank. He has a bachelor’s degree in economics from Universidad de Buenos Aires and holds a PhD in economics from Sciences Po, Paris. Mitchell Wiener is Senior Social Protection Specialist in the Eastern Europe and Central Asia region at the World Bank. He is a pension and social security actuary with more than 40 years of experience with public and private pension programs. He specializes in the design, financing, and administration of social security systems. Soonhwa Yi works on identifying good policies to facilitate internal and interna- tional labor mobility in low- and middle-income countries. Prior to joining the Social Protection and Jobs Global Practice, she managed multi-institutional teams to take forward the global migration agenda of KNOMAD (Global Knowledge Partnership on Migration and Development) at the World Bank. Areas of her current research interest include labor policy responses to aging populations and jobs.  Acknowledgments This regional flagship report is a joint product of ECA’s Office of the Chief Economist, the Social Protection and Jobs Global Practice, and the Poverty and Equity Global Practice of the World Bank. The work was carried out under the direction of Hans Timmer, Chief Economist of the Europe and Central Asia region, and with the guidance of Cyril Muller, Europe and Central Asia Regional Vice President; Michal Rutkowski, Senior Director of Social Protection and Jobs; and Carolina Sánchez-Páramo, Senior Director of the Poverty and Equity Global Practice. This report was prepared by a team led by Maurizio Bussolo (Chief Economist Office for Europe and Central Asia), María E. Dávalos (Poverty and Equity Global Practice), Vito Peragine (University of Bari), and Ramya Sundaram (Social Protection and Jobs Global Practice), with support from Luís F. López-Calva, Practice Manager of the Poverty and Equity Global Practice, and Andy Mason and Cem Mete, Practice Managers of the Social Protection and Jobs Global Practice. The authorship of the chapters is as follows: • The Overview was written by Maurizio Bussolo, María E. Dávalos, Vito Peragine, and Ramya Sundaram, with inputs from Iván Torre. • Chapter 1 was written by Maurizio Bussolo, María E. Dávalos, Vito Peragine, and Ramya Sundaram, with inputs from Iván Torre. • Chapter 2 was written by Maurizio Bussolo, María E. Dávalos, Jonathan George Karver, Xinxin Lyu, Vito Peragine, and Iván Torre, with inputs from Ignacio Apella, Damien Capelle (Princeton University), Lidia Ceriani (Georgetown University), Daniele Checchi (Università di Milano), Hai-Anh H. Dang, Ernest Dautovic, Carola Gruen, Tullio Jappelli (University of Naples Federico II and CSEF), Roberto Nisticò (University of Naples Federico II), Stefan Thewissen (OECD), Sailesh Tiwari, Hernan Winkler, and Gonzalo Zunino (CINVE). • Chapter 3 was written by Maurizio Bussolo, Aylin Isik-Dikmelik, Mattia Makovec, Ramya Sundaram, Iván Torre, Mitchell Wiener, and Soonhwa Yi, with inputs from Florentin Philipp Kerschbaumer, Carla Krolage (CESifo), Laura Maratou- Kolias, Renata Mayer Gukovas, Atul Menon, María Laura Oliveri, Andreas Peichl (CESifo), Marc Stoeckli (CESifo), and Christian Wittneben (CESifo). • Chapter 4 was written by Maurizio Bussolo and Iván Torre, with inputs from Esther Bartl, Ada Ferrer-i-Carbonell (IAE-CSIC), Anna Giolbas (GIGA Institute of African Affairs), Bingjie Hu, Jonathan George Karver, and Mathilde Lebrand. • Chapter 5 was written by Maurizio Bussolo, María E. Dávalos, Vito Peragine, and Ramya Sundaram, with inputs from Joe Chrisp (University of Bath), xvii xviii  ●   Toward a New Social Contract Mattia Makovec, Luke Martinelli (University of Bath), Mabel Martinez, Alice Scarduelli (CESifo), Marc Stoeckli (CESifo), Mitchell Wiener, and Jurgen De Wispelaere (University of Bath). The report’s advisory committee, comprising Francois Bourguignon, Francisco H. G. Ferreira, Marc Fleurbaey, and Ravi Kanbur, was a source of knowledge and inspiration. During multiple occasions—in long discussions at the authors’ work- shops, in bilateral communications, and at conferences and seminars—the advi- sors have been always available and generous in providing critical feedback that contributed to shaping the report. We are grateful for invaluable comments from our reviewers, including Arup Banerji, Elena Ianchovichina, and Ana Revenga. The team also received useful comments at various stages from Margaret Ellen Grosh, Carl Patrick Hanlon, Ugo Gentilini, and Truman G. Packard. We want to express gratitude to the wide range of participants at our authors’ workshops, and at seminars and conferences in which we presented background research for the report. Feedback received at the 2016 OECD–World Bank high- level “The Squeezed Middle-Class in OECD and Emerging Countries” conference in Paris, the ECINEQ 2017 conference in New York, the 2017 Stanford-Cornell “Commodification and Inequality” conference in Palo Alto, the 2017 IBS Jobs Conference in Warsaw, the 2018 IZA–World Bank Jobs and Development Conference in Bogotá, and the 2018 IZA World Labor Conference in Berlin has been very useful in strengthening the report. The team appreciates the writing skills, patience, and availability of William Shaw, who helped in the drafting and redrafting of all the chapters. Valuable con- tributions were offered by Mukaddas Kurbanova and Ekaterina Ushakova, who provided exceptional administrative support and oversaw the production of this report; Bruce Ross-Larson, who facilitated the development of the storyline through several hours of discussion among team members; and Robert Zimmerman and Thomas Cohen, who provided insightful comments and changes to the report and whose editing skills improved the report substantially. It also expresses grati- tude to the communications team, including Carl Patrick Hanlon, Paul Clare, Artem Kolesnikov, John Mackedon, and Kym Smithies, for their support in preparing the outreach for and dissemination of the report, and the dedicated webpage. Mary Fisk was the production editor for the report, working with acquisitions editor Jewel McFadden and production manager Aziz Gokdemir. Carlos Reyes designed the cover image, and Datapage International prepared the typeset pages. The team is grateful for their professionalism and expertise. Abbreviations CCT conditional cash transfer C-UCT categorical unconditional cash transfer EU European Union EU13  Bulgaria, Croatia, Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Romania, Slovak Republic, and Slovenia EU15  Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom EU28  the current full membership: Austria, Belgium, Bulgaria, Croatia, Cyprus, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Poland, Portugal, Romania, the Slovak Republic, Slovenia, Spain, Sweden, and the United Kingdom GDP gross domestic product GMI guaranteed minimum income ISCO  International Standard Classification of Occupations (International Labour Organization) LiTS Life in Transition Survey OECD Organisation for Economic Co-operation and Development PISA Programme for International Student Assessment (OECD) PPP purchasing power parity PPS purchasing power standard PRUBI progressive realization of the coverage of universal basic income PT-UCT poverty-targeted unconditional cash transfer PUBI pure universal basic income PUBI-AO pure universal basic income for adults only TUBI  tapered universal basic income or progressive realization of the coverage of the tapered universal basic income UBI universal basic income UCT unconditional cash transfer Note: All dollar amounts are U.S. dollars (US$) unless otherwise indicated. xix Regional Classifications Used in This Report European Union Western Balkans Western Europe Southern Europe Central Europe Northern Europe Austria Greece Bulgaria Denmark Albania European Belgium Italy Croatia Finland Bosnia and Herzegovina Union France Portugal Czech Republic Sweden Kosovo and Western Germany Spain Hungary Estonia Macedonia, FYR Balkans Ireland Cyprus Poland Latvia Montenegro Europe and Luxembourg Malta Romania Lithuania Serbia Central Netherlands   Slovak Republic     Asia United Kingdom   Slovenia               South Caucasus Central Asia Russian Federation Turkey Other Eastern Europe Eastern Armenia Kazakhstan     Belarus Europe and Azerbaijan Kyrgyz Republic     Moldova Central Georgia Tajikistan     Ukraine Asia   Turkmenistan         Uzbekistan       Other country groups mentioned in this report: EU13: Bulgaria, Croatia, Cyprus, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Romania, the Slovak Republic, Slovenia EU15: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden, the United Kingdom EU28: Austria, Belgium, Bulgaria, Croatia, Cyprus, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Poland, Portugal, Romania, the Slovak Republic, Slovenia, Spain, Sweden, the United Kingdom xxi Overview The Europe and Central Asia region stands out as the most equal region in the world. Of the 30 countries around the globe with the lowest Gini coefficient—a measure of income inequality whereby a lower coefficient corresponds to a more equal distribution—23 are in the region.1 Compared with other regions, the coun- tries in Europe and Central Asia redistribute income on a larger scale and have more extensive welfare systems, more progressive taxation, and more generous social protection. This reflects the strong preference of Europeans for egalitarian societies. Yet, people in Europe and Central Asia are dissatisfied with the status quo and, as in regions that exhibit greater inequality, demand changes.2 More people are either voting for populist parties that promise to get rid of current policies and establish a new social order, or they are not voting at all. Separatist movements are on the rise, while trust in political institutions is on the decline. The primary goal of this report is to analyze changes in the distribution of incomes and resources that, even if not fully reflected in changes of inequality among individuals and households, are affecting people’s security, aspirations, and sense of well-being and identity. When asked in opinion surveys, a large majority of people across all countries in the region expresses concerns about rising inequality. It is important to investigate the potential sources of these beliefs and views. 1 2  ●   Toward a New Social Contract The report emphasizes the relevance of distributional tensions among groups and of u ­ nfairness. These reflect the economic drivers of the rising dis- content with the political and social order in the r­egion. The clash between these distributional tensions and the preferences for equity is posing a serious challenge to the social contract in Europe and Central A­ sia. Distributional Tensions and the Need to Rethink the Social Contract The weakening of the social contract is occurring in the context of a rapidly chang- ing economic l ­andscape. The entry of China, India, and the transition countries of Eastern Europe and Central Asia in the global market in the 1990s expanded the size of globally integrated labor markets from 1 ­ .5 billion workers to 2 ­ .9 billion workers, the “great doubling” as Freeman (2007, 55) calls i 3 ­t. Recent technological progress is increasing the demand for advanced problem-solving and interper- sonal skills, while the demand for less-advanced skills decreases as routine jobs become ­ automated. Digital transformation allows new technologies and start-up firms to scale up quickly and is rapidly altering production p ­ atterns. These global forces continue to roil labor markets and cause uneven economic impacts through- out societies in Europe and Central A ­ sia. These pervasive changes are affecting specific groups d ­ ifferentially. Although some are benefiting from the transforming economic landscape, ­ others are n ­ ot. The report describes in detail the key distributional tensions among groups that are identified according to four criteria: birth cohort, occu- pation, place of residence, and, following the literature on inequality of oppor- tunity, circumstances beyond the control of the individual, such as parental background and gender. Horizontal inequality among groups—which affects young workers, people in vanishing occupations, individuals lacking good social networks, and people living in lagging regions—is not captured by the vertical inequality in income among individuals and households that is measured by the Gini ­ coefficient. The changes in the Gini coefficient may even be at odds with the deepening dis- tributional ­ tensions. Widening horizontal inequality makes people feel they lack opportunity in an unfair ­ system. A shift to part-time, temporary, or gig jobs, such as driving for Uber, provides income, but not the benefits offered through full-time employment in this ­ region. The value placed on noneconomic factors, such as autonomy and status, is also threatened by the rise of nonstandard forms of e ­ mployment. This leaves workers more vulnerable to economic shocks and, regardless of income, perceptions that they are less well o ­ ff. Individuals who expected to join the middle class through educational attain- ment or through work experience find themselves instead struggling for financial stability and ­ security. The steady size of the region’s middle class masks the pres- ence of considerable disappointment among working-age individuals who may still enjoy middle-class incomes but do not have middle-class economic ­ security. The report shows that government policies and institutions, which were designed in the twentieth century and had been working satisfactorily for quite Overview ●  3 some time, are not equipped to handle the emerging distributional ­ tensions. This inadequate response clashes with the value that people place on equity and region and creates an imbalance. This imbalance—across markets, stability in the ­ policies, and preferences in the distribution of resources—is a major reason for the appeal of populism and is exerting pressure on the social c ­ ontract. Based on an analysis of the rising distributional tensions in the region, this report calls for a fundamental rethinking of the principles behind the policies and institutions that regulate markets, define responsibilities and benefits, and redis- tribute incomes—a rethinking of the social contract where equity, progressivity, and universality are reevaluated. Equity: A Key Aspiration in the Region The desire for social equity is a characteristic of European civilizations dating back more than 2,000 y ­ ears. “There should e ­ xist . . . neither extreme poverty n ­ or . . . excessive wealth, for both are productive of great evil,” wrote Plato (Tanzi 2018, ­ 302). “An imbalance between rich and poor is the oldest and most fatal ailment of all republics,” Plutarch later affirmed (Tanzi 2018, 3 ­ 02). Following the Great Depression and the devastation of World War II, societies in Europe greatly expanded the welfare s ­ tate. In Western Europe, free markets were combined with broad participation in education, social safety nets, and income redistribution, as well as universal access to health c ­ are. During the same period, countries in the eastern part of Europe and Central Asia featured state- controlled economic activity, alongside universal, state-provided access to ser- vices and to guaranteed ­ work. While political, ideological, and economic perspectives differed significantly across countries, a common theme was the aspi- ration for equity and social ­ cohesion. Such a commitment to equity is not evident across all regions of the w ­ orld. For example, in North America, the United States does not have a European-style welfare system because of different social preferences and degrees of aversion to inequality (Alesina, Glaeser, and Sacerdote 2 ­ 001). About 70 percent of people in the United States believe the poor can help themselves to improve their s ­ ituation. In Western Europe, only 40 percent of individuals believe that poor people have a chance to escape poverty on their own; in Eastern European transition countries, the share drops to 24 p ­ ercent. As a result, a majority in Europe supports govern- ment policies to ensure well-being and redistribute i ­ncome. Balancing Markets, Policies, and Preferences The term “social contract” originated in political philosophy in reference to the agreement of individuals to give up part of their freedom in return for protection provided by the state (for example, see Hobbes 2012; Locke 1988; Rousseau 1968). This report puts an economic interpretation on the c ­ ­ oncept. Individuals accept the broad outline of economic policies if the outcomes of these policies ­ references. This dynamic is similar, although not identical, to coincide with their p 4  ●   Toward a New Social Contract the approach of Binmore (1998), who sees the social contract as an equilibrium of a game between social entities and individuals, as well as the analytical approach proposed by Kanbur (1999) in the context of optimal t ­ axation. It also resembles the recent effort to evaluate social progress, including distributional issues, by the International Panel on Social Progress (IPSP 2 ­ 018). According to Rodrik (1999), well-functioning social contracts allow countries to manage shocks effectively and adapt to new, efficient equilibria. Countries that have unresolved distributional conflicts may experience inefficient outcomes because the losers do not trust the system, opt out, and resist the needed a ­ djustments. Distributional tensions, if not balanced by corrective policies, institutional arrangements, or a shift of prefer- ences on equity and fairness, can generate cracks in the social c ­ ontract and stop or severely hinder economic growth. Thus, a stable social contract finds a balance among the following (figure O.1): • The market-generated distribution of resources and incomes • Public policies, including taxes and transfers, regulation, and the provision of goods and services, that alter this distribution • Individual and societal preferences for equity, perceptions of inequality, and the demand for the redistribution of opportunities and ­ outcomes Temporary deviations from an equilibrium among these three elements are normal and can be t ­olerated. However, a long-term imbalance risks generating ruptures in the social c­ ontract. This conceptual framework is an organizing principle of the ­ report. The report first describes the rise in horizontal inequality in the market-generated distribution of income and examines how policies (regulations, redistribution through taxes and transfers, and public expenditures) fail to fully address this. It also shows peo- ple’s preferences for equity and the increasingly negative perceptions of the situ- ation in income distribution and ­ fairness. A main contribution of the report is the organization of a wealth of data and empirical research around the three elements shown in figure O.­ 1. This structure also highlights a growing imbalance between the distribution of income generated by the market and the policy regime in responding to the FIGURE ­O.1 The social contract as a dynamic equilibrium Market-generated Perceptions and distribution of resources societal preferences Public policies Overview ●  5 desires of individuals about e­ quity. A failure to resolve this imbalance can under- mine social cohesion and have serious implications for the stability of the ­ social contract. The polarization in recent voting behavior in several countries of the region is a symptom of the d ­ iscontent. The final section of the report thus considers changes in the policy framework that could support a return to a long-term equilibrium and a renewed and stable social ­contract. The Market-Generated Distribution of Incomes The first part of the report considers four distributional tensions generated by the market: • The intergenerational divide, or disparities between young and old generations • Inequalities among workers engaged in different occupations, such as office clerks and machine operators versus nurse’s aides, private security guards, or the more highly skilled engineers and scientists • Inequality in access to economic opportunities based on geographical location • Inequalities of opportunity based on gender, ethnicity, background, or other characteristics rather than individual effort (figure O.2) FIGURE O.2 Distributional tensions along four dimensions are Between explored occupations Economic insecurity Inequality of and unfairness Between and opportunity Crisis of the within cohorts middle class Between geographic areas 6  ●   Toward a New Social Contract Some groups are on the losing side of more than one of these distributional ­ensions. Because it supports economic and political stability, the middle class is t an important g ­ roup. The first part of the report analyzes the extent to which the ­ lass. four distributional tensions are linked to the malaise of the middle c The four distributional tensions have emerged amid concerns and resentment over the falling share of labor relative to capital in total income and over the increasing concentration of top incomes and w ­ ealth. In the United Kingdom, the share of income held by the top 1 percent has risen by 7 percentage points in the past 25 years, reaching 14 percent in ­ 2014. The number of billionaires in Western Europe rose from 90 in 1996 to 379 in 2017, and the number of Russian Federation billionaires rose from 8 in 2001 to 96 in 2 ­ 017. A Growing Intergenerational Divide In Western Europe, relative to older cohorts, younger cohorts include a larger share of workers who are unemployed or in low-quality j ­obs. In 2015, temporary contracts represented close to 50 percent of employment among workers ages 15–24 in France and the Netherlands, compared with around 20 percent among the overall population in both c­ ountries. The young will likely have to work more years and will likely have less savings to finance retirement despite longer work histories compared with preceding g ­ enerations. For these younger workers, lower earnings and fewer old-age income prospects imply a widening intergenerational divide, which is an important source of distributional tension even if it is masked by positive income trends more g ­ enerally. In addition, younger workers in Southern and Western Europe are facing higher income inequality at every point of the life cycle compared with older generations (figure O.3). For example, income inequality among Italians born in FIGURE O.3 Income inequality by birth cohort Income inequality is much 0.50 higher among cohorts born in the 1980s Implied Gini coefficient at age 40 Household equivalized income 0.45 0.40 0.35 0.30 1930 1940 1950 1960 1970 1980 Birth cohort born in... France Germany Italy ­ l. ­ Source: Bussolo et a 2018. Overview ●  7 the 1930s was similar to that in (fairly equal) Japan (Gini coefficient of about ­ 0.31). In contrast, income inequality among the cohorts born in the 1980s was at the ­ .48). This greater income level of (highly unequal) Chile (Gini coefficient of about 0 dispersion can be interpreted as a sign of greater insecurity and v ­ ulnerability. Because inequality tends to rise as cohorts age, starting the life cycle with high inequality increases the likelihood of even greater inequality in the ­ future. Together with slower growth, this creates more insecurity, along with the serious risk that populations in Europe and Central Asia will age ever more unequally (OECD ­2017). Polarization in Occupations Occupational polarization has increased because economic transformation favors some sectors and ­ occupations. More broadly, occupations intensive in routine tasks, typically in the middle of the wage spectrum, have shrunk across Europe: their share of employment has fallen by more than 10 percentage points in Southern and Western Europe and by close to 5 percentage points in Central and Eastern Europe (figure O.4). This has forced many middle-skilled workers into lower-skilled occupations, thereby reducing the incomes of low-skilled ­ workers. At the same time, occupations at the top of the wage distribution—typically inten- sive in nonroutine cognitive tasks—have ­increased. This has been associated with an upward pull in incomes among highly skilled w ­ orkers. Overall, the polarization of occupations in Europe has translated into greater labor income inequality: the Gini index of labor earnings rose by 8 points in Germany and Spain from the mid- 1990s to 2013 and by about 5 points in Poland during the same ­ period. More seriously affected by the occupational changes were workers already at the bot- tom of the income distribution, but workers in the middle also faced reductions in earnings growth and greater job insecurity because mid-income occupations are ­disappearing. FIGURE O.4 a. Baltic States b. Central Europe c. Northern Europe 10 The employment share of 5 routine task-intensive Change in percent of regular employees 0 occupations has fallen in −5 Europe −10 Change in the share of employment, by occupation d. Southern Europe e. Western Balkans f. Western Europe category, late 1990s to early 10 2010s 5 0 −5 −10 Region Nonroutine, manual Routine Nonroutine, cognitive ­ urveys. Source: World Bank calculations based on household surveys and labor force s Note: Northern Europe: Denmark, Finland, Norway, Sweden. The Baltic States: Estonia, Latvia, Lithuania. 8  ●   Toward a New Social Contract FIGURE O.5 a. Armenia b. Georgia c. Kyrgyz Republic The share of employment, 10 by occupational category, Change in percent of regular employees 5 early 2000s to mid-2010s 0 −5 −10 d. Moldova e. Russian Federation f. Turkey 10 5 0 −5 −10 Country Nonroutine, manual Routine Nonroutine, cognitive ­ urveys. Source: World Bank calculations based on household and labor force s In the eastern part of Europe and Central Asia, particularly in the former Soviet Union economies, the picture is more ­ nuanced. Occupational change has been less significant, and, with the exception of Moldova, this has meant a reduction in nonroutine cognitive task-intensive ­occupations. Highly skilled workers, usually prevalent in this occupation type, experienced an average decline of about 5 percentage points in their share of employment in Armenia, Georgia, the Kyrgyz Republic, and the Russian Federation (figure O.5). This occupational transformation in the former Soviet Union economies risks frus- trating the aspirations of the well-educated younger cohorts that are entering the job ­market. A Spatial Divide Differences in income levels and poverty rates persist among regions in many countries of Europe and Central Asia, and, despite increases in average consump- tion among households over the past decade, inequalities between geographical areas have widened in several c ­ ountries. In Armenia, for example, the difference in poverty rates between the less well-off and the more well-off regions rose from 25 percentage points to 38 percentage points between 2005 and ­ 2014. In 2.5 times higher than the Romania, the poverty rate in the least well-off region was ­ ­egion. The poorest region in France had an at-risk-of rate in the most well-off r region. In the European poverty rate three times higher than the rate in the richest ­ Union (EU), despite a reduction in country-level inequalities, differences in output across regions have been widening (figure O.6). Differences in educational attainment are a key determinant of spatial gaps in welfare and undermine equality of o ­ pportunity. Across the region, in both the east and the west, there are gaps in the quality of education both between socioeco- nomic groups and between rural and urban ­ areas. The spatial divide in learning between youth in urban areas and youth in rural areas in Bulgaria and Moldova is equivalent to around two years of s ­ chooling. Overview ●  9 FIGURE O.6 115 Between-region spatial 110 inequalities within countries Coefficient of variation in gross domestic product have increased in the 105 European Union (GDP) per capita, Index 2000 = 100 100 95 90 85 80 75 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 00 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Year Country Within-country NUTS-3 (unweighted) Source: Farole, Goga, and Ionescu-Heroiu 2 ­ 018. Note: Country refers to the coefficient of variation across European Union countries, signaling convergence in gross domestic product ( ­GDP). The within-country coefficient of variation measure is at ­NUTS-3). GDP per capita is measured in the Nomenclature of Territorial Units for Statistics–3 level ( Purchasing Power Standard (PPS). Inequality of Opportunity Inequality is often measured in outcomes, such as consumption, income, wealth, or even education, while fairness refers to the process generating these o ­ utcomes. Even in a context of stable income inequality, opportunity inequality—the propor- tion of the overall inequality deriving from circumstances beyond the control of individuals—may ­ rise. Finding a good job—according to many, a crucial step in accessing a stable, middle-class standard of living—is becoming more ­ difficult. It involves possessing favorable connections more than possessing ability or effort (Dávalos et a ­ 016). Inequality of opportunity or changes in fairness may be ­ l. 2 emerging as key distributional tensions in the r ­egion.4 In Western Europe, the transmission of education privileges from parents to offspring has decreased (a result of the mass education effort), and the education premium in wages has also been trending d ­ ownward. Together, these phenomena should have reduced overall inequality of ­ opportunity. Instead, inequality of opportunity in incomes has been generally stable at high l ­evels. Parental back- ground still counts in explaining inequality in the earnings of offspring through a networking mechanism, analogous to the social separatism of the upper classes, as reflected in the growing importance of private education, private health plans, and private pensions (Milanovic ’­ 2017). This means that networking among well-off parents buys better positions for the offspring in the income distribution, thereby achieving the same objectives promised by private e ­ ducation. 10  ●   Toward a New Social Contract In Eastern Europe, by contrast, inequality of opportunity in education is increas- ing, which translates into greater inequality of opportunity in the labor m ­ arket. Birth circumstances, especially parental background among individuals, are more important determinants of access to tertiary education among the generation that came of age in the early 2000s than among the generation that entered educa- ­ conomy. Indeed, tional institutions before the subregion’s transition to the market e a large portion of inequality of opportunity in education among the youngest cohorts in Eastern Europe is explained solely by parental background: access to education has become more unfair over time because it is increasingly linked to parental educational a ­ chievement. Increased Vulnerability in the Middle Class Policies are often justified by reference to the needs of the middle class partly because a large, thriving middle class has been associated with political stability and sustained economic growth (for example, see Birdsall 2010; Birdsall, Graham, and Pettinato 2000; Easterly ­ 2001). Overall, the rise of distributional tensions and persistent unfair economic ­ processes have altered the complexion of the region’s middle class, reducing eco- nomic security and disappointing the expectations of many workers who had anticipated that they would be able to enjoy a middle-class ­ lifestyle. While the changes in the size of the middle class have been quite slow, there has been a pronounced deterioration in the sense of security and an expansion in the risk of dropping out of the middle class and into p ­ overty. For example, the income necessary to guarantee a small probability of falling into poverty has risen from an average of US$34-a-day purchasing power parity (PPP) to an average of US$40-a-day PPP in the last decade (Bussolo, Karver, and López-Calva 2018) (­figure  O.7).5 This additional US$6 can be interpreted as an increase in the insur- ance premium to mitigate the growing risk of falling into ­ poverty. In some coun- tries, the cost of the premium climbed by 100 percent or m ­ ore. Thus, it rose from US$14 to US$32 in Bulgaria and from US$22 to US$44 in L ­ atvia. This surge in vul- nerability, linked to the changing profile of the middle class, is in line with the per- ception that the middle class is losing ­ out. It has provoked heated policy debates and proposals for a full overhaul of taxation and social ­protection. It also has impli- cations for the political platforms that can gain support from the middle c ­ lass. Public Policy Responses Public policies are struggling to cope with rising inequality between groups in the ­region. The significant progress in economic and social equality during the second half of the 20th century, mainly in Western Europe, was supported by mass education, labor unions, and substantial redistribution through taxation and public transfers (Atkinson 2016; Milanovic ’­ 2017). Government policy and the welfare state were crucial in the effort to achieve equity and still deliver a considerable reduction in vertical inequality. For the 28 countries in the EU, the difference between the Gini Overview ●  11 FIGURE O.7 Low probability to The middle class in the 80 fall into poverty European Union has become more vulnerable Daily income per capita (2011 US$ PPP) 60 Increase in the level of income to 40 keep the same low probability (i.e, to still be in the middle class) 20 0 0 20 40 60 80 100 Probability of falling into poverty (%) 2005–2008 2011–2014 Source: Calculations based on data of the Longitudinal–User Database of EU-SILC (European Union Statistics on Income and Living Conditions) (database), Eurostat, European Commission, Luxembourg, ­http://ec.europa.eu/eurostat/web/microdata/european-union-statistics-on-income-and-living-conditions. Note: The two curves in the figure have been obtained using pooled data for Austria, Belgium, Bulgaria, Cypress, Denmark, Estonia, France, Greece, Hungary, Iceland, Latvia, Lithuania, the Netherlands, Norway, Poland, the Slovak Republic, Slovenia, and Spain during the two periods indicated. of market incomes and the Gini of disposable incomes averages the equivalent of 20 Gini points—an amazing ­ feat. However, the institutions and the policies face significant challenges in adapt­ ing to the profound global changes of the past few decades. The reaction of the welfare systems in most European countries to the emerg­ ing distributional ten- sions was partial and sometimes inconsistent. Losers were compensated by increases in transfers, but not by a significant decline in taxes. Several countries in Central and Eastern Europe introduced a flat tax on personal income, starting with Estonia in 1994 and followed by Bulgaria, the Czech Republic, Hungary, Latvia, Lithuania, Romania, and the Slovak Republic.6 These policies were largely regressive in terms of the vertical inequality of the income distribution. In Hungary, which was the last of the group of countries to institute the flat tax, in 2011, the average tax rate paid by the top three deciles of the income distribution fell by 2 percentage points between 2007 and 2014, while the rate paid by the bottom three deciles remained practically unchanged. However, these and other policy changes had an even greater impact on horizontal inequality or inequality between groups. For instance, average tax rates in Bulgaria, Hungary, and Poland were reduced signifi­ cantly for the win- ners of the shifts in occupations, while average tax rates were reduced by a smaller amount or remained unchanged for workers in occupations for which demand was falling. These changes accentuated the widening divide between the winners and losers of the changes in occupations. In Hungary, the 12  ●   Toward a New Social Contract distribution of income across age-groups was adversely affected. The aver­ age tax rate among tax-paying individuals ages 18–24 rose by more than 8 percent- age points; among individuals ages 35–44, however, it dropped by 2 percent- age points, and, among the 45–54 age-group, it did not change. In an already polarized society, these tax policy changes, on top of the initial disparities in market incomes and the reduced job security affecting younger workers, wid- ened the intergenerational divide.7 Although most politicians pay lip service to the needs of the middle class, little has been done to protect vulnerable workers through changes in tax and transfer policies. Support has shrunk for households that rely on a single source of market ­ income, and such households are facing a growing risk of falling into ­ poverty. This contrasts with the support for households with several earners or pensioners, groups with greater economic security that rely on multiple sources of income or on steady public ­transfers. Dual earner households in Poland obtained a tax cut of close to 5 percentage points, a pattern similar to that in other Central and Eastern European ­ countries. Similarly, in Belgium, Finland, and Sweden, house- holds that were dependent on transfers and that enjoyed relatively high levels of income security also benefited from tax c ­ hanges. In contrast, the economically insecure have not benefited from tax changes in any of these c ­ ountries. Preferences for Equity Rising income inequality among groups runs counter to the strong preferences for equity and fairness in the ­ region. If inequality in a country is not in line with the preferences of the population, there will be a demand for corrective ­ action. The government may respond with changes in redistributive ­ policies. However, if the policies do not address the dimensions of inequality that people care about and perceive as unfair, the policies are likely to f ­ail. The gap between perceptions of inequality and the inequality measures econo- mists use is substantial and p ­ ersistent. This may be because perceptions do not reflect reality, but it may also mean that individuals are concerned with types of inequality that are not readily or accurately measured by traditional objective ­ indicators. In any case, perceptions m ­ atter. “We suggest that most theories about political effects of inequality [demand for redistribution, the political participation of citizens, democratization] need to be reframed as theories about effects of perceived inequality,” note Gimpelson and Treisman (2018, 2 ­ 7). Indeed, the demand for redistribution is much more closely correlated with the perceptions of individuals on inequality than with tradi- tional measures of inequality (figures O.8 and O.9). What Drives Perceptions? People form their perceptions of inequality by considering the actual dispersion of ­ ispersion. incomes (or resources), as well as the process that generates this d How much do people value the security afforded by stable employment, and how does this influence their views on inequality? When individuals are Overview ●  13 FIGURE O.8 Percentage of people demanding more redistribution minus 60 Measured changes in HUN inequality explain little of percentage of people demanding less, in 2009 FRA SVN the demand for ISR BGR 40 redistribution POL CZE AUT ESP 20 DEU SWE NOR 0 USA –20 –0.02 0 0.02 0.04 0.06 0.08 Change in Gini coefficient 1999–2009 Source: Bussolo, Ferrer-i-Carbonell, Giolbas, and Torre (2018). FIGURE O.9 Percentage of people demanding more redistribution minus 60 Perceived inequality HUN RUS correlates strongly with the percentage of people demanding less, in 2009 PRT FRA SVN demand for redistribution BGR ISR 40 SVK POL AUT CZE ESP 20 CHL DEU JPN SWE GBR CHE NOR 0 USA –20 –50 0 50 Net equality perception in 2009 (%) Source: Bussolo, Ferrer-i-Carbonell, Giolbas, and Torre (2018). Note: Net equality perception is defined as the difference between the share of people believing their country is equal and the share of people believing their country is unequal. asked to place themselves on a 10-step income ladder on which the bottom step represents the poorest 10 percent of the population and the top step the richest 10 percent, individuals who are not in stable, full-time employment are more likely to report that they feel poor (that is, that they belong to the lowest deciles or steps of the ladder) compared with those who have such employ- ment (figure O.10). Declining job security is clearly an important source of dissatisfaction among middle-class w ­ orkers. Similarly, for a given income, people reporting that they are in good health place themselves higher in the income distribution than do people who report they are in bad h ­ ealth. 14  ●   Toward a New Social Contract FIGURE O.10 At any decile of 15 consumption, Individuals more likely feel poor when they are not in full-time Probability of feeling poor (%) employment 10 Change in probability of feeling poor due to lack of full employment 5 Equivalent change in probability of feeling poor due to a change in consumption decile 0 1 2 3 4 5 6 7 8 9 10 Decile of consumption has worked in the last 12 months has not ­ 017. Source: Bussolo and Lebrand 2 Fissures in the Social Contract Labor market regulations and redistribution systems in Europe and Central Asia have not been effective in protecting important segments of the population from the rise in social tensions driven by market ­ forces. This means that societies are becoming less equitable, while people continue to value equity, which is evident from their preferences for fairness and their assessments of the impact on their welfare of the c­ hanges. This imbalance may be reaching a critical l ­evel. Voting is becoming more polar- ized, and populist parties have achieved success in recent e ­ lections. Separatist movements have spread in Catalonia and ­ Scotland. The 2018 appointment of a government led by the League and the Five Star Movement in Italy; the 30 percent of votes achieved by Marine Le Pen, an extreme right-wing candidate, in the runoff of the 2017 French presidential election; and the emergence of the euroskeptic Alternative for Germany in the 2017 German election are ­ examples. Meanwhile, the already low level of trust in institutions has continued to trend ­ downward. In 2015, only 11 percent of the respondents to the Life in Transition Survey expressed com- plete trust in their national government, and only 10 percent in their national parliament. This calls for a reexamination of the social ­ ­ contract with a focus on rem- edying the emerging distributional tensions and reestablishing social cohesion. Analysis of recent data show that there is a direct correlation between these manifestations of the imbalance, or of the cracks in the social contract, and the emerging distributional tensions described above. For example, the group of workers penalized by recent shifts in the demand for skills appear to be voting more regularly for extremist parties. There is also evidence that polarization of the voting is related to regional welfare disparities. And younger generations are opt- ing out of the system by not voting, as shown by their declining turnout at elec- tions across Europe. Overview ●  15 Looking Ahead: Public Policies for a Stable Social Contract Market-driven inequalities, absent or delayed adjustments in public policy and institutions, and strong preferences for equity are contributing to instability in the region. social contract in the ­ The countries of Europe and Central Asia differ in many respects, and policy prescriptions ought to be context s­ pecific. Even so, three principles are relevant to any consideration of policy instruments to achieve people’s aspirations for an equi- table and cohesive society and to build a stable social contract: • Promote labor market flexibility, while maintaining protection for all types of labor contracts • Seek universality in the provision of social assistance, social insurance, and good-quality basic services • Expand the tax base by complementing progressive taxation on labor incomes capital with taxation on ­ These principles can contribute to tackling the emerging distributional tensions contract. Any approach should incorporate all affecting the stability of the social ­ ­ three. Acting on one or two alone might exacerbate t ­ ensions. Labor Market Flexibility and Protection The dynamic labor markets of today call for greater efficiency in job matching that helps workers embrace better opportunities and assists firms in finding appropriate skills. This helps everyone in adapting and benefiting from the new world of w ­ ­ ork. The traditional employer-employee relationship has eroded in Europe over the last two ­ decades. The erosion has been more dramatic in some countries, such as Poland. It has been accompanied by a proliferation in alternate types of ­ ­ contracts. Labor regulations should keep pace and avoid creating divisions among groups that may fuel distributional tensions and undermine the equality of o ­ pportunity. Efforts to achieve flexibility cannot be undertaken only at the margin, which would result in protecting some workers, but not ­ others. Partial reforms would mean that a majority of the people entering the labor market or starting new jobs will be active in nonstandard ­ employment. In several countries, graduating from temporary to permanent employment is ­ difficult. Some workers therefore experi- ence greater economic insecurity, while others are in permanent employment with strong ­protections. In the western part of the region, efforts to foster more flexibility should be aimed at closing the divide in protection across types of employment, thereby reducing labor market ­ segmentation. The Jobs Act in Italy sought to reach this goal by simplifying the types of labor contracts and offering protection for all ­ workers. In the eastern part of the region, informality is widespread in several countries, and a large share of the workforce does not benefit from the protections offered by labor regulations or by social ­ insurance. If informality remains substan- tial, the key is to reform labor market institutions and other business regulations to promote greater f ­ormalization. 16  ●   Toward a New Social Contract Social Assistance, Social Insurance, and Key Services Social assistance is still an important policy arm in efforts to reduce poverty in many countries in the r ­egion. More nonpoor households are becoming vulnerable. This is incompatible with the aspiration to end poverty and vulnera- ­ bility and promote a middle-class ­ society. Extending the reach of social assis- tance programs ought to be a key feature of any new social contract among countries in the r ­ egion. The nature of the initiatives implemented to realize the objective of providing guaranteed minimum protection among the population will vary by c ­ ountry. Fiscal and political considerations are c ­ rucial. There are advantages and disadvantages to means testing and to universal ­ approaches. Income-based targeted schemes, well established in many countries in the region, can be used to supply generous transfers by assisting the people most in ­ need. However, that may leave many people unprotected, including the many nonpoor who are v ­ ulnerable. Complex eligibility rules, stigma effects, a lack of knowledge among potential beneficiaries, and the administrative burden of delivering and receiving the benefits are some of the ­obstacles. Universal approaches to social assistance may address some of these ­challenges. The universal basic income (UBI) being discussed in many forums could provide broader protection and security to the population through greater coverage and take-up, and it would reduce disincentives to w ­ ork. Yet, a UBI may be associated with other ­ challenges. Depending on the design, it might entail a substantial fiscal burden, and the feasibility and equity impacts of implementing a UBI relative to other approaches must be ­ weighed. A pure UBI—a minimum income transfer to all individuals—does not exist in the region, but categorical unconditional cash transfers are being provided as a benefit among population groups such as chil- dren and the e ­ lderly. The emergence of distributional tensions represents a clear message: the growing economic insecurity affecting nonpoor households is a call for a review of the design and coverage of social ­ assistance. The changing nature of work is likewise a call for a reexamination of social insurance. In Europe and Central Asia, pension systems are the main channel for ­ social ­insurance. However, the systems in many countries do not supply adequate protection in old age to individuals who have been active in nonstandard forms of employment or in informal work or who have been out of ­ work. Aging populations threaten the sufficiency of the coverage and financial sustainability of the ­ systems. The poverty-preventing objective of social insurance among the elderly, chroni- cally ill, unemployed, or disabled should be separate from the consumption- smoothing ­ objective. Insurance against the catastrophic risk of illness, injury, job loss, and other shocks that could drive households into poverty could be provided directly by government in conjunction with income support for all people in need as part of a guaranteed minimum poverty prevention p ­ ackage. This minimum package could cover everyone and would be financed through general tax reve- nue, thereby avoiding reliance on employment relationships and mandatory pay- contributions. The decoupling of social insurance from employment could roll ­ facilitate the expansion of coverage to all, reduce the adverse impact on work Overview ●  17 incentives and the labor demand of firms that is associated with the financing of social insurance through payroll taxes, and enhance the sustainability of social insurance ­ systems. In a dynamic labor market, such an insurance scheme could encourage people to seek out and take on better jobs without fear of losing ­ coverage. Meanwhile, a mandated insurance plan could address consumption- smoothing if the provider of the financing for program benefits is identified and the benefits are reasonable in relation to the c ­ ontributions. Public policies in Europe and Central Asia also need to aim at recognizing a universal right to quality services to ensure that everyone can build their human capital and access economic ­ opportunities. Key services—water, sanitation, trans- portation, education, health care, childcare, and eldercare—are provided in most ­ countries. Yet, these services are not available to ­ all. Under a stable social contract, they should not be out of reach of segments of the ­ population. Universal provision of these services as a premarket intervention could represent great progress in ensuring equal opportunity for ­ all. Education, in particular, has been a great e ­ qualizer. Education systems can help level the playing field by addressing the concern over the widening inequality of opportunity and the persistent spatial inequalities in many ­ countries. However, simply expanding access to education no longer guarantees equal ­ impacts. The focus should be not only universal access to schooling, but also universal access to learning as a key feature of a new social ­ contract. Throughout education sys- tems, learning should include the development of cognitive skills (numeracy and literacy) and socioemotional skills so that younger generations, regardless of their socioeconomic background or the location of their residence, leave school pre- pared to lead productive lives and able to adapt to the changing nature of w ­ ork. Developing these skills starts early in ­life. So, the gaps in the access to early child- hood education that affect the most disadvantaged need to be c ­ losed. Education and training services accessible to all adults that allow for learning new skills or for upskilling require strong partnerships between public and private providers. Employers should be encouraged to participate, which may require ­ incentives, especially if more flexible labor markets and shorter job tenure reduce the returns to investments by firms in ­ employees. Firms could contribute to build- ing training systems that are more flexible in responding to labor market demands and provide more work-based l ­earning. Progressive Tax Systems Public policies need to expand the tax base, raise tax rates on top earners, and implement more progressive taxation that does not target only i ­ncome. Higher taxes on capital income and higher taxes on wealth (for example, on inheritance or bequests) could underpin a more equitable fiscal system in the ­region. Because capital and the returns to capital are concentrated among a smaller share of the population, taxes on capital could enhance the progressivity of tax systems and reduce the inequalities between economic ­ groups. They could also promote equality of opportunity among people whose lack of endowments mean that they footing. They can also supply a source of financing to do not start life on an equal ­ expand and strengthen the social c ­ ontract. 18  ●   Toward a New Social Contract Increasing progressivity in the inheritance tax and in capital income taxation represent ways to promote equity and boost financing s ­ ources. In a globalized world where capital is highly mobile, capital taxation would be difficult to establish without coordination across ­countries. Recent proposals include global or regional 2014). taxes on capital (Atkinson 2016; Piketty ­ Conclusion The widening economic fissures in the societies of Europe and Central Asia are affecting young workers, people in vanishing occupations, individuals lacking good networks, and residents of lagging regions, and they are threatening the contract. Institutions that have achieved a remarkable sustainability of the social ­ degree of equity and prosperity over the course of several decades now face con- siderable difficulty in coping with the associated c ­ hallenges. Surveys reveal grow- ing concerns about the inequality of opportunity, while electoral results show a marked shift in favor of populist parties that offer radical solutions to voters dis- quo. satisfied with the status ­ There is no single solution to all the ills in every country, and the response to these problems varies considerably across the ­ region. However, this report pro- poses three broad policy principles: • Promote labor market flexibility, while maintaining protection for all types of labor contracts. • Seek universality in the provision of social assistance, social insurance, and basic quality services. • Expand the tax base by complementing progressive labor income taxation with capital. the taxation of ­ These principles could guide the rethinking of the social contract and fulfill the aspirations for growth and equity among the peoples of Europe and Central ­Asia. Notes 1. Calculations based on data in Milanovic ´ 2016; PovcalNet (online analysis tool), World Bank, Washington, DC, ­ http://iresearch.worldbank.org/PovcalNet/. 2. Other World Bank reports have analyzed the need to adjust the social contract in other regions and have also provided evidence on the changing nature of intergenerational mobility (Ferreira et ­ al. 2013; Narayan et a ­ 015). The challenges of ­ l. 2018; World Bank 2 new distributional tensions seem even bigger in Europe and Central Asia given the limited tolerance for inequality in this ­ region. Ridao-Cano and Bodewig (2018) analyze the impact of emerging inequalities on economic growth in the European Union ­ (EU). The current report focuses on additional distributional tensions and challenges facing taxation and social protection ­ systems. 3. Freeman (2007, 55), writing about the effect of this doubling on the United States, asserts that it “presents the U ­ epression.” ­ .S. economy with its greatest challenge since the Great D He adds that, “if the country does not adjust well, the next several decades will exacerbate economic d globalization.” ­ ivisions . . . and risk turning much of the country against ­ Overview ●  19 4. Recent studies document this phenomenon in the United ­ States. For example, Chetty et ­al. (2016) show that intergenerational mobility, a special case of equality of opportunity, ­ ecades. For a recent global perspective, see Narayan has fallen dramatically in the last few d et ­ (2018). Also see ­ al. ­ EqualChances.org (database), World Bank, Washington, DC, ­ http:// www.equalchances.org/. The database is the first online repository of internationally comparable information on inequality of opportunity and socioeconomic ­ mobility. 5. For more on the definition of the middle-class income thresholds in terms of vulnerabil- ity, see López-Calva and Ortiz-Juárez ­ (2014). 6. The Czech Republic and the Slovak Republic abandoned the scheme in 2013 after hav- ing introduced it in 2008 and 2004, ­ respectively. 7. The estimates refer to ­ 2007–14. This period is not long, but the trend observed is in line with the trajectory observed in the longer period, for e taxation. The data are ­ xample. in ­ based on EUROMOD (Tax-Benefit Microsimulation Model for the European Union) (database), Institute for Social and Economic Research, University of Essex, Colchester, UK, h ­ ttps://www.euromod.ac.uk/using-euromod/access; EU-SILC (European Union Statistics on Income and Living Conditions) (database), Eurostat, European Commission, Luxembourg, ­http://ec.europa.eu/eurostat/web/microdata/european-union-statistic​ son-income​-and-living-conditions. References Alesina, Alberto ­ I. ­ F., Edward Glaeser, and Bruce ­ 2001. “Why Doesn’t the Sacerdote. ­ United States Have a European-Style Welfare State?” Brookings Papers on Economic 187–254. Activity 2: ­ Atkinson, Anthony ­ 2016. “How to Spread the Wealth: Practical Policies for Reducing B. ­ ­ 9–33. ­Inequality.” Foreign Affairs 95 (1): 2 Binmore, ­Ken. ­1998. Just ­Playing. ­Vol. 2 of Game Theory and the Social ­Contract. Economic Learning and Social Evolution ­ Series. Cambridge, MA: MIT ­ Press. ­ ancy. 2 Birdsall, N ­ 010. “The (Indispensable) Middle Class in Developing Countries; or, The Rich and the Rest, Not the Poor and the ­ Rest.” Working Paper 207 (March), Center for Global Development, Washington, ­ DC. Birdsall, Nancy, Carol Graham, and Stefano ­Pettinato. ­2000. “Stuck in the Tunnel: Is Globalization Muddling the Middle Class?” CSED Working Paper 14 (August), Center DC. on Social and Economic Dynamics, Brookings Institution, Washington, ­ Bussolo, Maurizio, Ada Ferrer-i-Carbonell, Anna Giolbas, and Iván Torre. 2018. “Perceptions, Reality and Demand for Redistribution.” Background paper, World Bank. Washington, DC. Torre. ­ Bussolo, Maurizio, Tulio Jappelli, Roberto Nisticò, and Iván ­ 2018. “Inequality across Generations in E­ urope.” Background paper, World Bank, Washington, ­ DC. ­ 018. “Is There a Middle- Bussolo, Maurizio, Jonathan Karver, and Luís F. López-Calva. 2 Class Crisis in Europe?” Future Development (blog), March 2­ 2 and forthcoming work- ing paper. Lebrand. ­ Bussolo, Maurizio, and Mathilde Sylvie Maria ­ 2017. “Feeling Poor, Feeling Rich, Investigation.” Working paper (May 29), World or Feeling Middle Class: An Empirical ­ Bank, Washington, D­ C. Chetty, Raj, David Grusky, Maximilian Hell, Nathaniel Hendren, Robert Manduca, and Jimmy N ­ 016. “The Fading American Dream: Trends in Absolute Income ­ arang. 2 Mobility since 1­ 940.” NBER Working Paper 22910 (December), National Bureau of Economic Research, Cambridge, M ­ A. ­ . Santos, Barbara Kits, and Isil O Dávalos, María E., Giorgia DeMarchi, Indhira V ­ 016. ­ ral. 2 “Voices of Europe and Central Asia: New Insights on Shared Prosperity and ­ Jobs.” World Bank, Washington, D ­ C. 20  ●   Toward a New Social Contract Easterly, William ­Russell. ­2001. “The Middle Class Consensus and Economic ­Development.” Journal of Economic Growth 6 (4): ­ 317–36. Farole, Thomas, Soraya Goga, and Marcel ­ Ionescu-Heroiu. ­ 2018. Rethinking Lagging Regions. World Regions: Using Cohesion Policy to Deliver on the Potential of Europe’s ­ Bank Report on the European ­Union. Washington, DC: World B ­ ank. ­ rancisco. ­ Ferreira, F H. ­G., Julián Messina, Jamele Rigolini, Luís F. López-Calva, María Ana Lugo, and Renos V ­ 013. Economic Mobility and the Rise of the Latin American ­ akis. 2 Middle ­Class. Washington, DC: World B ­ ank. Freeman, Richard B­.­2007. “The Great Doubling: The Challenge of the New Global Labor ­Market.” In Ending Poverty in America: How to Restore the American Dream, edited by John Edwards, Marion Crain, and Arne ­ ­ 5–64. New York: New P L. Kalleberg, 5 ­ ress. ­ reisman. 2 Gimpelson, Vladimir, and Daniel T ­nequality.” Economics ­ 018. “Misperceiving I 27–54. and Politics 30 (1): ­ Hobbes, T ­ 012. L ­ homas. 2 ­ eviathan. Edited by Noel M­ alcolm. Clarendon Edition of the Works of Thomas Hobbes S Press. First published ­ eries. New York: Oxford University ­ ­1651. Progress). ­ IPSP (International Panel on Social ­ 2018. Rethinking Society for the 21st Century: Report of the International Panel on Social ­ vols. Cambridge, UK: Cambridge Progress. 3 ­ University ­Press. Ravi. ­ Kanbur, ­ “Comments.” In Economic Policy and Equity, edited by Vito Tanzi, 1999. ­ ­ 39–42. Washington, DC: International Monetary Ke-young Chu, and Sanjeev Gupta, 2 ­Fund. Locke, ­John. ­1988. Two Treatises of Government, 3rd ­ ed. Edited by Peter ­Laslett. ­ eries. New York: Cambridge Cambridge Texts in the History of Political Thought S Press. First published 1 University ­ ­ 689. López-Calva, Luís F­ ., and Eduardo O ­ 014. “A Vulnerability Approach to the ­ rtiz-Juárez. 2 Definition of the Middle ­Class.” Journal of Economic Inequality 12 (1): ­23–47. ´, ­Branko. ­2016. Global Inequality: A New Approach for the Age of G Milanovic ­ lobalization. Cambridge, MA: Harvard University ­ Press. ———. ­2017. “The Welfare State in the Age of G ­ lobalization.” globalinequality (blog), March ­26. ­http://glineq.blogspot.com/2017/03/the-welfare-state-in-age-of.html. Narayan, Ambar, Roy van der Weide, Alexandru Cojocaru, Christoph Lakner, Silvia Redaelli, Daniel Gerszon Mahler, Rakesh Gupta N ­ hewissen. ­ ­ . Ramasubbaiah, and Stefan T 2018. Fair Progress? Economic Mobility across Generations around the ­World. Washington, DC: World B­ ank. ­ evelopment). 2 OECD (Organisation for Economic Co-operation and D ­ 017. Preventing OECD. Ageing ­Unequally. Paris: ­ Century. Cambridge, MA: Belknap ­ Piketty, ­Thomas. ­2014. Capital in the Twenty-First ­ Press. Bodewig. ­ Ridao-Cano, Cristobal, and Christian ­ 2018. Growing United: Upgrading Europe’s Convergence ­Machine. Washington, DC: World B ­ ank. Rodrik, Dani. 1999. “Where Did All the Growth Go? External Shocks, Social Conflict, and Growth Collapses.” Journal of Economic Growth 4(4): 385–412. Rousseau, ­Jean-Jacques. ­1968. The Social ­Contract. Translated by Maurice ­ Cranston. Penguin Books for Philosophy S Books. First published ­ ­ eries. London: Penguin ­ 1762. ­nequality. New York: Tanzi, ­Vito. ­2018. Termites of the State: Why Complexity Leads to I Cambridge University ­ Press. ­ ank. 2 World B ­ ontract.” MENA Economic Monitor (April), ­ 015. “Towards a New Social C DC. Middle East and North Africa Region, World Bank, Washington, ­ 1 Introduction Emerging Distributional Tensions in Europe and Central Asia Globalization and technological change are altering the day-to-day lives of people across Europe and Central Asia. A large share of jobs is susceptible to automation in the region, including, for example, nearly 40 percent of jobs in Georgia and Tajikistan and around 60 percent in Croatia and Latvia (World Bank 2016). Different jobs requiring different skills are being created. Increasingly, workers are entering into nonstandard labor contracts or finding opportunities to become self-employed. Changing opportunities generated by markets can lead to new distributional tensions. Those people who are ready to take advantage of emerging opportuni- ties because of their skills, access to markets and digital technologies, and location will gain from these changes, while others may have fewer economic opportunities within reach. These dynamics create new divides that can affect social cohesion by making inequalities more salient across groups described by different skills and occupations, of different generations, living in different regions, or born to parents of different educational backgrounds. Public policies play an important role in taming market-generated inequali- ties in Europe and Central Asia. Available evidence shows that the lower levels of inequality in the region relative to other regions around the world derive mostly from public redistribution systems. A look at market incomes shows that the countries of Europe and Central Asia exhibit inequality gaps similar to 21 22  ●   Toward a New Social Contract those in Latin America, where the redistributive effect of taxes and transfers is more limited (Lustig 2017). In Europe and Central Asia, nonmarket income ­represents a substantial part of total household income, particularly among the poorest 40 percent of the population (the bottom 40) (Bussolo and López-Calva 2014). The Potential Implications for the Social Contract In a changing economic context that might be giving way to increasing distribu- tional tensions, public policies may no longer be equipped to respond in line with people’s preferences and perceptions about equality. Europe and Central Asia possesses some of the oldest and most developed models of the welfare state in the world. The preferences in the region seem to value economic security. Evidence from the eastern part of the region, for example, shows that people aspire to the stability provided by a full-time job, strongly associated with public sector jobs (Dávalos et al. 2016). These preferences seem to be at odds with the changing nature of employment. Moreover, many welfare state mechanisms were not designed to operate in the dynamic context of today’s world nor to deal with potential distributional tensions that may be emerging. For instance, if more jobs are in temporary contracts or of shorter tenure than in the past, insurance systems tied to standard open-ended employment contracts might no longer be adequate. The result is that people across countries perceive that the playing field in access to economic opportunities is increasingly less level (EBRD 2018). Voice is being given more frequently to the word “unfairness,” even if inequality across individu- als and households is less in the region than in other parts of the world (Dávalos et al. 2016). Europe and Central Asia accounts for 23 of the 30 countries with the lowest Gini coefficient among 158 countries across the world (figure 1.1). Underlying all stable societies is some form of social contract, an implicit agree- ment among the members of a society. This agreement is backed by institutional arrangements that influence how markets function, how responsibilities and ben- efits are defined, and how resources are redistributed (World Bank 2017). This report defines a stable social contract as one in which, in the context of a given distribution of resources generated by market forces, the way that distribution is affected by public policies (through the fiscal system, but also through rules and institutions that affect the functioning of the markets) is aligned with people’s per- ceptions and preferences, a combination of people’s beliefs and social values and norms (figure 1.2).1 The elements that bring the social contract into equilibrium— the role of the market, public policies, and people’s preferences—can be vastly different across countries. A mismatch between what the market delivers and what people expect and value can tear the social contract. Despite different economic, social, and political paths, particularly between Western Europe and the countries of Eastern Europe and Central Asia, a drive for equity has underpinned the social contract in the region. The formation of the social contract in Europe and Central Asia has a long history, beginning with the Elizabethan Poor Relief Act of 1601 that created a Poor Law system in England and Wales. The major foundations for social contracts that prevail in most countries in Introduction ●  23 FIGURE 1.1 Income inequality is lower in Europe and Central Asia than in most of the rest of the world Europe and Central Asia Gini index of inequality, 2014 or latest year available South Asia Middle East and North Africa East Asia and Pacific Africa Latin America and the Caribbean 20 40 60 80 Gini coefficient Europe and Central Asia Rest of the world Regional average Sources: Calculations based on data in Milanovic 2016; PovcalNet (online analysis tool), World Bank, Washington, DC, http://iresearch.worldbank.org/PovcalNet/. Note: The Gini index has been calculated on income whenever possible; alternatively, consumption is used. Rest of the world = all countries, including high-income countries. Europe and Central Asia includes Western Europe. Regional averages are unweighted averages of individual countries. FIGURE 1.2 The social contract as a dynamic equilibrium Market-generated Perceptions and distribution of resources societal preferences Public policies Europe and Central Asia developed following World War II, with a strong compo- nent of equity. Countries behind the iron curtain had a system based on state ownership of many of the means of production, and state enterprises were the main suppliers of welfare services to the population. Western European economies have 24  ●   Toward a New Social Contract supported economic and social interventions to provide social justice within the framework of a capitalist economy and representative democracy. The social contracts in Europe and Central Asia have faced serious challenges in recent decades. In Western Europe, the end of full employment in the 1970s led to changes and some retrenchment in the welfare state. The fall of the Berlin Wall and the varying speed in the transition experienced by the formerly centrally planned economies also drove a transformation in welfare systems; the social contract in some transition countries remains strongly influenced by the legacy of the arrangements under the Soviet system. Nonetheless, a preference for the fair distribution of incomes is at the core of the social contract and of societies in Europe and Central Asia (Alesina and Glaeser 2004; Alesina, Glaeser, and Sacerdote 2001). Is a Rethinking of the Social Contract in the Region Warranted? The report uses the three parts of a stable social contract—the distribution of resources by the market, public policies, and social preferences—as an organizing principle in exploring the equity dimension of social contracts in the region. Chapter 2 focuses on market forces and analyzes four key areas of emerging distributional tension (figure 1.3). First, the chapter examines whether the labor FIGURE 1.3 Distributional tensions along four dimensions are explored Between occupations Economic insecurity Inequality of and unfairness Between and opportunity Crisis of the within cohorts middle class Between geographic areas Introduction ●  25 market engagement of certain occupations is changing more relative to other occupations (the distributional tensions between workers in different occupations); ­ specifically, it looks at how technological progress is affecting the demand for dif- ferent skills, altering the occupational structure of employment, and impacting the earnings and economic security of certain groups. Second, the chapter explores whether economic and institutional forces (for example, pension arrangements or the availability of new types of jobs) are reducing the earnings prospects of some cohorts compared with others (younger and older generations) and increasing the dispersion of incomes within the same cohort, potentially opening a generational divide (distributional tensions between and within cohorts). Third, the chapter examines whether people in certain geographical areas have more limited access to opportunities, including in building up their productivity through access to ser- ­ pecifically, it looks at whether spatial inequalities exist, and whether they vices; s are persisting or widening (distributional tensions between geographic areas). Fourth, the chapter explores whether the share of total inequality that can be attributed to circumstances outside an individual’s control is rising and how this affects the ability of workers to benefit from economic opportunities (distributional tensions from inequality of opportunity). Chapter 2 also considers how these four areas of distributional tension are related to a crisis of the middle class that has become more manifest in growing economic insecurity than in an actual reduction in the share of populations in the middle of the income distribution. Chapter 3 examines the second element of the social contract and discusses the tax and benefit systems and labor market policies in the region. An array of policy instruments is assessed in the report, with the objective of understanding whether they are able to respond to emerging inequalities and contribute to eco- nomic security and equal access to opportunity for all. Performance and chal- lenges in three policy areas that are particularly relevant to the distributional tensions are investigated. First is the evolution of labor market regulations, inter- ventions, and institutions and their role in managing or influencing distributional tensions among workers. Second, what has been the impact of taxes and transfers in reducing inequality (not only across income groups, but also, for example, between groups of workers in different occupations) and in compensating for the growing vulnerabilities among some groups. Third, what has been the role of poli- cies in facilitating labor mobility within countries. The report explores whether housing policies, for instance, are having an impact on labor mobility and thus access to economic opportunities among some groups, particularly groups in remote rural areas.2 Following the discussion of market and policy trends in the previous chapters, chapter 4 explores the third element of the social contract, people’s perceptions of recent changes and societal preferences for equity. It looks at how the prefer- ences of individuals are shaped and, more generally, the type of society in which they would like to live. An important dimension is introduced: people’s percep- tions of inequality and how these are associated with demands for redistribution. The chapter brings together the three elements of the social contract: market- driven inequality, public redistribution, and societal preferences for equity and shows that imbalances in the social contract may be emerging. Signs of these imbalances include polarization in voting and in the negative trends in the trust 26  ●   Toward a New Social Contract exhibited by individuals in public institutions. Voting for populist parties, lower turnouts in elections, and a substantial increase in distrust are strongly associated with those people who are on the losing end of distributional tensions, such as workers in more precarious positions, individuals in young cohorts, or people who have been cut off from economic opportunity. Chapter 5 presents a rethinking of public policies to tackle the distributional tensions and strengthen the social contract. Although the policy agenda for strengthening the social contract is country specific in that it aligns with each coun- try’s markets, preferences, and policies, general policy principles are set forth that may contribute to the debate on a new social contract in the countries of Europe and Central Asia. These principles are focused on the balance between labor mar- ket flexibility and protection, the role of social assistance and social insurance in providing economic security, the provision of good-quality services, and tax poli- cies that foster more equitable fiscal systems. Notes 1. For a treatment of the social contract as a bargaining process between conflicting indi- viduals and interests, see Binmore (1994, 1998). The text view of the social contract is related to the analytical approach proposed by Kanbur (1999) in the context of optimal taxation. Kanbur (1999) sets out a framework for thinking about optimal taxation that includes (a) the degree of inherent inequality as reflected in the extent of productivity distribution; (b) the incentive effects, captured in that model by the elasticity of labor supply; and (c) the degree of egalitarianism, captured by the inequality aversion parameter. 2. A somewhat different categorization of policies that affect the distribution of income is proposed by the International Panel on Social Progress (IPSP 2018), which distinguishes among premarket (for example, the provision of education), in-market (for instance, labor market regulation), and postmarket interventions (such as taxes and transfers). References Alesina, Alberto F., and Edward L. Glaeser. 2004. Fighting Poverty in the US and Europe: A World of Difference. Oxford, UK: Oxford University Press. Alesina, Alberto F., Edward L. Glaeser, and Bruce I. Sacerdote. 2001. “Why Doesn’t the United States Have a European-Style Welfare State?” Brookings Papers on Economic Activity 2: 187–277. Binmore, Ken. 1994. Playing Fair. Vol. 1 of Game Theory and the Social Contract. Economic Learning and Social Evolution Series. Cambridge, MA: MIT Press. ———. 1998. Just Playing. Vol. 2 of Game Theory and the Social Contract. Economic Learning and Social Evolution Series. Cambridge, MA: MIT Press. Bussolo, Maurizio, and Luis F. López-Calva. 2014. Shared Prosperity: Paving the Way in Europe and Central Asia. Europe and Central Asia Studies. Washington, DC: World Bank. Dávalos, María Eugenia, Giorgia DeMarchi, Indhira V. Santos, Barbara Kits, and Isil Oral. 2016. “Voices of Europe and Central Asia: New Insights on Shared Prosperity and Jobs.” World Bank, Washington, DC. Introduction ●  27 EBRD (European Bank for Reconstruction and Development and World Bank). 2018. Transition Report 2017–18: Sustaining Growth. London: EBRD. IPSP (International Panel on Social Progress). 2018. Rethinking Society for the 21st Century: Report of the International Panel on Social Progress. 3 vols. Cambridge, UK: Cambridge University Press. Kanbur, Ravi. 1999. “Comments.” In Economic Policy and Equity, edited by Vito Tanzi, Ke-young Chu, and Sanjeev Gupta, 239–42. Washington, DC: International Monetary Fund. Lustig, Nora C. 2017. “The Impact of the Tax System and Social Expenditure on the Distribution of Income and Poverty in Latin America: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Peru, Dominican Republic, Uruguay, and Venezuela.” CGD Working Paper 450 (March), Center for Global Development, Washington, DC. Milanovic, Branko. 2016. Global Inequality: A New Approach for the Age of Globalization. Cambridge, MA: Belknap Press: An Imprint of Harvard University Press. World Bank. 2016. World Development Report 2016: Digital Dividends. Washington, DC: World Bank. ———. 2017. “Leveling the Playing Field: Rethinking the Social Contract in Europe and Central Asia (ECA).” ECA Concept Note (March), World Bank, Washington, DC. 2 Are Distributional Tensions Brewing in Europe and Central Asia? This chapter first briefly assesses the inequality in the distribution of income and wealth across individuals. A rising inequality and a concentration of income and wealth for the top 1 percent of the population are important, especially when more economic power is accompanied by more political power, but those are also, by now, quite well-studied issues. In fact, focusing on this type of (vertical) inequality alone misses a lot. In Europe and Central Asia, inequality between groups (hori- zontal inequality) is growing. This, together with heightened economic insecurity and, at times, a more unfair process generating the distribution of incomes, is behind the brewing distributional tensions in the region. The chapter identifies and considers in detail four key distributional tensions. First, in some countries, technological progress is boosting the returns to higher- level skills, but also driving complex changes in the distribution of skills. The polar- ization of occupations is forcing a large group of middle-skilled-level workers employed in occupations intensive in routine tasks into lower-skill jobs, thereby reducing economic security and the incomes of low-skilled workers. Second, eco- nomic and institutional forces (such as globalization, labor market regulations, and pension arrangements) are reducing the earnings prospects of young workers relative to older workers and retirees and increasing inequality among the young. These forces are driving a generational divide because many young workers may not be able to join or maintain themselves in the middle class. Third, spatial inequalities in Europe and Central Asia reduce the economic opportunities of groups living in lagging subregions and underserved rural areas. Fourth, inequality of opportunity—the share of total inequality that can be attributed to 29 30  ●   Toward a New Social Contract circumstances outside an individual’s control—still represents an important share of total inequality. This share is rising in some countries in the region, reducing worker social mobility and adding to a perception of unfairness. These four distributional tensions—among workers, across generations, among geographical areas, and in opportunity—are not necessarily highlighted when ver- tical inequality, or disparities across individual incomes, is considered. However, these tensions are relevant for the stability of the social contract. If individuals belonging to groups that are losing, or that are exposed to more intense economic insecurity, cannot move to other groups, then social cohesion suffers. The chapter concludes with an analysis of the situation of the middle class, a class that repre- sents the bulk of Europe’s population. The expansion of the middle class has slowed in most countries and even reversed in some. Even more importantly, however, decreasing economic security, rather than a shrinking population share, characterizes the “crisis” of the middle class. Many people still have middle-class incomes, but they are, or feel, vulnerable to fall into poverty, as their incomes are less stable and less secure. The analysis shows these vulnerable people belong to the groups that are on the losing side of the distributional tensions: younger cohorts, workers in middle-skilled occupations, residents in lagging regions. A society with a shrinking or vulnerable middle class is a symptom of a more polarized society and, in turn, a society with a lower support for the current social contract. Inequality Across Individuals in Europe and Central Asia Inequality has increased worldwide in recent decades. In most countries, the inequality gap is widest today relative to the last 30 years. In the countries of the Organisation for Economic Co-operation and Development (OECD), the richest 10 percent of the population earn 9.6 times the income of the poorest 10 percent, a significant increase since the 1980s when the ratio was 7 to 1. In emerging coun- tries, the income gaps are even greater. The Europe and Central Asia region is no exception. Income inequality in the region has risen during the last 25 years, although the trends and magnitudes vary across the region.1 Income inequality in the European Union (EU) widened during the 1990s, but has been relatively stable since then. Figure 2.1 illustrates trends in the average Gini index of per capita household income in various regions of the current EU. In 1988–98, income inequality widened across all regions. The biggest increases occurred in the Baltic States and Central Europe (with an average increase of 12 and 8 Gini index points, respectively). This change, associated with the transition from a planned to a market economy, was driven mostly by increased inequality in labor income.2 Inequality also increased in continental Europe and Northern Europe, though the pattern was heterogeneous in the latter. Inequality remained roughly stable in Southern Europe in 1988–98 with the exception of Italy, where it increased. Income inequality presented a slight U shape in the case of the Baltic States.3 The following period shows a more mixed pattern, with a stable trend in Are Distributional Tensions Brewing in Europe and Central Asia? ●  31 FIGURE 2.1 40 Trends in income inequality, per capita household income 35 European Union, 1988–2015 Average Gini index 30 25 20 15 1988 1993 1998 2003 2008 2013 2015 Year Baltics—WYD/Indie Baltics—EU–SILC Central Europe—WYD/Indie Central Europe—EU–SILC Continental Europe—WYD/Indie Continental Europe—EU–SILC Northern Europe—WYD/Indie Northern Europe—EU–SILC Southern Europe—WYD/Indie Southern Europe—EU–SILC Sources: Data for 1988–98: independent databases and WYD (World Income Distribution Dataset), Stone Center on Socio-Economic Inequality, Graduate Center, City University of New York, New York, https://www.gc.cuny.edu/Page-Elements/Academics-Research-Centers-Initiatives/Centers-and​ -Institutes/Stone-Center-on-Socio-Economic-Inequality/Core-Faculty,-Team,-and-Affiliated-LIS-Scholars​ /­Branko-Milanovic/Datasets. Data for 2003–15: EU-SILC (European Union Statistics on Income and Living Conditions) (database), Eurostat, European Commission, Luxembourg, http://ec.europa.eu​ /­eurostat/web/microdata/european-union-statistics-on-income-and-living-conditions. Note: The independent databases represent a collection of household surveys gathered by academic scholars for the study of income and wealth inequality. EU-SILC is a harmonized EU–based survey carried out annually since 2003 in most EU countries; Central Europe = Bulgaria, Croatia, Czech Republic, Hungary, Poland, Romania, Slovak Republic, and Slovenia; Northern Europe = Denmark, Finland, Norway, and Sweden; the Baltic States = Estonia, Latvia, and Lithuania; Continental Europe = Austria, Belgium, France, Germany, Ireland, Netherlands, Switzerland, and United Kingdom; Southern Europe = Cyprus, Greece, Italy, Malta, Portugal, and Spain. 2003–08, with the exception of the Baltic States, in which inequality narrowed, and a mild increase in inequality during the last recession (2008–13) only in Southern and Central Europe. Inequality in the economies of the former Soviet Union, Turkey, and the Western Balkans has narrowed in the last decade after increasing during the transition. Data on household incomes from the initial years of transition in the former Soviet Union economies are scarce and, because of dramatic changes in relative prices, unreli- able as a means of measuring household welfare. For this reason, estimates on inequality rely on household consumption information. Figure 2.2 shows that Central Asia saw a strong decrease in inequality between 1993 and 2003, although this is possibly a rebound from the transition period, on which there is limited data. Belarus, Moldova, the Russian Federation, and Ukraine saw a decline in the same period, particularly because of the strong performance of Russia. In the Western Balkans, which is only fully observed after 2003, inequality has shown a slight decrease. In the case of the South Caucasus, an initial decrease between 1998 and 2003 was later reversed, with an increase until 2013. Turkey experienced a long- term decline in inequality, albeit with relative stagnation during the 1990s and a slight increase in 2008–13. 32  ●   Toward a New Social Contract FIGURE 2.2 45 per capita household consumption Trends in consumption inequality, former Soviet 40 Average Gini index Union economies, Turkey, and Western Balkans, 35 1988–2013 30 25 20 1988 1993 1998 2003 2008 2013 Year BMU and Russian Federation––Povcal BMU and Russian Federation—ECAPOV Central Asia—Povcal Central Asia—ECAPOV South Caucasus—Povcal South Caucasus—ECAPOV Western Balkans—Povcal Western Balkans—ECAPOV Turkey—Povcal Turkey—ECAPOV Sources: Household consumption, 2003–13: ECAPOV database harmonization as of April 2018, Europe and Central Asia Team for Statistical Development, World Bank, Washington, DC; 1988–98: PovcalNet (online analysis tool), World Bank, Washington, DC, http://iresearch.worldbank.org/PovcalNet/. Note: Western Balkans = Albania, Bosnia and Herzegovina, the former Yugoslav Republic of Macedonia, Montenegro, and Serbia; Central Asia = Kazakhstan, Kyrgyz Republic, Tajikistan, and Uzbekistan; South Caucasus = Armenia, Azerbaijan, and Georgia; BMU and Russia = Belarus, Moldova, Russian Federation, and Ukraine. FIGURE 2.3 20 Gini index adjusted for the top incomes, 2011 Increase in inequality, Gini points 15 10 5 0 S nia Sl pain ng a hu a Ne Port nia La ds ec Es tvia ep nia Bu ublic Ire ria mb ria Fin ce Ice den ite F prus ak ng e Re dom um rm g Be blic Cr ary Po ny It k Sw or aly er y er l Swland nm d Cy nd Lu Auand Gr and Hu eni Lit oati ov Ki nc th uga Ge our ar itz wa De lan lan ee lga xe st a ma a h R to lgi la Sl d ra pu e ov l l Ro N Cz Un Country Source: Calculations based on Hlasny and Verme 2018. This analysis, based on household surveys, may understate the level of inequal- ity or overstate the recent decline in inequality because surveys typically do not include the income of the richest individuals in the region.4 Increasing attention is being paid in the literature, in the media, and in many forums around the globe to income concentration among the richest. Indeed, accounting for top incomes in traditional household survey–based measures of inequality such as the Gini index reveal higher income inequality (figure 2.3). Are Distributional Tensions Brewing in Europe and Central Asia? ●  33 Available evidence indicates that concentration of income is growing in many countries in the region.5 Among the countries on which data are available, the income share of the top 1 percent of the population is highest in Russia and Turkey and has increased in many countries (table 2.1). The World Inequality Database provides estimates for the top 1 percent income share in 17 countries of Europe and Central Asia, mainly Western European countries, but including Hungary, TABLE 2.1  Top 1 Percent Income Shares Vary Across the Region, but Have Risen in Many Countries a. Top 1 percent income shares, latest available estimates Year Top 1% income share Netherlands 2012 6.3% Low Denmark 2010 6.4% Finland 2009 7.5% Norway 2011 7.8% Spain 2012 8.6% Sweden 2013 8.7% Medium Low Italy 2009 9.4% Hungary 2008 9.6% Portugal 2005 9.8% Ireland 2009 10.5% France 2014 11.0% Medium High Germany 2011 13.0% Poland 2015 13.3% United Kingdom 2014 13.9% High Russian Federation 2015 20.2% Turkey 2016 23.4% b. Average percentage point change in top 1 percent income shares, circa 1980–2014 Change in top 1% income share Netherlands 0.48 Denmark 0.94 Spain 0.95 France 2.00 Germany 2.26 Italy 2.48 Finland 3.14 Norway 3.20 Ireland 3.85 Sweden 4.60 Portugal 5.45 Hungary 7.01 United Kingdom 7.21 Poland 9.14 Russian Federation 16.78 Source: Based on data of WID (World Inequality Database), Paris School of Economics, Paris, https:// wid.world/. Note: Panel a: the population is comprised of individuals more than 20 years old. The base unit is the tax unit defined by national fiscal administrations for the measurement of personal income taxes. It may refer to individuals or be equally split across adults according to the available data in each country. Panel b: Denmark (1980–2010), Finland (1980–2009), France (1980–2014), Germany (1980–2011), Hungary (1980–2008), Ireland (1980–2009), Italy (1980–2009), Netherlands (1981–2012), Norway (1980–2011), Poland (1983–2015), Portugal (1980–2005), Russian Federation (1980–2015), Spain (1981–2012), Sweden (1980–2013), and United Kingdom (1981–2014). 34  ●   Toward a New Social Contract Poland, Russia, and Turkey.6 The top 1 percent income share has increased sub- stantially in countries such as Poland and Russia and stands at around 20 percent in Russia and 23 percent in Turkey. This big expansion in the share of total income captured by top earners is related to the fact that, while, for the vast majority of individuals, wages are by far the largest component of income, a distinguishing feature of the incomes of top earners is the share of capital income. This share has grown in the last two decades. For instance, in France, the top 0.01 percent receives about 20 percent of their income from capital (OECD 2014). This suggests that income inequality may be related to inequality in capital holdings, that is, wealth inequality. Wealth has also become more concentrated. Data on wealth are scarce, and the coverage of the World Inequality Database, the source of table 2.1, is limited. The Forbes list of billionaires indicates that the number of billionaires in Europe and Central Asia rose from 106 in 1996 to more than 500 in 2017 (figure 2.4). The number of bil- lionaires in Western Europe increased from 90 in 1996 to 379 in 2017, and the number of Russian billionaires rose from 8 in 2001 to 96 in 2017. While 62 percent of the region’s billionaires are concentrated in Western Europe, the billionaires in Eastern Europe are far richer: three to four times richer in the case of Russia because of high growth rates in net worth over the past decade and two to three times richer in the case of billionaires in other Eastern European countries. The sources of wealth vary across the region. There is a concentration in mining among Russian billionaires, the financial sector in other Eastern European countries, and manufacturing in Western Europe. Labor income is losing ground as a share of total income. The labor share of income declined in Western Europe particularly during the 1980s, though it remained relatively stable thereafter. However, since the mid-1990s, a growing number of countries in the region, especially those that transitioned out of a planned economy, have witnessed declining labor income shares (figure 2.5). The largest decline in 1994–2014 occurred in Azerbaijan (−34 percentage points), followed by the former Yugoslav Republic of Macedonia and Serbia (−19 percentage points), Armenia and Tajikistan (−16 percentage points), Estonia (−12 percentage points), Luxembourg and Turkey (−11 percentage points), and Kazakhstan (−10 points). The current literature has been exploring various hypotheses to explain this, from advances in information technology and the decline of the relative price of investment goods, to automation, finan- cial deregulation, and an increase in industry concentration, which allows incumbent superstar firms to exploit monopolistic rents (Autor et al. 2017; Eden and Gaggl 2015; Karabarbounis and Neiman 2014; and Stiglitz 2012). The greater concentration of wealth and the decline in the share of labor income may result in widening income inequality. Piketty (2014) argues that the main driver of inequality is the tendency of returns on capital to exceed the rate of economic growth. As an economy expands, a larger share of gross domestic product (GDP) is represented by profits, while the share of GDP accounted for by worker wages shrinks. This may explain the parallel trends in wealth and income inequality. Are Distributional Tensions Brewing in Europe and Central Asia? ●  35 FIGURE 2.4 a. Number of billionaires 400 The number of billionaires and their net worth have 350 increased 300 Number of billionaires 250 200 150 100 50 0 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 Year b. Net worth as a share of GDP and country and group average 35 30 Average net worth as share of GDP (%) 25 20 15 10 5 0 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 Year East Russian Federation West Source: Based on “The World’s Billionaires,” Forbes Media, Jersey City, NJ, https://www.forbes.com​ /billionaires/list/#version:static. Note: West = Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Lithuania, Netherlands, Norway, Poland, Portugal, Romania, Serbia, Spain, Sweden, and United Kingdom; East = Georgia, Kazakhstan, Turkey, and Ukraine. Panel b: east and west data reflect the group-level average of the country-level average of net worth as a share of GDP and can therefore be interpreted as the net worth of the average billionaire as a share of GDP in each group of countries. GDP = gross domestic product. 36  ●   Toward a New Social Contract FIGURE 2.5 70 The declining share of labor income, particularly in transition economies 65 Labor share on income (%) 60 55 50 1970 1980 1990 2000 2010 Year Western Europe Transition countries Source: Based on data of Karabarbounis and Neiman 2014. Note: Western Europe = Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and United Kingdom; Transition countries = Central and Eastern Europe, the Baltic States, the Western Balkans, and the remaining economies of the former Soviet Union. Turkey, though not strictly a transition country, is also included in this last group. While the previously mentioned measures of inequality across incomes of indi- viduals reveal no reason for alarm in Europe and Central Asia, the adjustment showing the share of top incomes is more in line with people’s perceptions that inequality is widening (chapter 4).7 Moreover, incomes may be clustering at the extremes of the distribution and, while vertical inequality may narrow, polarization may become more severe (Zhang and Kanbur 2001). So, a look at the polarization of distribution or, equivalently, at what is happening at the middle of the distribution may reveal that clustering is a relevant distributional shift. While inequality across individuals or polarization are indeed important, most of the focus of this report is the distributional tensions that may arise because of between-group inequality, that is, horizontal inequality (box 2.1). This perspective is critical to social stability and, as highlighted by Stewart (2002, 4), is also a “precondi- tion of economic development.” This is because individuals within disadvantaged groups may not be able to contribute to society’s prosperity if between-group inequality persists. For example, the children of parents in disadvantaged groups may not be able to accumulate sufficient human capital or, even if they can do so, may not have access to the most productive occupations. Inequality across occupa- tion groups, generations, and regions and inequality of opportunity are the focus of this chapter. Are Distributional Tensions Brewing in Europe and Central Asia? ●  37 BOX 2.1 Horizontal Inequality Horizontal inequality refers to inequality between The potential for economic policy to address hori- groups. These groups may be culturally defined (or zontal inequality depends, in part, on the group constructed), such as racial, ethnic, or religious involved. For example, governments can have a direct groups; they may be defined by situation, such as impact on the relative incomes across groups through regional location or age; or they may be defined policies on social protection, and governments can based on economic criteria, such as occupation. address regional inequality through public investment This is in contrast with vertical inequality, which is and service provision. By contrast, improving the posi- inequality across all households or individuals. tion of racial or ethnic groups may require changes Vertical inequality can be decomposed into between- in norms, culture, and social relations to reduce group inequality (horizontal inequality) and within- boundaries between groups (for example, through group inequality. Within-group inequality typically intermarriage) or encourage greater acceptance of accounts for a larger share of vertical inequality, but underprivileged groups. horizontal inequality may nonetheless be substantial. Government policies can support or hinder such Inequality can be measured along many dimen- changes. Policies that prohibit discrimination against sions, for example income, assets, life expectancy, disadvantaged groups in the accumulation of assets health, educational attainment, opportunities for (for example, in lending) or human capital (access to political participation, or access to public services. education) can eventually narrow horizontal inequal- The classifications used to define groups should be ity. Such policies may need to be supported by affir- meaningful and recognizable to the individuals so mative action in favor of underprivileged groups to grouped and to society at large, and rapid move- counteract the effects of past discrimination. For ment from one such group to another should be example, educational admission policies based on difficult for individuals. competitive examinations will discriminate against Horizontal inequality often endures, even for young people from less well-educated backgrounds centuries, for example, inequality between blacks or those that do not speak the dominant language. and whites in the United States. But horizontal However, there is a risk that affirmative action may inequality may also disappear within a few gener- add to the boundaries between groups, thereby dis- ations because, for instance, of the upward mobil- couraging the social acceptance of some groups. ity of immigrant groups in some countries. The Comprehensive programs may be necessary welfare cost of horizontal inequality tends to be because deprived groups often suffer from priva- high because people become trapped in tions across several dimensions. Thus, improving deprivation. secondary schools that serve underprivileged This contrasts with vertical inequality, which groups may be of little help if households cannot may, at least partially, be characterized by churn- survive without the full-time labor of their older chil- ing, as households rise into or fall from the upper dren. Raising contacts with more privileged groups segments of the income distribution over time. in an equal setting—for example, avoiding segre- Horizontal inequality also matters because the gated schooling—can improve the opportunities of members of the groups may identify with their the less privileged. Strong group organizations can groups, and the well-being of the group affects their promote self-respect, provide mutual insurance, and identity. Likewise discrimination based on group strengthen the bargaining position of the group, but identity can lower efficiency, and persistent horizon- may also reduce beneficial contacts and, similar to tal inequality can undermine political stability. affirmative action, may highlight group divisions. Source: Stewart and Langer 2007. 38  ●   Toward a New Social Contract Labor Market Polarization and the Shifting Demand for Skills Technological change, globalization, and institutional or policy changes are trans- forming labor markets in Europe and Central Asia.8 Advances in technology, including the Internet and various forms of mechanization, are reducing the demand for work that is intensive in routine tasks (Autor and Dorn 2013; Autor, Katz, and Kearney 2006, 2008; Autor, Levy, and Murnane 2003; Goos and Manning 2007; Goos, Manning, and Salomons 2014). For example, airline ticket agents are being replaced by websites, and assembly line workers are being replaced by robots. The rise of developing countries in the global economy has intensified the competition facing low- and mid-skilled workers in advanced countries who have traditionally relied on now-declining manufacturing industries. Many countries have eased regulatory restrictions on firing and reduced other worker protections to facilitate adjustment to shocks by making markets more flexible. However, in many cases, this has created dual labor markets, whereby only skilled and experi- enced workers continue to benefit from high levels of protection, while the rest face more vulnerable employment conditions (chapter 3). Many of the displaced jobs tend to be in medium-earning occupations, generating a shift into low-paid and, to a lesser extent, high-paid occupations. This polarization of occupations has implications for the distribution of earnings that may not be reflected in a rise in the Gini coefficient, but could still cause rifts. Job polarization, understood as the simultaneous growth of occupations at the extremes of the wage spectrum and the hollowing out of mid-skill jobs, has driven significant changes in the distribution of wage income. A worker’s occupation— not simply the skill level—has become an important determinant of labor market earnings in the United States in recent decades (Acemoglu and Autor 2011). What is the situation in Europe and Central Asia? Do countries in the region experience occupational and earnings polarization? The next two subsections address these issues, considering, first, occupational changes in the region and then diving deeper in country case studies. For these studies, a formal decomposition analysis aims at determining the importance of occupational changes in changes in the earnings distribution and, ultimately, income distribution. A brief outline of the main results paves the way for the detailed analysis. Results: polarization in occupations. From the fall of the Berlin Wall to the early 2010s, the western part of the region experienced job polarization. Jobs in the middle of the distribution, intensive in routine tasks, are becoming less available. In the eastern part of the region and from about 2000 to 2010, the occupational transformation was more mixed. The share of jobs at the low end and in the middle of the distribution expanded. Routine biased technological change, auto- mation, appears not to be widespread in the eastern part of the region, where occupational changes are more closely linked with the formalization of jobs (chapter 3). Especially in the lower half of the distribution, unpaid family work or self-employment was being replaced by more wage work. But this is likely to be a transitional effect, partially caused by the temporarily high growth rates. The shifts in the occupational structure meant that many workers had to move to dif- ferent jobs or, especially in the east, enter the labor market in the expanding number of low-end occupations, which may not have been their aspiration. Are Distributional Tensions Brewing in Europe and Central Asia? ●  39 Results: changes in the earnings distribution. The polarization of jobs in the western part of the region was not accompanied by polarization in earnings (as in the United States; see Acemoglu and Autor 2011). Instead, there was a regressive change in the distribution of earnings. Workers displaced in the middle occupations added to the supply of workers for jobs at the bottom more rapidly than the demand for such jobs, and the greater competition drove wages down. At the higher end, demand outstripped supply, and wages rose. The reverse occurred in the eastern part of the region. Wages at the bot- tom of the distribution rose, along with the number of jobs. A detailed decom- position analysis confirms that occupational change was especially relevant in the regressive change in the earnings distribution in the west, while other fac- tors, such as demography and higher educational attainment, were at play in the east.9 Both halves of the region experienced increasing distributional tensions. In the west, deterioration in economic security and in the incomes of low-skilled workers put pressure on redistribution systems (chapter 3). One result was a growth in dual earner households as households boosted their participation in the labor market to offset the deterioration in incomes. In the east, the aspirations of the more well educated to obtain high-end jobs were frustrated, similar to the outcome in the Middle East (Arampatzi et al. 2015). Trends in Occupational Change in the Region Individuals are employed in occupations, and any given occupation can be under- stood as a bundle of tasks. Six tasks may be defined (box 2.2). Each occupation is intensive in each of these tasks, but in different ways. Changes in the labor market may be described in terms of how the intensity of each of these tasks is evolving in overall employment. This is the approach followed, for instance, by Hardy, Keister, and Lewandowski (2016). Overall employment may become more routine intensive because the occupational structure changes and occupations that are more intensive in routine tasks come to account for a larger share of total employ- ment. An alternative approach, followed by Autor (2014), the World Bank (2016a), ­ ntensity.10 and this report, involves grouping occupations according to relative task i This allows the change in occupational structure to be examined directly rather than inferring the change according to the change in the average task content of employment. This is more useful in evaluating the distributional tensions in the labor market because individuals choose and employers demand occupations rather than tasks.11 In this approach, occupations are classified based on the inten- sity of each task involved. For simplicity and given the high correlation that exists between some tasks, the classification used in this report groups occupations into three categories based on their intensity in routine tasks; nonroutine, cognitive tasks; and nonroutine, manual, physical tasks (box 2.2; see annex 2A). Occupations can be classified into three groups based on the type of task involved. Occupations intensive in routine tasks may include either routine cogni- tive tasks, such as filling out forms or performing repetitive administrative assign- ments (an office clerk), or routine manual tasks, such as operating a machine in a factory (a metal molder). These occupations are typically found in the middle of the wage distribution in high-income countries (Goos, Manning, and Salomons 2014). 40 ● Toward a New Social Contract BOX 2.2 Construction of Occupational Categories Grouping occupations according to task content (9 groups) with the highest index value. The implies making a decision on which task dimension remaining 18 submajor occupation groups are to prioritize. Because the potential number of tasks divided into two groups according to their value on characterizing an occupation may be large, this the nonroutine, cognitive, analytical index.c The report relies on task content indexes formulated by half with the highest values on the nonroutine, cog- the Institute for Structural Research that originate nitive, analytical index is included in the second from O*NET and follow Acemoglu and Autor category, occupations intensive in nonroutine, cog- (2011). a There are six task content indexes: (a) nitive tasks. The bottom half is included in the third nonroutine, cognitive, analytical; (b) nonroutine, category, occupations intensive in nonroutine, cognitive, personal; (c) routine, cognitive; (d) rou- manual tasks. Annex 2A, table 2A.1 presents a sta- tine, manual; (e) nonroutine, manual, physical; and tistical summary of the categories. The categoriza- (f) nonroutine, manual, personal. Additionally, tion of occupations is based on the relative indexes (c) and (d) can be combined into a routine- intensity of some tasks. Thus, nonroutine, manual, task intensity index based on Autor, Levy, and physical task content is high in both the first and Murnane (2003). Each occupation at the 4-digit third groups, but the first group also exhibits high level (unit group titles) of the International Standard routine-task intensity, whereas the third group Classification of Occupations (ISCO) 88 has a value shows a low value for routine tasks. In this sense, in every task content index.b For the purpose of the first group is relatively more routine-intensive this work, occupations are aggregated at the ISCO than the third group, which is relatively more inten- 88 2-digit level (submajor group titles) by taking a sive in nonroutine, manual, physical tasks. simple average of the indexes of the unit groups This classification is possible if occupation data included in the corresponding submajor group. are available at the ISCO 2-digit level. In some This is done to have a common aggregation level instances, the relevant data are available only at across countries because not all surveys record the ISCO 1-digit level (major groups). In this case, occupations at the 4-digit level. the first occupation category (occupations intensive There are 27 submajor occupation groups in the in routine tasks) includes ISCO major groups 4, ISCO 88 classification. These are divided into three 7, and 8; the second occupation category groups as follows. First, the 27 groups are ranked ( occupations intensive in nonroutine, cognitive according to the routine-task intensity index. The tasks) includes ISCO major groups 1, 2, and 3; and first category—occupations intensive in routine the third occupation category includes ISCO major tasks—includes the top third of the groups groups 5, 6, and 9. a. A caveat involved in using O*NET data is the assumption that the task content of each occupation is the same across all countries and that it is the same as the content for each occupation in the United States, for which O*NET was specifically constructed. There is evidence that the tasks performed in a same occupation, for example, an office clerk, differ across countries (Dicarlo et al. 2016). See Occupation Classifications Crosswalks: From O*NET-SOC to ISCO (database), Institute for Structural Research, Warsaw, April 6, 2016, http://ibs.org.pl/en/resources/occupation-classifications-crosswalks-from-onet-soc-to-isco/; O*NET OnLine (database), Employment and Training Administration, U.S. Department of Labor, Washington, DC, https://www .onetonline.org/. b. The current version of the classification is ISCO 08, and most of the surveys undertaken after 2010 have used this classification instead of ISCO 88. The categorization here is based on ISCO 88; correspondence tables allow ISCO 08 occupations to be mapped onto this categorization. c. Results practically do not change if the nonroutine, cognitive, personal index is used. Occupations intensive in nonroutine manual tasks involve low-skilled work, for example, work as a nurse’s aide or private security guard, that cannot easily be replicated by a machine. These jobs are among the lowest paid in modern economies.12 By contrast, occupations intensive in nonroutine cognitive tasks require high-skilled professionals, such as scientists, engineers, or managers, Are Distributional Tensions Brewing in Europe and Central Asia? ●  41 and are usually the highest paid in modern economies. In the literature on job polarization, the second and third categories are referred to, respectively, as lousy and lovely jobs (Goos and Manning 2007). Technological progress has sharply reduced the share of jobs involving routine tasks, usually middle-paid jobs, in Europe over the last two decades (figure 2.6). The share of routine-task–intensive occupations in total wage employment fell from an average of around 40 percent in 1995 (ranging from 23 percent in Albania to 50 percent in Italy) to around 33 percent in 2013 (ranging from 23 percent in Montenegro to 41 percent in the Czech Republic).13 The decline in the share of routine-task–intensive occupations was as high as 11 percentage points in Southern Europe and as low as 2 percentage points in the Baltic States. By contrast, the employment share of nonroutine cognitive-task–intensive occupations, usually well-paying jobs, has risen substantially. The share of these occupations in employment grew from an average of 25 percent in 1995—though in countries such as Italy and Portugal, this share was below 20 percent—to 32 percent in 2013; several countries, such as Luxembourg, Norway, and Switzerland, had shares above 40 percent. The increase in the share across regions ranged from 8.0 percentage points in Scandinavia and Southern Europe to 4.5 percentage points in Central Europe, and the employment share in 2013 ranged from almost 38 percent in Northern Europe to around 26 percent in Southern Europe. Changes in the share of employment in occupations intensive in nonroutine, manual tasks, usually low-paying jobs, have varied across Europe. The employ- ment share of these occupations rose from 35 percent in 1995 to 38 percent in 2013 in Southern Europe and from 31 percent to 32 percent in Western Europe. Thus, these subregions experienced a rise in both kinds of nonroutine tasks at the expense of routine tasks. The employment share of nonroutine, manual tasks remained stable at 33 percent in Central Europe and 37 percent in the Western FIGURE 2.6 a. Baltic States b. Central Europe c. Northern Europe 10 The employment share 5 in routine task-intensive occupations has fallen in Change in percent of regular employees 0 −5 Europe −10 Change in the share of employment, by occupation Region Region Region category, late 1990s to early d. Southern Europe e. Western Balkans f. Western Europe 2010s 10 5 0 −5 −10 Region Region Region Nonroutine, manual Routine Nonroutine, cognitive Source: World Bank calculations based on household surveys and labor force surveys. 42  ●   Toward a New Social Contract Balkans over this period, while this share fell from close to 38 percent to 35 percent in the Baltic States and from 35 percent to 34 percent in Northern Europe. Overall, these data illustrate significant shifts in occupational structure in Europe, as well as job polarization in some subregions. All European subregions experienced a fall in the employment share of routine-task–intensive occupations and a rise in the share of nonroutine cognitive-task–intensive occupations. In coun- tries where nonroutine cognitive-task–intensive occupations are the only ones to enjoy a rise in demand, high-skilled workers already at the top of the distribution may experience a greater increase in wages relative to low-skilled workers. However, this tendency to greater inequality could be addressed by expanding the supply of high-skilled workers through education and training. However, in Southern and Western Europe, the employment share of both rose in the most highly paid occupations (involving nonroutine, cognitive tasks) and the least well- paid occupations (involving nonroutine, manual tasks). This job polarization can drive greater distributional tensions because many middle-paid workers who per- form routine tasks may be displaced to less well-paid jobs.14 The eastern part of Europe and Central Asia—the former Soviet Union econo- mies and Turkey—did not experience the job polarization seen in Western and Southern Europe. The employment share of nonroutine cognitive-task–intensive occupations fell by an average of 5 percentage points from the early 2000s to the mid-2010s, with declines in all countries on which there are consistent data, except Moldova and Turkey (figure 2.7).15 The expansion in wage employment at the expense of unpaid family work or self-employment may explain part of the decline in the share of nonroutine cognitive-task–intensive occupations. The employment share of routine-task–intensive occupations increased or remained stable in all coun- tries on which there are consistent data, except Moldova and Turkey. The employ- ment share of nonroutine manual-task–intensive occupations rose in all countries except Armenia and Moldova. The expansion of elementary occupations in a con- text of economic growth suggests there was considerable growth in the demand for low-skilled services. FIGURE 2.7 a. Armenia b. Georgia c. Kyrgyz Republic The share of employment, 10 by occupational category, 5 Change in percent of regular employees early 2000s to mid-2010s 0 −5 −10 Country Country Country d. Moldova e. Russian Federation f. Turkey 10 5 0 −5 −10 Country Country Country Nonroutine, manual Routine Nonroutine, cognitive Source: World Bank calculations based on household surveys and labor force surveys. Are Distributional Tensions Brewing in Europe and Central Asia? ●  43 Thus, changes in occupational structure have differed considerably across Europe and Central Asia. Some regions, particularly Western and Southern Europe, are undergoing a process of job polarization, while the economies of the former Soviet Union have seen a growth in nonroutine manual-task–intensive occupations and a fall in the share of highly skilled nonroutine cognitive-task–intensive occupa- tions. The impact of occupational change on the distribution of earnings also likely differs across the region. The next subsection analyzes this impact in seven coun- tries in Europe and Central Asia (Bussolo, Torre, and Winkler 2018).16 Job Polarization and Earnings in Selected Countries This subsection reviews detailed information on trends in earnings in selected countries in the western and eastern parts of Europe and Central Asia. It examines earnings data on three countries in the west (Germany, Poland, and Spain) from the early 1990s to 2013. In the east, it investigates data on four countries (Georgia, Kyrgyz Republic, Russia, and Turkey) over a slightly shorter period. Job polarization in the three countries in the west was accompanied by a decline in earnings among low-wage workers relative to the earnings of high-wage workers. Most EU countries experienced strong growth in the years between the fall of the Berlin Wall and the global financial crisis. While income inequality did not change much overall (chapter 1), labor incomes became more unequal in several countries. From the early 1990s to 2013, the Gini index of labor income rose by about 8 points in Germany and Spain and by about 5 points in Poland. This increase in inequality mainly reflected slower earnings growth among low-wage workers; the earnings of workers in the two bottom deciles of the wage distribution rose at least 10 percent- age points less than the earnings of workers at the median of the distribution and more than 30 percentage points less than the earnings of workers at the top. In the United States, job polarization was accompanied by wage polarization, that is, growth in wages at the two extremes of the earnings distribution. In the EU, how- ever, the polarization in occupations did not translate into a greater rise in the wages of low-paid workers relative to workers at the median; rather, the distribution of wages became more regressive in general (­ figure 2.8, panels a, c, and e). The deterioration in earnings inequality in Europe was partly driven by job polarization. In Germany, Poland, and Spain, occupational changes played a big role in accounting for the relative wage decline among low-paid workers (see ­figure 2.8, panels b, d, and f). This analysis separates changes in the overall distribution of wages (figure 2.8, panels a, c, and e) into changes deriving from occupational shifts and changes deriving from other factors (figure 2.8, panels b, d, and f), for example the entry of new workers with better skills or shifts in demand that increase the wages for some skill occupations (box 2.3). In these three coun- tries, declining relative demand for occupations intensive in routine tasks dis- placed many workers who could not compete for high-skill jobs. These workers were forced to compete for jobs at the bottom of the wage distribution, resulting in a relative reduction in wages and an expansion in employment in jobs intensive in nonroutine manual skills (the least well-paid workers). Simulation results show that moving from a routine-task–intensive job to a job intensive in nonroutine manual tasks—the usual transition for the relatively low skilled employed in routine intensive occupations—implied a reduction of almost 30 percentage points in 44  ●   Toward a New Social Contract FIGURE 2.8  Changes in wages, Germany, Poland, and Spain, 1990s to 2013 a. Germany b. Germany Change in wages, 1994–2013 Decomposition of change in wages, 1994–2013 Log change in wages with respect to median Log change in wages with respect to median 0.25 0.25 0.20 0.20 0.15 0.15 0.10 0.10 0.05 0.05 0 0 –0.05 –0.05 –0.10 –0.10 –0.15 –0.15 –0.20 –0.20 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 Quantiles of wage distribution Quantiles of wage distribution c. Poland d. Poland Change in wages, 1992–2013 Decomposition of change in wages, 1992–2013 Log change in wages with respect to median Log change in wages with respect to median 0.3 0.3 0.2 0.2 0.1 0.1 0 0 –0.1 –0.1 –0.2 –0.2 –0.3 –0.3 –0.4 –0.4 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 Quantiles of wage distribution Quantiles of wage distribution e. Spain f. Spain Change in wages, 1990–2013 Decomposition of change in wages, 1990–2013 Log change in wages with respect to median Log change in wages with respect to median 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 –0.1 –0.1 –0.2 –0.2 –0.3 –0.3 –0.4 –0.4 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 Quantiles of wage distribution Quantiles of wage distribution Occupational structure component Total change Returns and characteristics Source: Bussolo, Torre, and Winkler 2018. Are Distributional Tensions Brewing in Europe and Central Asia? ●  45 BOX 2.3 Decomposing the Change in Wages: The Role of Occupational Change Analysis of the factors explaining changes in wages parameters defining the occupational structure of the requires, initially, establishing the possible drivers. earnings distribution in the initial year. Thus, the In this sense, characteristics such as educational decomposition relies on simulating the earnings dis- attainment, age, sector, or task-specific skills can tribution in 2013 as if the occupational structure be thought of as assets that workers accumulate parameters—for instance, the probability of an indi- and for which they receive returns on the labor vidual with tertiary educational attainment to be in a market. The simultaneous accumulation of these nonroutine cognitive-task–intensive occupation—had assets and changes in the associated returns affect been the same as those in 1993. The change trends in the distribution of earnings. Moreover, between the actual earnings distribution and the sim- the returns to these assets can be thought of as ulated earnings distribution is explained in this case specific to each job. Returns to education, for by the change in the parameters of the occupational instance, need not be the same across occupa- structure. A similar exercise is carried out with the tions. But occupations are not distributed randomly characteristics of individuals and the respective within the population. Individuals with a certain set returns. A residual component is needed because the of characteristics may be more likely to be found in counterfactual simulation can only be carried out on certain occupations. This is a representation of the observed characteristics and cannot account for occupational structure and is the result of the inter- changes in unobservable variables. action of both labor demand and the supply of Figures 2.8 and 2.9 in the text show the results of skills. Bussolo, Torre, and Winkler (2018) carry out a the decomposition of the changes in wages. In blue decomposition of changes in wages in seven coun- is indicated the occupational structure component, tries of Europe and Central Asia over 20 years that is, the part of the change in wages that can be within such a framework and provide estimates of accounted for by changes in the occupational struc- the extent to which changes in the characteristics ture. The remaining two components—the change of individuals, changes in the returns to these char- accounted for by variations in the characteristics of acteristics, and changes in the occupational struc- individuals, such as educational attainment, and ture account for trends in earnings. their returns—is shown, added up, in green. The A standard method for decomposing changes in orange line indicates the actual changes observed wages between two periods (for instance, 1993 and between the initial year of the analysis and the last 2013) is the Oaxaca-Blinder method, which decom- year of the analysis. For presentational purposes the poses the change between two earnings distribu- residual component, which would account for the tions by analyzing the changes in the means of the remaining difference between the actual change relevant factors, such as individual characteristics, and the explained components, is not depicted. returns to characteristics, and occupational structure. The results indicate that, in Germany, Poland, Bussolo, Torre, and Winkler (2018) perform a and Spain, changes in the parameters defining the decomposition inspired by Bourguignon and Ferreira occupational structure were particularly damaging (2005) and Inchauste et al. (2014) that generalizes the for the earnings of those at the bottom of the wage Oaxaca-Blinder methodology to changes in distribution. For instance, the probability of house- the whole earnings distribution, rather than only the hold heads with only secondary educational means. This decomposition is carried out with the attainment to working in nonroutine manual-task– use of counterfactual simulations in which the earn- intensive jobs—typically the lowest paid in the ings distribution in the final year is simulated by economy—rose by 12 percentage points in Spain, retaining, alternatively, the characteristics of individu- while, among household spouses with similar edu- als, the returns to these characteristics, and the set of cational profiles, it increased by 19 percentage (Continued) 46 ● Toward a New Social Contract BOX 2.3 Decomposing the Change in Wages: The Role of Occupational Change (continued) points in Germany and 14 percentage points in Turkey show that the occupational structure com- Poland. Individuals with secondary education are ponent accounts for a small part of the change in found more often in low-paid occupations now wages. The relative improvement in wages at the than before, explaining part of the relative decline bottom of the distribution in these countries is in wages at the bottom of the distribution. explained more by the changes observed in the The results of the decomposition for Georgia, characteristics of individuals and in the associated the Kyrgyz Republic, the Russian Federation, and returns. labor market earnings. Conversely, the transition to a job intensive in nonroutine cognitive tasks implied an increase of around 25 percentage points. Thus, many formerly middle-paid workers experienced a significant cut in earnings. By contrast, earnings inequality fell in the former Soviet Union economies because the employment share and relative earnings of high-skilled workers declined. The employment share of occupations intensive in nonroutine cogni- tive skills fell (see above). This was accompanied by a drop in the earnings of high-skilled workers, who are typically the most well-paid workers, relative to the earnings of other workers. For example, in Georgia and the Kyrgyz Republic from the early 2000s to the mid-2010s, the top two deciles of the distribution of earnings experienced earnings growth about 20 to 40 percentage points lower than the median (figure 2.9). In Russia, the labor market incomes of high earners rose by about 50 percentage points less than the corresponding incomes of the median between 1994 and 2014. Most of the relative loss in earnings at the top of the wage distribution in the east can be explained by a reduction in the returns to education (Bussolo, Torre, and Winkler 2018). In contrast, low earners experienced earnings growth significantly above the median. These changes resulted in a strong decrease in the inequality of labor income in Georgia (from a Gini coefficient of 0.48 in 2002 to 0.45 in 2015), Russia (from a coefficient of 0.55 in 1994 to 0.39 in 2014) and Turkey (from a Gini coefficient of 0.42 in 2002 to 0.36 in 2013). In the Kyrgyz Republic, the Gini coefficient of the labor market remained at around 0.44. Thus, economies of the former Soviet Union avoided the deterioration in wage inequality experienced in Western Europe. In Europe, the demand for low-skill workers could not keep up with the increase in supply caused by the influx of dis- placed routine workers, leading to rising inequality, while in the former Soviet Union countries, the demand for high-skill workers was not as strong as the grow- ing supply of skilled workers, resulting in falling inequality. While the absence of job polarization in the east and falling inequality may have helped avoid the kinds of distributional tensions experienced in the west, it may also indicate a lack of economic dynamism. Policy distortions in the east may mean that highly educated workers are not paid wages commensurate with their productivity (box 2.4). Moreover, the absence of job polarization in a subregion subject to the same Are Distributional Tensions Brewing in Europe and Central Asia? ●  47 FIGURE 2.9  Wage changes, Georgia, Kyrgyz Republic, Russian Federation, and Turkey, 1990s to 2010s a. Georgia b. Georgia Change in wages, 2002–2015 Decomposition of change in wages, 2002–2015 0.3 0.3 Log change in wages with Log change in wages with 0.2 0.2 respect to median respect to median 0.1 0.1 0 0 –0.1 –0.1 –0.2 –0.2 –0.3 –0.3 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 Quantiles of wage distribution Quantiles of wage distribution c. Kyrgyz Republic d. Kyrgyz Republic Change in wages, 2004–2014 Decomposition of change in wages, 2004–2014 0.2 0.2 Log change in wages with 0.1 Log change in wages with 0.1 respect to median 0 respect to median 0 –0.1 –0.1 –0.2 –0.2 –0.3 –0.3 –0.4 –0.4 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 Quantiles of wage distribution Quantiles of wage distribution e. Russian Federation f. Russian Federation Change in wages, 1994–2014 Decomposition of change in wages, 1994–2014 0.5 0.5 0.4 0.4 Log change in wages with Log change in wages with 0.3 0.3 respect to median respect to median 0.2 0.2 0.1 0.1 0 0 –0.1 –0.1 –0.2 –0.2 –0.3 –0.3 –0.4 –0.4 –0.5 –0.5 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 Quantiles of wage distribution Quantiles of wage distribution g. Turkey h. Turkey Change in wages, 2002–2013 Decomposition of change in wages, 2002–2013 0.4 0.4 Log change in wages with Log change in wages with 0.3 0.3 respect to median respect to median 0.2 0.2 0.1 0.1 0 0 –0.1 –0.1 –0.2 –0.2 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 Quantiles of wage distribution Quantiles of wage distribution Occupational structure component Total change Returns and characteristics Source: Bussolo, Torre, and Winkler 2018. 48 ● Toward a New Social Contract BOX 2.4 Teachers and Drivers: Low Wages in High-Skill Occupations in the former Soviet Union Economies One of the characteristics of the wage structure in deciles of the wage distribution, while drivers typi- former Soviet Union economies is that the occupa- cally earn wages in the middle deciles. The pattern tions of high-skill workers do not necessarily pay in Georgia and the Kyrgyz Republic is the oppo- high wages. Consider, for example, teaching pro- site: teaching professionals are found in the mid- fessionals (ISCO category 23), a job which typically dle to the bottom of the wage distribution, and requires at least a high school degree and is inten- drivers are mostly found from the middle to the sive in nonroutine cognitive tasks, versus drivers top of the distribution. The prevalence of low and mobile plant operators, International Standard wages among teachers may have resulted in an Classification of Occupations (ISCO) category 83, incentive for workers in these occupations to move jobs that usually do not require any formal school- to jobs where, even if overqualified, such as driv- ing qualification and are intensive in nonroutine ing jobs, they can earn higher wages. Indeed, the manual tasks. Figure B2.4.1 presents the distribu- distributions of teaching professionals and drivers tion of these occupations within the overall wage in the most recent year (not shown) have moved to distribution in three countries—Georgia, Germany, the right and the left, respectively, suggesting that and the Kyrgyz Republic—in the mid-1990s to shifts out of nonroutine cognitive-task–intensive early 2000s. In Germany, the expected pattern is occupations and into other occupations may have found. Teachers typically earn wages in the upper reduced this counterintuitive wage difference. FIGURE B2.4.1 Distribution of teaching professionals, drivers, and mobile plant operators, initial year a. Germany, 1994 15 Relative frequency (%) 10 5 0 0 5 10 15 20 Ventiles of the earnings distribution Teaching professionals Drivers and mobile plant operators (Continued) Are Distributional Tensions Brewing in Europe and Central Asia? ●  49 BOX 2.4 Teachers and Drivers: Low Wages in High-Skill Occupations in the former Soviet Union Economies (continued) FIGURE B2.4.1 Distribution of teaching professionals, drivers, and mobile plant operators, initial year (continued) b. Georgia, 2002 15 Relative frequency (%) 10 5 0 0 5 10 15 20 Ventiles of the earnings distribution Teaching professionals Drivers and mobile plant operators c. Kyrgyz Republic, 2004 15 Relative frequency (%) 10 5 0 0 5 10 15 20 Ventiles of the earnings distribution Teaching professionals Drivers and mobile plant operators Source: Bussolo, Torre, and Winkler (2018). Note: The figure plots the relative distribution of teaching professionals (ISCO code 23) and drivers and mobile plant operators, International Standard Classification of Occupations (ISCO) category 83 on the overall earnings distribution in the initial year of the analysis. All curves are smoothed by a locally weighted regression. All values include the self-employed. Similar patterns are observed if the self-employed are excluded. 50  ●   Toward a New Social Contract technological and globalization forces as the west may suggest a static labor mar- ket in which innovation and technological change are weak, and the process of creative destruction—whereby some occupations shrink and others expand—is muted. Economic insecurity increases in times of intense occupational change. Because there are returns to specialization in any employment activity, chang- ing occupations can represent a short-term and potentially also a long-term decrease in productivity and earnings for any given individual. Even if, from the perspective of society, occupational change represents dynamism and growth, a high turnover in occupations can be a source of economic distress from an individual’s point of view. Moreover, the fact that the distribution of the winners and losers of occupational change can be highly polarized adds a distributional dimension to the inherent tension emerging from the shift in jobs by individuals. An Increasing Generational Divide, and the Young Are Losing Ground The economic transformations in Europe and Central Asia in 1990–2010 had differing effects across generations. The changes impaired the economic prospects of youth relative to the changes experienced by older generations at the same age. Until early adulthood, the economic welfare of an individual is largely the same as that of the individual’s immediate family. Among peo- ple aged 16–30, however, the surrounding society becomes a more influen- tial factor, and the experiences of individuals during this formative period may shape the fortunes and attitudes of these individuals for a lifetime (Chauvel and Schröder 2014). As the economic environment changes, differ- ent birth cohorts will have different experiences, and the cohort to which an individual belongs becomes an important determinant of the individual’s welfare. This generational divide may be masked in analyses of vertical inequality. For instance, inequality may be increasing between generations, but inequality within each generation may be declining. Thus, measures of aggregate inequal- ity may not change, while tensions between or within generations rise. Also, young generations tend to be smaller in number than older ones in the aging countries of Europe and Central Asia and account for an even smaller share of total income (because earnings are typically lower at the beginning of one’s career). Thus, a decline in the earnings of young generations may not have a large impact on aggregate income distribution, while it may become an impor- tant source of distributional tension. Five stylized facts point to the difficulties facing younger workers and thus rais- ing distributional tensions between generations, as follows: • Nonstandard employment (part-time, temporary, and agency work) is becom- ing more common in the region, and younger cohorts are engaged in these types of employment more intensively relative to older cohorts. Are Distributional Tensions Brewing in Europe and Central Asia? ●  51 • Job tenure has decreased among young workers. • The declining fortunes of the young are associated with their labor market earnings: college graduates have seen the growth of their wages decrease ­ substantially. • The position of the young relative to the middle aged and, particularly, the elderly has been deteriorating in Southern and Western Europe during the last decades, while, in Central and Northern Europe, former Soviet Union economies, and Turkey, the situation has been relatively stable. ­ • In those regions where the income levels of younger generations have declined compared with that of older generations, inequality among the young has also widened. New Types of Jobs for Younger Workers Nonstandard employment is becoming more common in the region, and, together with shorter tenures, may partly explain the narrowing wage pros- pects and the greater within-cohort inequality among younger generations. Traditionally, security and stability in labor markets have been achieved through formal employment involving permanent contracts of indefinite duration. This is the benchmark against which workers in postwar societies in Europe and Central Asia have typically measured themselves. In recent decades, however, new forms of employment have become more common, partly because of changes in labor policies that have diversified the type of contracts available (chapter 3). This nonstandard employment includes part-time and temporary employment.17 The share of nonstandard employment in total employment rose steadily and substantially in Central, Southern, and Western Europe, while in Northern Europe, the share remained relatively stable, but at a high level (more than 30 percent) (figure 2.10). In Southern Europe, the share shot up from 8 percent in the early 1980s to 29 percent in the early 2010s, and in Western Europe from around 18 percent to close to 34 percent in the same period. In Central Europe, where data are available only from the late 1990s, the increase was from 10 percent in 1997 to almost 21 percent in 2013. In the Baltic States, the share of nonstandard employment hovered between 10 per- cent and 12 percent during the same period. Data on the economies of the former Soviet Union and Turkey cover a more limited time span and show mixed trends. Albania, Armenia, and Georgia have experienced declines in the share of nonstandard employment in total employ- ment, while in the Kyrgyz Republic, Moldova, and Turkey, the share has increased. In terms of composition, in many countries, such as Croatia, Hungary, Poland, Portugal, and Slovenia, the overall expansion in nonstandard employment was driven by the growing share of (full-time) temporary employment (figure 2.11). In several other countries, such as Austria, Ireland, and the Netherlands, the rise in permanent part-time employment was the bigger contributor. Yet, in others, such as Denmark, Sweden, and the United Kingdom, no significant change in the share of nonstandard employment was observed, but the composition of this employment changed, including a shift from permanent part-time to temporary 52  ●   Toward a New Social Contract FIGURE 2.10 a. European Union Nonstandard employment 40 (NSE) has expanded in most of Europe and Central Asia Share of NSE employment (%) Nonstandard employment as 30 percent of total employment 20 10 0 1982 1985 1990 1995 2000 2005 2010 2013 Year Western Europe Southern Europe Central Europe Northern Europe Baltic States b. Former Soviet Union economies and Turkey 60 Share of total employment (%) 50 40 30 20 10 0 Armenia Georgia Kyrgyz Turkey Albania Moldova Russian Republic Federation Temporary + Part-Time Employment Only Part-Time Employment Starting point Ending point Sources: World Bank calculations based on labor force surveys; Apella and Zunino 2018. Note: Panel a: each line depicts the smoothed (locally weighted regression) average of the prevalence of nonstandard employment (temporary and part-time employment) by region. Panel b: starting point and ending point, respectively: Albania, 2002, 2013; Armenia, 1998, 2015; Georgia, 2002, 2015; Kyrgyz Republic, 2004, 2014; Moldova, 1998, 2013; Russian Federation, 1994, 2014; Turkey, 2004, 2013. employment in Sweden and a reverse shift in Denmark and, to a lesser degree, the United Kingdom. The education and task profile of workers in nonstandard employment also changed (box 2.5). Younger workers are more engaged in nonstandard employment. Various groups may exhibit variations in their willingness to engage in temporary or part- time employment, and employers may vary in their willingness to hire certain groups of workers under such conditions. In the subregions with the largest rise in the share of nonstandard employment, Southern and Western Europe, a greater share of younger age-groups tend to take on nonstandard employment, and the expansion in the share of nonstandard employment was greater among the young (figure 2.12). In Southern Europe, the share of nonstandard employment among the 20–24 age- group rose from 15 percent in the early 1980s to well above 60 percent in 2013. In Western Europe, the share of nonstandard employment in the same age-group increased from around 15 percent to more than 40 percent in the same period, Are Distributional Tensions Brewing in Europe and Central Asia? ●  53 FIGURE 2.11 The composition of nonstandard employment differs in countries and regions 100 80 Total employment (%) 60 40 20 0 ia um e d g s m us ain a Po y l h R ria lic tia y Ro d nia Re a De ic k nia Lit d ia Sw a en ga d l ar ar nc lan Ne our alt lan ni lan tvi Ita bl str do an ite an pr ub ed a oa Sp rtu e lgi ng ma nm to Fra pu La M Cz ulg Ire ing Cy Po Sl Slov Fin hu Au mb Un herl ep Cr Es Be Hu B dK xe t ak Lu ec ov Western Europe Southern Europe Central Europe Northern Europe Temporary part-time 1995 Temporary full-time 1995 Permanent part-time 1995 Temporary part-time 2015 Temporary full-time 2015 Permanent part-time 2015 Source: World Bank calculations based on labor force surveys. BOX 2.5 The Changing Education and Task Profile of Nonstandard Employment Workers in nonstandard employment are more well heterogeneous. Overall, the task content of nonstan- educated today than in the 1990s (figure B2.5.1). dard jobs seems to follow a similar pattern, though However, this is not unique to nonstandard with notable exceptions. In Europe, standard employ- employment and reflects the more widespread ment has become more intensive in all nonroutine access to education in all countries in the west. The cognitive tasks, such as analyzing information or think- spread of education access in Southern Europe ing creatively, while in nonstandard employment, this appears to have been more pronounced relative to has been observed only among jobs requiring inter- nonstandard employment. Coupled with the differ- personal relationships, such as supervising subordi- ence in the education profile of workers in nonstan- nates or interacting with customers; in many countries, dard employment versus workers in standard it has not been observed among nonstandard jobs employment, this shift widened the divide, sustain- involving analytical tasks (figure B2.5.2). The broader ing, even deepening, the vulnerability of workers decline in manual tasks in standard employment has involved in nonstandard employment. also been observed in nonstandard employment.a The jobs of workers in nonstandard employment These parallel trends between the task content in non- increasingly involve tasks that require more complex routine cognitive and manual tasks are also found in skills, mimicking the broader trend in employment. In the economies of the former Soviet Union, although most of the European Union, the trend is toward occu- the changes are smaller in magnitude relative to those pations more intensive in nonroutine cognitive tasks, elsewhere in Europe (Apella and Zunino 2018). There while in the former Soviet Union economies, the pat- is a clear divergence with respect to routine cognitive tern of occupational change has been more tasks—those involving a need for precision in a (Continued) 54 ● Toward a New Social Contract BOX 2.5 The Changing Education and Task Profile of Nonstandard Employment (continued) structured work environment—in Europe. While there routine cognitive-task intensity in overall employment has been a consistent decline in these tasks among in many of these countries is explained mostly by the standard employees, there have been increases in spread of nonstandard employment. Thus, tasks that nonstandard employment, particularly in some coun- appear to be disappearing in Southern Europe are tries in Central and Eastern Europe. The expansion becoming more highly concentrated in more flexible identified by Keister and Lewandowski (2016) in forms of employment, particularly in certain countries. FIGURE B2.5.1 Changes in the education profile of workers, by employment type a. Standard employment (SE) workers: education profile change (1995–2015) 40 Percent change in share of workers 30 20 by skill level 10 0 –10 –20 –30 –40 Be ria um e Un xem nd ing g m Po ly l Bu in Cr ia Po a Sl Ro d Re ia Sl lic De ia Es rk Fin ia d Lit tvia Sw ia en ga nc d K ur ti lan lan Ita do a r ak an en a n an b Lu Irela ed st lga oa lgi rtu Sp nm to ite bo Fra pu La ov m ov hu Au Western Europe Southern Europe Central Europe Northern Europe Low-skilled Medium-skilled High-skilled b. Nonstandard employment (NSE) workers: education profile change (1995–2015) 40 Percent change in share of workers 30 20 10 by skill level 0 –10 –20 –30 –40 Be ria um e Un xem nd ing g m Po ly l Bu in Cr ia Po a Sl Ro d Re ia Sl lic De ia Es rk Fin ia d Lit tvia Sw ia en ga nc d K ur ti lan lan Ita do a r ak an en a n an b Lu Irela ed st lga oa lgi rtu Sp nm to ite bo Fra pu La ov m ov hu Au Western Europe Southern Europe Central Europe Northern Europe Low-skilled Medium-skilled High-skilled Source: Calculations based on data of EU-LFS (European Union Labour Force Survey) (database), Eurostat, European Commission, Luxembourg, http://ec.europa.eu/eurostat/statistics-explained/index .php/EU_labour_force_survey_%E2%80%93_data_and_publication. (Continued) Are Distributional Tensions Brewing in Europe and Central Asia? ●  55 BOX 2.5 The Changing Education and Task Profile of Nonstandard Employment (continued) FIGURE B2.5.2 Changes in task content, by employment type Similar trends in nonroutine cognitive analytical and manual tasks; diverging trends in routine cognitive and nonroutine cognitive interpersonal tasks a. Change in nonroutine, cognitive analytical task intensity, mid-1990s to 2015 0.6 0.5 Normalized task intensity index 0.4 0.3 0.2 0.1 0 –0.1 –0.2 –0.3 –0.4 Be tria Fra m Lu rel e m d Un th rg ing ds Cy m Sp s ain Po aly ec lg l ep ia Cr lic Hu atia Po ry Ro nd ov lo a Re nia De blic Es ark Fin ia Lit land La ia Sw tvia en Cz Bu ga u nc xe an Sl S ani u do h R ar n a an Ne bou d K an pr ub la ed It lgi rtu ak ve nm to ng s pu o m hu Au ite erl I Western Europe Southern Europe Central Europe Northern Europe Standard employment (SE) Nonstandard employment (NSE) b. Change in nonroutine, cognitive interpersonal task intensity, mid-1990s to 2015 0.7 0.6 Normalized task intensity index 0.5 0.4 0.3 0.2 0.1 0 Be tria Fra m Lu rel e m d Un eth ourg ing ds Cy m Sp s ain Po aly ec lg l ep ia Cr lic Hu atia Po ry Ro nd ov lo a Re nia De blic Es ark Fin ia Lit land La ia Sw tvia en Cz Bu ga u nc xe an Sl S ani u do h R ar a n an d K an pr ub la ed It lgi rtu ng ak ve nm to s pu o m hu Au N b ite erl I Western Europe Southern Europe Central Europe Northern Europe Standard employment (SE) Nonstandard employment (NSE) (Continued) 56 ● Toward a New Social Contract BOX 2.5 The Changing Education and Task Profile of Nonstandard Employment (continued) FIGURE B2.5.2 Changes in task content, by employment type (continued) c. Change in routine, manual task intensity, mid-1990s to 2015 0.4 0.3 Normalized task intensity index 0.2 0.1 0 –0.1 –0.2 –0.3 –0.4 –0.5 –0.6 –0.7 Be tria Fra m Lu Irel ce m d Un th rg ing ds Cy m Sp s ain Po aly ec lg l ep ia Cr lic Hu tia Po ry Ro d ov lo a Re nia De blic Es ark Fin ia Lit land La ia Sw tvia en Cz Bu ga u xe an lan Sl S ani u do h R ar n a an Ne bou n d K an pr ub ed oa It lgi rtu to ng ak ve nm s pu m hu Au ite erl Western Europe Southern Europe Central Europe Northern Europe Standard employment (SE) Nonstandard employment (NSE) d. Change in routine, manual physical task intensity, mid-1990s to 2015 0.3 0.2 Normalized task intensity index 0.1 0 –0.1 –0.2 –0.3 –0.4 –0.5 –0.6 Be tria Fra m Lu Irel ce m d Un th rg ing ds Cy m Sp s ain Po aly ec lg l ep ia Cr lic Hu tia Po ry Ro d ov lo a Re nia De blic Es ark Fin ia Lit land La ia Sw tvia en Cz Bu ga u xe an lan Sl S ani u do h R ar a n an Ne bou n d K an pr ub ed oa It lgi rtu ng ak ve nm to s pu m hu Au ite erl Western Europe Southern Europe Central Europe Northern Europe Standard employment (SE) Nonstandard employment (NSE) (Continued) Are Distributional Tensions Brewing in Europe and Central Asia? ●  57 BOX 2.5 The Changing Education and Task Profile of Nonstandard Employment (continued) FIGURE B2.5.2 Changes in task content, by employment type (continued) e. Change in routine, cognitive task intensity, mid-1990s to 2015 0.3 0.2 Normalized task intensity index 0.1 0 –0.1 –0.2 –0.3 –0.4 Be tria Fra m Lu rel e m d Un th rg ing ds Cy m Sp s ain Po aly ec lg l ep ia Cr lic Hu atia Po ry Ro nd ov lo a Re nia De blic Es ark Fin ia Lit land La ia Sw tvia en Cz Bu ga u nc xe an Sl S ani u do h R ar a n an Ne bou d K an pr ub la ed It lgi rtu ng ak ve nm to s pu o m hu Au ite erl I Western Europe Southern Europe Central Europe Northern Europe Standard employment (SE) Nonstandard employment (NSE) Source: Calculations based on data of EU-LFS (European Union Labour Force Survey) (database), Eurostat, European Commission, Luxembourg, http://ec.europa.eu/eurostat/statistics-explained/index .php/EU_labour_force_survey_%E2%80%93_data_and_publication. a. A notable outlier is Hungary, where intensity in manual tasks, both routine and nonroutine, has risen considerably in nonstandard employment. This derives from a surge in the share of nonstandard employment among agricultural laborers and garbage collectors, which represented 27 percent of nonstandard employment in Hungary in 2015, while the same occupations constituted only 1 percent in 1997. while the share of the remaining age-groups expanded from close to 20 percent to around 30 percent. A similar pattern is observed in Central and Northern Europe (figure 2.13). In the latter, while the overall share of nonstandard employment did not change, the share of nonstandard employment rose by around 20 percentage points among the 20–24 age-group. In Central Europe, younger workers always show a greater share of nonstandard employment relative to older individuals, while the rise in the share of nonstandard employment was greatest among the young. The significant increases in nonstandard employment among the young were largely driven by a rise in temporary employment in Croatia, Hungary, Italy, Poland, Portugal, and Slovenia and, to a lesser degree, in France and Sweden. Part-time employment also grew substantially as a share of youth employment (from 15 percent to 25 percent), including involuntary part-time employment.18 58  ●   Toward a New Social Contract FIGURE 2.12  Rising nonstandard employment (NSE), Southern and Western Europe a. Southern Europe b. Western Europe 80 80 60 60 Share of NSE (%) Share of NSE (%) 40 40 20 20 0 0 1982 1985 1990 1995 2000 2005 2010 2013 1982 1985 1990 1995 2000 2005 2010 2013 Year Year Age group 20−24 25−29 30−34 35−39 40−44 Source: World Bank calculations based on data of labor force surveys. Note: Each line depicts the smoothed (locally weighted regression) average of the prevalence of nonstandard employment (temporary and part- time employment) by age-group. FIGURE 2.13  Rising nonstandard employment (NSE), Central and Northern Europe a. Central Europe b. Northern Europe 80 80 60 60 Share of NSE (%) Share of NSE (%) 40 40 20 20 0 0 1997 2000 2005 2010 2013 1982 1985 1990 1995 2000 2005 2010 2013 Year Year Age group 20−24 25−29 30−34 35−39 40−44 Source: World Bank calculations based on data of labor force surveys. Note: Each line depicts the smoothed (locally weighted regression) average of the prevalence of nonstandard employment (temporary and part- time employment) by age-group. Are Distributional Tensions Brewing in Europe and Central Asia? ●  59 Shorter Job Tenure among Younger Workers One consequence of the greater prevalence of temporary contracts—one form of nonstandard employment—may be an increase in employee turnover and, thus, a reduction in average job tenure. Some studies argue that employment regula- tions that impose high costs for firing workers lower the incentives to either hire or fire workers so that job tenure becomes longer (Hopenhayn and Rogerson 1993; Lazear 1990). Employment protection legislation is strongly linked to cross-country differences in tenure levels (Eurofound 2015). Auer and Cazes (2000) find that ­differences in job tenure in Europe, Japan, and the United States derive from differences in labor market institutions and the labor market behavior of workers. ­ Analyzing employer–employee data in Germany, Boockmann and Steffes (2010) find that labor market institutions (mainly work councils) play a pronounced role in reducing mobility and thus prolonging tenure. Because the shift from permanent to temporary contracts reflects an easing of labor market protections for workers and involves reduced costs in shedding workers, it might be expected that this shift was accompanied by a reduction in job tenure. At first glance, however, job tenure seems to have expanded in Europe figure 2.14). In most subregions in 1992–2013, the average job tenure was (­ stable at close to 10.0 years. In the Baltic States, it was close to 7.5 years. The average job tenure rose by almost one year in Southern Europe, the region with the highest average job tenure, more than 12 years in 2013. However, job tenure tends to rise with the unemployment rate; so the rise in Southern Europe, the region affected the most by the 2008–09 financial crisis, is not surprising. Moreover, countries in that region have a high share of long tenured workers, who are generally more difficult to fire during recessions (Abraham and Medoff 1984; Jovanovic 1979). FIGURE 2.14 Average job tenure, by region 12 Average job tenure has been mostly stable in Europe and Central Asia 11 Years of tenure 10 9 8 7 1992 1995 2000 2005 2010 2013 Year Western Europe Southern Europe Central Europe Northern Europe Baltic States Source: Based on data of labor force surveys. Note: Each line depicts the smoothed (locally weighted regression) average of the prevalence of nonstandard employment (temporary and part-time employment) by region. 60  ●   Toward a New Social Contract The picture of overall stability in tenure across Europe may hide diverse trends among older and younger workers. The impact of recent transformations—rapid technological change and the easing of labor market protections—may have dif- ferent effects on job tenure across age-groups. Older workers tend to have longer job tenure relative to younger workers because older workers are often endowed with more specific human capital, and employers are thus less likely to fire them. However, older workers tend to have completed fewer years of education, and the more well-educated (younger) workers are likely to represent the lower costs or greater benefits associated with specific skills. Likewise, technological change may increase the need for retraining and thus drive greater demand for younger skilled workers, who are also further from retirement and thus more suitable for retraining (Rodriguez and Zavodny 2003). In countries with strict labor market regulations, allowing more temporary contracts may induce greater competition between those for whom short-term contracts are usually tailored, young people and the pool of the unemployed (Boeri 1999). Greater competition should raise turnover and reduce average tenure among these groups. Job tenure has decreased among younger workers in Europe. Among the 25–29 age-group, the average job tenure has declined in all regions (figure 2.15). In Southern Europe, for instance, job tenure fell from 4.2 years in 1993 to 3.6 years in 2013, while, in Western Europe, it narrowed from 4.4 years to 3.5 years, decreases of 15 percent and 20 percent, respectively. The reduction was smaller among older age-groups: around 10 percent (1.5 years) for the 45–49 age-group in the same regions and between 7 percent and no change among the 60–64 age-group. The decline in job tenure among younger generations is evident even after one con- trols for cyclical and composition effects (Bussolo, Capelle, and Winkler 2018). FIGURE 2.15  Tenure is decreasing among the young, but less among the middle and older age-groups Average job tenture, by age group a. Age 25−29 b. Age 45−49 c. Age 60−64 25 25 25 20 20 20 15 15 15 Logarithmic scale Years of tenure 10 10 10 5 5 5 2 2 2 1992 1999 2006 2013 1992 1999 2006 2013 1992 1999 2006 2013 Year Year Year Western Europe Southern Europe Central Europe Northern Europe Baltic States Source: Based on data of labor force surveys. Note: Each line depicts the smoothed (locally weighted regression) average of job tenure by age-group. The vertical axis is expressed in logarithmic terms. Are Distributional Tensions Brewing in Europe and Central Asia? ●  61 The stability of overall average job tenure results from a change in the composition of the labor force. While job tenure has fallen among younger individuals, older individuals with, on average, long job tenures have considerably boosted their labor force participation rate. In this sense, the trends in average tenure seem to be more in line with the predictions of Boeri (1999) in terms of changes in employ- ment regulation rather than those of Rodriguez and Zavodny (2003) with respect to technological change. The evidence that younger generations are facing shorter job tenures is in line with the findings of Eurofound (2015), which show that the trend is associated with the weakening of employment protection provided by law and by trade unions. Given the weak individual bargaining power of young work- ers, the trend toward the interpersonal employment relationship may affect them disproportionately. O’Higgins (2010) argues that the increased flexibility of employ- ment protection in transition economies in Europe and Central Asia has particularly affected the job stability of young people, among whom average tenure has fallen to the levels in Western European countries (figure 2.15). The Young Are Faring Worse than Older Generations In modern societies, children are expected to achieve, over the course of their lifetimes, a better living standard than their parents (Chetty et al. 2016). This is typically what happens in a growing economy because the average productivity of the young workers entering the labor force exceeds that of older workers when they entered the labor force. Thus, even though the greater experience of older workers means that their productivity and earnings will exceed that of younger workers at any given time, as the young workers age they will catch up and eventu- ally surpass the older generation. However, a slowdown in growth may interrupt this process and reduce the difference in earnings between younger and older generations. The recent slowdown in growth—not entirely caused by the global financial crisis—affected the income prospects of younger generations disproportionately. In Northern and Western Europe, the income growth rate among older household heads—tied to previous trends in income given the contributive nature of most pension systems—has been constant throughout the last two decades, while the income growth rate among younger household heads has declined. The income of middle-aged household heads—the 45–54 age-group—in these two regions showed a pattern similar to that of older household heads, suggesting that the slowdown in growth particularly affected the young (figure 2.16). In Southern Europe, moreover, the income of younger household heads has declined, while that of older household heads continued to grow. Thus, the youngest generations—today’s 25–34 age-group—have experienced much slower income ­ growth relative to older household heads across the EU15.19 Contrasting with the western, more developed countries of the region, in Central Europe, the Baltic States, Russia, and Turkey households incomes for all age groups have increased in the past decade (figure 2.17). Only in Turkey does the income of older heads of household show a slower growth rate, providing an explanation for the steady decline in the ratio of the average incomes of the older group to the younger group. The incomes of the 25–34 and 45–54 age-groups is 62  ●   Toward a New Social Contract FIGURE 2.16  Household income, by age of household head, Western, Northern, and Southern Europe a. Western Europe b. Northern Europe c. Southern Europe 30,000 30,000 30,000 Equivalized household income, in Euros at PPP 25,000 25,000 25,000 20,000 20,000 20,000 15,000 15,000 15,000 10,000 10,000 10,000 5,000 5,000 5,000 1995 2000 2005 2010 2015 1995 2000 2005 2010 2015 1995 2000 2005 2010 2015 Year Year Year Age of household head Age 25−34 Age 45−54 Age 65+ Sources: Based on data of ECHP (European Community Household Panel) (database), Eurostat, European Commission, Luxembourg, http://ec​ europa​ .­ eu/eurostat/web/microdata/european-community-household-panel; EU-SILC (European Union Statistics on Income and Living Conditions) (database), .­ -living-conditions. Eurostat, European Commission, Luxembourg, http://ec.europa.eu/eurostat/web/microdata/european-union-statistics-on-income-and​ Note: PPP = purchasing power parity. FIGURE 2.17  Household income, by age of household head, Central Europe, Baltic States, Russian Federation, and Turkey a. Central Europe b. Baltic States 15,000 15,000 Euros at PPP Euros at PPP 10,000 10,000 5,000 Equivalized household income 5,000 2004 2006 2008 2010 2012 2014 2004 2006 2008 2010 2012 2014 Year Year c. Russian Federation d. Turkey 12,000 40,000 RUB at constant prices TRL at constant prices 7,000 25,000 2,000 10,000 1995 2000 2005 2010 2015 2000 2005 2010 2015 Year Year Age of household head Age 25−34 Age 45−54 Age 65+ Sources: Based on data of EU-SILC (European Union Statistics on Income and Living Conditions) (database), Eurostat, European Commission, Luxembourg, http://ec.europa.eu/eurostat/web/microdata/european-union-statistics-on-income-and-living-conditions; household income, consumption, and expenditure surveys; RLMS–HSE (Russia Longitudinal Monitoring Survey–Higher School of Economics) (database), Higher School of Economics, National Research University, Moscow, http://www.hse.ru/en/rlms/. Note: PPP = purchasing power parity. Are Distributional Tensions Brewing in Europe and Central Asia? ●  63 similar in all transition countries, while, in Turkey and the rest of Europe, the middle age-group enjoys significantly higher incomes. This suggests that the wage returns to experience is relatively low in transition economies. The earnings prospects of young generations in Southern and Western Europe appear to be deteriorating. The flat profile of the earnings of workers with only a high school diploma suggests these individuals are receiving low returns to experi- ence (figures 2.18 and 2.19, panel a).20 Moreover, average earnings did not improve among successive cohorts. The impact of the 2008–09 financial crisis may FIGURE 2.18 a. High school only b. College graduates 18,000 18,000 Average annual earnings, 30–34 age-group, Southern Europe, 2004–14 15,500 15,500 Euros at PPP 13,000 13,000 10,500 10,500 8,000 8,000 2004 2006 2008 2010 2012 2014 2004 2006 2008 2010 2012 2014 Year Year Year of birth 1974 1975 1976 1977 1978 1979 1980 1981 Source: Based on data of EU-SILC (European Union Statistics on Income and Living Conditions) (database), Eurostat, European Commission, Luxembourg, http://ec.europa.eu/eurostat/web​ /­microdata/european-union-statistics-on-income-and-living-conditions. Note: Each line depicts the smoothed (locally weighted regression) average labor market earnings from age 30 to age 34 of each birth cohort. PPP = purchasing power parity. FIGURE 2.19 a. High school only b. College graduates 30,000 30,000 Average annual earnings, 30–34 age-group, Western 27,000 27,000 Europe, 2004–14 24,000 24,000 Euros at PPP 21,000 21,000 18,000 18,000 15,000 15,000 12,000 12,000 2004 2006 2008 2010 2012 2014 2004 2006 2008 2010 2012 2014 Year Year Year of birth 1974 1975 1976 1977 1978 1979 1980 1981 Source: Based on data of EU-SILC (European Union Statistics on Income and Living Conditions) (database), Eurostat, European Commission, Luxembourg, http://ec.europa.eu/eurostat/web​ /­microdata/european-union-statistics-on-income-and-living-conditions. Note: Each line depicts the smoothed (locally weighted regression) average labor market earnings from age 30 to age 34 of each birth cohort. PPP = purchasing power parity. 64  ●   Toward a New Social Contract be partly responsible for this lack of growth. However, the wage profile is similar even among cohorts that were 30–34 years old before the crisis, suggesting that low earnings prospects are a structural characteristic of the labor market among individuals with only a high school diploma. The wages of college graduates from older generations did rise substantially (the slope of the wage profile in figures 2.18 and 2.19 is steep), but the youngest cohorts that entered their prime earnings years during or after the financial crisis, experienced a considerably lower increase in wages. In Southern Europe, the youngest generations have a practically flat wage profile, similar to that of workers with only a high school diploma. The flat- tening of the wage profile among young college graduates may reflect either a decline in returns to experience or greater job turnover. More frequent shifts in and out of jobs or even between jobs reduce the average wage, especially among younger generations. Indeed, there is some evidence to support the second hypothesis: there has been a decline in the average job tenure of younger genera- tions relative to older generations in Europe (see above). Trends in wages across generations have been more varied in Central and Northern Europe (figures 2.20 and 2.21). The average income of workers with only a high school diploma has increased across successive cohorts. Among college graduates, the flattening of the wage profile is evident in Central Europe, albeit with a smaller magnitude than in Southern and Western Europe. In Northern Europe, cohorts entering the prime earnings period during the crisis (cohorts born in 1978 and 1979, for instance) faced only limited earnings growth thereafter, but this pattern is reversed among later generations, who enjoyed income increases similar to those of generations born in the early 1970s. FIGURE 2.20 a. High school only b. College graduates Average annual earnings, 14,000 14,000 30–34 age-group, Central Europe, 2004–14 11,500 11,500 Euros at PPP 9,000 9,000 6,500 6,500 4,000 4,000 2004 2006 2008 2010 2012 2014 2004 2006 2008 2010 2012 2014 Year Year Year of birth 1974 1975 1976 1977 1978 1979 1980 1981 Source: Based on data of EU-SILC (European Union Statistics on Income and Living Conditions) (database), Eurostat, European Commission, Luxembourg, http://ec.europa.eu/eurostat/web​ /­microdata/european-union-statistics-on-income-and-living-conditions. Note: Each line depicts the smoothed (locally weighted regression) average labor market earnings from age 30 to age 34 of each birth cohort. PPP = purchasing power parity. Are Distributional Tensions Brewing in Europe and Central Asia? ●  65 FIGURE 2.21 a. High school only b. College graduates 30,000 30,000 Average annual earnings, 30–34 age-group, Northern Europe, 2004–14 25,000 25,000 Euros at PPP 20,000 20,000 15,000 15,000 2004 2006 2008 2010 2012 2014 2004 2006 2008 2010 2012 2014 Year Year Year of birth 1974 1975 1976 1977 1978 1979 1980 1981 Source: Based on data of EU-SILC (European Union Statistics on Income and Living Conditions) (database), Eurostat, European Commission, Luxembourg, http://ec.europa.eu/eurostat/web​ /­microdata/european-union-statistics-on-income-and-living-conditions. Note: Each line depicts the smoothed (locally weighted regression) average labor market earnings from age 30 to age 34 of each birth cohort. PPP = purchasing power parity. Increased Inequality Among the Young Weak income growth among younger generations in Southern and Western Europe has been accompanied by widening inequality. While income inequal- ity across a given generation tends to rise over time, younger generations in Southern and Western Europe are facing higher income inequality at every point of the life cycle relative to older generations.21 For example, Bussolo, Jappelli, Nisticò, and Torre (2018) find that income inequality among Italians born in the 1930s was similar to that of a relatively equal country, such as Japan (Gini coefficient of approximately 0.31) (­ figure 2.22).22 By contrast, income inequality among Italians born in the 1980s was similar to a highly unequal country such as Chile (Gini coefficient of approximately 0.48). The equivalent intergenerational rise in the Gini coefficient was much smaller in Germany (4 points) and in France (1 point). The intergenerational rise in inequality is even higher if the analysis is restricted to labor income rather than total income.23 So, successive generations are experiencing an increase in inequality that exceeds the amount expected as generations age. Persistent Spatial Disparities across the Region As technological change, agglomeration economies, trade, and other market forces transform the economic and labor market landscape across countries, some individuals may be systematically excluded from economic opportunities. If place of birth or residence limits the access of people to quality education or good jobs, it will also limit their productive capacity and their opportunities to 66  ●   Toward a New Social Contract FIGURE 2.22 Income inequality by birth cohort Income inequality is much 0.50 higher among cohorts born in the 1980s Implied Gini coefficient at age 40 0.45 Household equivalized income 0.40 0.35 0.30 1930 1940 1950 1960 1970 1980 Birth cohort born in... France Germany Italy Source: Bussolo, Jappelli, Nisticò, and Torre 2018. join the middle class, and may thus fuel discontent and perceptions of unfair- ness. Evidence on Europe and Central Asia points to place of birth as an important factor in the inequality of opportunity in gaining access to tertiary ­ education, a job, and higher income (EBRD 2016). Moreover, despite the rela- tively high international emigration rates in many countries in Eastern Europe and Central Asia, internal mobility rates are low, reflecting limited opportunities to move to obtain better jobs (Arias et al. 2014). Recognizing that economic growth may be an unbalanced process (World Bank 2009), spatial disparities, particularly if persistent and not mitigated by targeted policies to promote con- vergence in living standards, can contribute to rising distributional tensions and populism. People in some places may feel left behind and sense their restricted ability to influence policy making and the allocation of resources in society (World Bank 2017a). This section explores trends in such spatial disparities in Europe and Central Asia. Spatial Disparities Are Common in the Region Differences in income persist between regions in many countries of Europe and Central Asia. National indicators of welfare may mask even vast differences across regions within countries. The use of the coefficient of variation as a measure of disparities in consumption or disposable income between regions reveals that spatial inequalities in welfare are common within countries. The greatest inequali- ties occur in the Slovak Republic, Tajikistan, and Russia and the lowest in Denmark (figure 2.23). In the European Union, the highest disparities are in Southern Europe, including Greece, Italy, and Spain, where levels are higher than the OECD average of 0.14 (OECD 2016). The varying geographical aggregations at which disparities Are Distributional Tensions Brewing in Europe and Central Asia? ●  67 FIGURE 2.23 Slovak Republic Tajikistan Spatial disparities in welfare Russian Federation are not uncommon in the Estonia region Montenegro Coefficient of variation of Moldova Belarus disposable income or Italy consumption, by region, Armenia circa 2013 Serbia Greece Spain Kazakhstan Kyrgyz Republic United Kingdom Czech Republic Georgia Country Portugal Romania Hungary Poland Finland Albania Germany Ireland Sweden Netherlands Belgium Poland France Slovenia Norway Ukraine Austria Denmark 0 0.10 0.20 0.30 0.40 0.50 Coefficient of variation Sources: OECD 2016; World Bank calculations rely on harmonized data on other, non-OECD countries. are measured across several countries may pose a challenge in comparing inequal- ities across regions. Nonetheless, the disparities are also evident between urban and rural areas. A comparison between urban and rural areas using a welfare index constructed based on information on durables and the socioeconomic character- istics of households in the 2016 round of the Life in Transition Survey (LiTS) shows that living standards are higher in urban areas than in rural areas in all countries of Europe and Central Asia covered by the data except Greece (figure 2.24).24 The greatest urban–rural disparities by this measure occur in Bulgaria, Georgia, Romania, and Tajikistan. The share of inequality explained by inequality between within-country regions has risen in some countries. One summary measure of this regional inequality indicates the inequality between geographical areas in average per capita consumption as a share of the maximum possible inequality between these areas, taking into account the size and number of regions.25 68  ●   Toward a New Social Contract FIGURE 2.24 0.15 Gaps between urban and Urban–rural gap in welfare index rural areas are largest in 0.12 Georgia and Tajikistan and 0.09 are negative only in Greece 0.06 Urban–rural gap in welfare index, 2015 0.03 0 –0.03 Es ece Hu onia R Cr ary ed a Fe Lat a de via Ka love n M akh nia te an Po gro Az oso d er vo Be ijan Ar larus Cy nia Uk rus Sl L Serb e ov ith ia Re nia rg Al blic Uz epub ia be lic ol n Bu dova ma a Ge nia jik a an ac ti i S tio K an in M ista Ro gari Ta orgi on R n on st ist M oa ng me p ra ak ua yz ba ne e pu l ba ra t Gr k l z ian FY Ky ss Ru Country Source: Calculations based on data of the 2016 round, LiTS (Life in Transition Survey) (database), European Bank for Reconstruction and Development, London, http://www.ebrd.com/what-we-do​ /­economic-research-and-data/data/lits.html. Notes: The welfare index ranges from 0 to 1 and is constructed using a principal component analysis and 12 variables of household durables, including phone (landline or cell), computer, washing machine, car, bike, motorbike, as well as proxies for household socioeconomic status, including Internet access, adequate heating, a week’s holiday each year, a meal of meat, chicken, or fish every second day, ability to meet unexpected expenses through own resources equivalent to the national poverty threshold, and access to a bank account. All variables are transformed to 0 if the household cannot afford the asset and 1 if the household is in possession of the asset or does not have it for other reasons, for example, Internet is not available at the location of the household. The importance of inequality between within-country regions and between urban and rural areas has increased in several countries, although data are not available for all countries of Europe and Central Asia (figures 2.25 and 2.26). Increases are noticeable in Armenia, Moldova, and Serbia. Between-region inequality has narrowed in the Kyrgyz Republic, in addition to inequality between urban and rural areas in Kazakhstan and Poland, for instance. Spatial disparities in welfare have increased in many countries in the region. Despite increases in average household consumption over the past decade, inequalities across geographical areas persist and have increased in several coun- tries. The gap in consumption between urban and rural areas has widened in 10 of the 14 countries depicted in figure 2.27 (panel a), mostly in the eastern part of the region, and the gap between the richest and poorest regions has increased in 12 of the countries (figure 2.27, panel b). Regional disparities have also widened in the EU. Regional disparities in dispos- able income within countries increased over the last two decades in some coun- tries in Southern Europe as well as in other European countries such as Belgium, the Czech Republic, the Netherlands, the Slovak Republic, and the United Kingdom (figures 2.28 and 2.29). Regional disparities declined in others, such as Finland and Germany. Focusing on regional output measured by per capita GDP within coun- try regions, the coefficient of variation shows an average rising trend in regional disparities within EU countries in 2000–15. Thus, some regions are lagging, despite a reduction in within-country inequality that led to a convergence in the EU. Pooling all within-country regions in the EU also indicates there was an increase in dispersion during this period. Are Distributional Tensions Brewing in Europe and Central Asia? ●  69 FIGURE 2.25 30 Between-region inequality has widened in some Share of maximum spatial inequality circa 2003 (%) 25 countries 20 BLR 15 KAZ MDA ARM SRB 10 KGZ MNE RUS TJK ROU GEO 5 POL ALB UKR 0 0 5 10 15 20 25 30 Share of maximum spatial inequality circa 2003 (%) Source: World Bank calculations based on harmonized mean consumption data on 14 countries in Europe and Central Asia. Note: Maximum spatial inequality corresponds to a scenario where, given the size and the ranking of the regions in terms of mean income, households with the lowest incomes are allocated to the poorest regions, while households with the highest incomes are allocated to the richest regions, and households in the middle are allocated in similar fashion to the remaining regions. The values illustrated in the figure express between-region inequality, that is, the average difference in mean incomes, as a ratio of maximum spatial inequality. See Elbers, Lanjouw, and Lanjouw (2003) for a detailed explanation of the methodology. Dotted line represents no change in values between 2003 and 2013. In line with spatial differences in living standards, the concentration of poverty also has a spatial dimension. Subnational poverty rates vary significantly within countries. In Tajikistan, for instance, the poverty rate—the share of people with incomes below US$5.50 a day in constant 2011 prices in U.S. dollars purchasing power parity (PPP)—in the poorest region is 72 percent, more than twice the rate of the region with the lowest rate (31 percent). In Romania, the poverty rate in the least well-off region is two and a half times higher than the rate in the wealthiest region. Similarly, the at-risk-of-poverty measure at the Nomenclature of Territorial Units for Statistics–3 level shows that the poorest region in France has a rate three times higher than the rate in the region with the lowest poverty rate; this ratio is around seven in the United Kingdom.26 The spatial concentration of poverty is rising. As living standards have improved, poverty rates have declined across countries in the last decade or so. However, particularly important for potential concerns over emerging distributional tensions is the accompanying rise in geographical dispersion. In seven countries in the eastern part of the region where the poverty rates—the share of people living on less than US$5.50 a day at 2011 constant PPP prices—are more than 10 percent, the difference in poverty rates across regions (measured by the coefficient of varia- tion) has increased (figure 2.30). In Armenia, for example, the difference in poverty rates between the less well-off and the more well-off regions rose from 25 percent- age points to 38 percentage points in 2003–2013. 70  ●   Toward a New Social Contract FIGURE 2.26 30 Inequality between urban and rural areas has increased in some countries 25 Share of maximum spatial inequality circa 2013 (%) 20 MDA RUS 15 ROU BLR GEO 10 KAZ SRB POL TJK KGZ 5 UKR ARM ALB MNE 0 0 5 10 15 20 25 30 Share of maximum spatial inequality circa 2003 (%) Source: World Bank calculations based on harmonized mean consumption data on 14 countries in Europe and Central Asia. Note: Maximum spatial inequality corresponds to a scenario where, given the size and the ranking of the regions in terms of mean income, households with the lowest incomes are allocated to the poorest regions, while households with the highest incomes are allocated to the richest regions, and households in the middle are allocated in similar fashion to the remaining regions. The values illustrated in the figure express between-region inequality, that is, the average difference in mean incomes, as a ratio of maximum spatial inequality. See Elbers, Lanjouw, and Lanjouw (2003) for a detailed explanation of the methodology. Dotted line represents no change in values between 2003 and 2013. FIGURE 2.27 a. Gaps, urban and rural areas Gaps in mean consumption, Ukraine circa 2003–13 Tajikistan Serbia Russian Federation Romania Poland Country Montenegro Moldova Kyrgyz Republic Kazakhstan Georgia Belarus Armenia Albania 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 US$ at 2011 purchasing power parity Earliest Latest (Continued) Are Distributional Tensions Brewing in Europe and Central Asia? ●  71 FIGURE 2.27 b. Gaps, regions Gaps in mean consumption, Ukraine circa 2003–13 (continued) Tajikistan Serbia Russian Federation Romania Poland Montenegro Country Moldova Kyrgyz Republic Kazakhstan Georgia Belarus Armenia Albania 0 1,000 2,000 3,000 4,000 5,000 6,000 US$ at 2011 purchasing power parity Earliest Latest Source: World Bank calculations based on harmonized mean consumption data on 14 countries in Europe and Central Asia. 115 FIGURE 2.28 Between-region spatial inequalities within countries Coefficient of variation in gross domestic product (GDP) 110 have increased in the 105 European Union per capita, Index 2000 = 100 100 95 90 85 80 75 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 00 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Year Country Within country NUTS-3 (unweighted) Source: Farole, Goga, and Ionescu-Heroiu 2018. Note: Country refers to the coefficient of variation across European Union countries, signaling convergence in gross domestic product (GDP). The within-country coefficient of variation measure is at the Nomenclature of Territorial Units for Statistics–3 level (NUTS-3). GDP per capita is measured in purchasing power standard (PPS). 72  ●   Toward a New Social Contract FIGURE 2.29 0.50 Regional disparities in disposable income rose, were unchanged, or 0.40 declined Coefficient of variation Coefficient of variation in 0.30 regional disposable income, 1995 and 2014 0.20 0.10 0 Au rk No ia Sl ay Fra a Be ce th m Sw s Ire n Ge and Fin y Po d Hu nd Cz Po ry Un Re gal d K blic m ain ce ov Est y Re ia c d an l bli i e lan Ita a str en a do ak on Ne lgiu rw n lan ee ed la nm ng ec rtu Sp rm ite pu pu l ov ing Gr er De h Sl Country 1995 2014 Source: OECD 2016. FIGURE 2.30 0.7 The spatial dispersion of poverty rates has increased Coefficient of variation of 0.6 poverty rates at US$5.50 a Regional coefficient of variation in latest period day (2011 purchasing power MDA 0.5 parity), circa 2003–13 0.4 0.3 ROU TJK ARM 0.2 ALB GEO KGZ 0.1 0 0 0.2 0.4 0.6 0.8 Regional coefficient of variation in earliest period Source: World Bank calculations using harmonized mean consumption data of 7 countries in Europe and Central Asia. Note: Dotted line indicates no change in value between earliest and latest period. Are Distributional Tensions Brewing in Europe and Central Asia? ●  73 Access to Opportunities Is More Limited among Residents of Certain Areas Differences in individual endowments are a major reason for disparities in welfare between regions and between rural and urban residents. Income may be lower because of the characteristics of individuals in an area; for example, they may not be as well educated—education, the age of the head of household, and house- hold demographic composition and size are considered here—or because the returns to these characteristics are lower given location-specific factors. In 14 countries in Europe and Central Asia on which harmonized data are available, characteristics play a key role in driving income disparities between geographical areas, such as between leading versus lagging regions, the richest versus the poor- est regions, and, especially, urban versus rural areas (figure 2.31). In most of the countries under consideration, the educational attainment of the household head accounts for a large share of the characteristics component, pointing to the influ- ence of education on gaps in living standards. The role of education is not surpris- ing, given the close correlation of educational attainment with welfare in the region. There are also spatial gaps in schooling quality. Europe and Central Asia stands out as one of the regions with the highest educational attainment and learning outcomes. Yet, gaps in access remain, particularly among certain groups and areas, and schooling does not always translate into learning. For example, across FIGURE 2.31  Differences in characteristics and in returns to characteristics help explain welfare gaps across geographical areas, circa 2013 Decomposition of spatial disparities in mean consumption per capita in characteristics vs. returns (share of the gap explained by each component) a. Rural vs. urban b. Leading vs. lagging region c. Richest vs. poorest region ALB UKR UKR MNE KGZ KGZ KGZ POL ALB POL KAZ KAZ KAZ MNE POL UKR ROU ROU SRB MDA RUS TJK RUS SRB ROU ALB MNE MDA SRB MDA RUS ARM ARM ARM GEO GEO BLR TJK BLR GEO BLR TJK –150 –100 –50 0 50 100 150 200 250 –40 –20 0 20 40 60 80 100 120 140 –25 –5 15 35 55 75 95 115 Percent Percent Percent Characteristics Returns Source: Calculations using harmonized mean consumption data on 14 countries in Europe and Central Asia. Note: Laggers are defined based on average harmonized mean consumption below the national average. Regressors include demographics (age and gender of the household head), educational attainment of household heads, and household composition (size, demographic composition). The panels indicate the dominant explanation for welfare disparities between geographical areas: characteristics, returns to characteristics, or both. In the Kyrgyz Republic, for example, 93 percent of the welfare difference between the richest and the poorest regions is explained by differences in the characteristics of households and individuals in these areas. In Georgia, 84 percent of the differences derive from difference in the returns to characteristics. Country codes: ALB = Albania; ARM = Armenia; BLR = Belarus; GEO = Georgia; KAZ = Kazakhstan; KGZ = Kyrgyz Republic; MDA = Moldova; MNE = Montenegro; POL = Poland; ROU = Romania; RUS = Russian Federation; SRB = Serbia; TJK = Tajikistan; UKR = Ukraine. 74  ●   Toward a New Social Contract all countries in Europe and Central Asia that implemented the 2015 round of the test of the Programme for International Student Assessment (PISA), a share of students were found not to have developed the foundational cognitive skills they need to succeed in the labor market.27 The share of functionally illiterate students (15-year-old students who scored below level 2 on the PISA reading section) was 77 percent in Kosovo and 71 percent in FYR Macedonia, but also quite high in EU countries such as Bulgaria (42 percent), Romania (39 percent), and the Slovak Republic (32 percent). Disadvantages in schooling quality are evident along the spatial dimension, creating a divide based on geographical location. Thus, chil- dren in a same grade in a same country are losing out on accessing quality educa- tion depending on where they live. The largest gaps in 2015 PISA scores in countries on which data are available occurred in Bulgaria and Moldova (around a two-year schooling gap between urban and rural areas), followed by the Slovak Republic and Romania (figure 2.32). In Romania, 83 percent of low-performing schools are in rural areas (World Bank 2018a). Other factors may also lead to lower productivity and lower returns in certain geographical areas. Across countries, differences in returns are relevant in explain- ing disparities in mean consumption between rural and urban areas and between regions (see figure 2.31). These differences in returns may be capturing spatial gaps in public service delivery, service infrastructure, communication networks, access to markets, local governance coverage, social capital, or the business envi- ronment. Access to basic services has expanded in many countries. Yet, a spatial divide exists, including differences in quality. In Russia, some regions are system- atically affected by the limited presence of the state (box 2.6). In Moldova, 95 ­percent of the urban population is connected to piped water, but this is so among only 54 percent of the rural population, of which only 39 percent have the service within the dwelling (World Bank 2016b). Albania presents a similar situa- tion. There, the share of households with access to a steady water supply ranges from 47 percent in Durrës County to 88 percent in Shkodër County (World Bank 2015). Access to services does not refer only to basic infrastructure, which is more relevant in the developing countries in the region, but also access to technology. The Czech Republic, France, Portugal, and Spain show regional gaps of around 20 percentage points in the share of households with broadband connections.28 Quality may also vary. In Moldova, firms face lower-quality services depend- ing on where they are located (figure 2.33). Other region-related factors may likewise keep some areas from providing residents with access to opportuni- ties. Evidence from the eastern and the western parts of the region—Kazakhstan and Italy—shows that firms face a business environment that differs depending on location.29 Even if gaps in access to education and other services were to be addressed, the difference in returns across regions cannot be bridged if barriers to internal mobility limit the ability of residents in some areas to benefit from agglomera- tion and urbanization. This is an example of horizontal inequality whereby there are significant impediments in switching between groups. Evidence on the region points to low internal mobility compared with populations in other countries, despite evidence of agglomeration economies and gaps in unem- ployment rates between regions (Arias et al. 2014; Restrepo Cadavid Are Distributional Tensions Brewing in Europe and Central Asia? ●  75 FIGURE 2.32 Bulgaria Moldova Gaps in PISA reading scores: Slovak Republic often equivalent to a year of Romania schooling, urban and rural Hungary Georgia areas Turkey Albania Serbia Lithuania Poland Portugal France Latvia Russian Federation Kazakhstan Czech Republic Netherlands Italy Country Kosovo Greece Switzerland Sweden Norway Macedonia, FYR Finland Estonia Ireland Austria Croatia Spain Iceland Luxembourg Montenegro Denmark Slovenia Germany Belgium United Kingdom –30 –20 –10 0 10 20 30 40 50 60 70 80 Gap in PISA score, points Source: Calculations based on 2015 (2012 for Kazakhstan, Serbia, and Sweden) test scores in PISA (Programme for International Student Assessment) (database), Organisation for Economic Co- operation and Development, Paris, http://www.oecd.org/pisa/pisaproducts/. Note: Urban schools are located in a city or large city (more than 100,000 people). Rural schools are located in a town, a small town, or village, hamlet, rural area (fewer than 100,000 people). A gap in PISA scores of 30 points, covered by the red bracket, is estimated as the equivalent of one year of schooling. See Woessmann (2016). BOX 2.6 A Closer Look at Spatial Disparities in the Russian Federation An understanding of the obstacles in the Russian the Soviet era, labor and capital were forcibly moved Federation’s quest for development begins with the toward the east to exploit Siberia’s vast natural country’s expansive geography and the difficulties in resources, develop military capabilities, and support governing such a vast territory. Russia is the world’s a more even distribution of population and eco- largest country, and its geographical endowments nomic activity. The resulting economic structure was encompass harsh climatic conditions and a domi- physically more dispersed throughout the territory, nance of natural resources in peripheral regions that yet inefficient and distorted. Efforts to reverse this have shaped Russia’s development policies. During policy legacy have often been undermined by the (Continued) 76 ● Toward a New Social Contract BOX 2.6 A Closer Look at Spatial Disparities in the Russian Federation (continued) inherited economic, social, physical, and relational and most populous regions, including Moscow, networks that hindered progress toward more effi- St. Petersburg, and natural resource–rich regions, cient and equitable regional development. inequality is high, meaning the numbers of the Today, Russia has the highest level of inequal- poor are large, though the areas do not exhibit ity among large, emerging economies such as the highest poverty rates. Brazil, China, and India. Russian regions experi- The transformation from unbalanced growth to enced some convergence in income in the last inclusive development requires a shift in policies, decade as poorer regions grew more quickly (con- including a focus on richer regions where poverty trolling for other factors). Moreover, there appear and inequality are concentrated. Russia’s prevailing to be positive spillovers from one region to policy approach since the transition has been more another, that is, factors that raise incomes and equalizing than other countries. Poor regions reduce poverty in one region raise incomes and depend heavily on federal transfers. These drivers reduce poverty in neighboring regions. However, of convergence have become less sustainable, immense disparities in living standards persist. which became evident when Russia underwent the Households in Sakhalin Oblast (which has the recent oil price crisis and sanction regime. The pol- highest gross regional product per capita) experi- icies also appear to have hindered the ability of ence living standards similar to those in Singapore, poor regions to boost their comparative advan- while households in Ingushetia (which has the low- tage. Regions are still characterized by significant est gross regional product per capita) experience disparities in access to services, and some regions living standards closer to those in Honduras. are affected systematically by the low profile of the Poverty rates range from less than 10 percent in state. This invariably translates into disparities in resource-rich Tatarstan and large metropolitan outcomes. Addressing disparities in access to ser- areas of Moscow and St. Petersburg to almost 40 vices and thereby leveling the playing field remains percent in the poorest regions in the North at the heart of policies seeking to improve both Caucuses, Siberia, and the Far East. In the richest efficiency and equity. Source: World Bank 2017b. et al. 2017). Less than 30 percent of the population reported they would be willing to move to another part of the country for a job; younger, single, more well-educated men were more likely to move. Barriers to internal mobility, including those related to weaknesses in housing and land markets, can leave some trapped in lagging areas (chapter 3). Closing spatial disparities by ensuring that people build the human capital they need and that they can access opportunities will lead to more inclusive growth. There are important reasons to pay attention to spatial inequalities, especially their potential to foster location-related discontent. One reason is equity, which implies that location of residence should be neutral with respect to income, educational attainment, ownership of assets, and access to economic opportunity. Another reason is voice and accountability: spatial disparities may affect the agency and bargaining power of people living in different places and their ability to influence policy making and the allocation of resources in society (World Bank 2017d). If individuals in some areas are systematically excluded from Are Distributional Tensions Brewing in Europe and Central Asia? ●  77 FIGURE 2.33  Moldova: indicators of service quality, by region, 2013 a. Electricity services b. Water services 40 30 30 Days or incidents Days 20 20 10 10 0 0 North South Centre Chisinau Balti South North Centre Chisinau Balti Electric connection wait (days) Water connection wait (days) Water outages (number of incidents) Power outages (duration in days) Water outages (duration in days) Source: World Bank 2016b. economic gains and from emerging and changing opportunities given their lower skills, limited labor mobility, or other reasons, spatial inequalities may widen, and some groups will be left further behind. Rising Inequality of Opportunity, Particularly in the East If the access of people to opportunities is determined by circumstances beyond their control, this may lead to distributional tensions and to a growing sense of unfairness. Widening inequality of opportunity can also impair aggregate growth. It has been suggested that the existence of strong and persistent inequalities in the initial opportunities open to individuals can generate inequality traps that rep- resent severe constraints to the future growth of an economy by preventing entire groups from full participation in economic and social life (Bourguignon, Ferreira, and Menéndez 2007; World Bank 2006).30 Inequality of Opportunity Is Declining, but Is Still Evident in the West Inequality of opportunity—the impact of circumstances at birth on income and other welfare outcomes—is an important reason behind the existence of income inequal- ity in Europe. The contribution of inequality of opportunity to income accounts for between 25 percent and 60 percent of total income inequality (measured according to the Gini index) in most European countries (Checchi, Peragine, and Serlenga 2016).31 Moreover, inequality of opportunity and income inequality were strongly related across Europe in both 2005 and 2011 (figure 2.34). The Nordic countries had low levels of income and opportunity inequality; the Mediterranean and Continental European countries exhibited intermediate levels in these inequality dimensions; and the Eastern European countries showed a relatively high level of income inequality, but were more mixed with respect to inequality of opportunity. There were no significant changes in the estimated level of inequality of oppor- tunity in income in 2005–11 (comparing the two panels of figure 2.34). Measuring inequality of opportunity over longer periods is difficult, given the lack of data. 78  ●   Toward a New Social Contract FIGURE 2.34 Total inequality and opportunity inequality—disposable income Income inequality, Europe, a. 2005 b. 2011 2005 and 2011 0.20 0.20 CH DE Ex−ante opportunity inequality—Gini Ex−ante opportunity inequality—Gini LU DE LU UK NL disposable incomes disposable incomes 0.15 EL 0.15 AT IE EL ES UK AT IT EE PT NL RO BE NO PL IT BG PL HU LV CZ HU HR EE FR BE CZ LT FR ES IE LV NO 0.10 FI 0.10 PT LT SE FI DK SI SE SI DK 0.05 0.05 0.20 0.25 0.30 0.35 0.40 0.45 0.20 0.25 0.30 0.35 0.40 0.45 Total inequality—Gini disposable incomes Total inequality—Gini disposable incomes Source: Checchi, Peragine, and Serlenga 2016. However, for the four largest economies in the EU, namely, France, Germany, Italy, and the United Kingdom, the data are sufficient for this measurement.32 Depending on how inequality is calculated, inequality of opportunity accounts for about a third (based on the mean log deviation calculation) to a half (based on the Gini coefficient) of total income inequality in the four countries (figure 2.35, panel a, for the Gini). Inequality of opportunity in income in these countries was either stable or weakly decreasing (in Germany and the United Kingdom) in 1993–2014, sug- gesting that inequality of opportunity reflects embedded features of national socioeconomic systems that are little affected by temporary changes in economic activity. Inequality of opportunity with respect to education declined steadily, especially in Italy, over the period (figure 2.35, panel b). Inequality of opportunity tends to decline with age (figure 2.36).33 The observed inequality of opportunity exhibits an inverted U-shaped pattern over the life cycle; in France and Italy, it has a clearer decreasing pattern. After a certain age, which varies across countries, the effect of the circumstances at birth seem to weaken. This differs from the pattern of income or consumption inequality, which generally rises with age. The cohort analysis shows a more mixed picture: in Germany and the United Kingdom, inequality of opportunity declines across generations, that is, the younger generation experience less inequality of opportunity, while the data on France and Italy are characterized by an inverted U. Improved access to education in Europe is not always associated with declines in inequality of opportunity across generations. Figure 2.37 shows, for Germany and Italy, the contribution to inequality of opportunity of (1) the intergenerational persistence in education (whether an individual’s educational attainment is strongly related to the educational attainment of the parents), (2) the returns to education in the labor market, and (3) networking activity associated with parental back- ground.34 In Italy, the first two variables decline, which, other things being equal, should produce a decline in inequality of opportunity, but the third variable rises; so inequality of opportunity is roughly the same at the end of the period as at the beginning. In Germany, the contribution of the returns to education falls, while the Are Distributional Tensions Brewing in Europe and Central Asia? ●  79 FIGURE 2.35 a. Income Trends in inequality of France Germany opportunity: France, 0.6 0.6 Germany, Italy, United Kingdom 0.4 0.4 Gini coefficient of income 0.2 0.2 1980 1990 2000 2010 1980 1990 2000 2010 Year Year Great Britain Italy 0.6 0.6 0.4 0.4 0.2 0.2 1980 1990 2000 2010 1980 1990 2000 2010 Year Year Total inequality Absolute inequality of opportunity b. Education France Germany 0.4 0.4 0.3 0.3 0.2 0.2 Gini coefficient years of education 0.1 0.1 0 0 1980 1990 2000 2010 1980 1990 2000 2010 Year Year United Kingdom Italy 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 1980 1990 2000 2010 1980 1990 2000 2010 Year Year Total inequality Absolute inequality of opportunity Source: Bussolo, Checchi, and Peragine 2018. Note: The education outcome is measured by years of education. contribution of parental networking is roughly unchanged by the end of the period; so inequality of opportunity declines. Thus, increased equality of opportunity in education, mainly because of the expansion of tertiary education in most countries in recent decades, has not always resulted in greater equality of opportunity in income. Three of the many possible explanations are likely important. First, parental networking may be playing a role in helping young workers find desirable jobs. Second, the decline in the returns to education may be loosening the link between educational 80  ●   Toward a New Social Contract FIGURE 2.36  Decomposition of inequality of opportunity in age and cohort effects, France, Germany, Italy, United Kingdom Inequality of opportunities––Gini coefficient personal income a. France b. Germany Age contribution Age contribution 0.06 0.4 0.04 0.3 0.2 0.02 0.1 0 0 30 40 50 60 70 80 30 40 50 60 70 80 Age Age Cohort contribution Cohort contribution 0.20 0.4 0.15 0.3 0.10 0.2 0.05 0.1 0 0 1920 1940 1960 1980 2000 1920 1940 1960 1980 2000 Born in... Born in... c. United Kingdom d. Italy Age contribution Age contribution 0.25 0.30 0.20 0.25 0.15 0.20 0.15 0.10 0.10 0.05 0.05 30 40 50 60 70 80 30 40 50 60 70 80 Age Age Cohort contribution Cohort contribution 0.3 0.10 0.2 0.05 0.1 0 0 –0.05 1920 1940 1960 1980 2000 1920 1940 1960 1980 Born in... Born in... Source: Bussolo, Checchi, and Peragine 2018. Note: Estimated contributions of birth cohorts and age-groups in explaining the dynamics of inequality of opportunity, according to Deaton’s decomposition; see Deaton (1997). Blue line and markers indicate the actual values, and green line indicates the smoothed values. FIGURE 2.37 a. Germany Decomposition of inequality Relative inequality of opportunity Return to education Standard deviation logs of opportunity 0.55 0.5 0.09 Mean log deviation Coefficient 0.50 0.4 0.08 0.45 0.3 0.07 0.40 0.2 0.06 1980 1990 2000 2010 2020 1980 1990 2000 2010 2020 Year of survey Year of survey Standard deviation logs Mean log deviation Parental networking Intergenerational persistence in education 0.03 0.75 0.02 Coefficient Coefficient 0.01 0.70 0 0.65 −0.01 −0.02 0.60 1980 1990 2000 2010 2020 1980 1990 2000 2010 2020 Year of survey Year of survey (Continued) Are Distributional Tensions Brewing in Europe and Central Asia? ●  81 FIGURE 2.37 b. Italy Decomposition of inequality Relative inequality of opportunity Return to education 0.50 of opportunity (continued) Standard deviation logs 0.09 0.40 Mean log deviation 0.49 0.38 0.08 Coefficient 0.48 0.36 0.07 0.47 0.34 0.46 0.32 0.06 0.45 0.30 0.05 1995 2000 2005 2010 2015 1995 2000 2005 2010 2015 Year of survey Year of survey Standard deviation logs Mean log deviation Parental networking Intergenerational persistence in education 0.04 0.50 Coefficient 0.03 Coefficient 0.45 0.02 0.01 0.40 1995 2000 2005 2010 2015 1995 2000 2005 2010 2015 Year of survey Year of survey Source: Bussolo, Checchi, and Peragine 2018. Note: The first quadrant plots the relative inequality of opportunity, while the other three quadrants report the estimated structural parameters (with one standard error confidence interval) that contribute to the measured inequality of opportunity. attainment and incomes. Third, while access to education measured in years of education or the degrees obtained has become more equal, the quality of education may still vary greatly. There are still spatial gaps in quality. Deficiencies are also evident among children of disadvantaged socioeconomic back- grounds, as revealed by the results of the PISA test in 2015. In the Slovak Republic, for example, the gap between socioeconomic groups is the equiva- lent of five years of schooling; the gaps in Austria, the Czech Republic, and France are also large (Ridao-Cano and Bodewig 2018). Increasingly Unequal Access to Opportunities in the East Inequality of opportunity is greater in the east. In three-quarters of the transition countries studied by Brock, Peragine, and Tonini (2016), a third or more of total income inequality is associated with individual circumstances or inequality in oppor- tunity (figure 2.38).35 In general, inequality of opportunity is greater in the formerly planned economies than in the western countries in the LiTS 2016 sample.36 On average, the inequality of opportunity in acquiring labor income averages 0.11 in the 15 countries that are part of the EU, compared with 0.12 in the remaining 18 countries, despite the relatively high estimates for Bulgaria, Estonia, Greece, Hungary, and Latvia, where inequality of opportunity is above the regional average. However, inequality of opportunity is much lower in the transition economies than in other emerging economies (for example, Brazil and India) or in the United States.37 82  ●   Toward a New Social Contract FIGURE 2.38 55 Income inequality and Income inequality (Gini index) 50 MKD inequality of opportunity in KOS GEO 45 UZB obtaining income MON TUR 40 LTU MDA RUS ARM KAZ BLR TJK LVA 35 SRB MNE CYP EST BGR 30 DEU KGZ GRC ROU BIH POL UKR 25 HRV HUN SVN CZE AZE ITA SVK 20 5 7 9 11 13 15 17 19 Total inequality of opportunity (Gini index) CEB Central Asia EEC Russian Federation SEE Turkey Western Europe Source: Brock, Peragine, and Tonini 2016. Note: CEB = Central Europe and the Baltic States; EEC = Eastern Europe and the Caucasus; SEE = South-Eastern Europe. Inequality of opportunity varies substantially across the transition countries and often between neighboring countries. It is high in several transition countries that are now EU members and that also have more well-developed institutions. Differences in inequality of opportunity within the eastern subregions of Europe and Central Asia are largest in southeastern Europe, where Bosnia and Herzegovina, Montenegro, and Serbia display some of the lowest estimates, comparable with that of Germany. By contrast, inequality of opportunity in Bulgaria, Kosovo, and Romania is estimated to be above the median of the transition region. Inequality of opportunity is generally high in Eastern Europe, the Caucasus, and Central Asia. Transition countries with high inequality of opportunity also tend to exhibit high income inequality (see figure 2.38). The relationship is stronger among countries with higher inequality and weaker in countries with lower inequality. Some coun- tries show high inequality of opportunity, but moderate or low income inequality. However, low inequality of opportunity and high income inequality together, as in FYR Macedonia, are rare. Access to education is an important determinant of inequality of opportunity in the east, and it has become more unequal. In the transition region, workers with a tertiary degree earn, on average, 31 percent more in income than those workers with only a secondary degree (Brock, Peragine, and Tonini 2016). Returns to edu- cation of this magnitude are comparable with returns in some Western European countries, such as the Netherlands and Spain, but are lower than the correspond- ing returns in Eastern Europe in the early years of transition (Badescu, D’Hombres, and Villalba 2011; Bartolj et al. 2012). Individual birth circumstances are more important determinants of access to tertiary education among the generation that came of age in the early 2000s than among the generation that started education before the transition (Brock, Peragine, and Tonini 2016). This result is confirmed by the data of the three waves of the LiTS and by making a finer partition of the popu- lation into five separate cohorts to describe 40-year trends in equality of opportu- nity in education.38 Figure 2.39 reports, for each of five cohorts, the inequality of opportunity in tertiary education, as measured by the dissimilarity index.39 The panels show that, although some differences within subregions exist, access to Are Distributional Tensions Brewing in Europe and Central Asia? ●  83 FIGURE 2.39  Inequality of opportunity in tertiary education a. Eastern Europe and the Russian Federation Belarus Czech Republic Estonia Latvia 0.5 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.3 0.2 0.2 0.2 0.2 Inequality of opportunity for tertiary education (D-index) 0.1 0.1 0.1 0.1 45 55 75 65 65 84 45 55 65 75 84 45 55 65 75 84 45 55 75 84 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 Year Year Year Year Lithuania Moldova Poland Russian Federation 0.5 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.1 45 55 65 75 84 45 55 65 75 84 45 55 65 75 84 45 55 65 75 84 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 Year Year Year Year Slovakia Ukraine 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 45 55 65 75 84 45 55 65 75 84 19 19 19 19 19 19 19 19 19 19 Year Year b. Central Asia and Caucasus Armenia Azerbaijan Georgia 0.4 0.4 0.4 Inequality of opportunity for tertiary education (D-index) 0.3 0.3 0.3 0.2 0.2 0.2 0.1 0.1 0.1 45 55 65 75 84 45 55 65 75 84 45 55 65 75 84 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 Year Year Year Kazakhstan Kyrgyz Republic 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 45 55 65 75 84 45 55 65 75 84 19 19 19 19 19 19 19 19 19 19 Year Year (Continued) 84  ●   Toward a New Social Contract FIGURE 2.39  Inequality of opportunity in tertiary education (continued) Mongolia Tajikistan Uzbekistan 0.4 0.4 0.4 Inequality of opportunity for tertiary education 0.3 0.3 0.3 (D-index) 0.2 0.2 0.2 0.1 0.1 0.1 45 55 65 75 84 45 55 65 75 84 45 55 65 75 84 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 Year Year Year c. Southeastern Europe Albania Bosnia Bulgaria Croatia 0.5 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.1 0 0 0 0 Inequality of opportunity for tertiary education (D-index) 45 55 65 75 84 45 55 65 75 84 45 55 65 75 84 45 55 65 75 84 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 Year Year Year Year Macedonia, FYR Hungary Kosovo Montenegro 0.5 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.1 0 0 0 0 45 55 65 75 84 45 55 65 75 84 45 55 65 75 84 45 55 65 75 84 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 Year Year Year Year Romania Serbia Slovenia Turkey 0.5 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.1 0 0 0 0 45 55 65 75 84 45 55 65 75 84 45 55 65 75 84 45 55 65 75 84 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 Year Year Year Year Source: Calculations based on data of the 2016 round, LiTS (Life in Transition Survey) (database), European Bank for Reconstruction and Development, London, http://www.ebrd.com/what-we-do/economic-research-and-data/data/lits.html. Note: The panels show a dissimilarity index of inequality of opportunity, based on a probit regression of the variable indicating the completion of some tertiary education on individual circumstances. tertiary education in most countries in Eastern Europe has become more unfair over time, that is, more dependent on the individual’s circumstances. This is par- ticularly true of Eastern Europe. Birthplace and gender contribute less to the mea- sured inequality affecting the oldest cohort than to that affecting the younger cohorts. Are Distributional Tensions Brewing in Europe and Central Asia? ●  85 The increase in inequality of opportunity in education is confirmed in figure 2.40, which shows the trend in intergenerational mobility in education, calculated across cohorts, for countries in Europe and Central Asia (box 2.7). Examination of mobility trends among cohorts grouped by income reveals that the lower-middle- and high-income countries in the region exhibit greater mobil- ity, on average, relative to upper-middle-income countries. However, all three FIGURE 2.40 35 Intergenerational persistence Intergenerational persistence in education, Europe and Altham statistic 30 Central Asia 25 20 <1950 1951–60 1961–70 1971–80 1981–90 Birth cohort High income Upper-middle income Includes countries formerly in transition Lower-middle income In transition Source: Tiwari et al. 2018. Note: The figure shows the Altham (1970) measure of mobility applied to a pooled sample of the 2006, 2010, and 2016 rounds of the LiTS. See LiTS (Life in Transition Survey) (database), European Bank for Reconstruction and Development, London, http://www.ebrd.com/what-we-do/economic-research-and -data/data/lits.html. BOX 2.7 Calculating Measures of Intergenerational Mobility Intergenerational mobility—the extent to which indi- mobility, and, conversely, a weaker association viduals may expect to do better in life, usually in would imply greater mobility. terms of income or profession, than their initial cir- There are several challenges in calculating inter- cumstances might suggest—is one of the funda- generational income and earnings elasticities. First, mental cornerstones of development. In an absolute there are only a few countries in the world in which sense, intergenerational mobility is understood in one may construct matched parent-child pairs in terms of the attainment of a higher status or higher income or earnings. Even if this is possible, it can- standing (education, income, employment, living not be done for a sufficiently long period to cap- standards) relative to one’s parents. It involves free- ture unbiased estimates of persistence satisfactorily. ing oneself from the tethers of the social position of Second, even if matched data on parental income one’s parents to reach a higher position on the or earnings are available, they are often available socioeconomic ladder through one’s effort. only for a particular point in time. This forces one The ideal dataset for the analysis of intergener- to make inferences about the relationship based ational mobility would consist of a vector of the on a snapshot of income rather than long-term income or earnings of the children, paired with the income. Third, in many developing and transition corresponding vector of the income of the parents. countries where formal labor markets are not well (The literature has typically used the father’s developed and there is still a heavy reliance on income.) This would allow an estimation of the often informal self-employment, data on income association between the two vectors. A stronger and earnings are likely to exhibit significant mea- association between the two would imply lower surement errors. (Continued) 86 ● Toward a New Social Contract BOX 2.7 Calculating Measures of Intergenerational Mobility (continued) For these reasons, educational attainment is the measurement of educational attainment than in often used to analyze intergenerational mobility. calculating other direct measures of income in the There are several advantages to this. First, educa- settings under study. tional attainment is a good proxy for overall eco- Tiwari et al. (2018) measure intergenerational nomic status. There are positive and nontrivial mobility of education using the Altham statistic. returns to education in almost all labor markets. Originally proposed by Altham (1970), the statistic Even in the post-transition countries in the sample essentially summarizes the degree of association here, where there is some degree of notional between the rows and columns of any given matrix, universality of education at least to secondary relative to the degree of association between the education, there is significant heterogeneity in rows and columns of another matrix. Specifically attainment, and there are positive returns to higher applying this statistic to the transition matrices of educational attainment. Second, education is often educational attainment among parents and chil- also the most important conduit for the transmis- dren relative to a hypothetical matrix that would sion or reproduction of societal privilege or advan- denote perfect mobility yields a measure of the tage, and it is therefore a useful way to examine degree of persistence between parental and child social mobility. Third, there is much less noise in educational attainment. Source: Based on Tiwari et al. 2018. country groups have witnessed a decline in mobility. Among the lower-middle- income countries, the decline is steady among all cohorts born after the 1950s, though the decline appears to be slightly sharper among the youngest cohort. The mobility of the youngest cohort in the upper-middle-income countries has increased slightly. However, this group is still less well off than the generation of their parents and grandparents who would have been born before the 1950s or in the 1960s. In countries still undergoing a transition from the planned to the market economy and in countries that have now completed the transition and joined the EU, there appears to be a steady downward slide among all cohorts and a some- what sharper deterioration among the two youngest cohorts. Greater inequality of opportunity in access to education among young cohorts may be traced to the impact of the transition on university systems. First, tertiary education, which used to be universally free, is often now associated with non- trivial costs. Even where education is still nominally free, scholarships to cover the cost of living, generous before the transition, have been effectively phased out, resulting in much higher opportunity costs for an education. Second, the once strong and closely controlled link between tertiary education and jobs has effec- tively disappeared, while the transition has placed a premium on new skills. Third, parents with tertiary educational attainment have gained from the transition to the extent that, before the transition, manufacturing jobs, which did not typically require a university degree, were relatively high status. As a result of these forces, parental education has become more important in explaining access to tertiary education, signaling a deterioration in the equality of opportunity. The persistence or widening of inequality of opportunity in the east may have implications for the social contract. The transition has been accompanied by the expectation that the inequality linked to individual effort and talents would Are Distributional Tensions Brewing in Europe and Central Asia? ●  87 increase. In a sense, beyond the popular support for a free market economy and political liberalization, a fair return on individual effort was a main motivation behind supporters of the transition. However, the persistence of inequality because of exogenous factors, such as social background and ethnicity, is a violation of the principle of fair returns and may weaken popular support for the market economy and the implicit social contract. Distributional Tensions and the Path to a Middle-Class Society Emerging distributional tensions are making the middle class more fragile and less attainable for some groups. The distributional tensions generated by labor market polarization, generational differences, spatial disparities, and inequality of oppor- tunity are reducing the productive capacity and ability of some groups to benefit from economic opportunity. Relative to older generations, younger generations are facing a deterioration in economic security—a defining element of middle- class status—because of shorter average job tenure and greater reliance on tem- porary and part-time jobs. Many middle-class workers who depend on jobs that are intensive in routine tasks have become unemployed or are experiencing lower earnings because of technological change. The middle class is composed of “those who work from 8 am to 5 pm and receive a monthly salary,” said a Turkish man (Dávalos et al. 2016, 33). “People who belong to the middle class should not be under constant stress that they might lose their jobs,” said a Serb (Dávalos et al. 2016, 13). The rise in inequality of opportunity is also reducing the labor market prospects of individuals of lower social or family backgrounds. In the east, the access of such individuals to higher education is becoming more limited because parental back- ground is an increasingly important determinant of access. People living in remote regions may experience difficulties in access to key services, particularly quality education, that reduce their job prospects. Has the Middle Class Declined in the Region? The middle class accounts for a large share of the population in Europe and Central Asia. There are several ways of measuring the size of the middle class (box 2.8). Based on data on 20 countries in the Luxembourg Income Study (LIS) database and a definition of the middle class in relative terms as all persons with incomes between 75 percent and 125 percent of the median income, the middle class accounted for an average 38 percent of the population in Europe and Central Asia in around 2013.40 (For information on country groups and more data, see annex 2A, tables 2A.2 and 2A.3.) The lower and upper ends of the distribution accounted for 29 percent and 33 percent, respectively. The middle class generally comprises the largest group in Western Europe, though the three groups are roughly equal in size in Southern Europe. In Continental and Nordic Europe, the middle class is substantially bigger. In about half the Eastern European countries, the upper end of the distribution represents the largest group. The distribution of population in 88 ● Toward a New Social Contract BOX 2.8 Defining the Middle Class The composition of the middle class has been the 2008; Birdsall 2010; Loayza, Rigolini, and Llorente focus of academic research, mainly in sociology, at 2012; Ravallion 2010). Another absolute approach least since Max Weber’s (1922) work on status involves deriving an appropriate income level from groups and classes. There are multiple approaches a measure of vulnerability to relate middle-class to defining the middle class (see Atkinson and status and a feeling of economic security. Brandolini 2013; Banerjee and Duflo 2008; Foster Population space. The middle class can also and Wolfson 2010; Vaughan-Whitehead 2016). Two be defined according to a selected part of the main approaches can be identified: the objective income distribution within the population space. and the subjective approaches.a Unlike the income space approach, the relative size of the population is fixed in this approach. The The objective approach approach thus conceptualizes the middle position In this approach, the research defines ex ante cer- in the distribution in terms of the enjoyment of a tain thresholds to demarcate the middle from the particular social status. lower and upper classes. This approach is objective The subjective approach in the sense that certain thresholds are applied, The middle class can also be defined based on the though the choice of the thresholds is subjective. subjective perception of what it means to belong The thresholds can be selected in the income to the middle class. Ravallion (2012) discusses the space or in the population space. use of subjective welfare indicators as a way of Income space. A first set of objective approaches determining a socially subjective poverty line. defines the middle in the income space according Ferreira et al. (2013) take this a step further by to selected income thresholds. These thresholds can inferring absolute thresholds for the middle class be defined in a relative or absolute way. Relative based on the probability individuals will answer objective indicators tend to define the middle middle class to a question about their own social according to income bounds around the median. status. The main limitations of the subjective For instance, the International Labour Organization approach derive from the fact that individual defines the middle class as the population between answers vary considerably depending on the way 60 percent and 200 percent of the median (Vaughan- the question is framed and that inferences based Whitehead 2016). The Pew Research Center (2015) on the answers may be highly biased because of applies the range of 67 percent to 200 percent of latent heterogeneity. the median. Other authors use a more dense defini- tion of the middle class, for example, 75 percent to Advantages and disadvantages of selected 125 percent of the median, as a threshold (Birdsall, definitions Graham, and Pettinato 2000; Gornick and Jäntti All definitions of the middle class involve compro- 2013; Thurow 1987). mises and are characterized by advantages and A problem involved in the application of relative disadvantages (for an elaborate discussion, see thresholds is that the income bounds vary across Foster and Wolfson 2010). In the objective countries depending on the level of the median. To approach, the researcher defines cutoff points that relate the middle class to a specific purchasing are arbitrary to a certain extent. This is not true of power, other researchers have applied absolute the subjective approach, which relies on the per- thresholds. In this case also, different thresholds ceptions among individuals that they belong to the have been used, for instance, people living above middle class; however, these perceptions may not the median poverty line in developing countries or be confirmed by relative or absolute income mea- above another cutoff, such as US$10-a-day pur- sures and may be culturally determined. This is a chasing power parity (PPP) (Banerjee and Duflo particular disadvantage in cross-country research (Continued) Are Distributional Tensions Brewing in Europe and Central Asia? ●  89 BOX 2.8 Defining the Middle Class (continued) and is the reason the objective approach is favored middle class is defined on the basis of an absolute in this chapter. income threshold. Yet, in the latter approach, the rel- In the objective approach, the income position ative position of an individual in a society is ignored. of the middle class in the population space does For instance, applying a fairly low absolute income not depend on differences in average or median threshold to a rich country could result in placement incomes across countries. Thus, income growth of the entire population in the upper class. does not affect the size of the population. In acknowledgment of the advantages and dis- Moreover, inequality is also ignored because the advantages of each definition of the middle class, size of the middle class is fixed, regardless of the both a relative and an absolute objective income relative income position. The income space space approach are adopted in this chapter. For the approach is therefore applied in this chapter. relative definition, the thresholds at 75 percent to However, the income space approach is also asso- 125 percent of the equivalized disposable house- ciated with disadvantages. While income may be hold median income are used. In the absolute defi- defined relative to the median income, the size of the nition, the middle class represents the share of population may vary depending on trends in inclusive households with disposable incomes between growth. Moreover, in cross-country research, the mid- US$11- and US$28-a-day PPP. Vulnerability is taken dle class may be characterized by different incomes into account by linking the middle class to a certain across countries. This problem does not arise if the level of economic security. a. Another approach is to follow the median over time (for example, Aaberge and Atkinson 2013; Nolan, Roser, and Thewissen 2016; Thewissen et al. 2015). This approach can be useful for tracking trends in living standards, but it does not allow an exploration of the composition of the middle class in terms of income or demographics. It is therefore not discussed in more detail here. the Czech Republic, the Slovak Republic, and Slovenia resembles that in Nordic European countries, that is, smaller poorer groups and a large middle class, whereas the two extremes of the distribution are larger in Russia and Serbia and, especially, Estonia and Georgia. Defined by the relative measure, the size of the middle class in Europe and Central Asia did not change significantly from the mid-1990s to the early 2010s. The middle class shrank in half the countries and expanded in the other half of the countries (figure 2.41). In many countries in which the population share of the middle class declined, the reduction was not substantial. In Germany, for example, the share fell from 43 percent in 1994 to 39 percent in 2014. In Slovenia, the coun- try with the steepest decline, the share dropped from 48 percent in 1997 to 40 percent in 2012. It is also useful to determine a fixed level of income that defines the middle class and assess the changes in the composition of the middle class in absolute terms. Because median incomes differ greatly across countries, the relative middle-class definition used above groups people at quite different levels of welfare. For example, 75 percent of the median income in Georgia in 2013 was US$1,594 PPP, whereas the corresponding income in Finland in the same year was US$11,951 PPP. Defining the middle class in absolute terms thus enables comparisons among individuals at the same levels of welfare across countries. An absolute definition can be used to include in the middle class people who are 90  ●   Toward a New Social Contract FIGURE 2.41  Trends in the relative size of the middle class, Europe and Central Asia Change in population share 1.0 0.8 Average annual change, percentage points 0.6 0.4 0.2 0 –0.2 –0.4 –0.6 ia d k rg y nia ds ly ain d ia c e lic ce m ia d n gia ar an bli lan lan nc lan tio Ita en str do rb ou lan ub ee nm to Sp or rm pu Fra Se ra ov Fin Po ing Ire mb Au ep Gr Es Ge er de Re De Ge Sl hR th dK xe Fe ak Ne Lu ec ite ian ov Cz Un Sl ss Ru Source: Calculations based on data of LIS Database (Luxembourg Income Study Database), LIS Cross-National Data Center in Luxembourg, Luxembourg, http://www.lisdatacenter.org/our-data/lis-database/. Note: See annex 2A, table 2A.2 for more detail. The middle class is defined as the share of the population with incomes between 75 percent and 125 percent of the median income. The middle-class population shares were measured around 1995 and 2014. able to avoid falling into poverty in the face of unexpected shocks, which is an attribute that people likely view as essential to a middle-class lifestyle. In this analysis, the lower threshold defining the middle class is the income associated with a 5 percent probability of poverty—defined as falling below an income of US$5.50 a day in constant 2011 PPP prices—over a four-year period, calculated on the basis of the observed characteristics of households, including assets (see box 2.8 for details).41 Based on this analysis, individuals with incomes between US$11- and US$28-a-day PPP are included in the middle class. Another vulner- able group consists of people with incomes between US$5.50- and US$11.00-a-day PPP. This group is not poor, but, because it experiences a high probability of falling into poverty (given its distance to the poverty line), it is not considered part of the middle class. Around 2015, the middle class determined by this absolute definition included 45 percent of the population in developing Europe and Central Asia.42 However, the vulnerable and the poor, together, represented more than 50 percent of the population (figure 2.42).43 Apparent from this latter statistic alone, Europe and Central Asia may be a middle-income region, but it is not a middle-class society. Of course, the shares of the poor and vulnerable vary a great deal across countries. In 2015, the poor accounted for 37 percent, and the vulnerable 45 percent of the population of the lower-middle-income Are Distributional Tensions Brewing in Europe and Central Asia? ●  91 FIGURE 2.42  Income classes, subregions of Europe and Central Asia, excluding the EU15 a. Lower-middle-income countries, consumption based b. Upper-middle-income countries, consumption based 100 100 90 90 80 80 70 70 60 60 Percent Percent 50 50 40 40 30 30 20 20 10 10 0 0 00 01 02 03 04 05 06 07 08 09 10 11 12 00 01 02 03 04 05 06 07 08 09 10 11 13 12 14 13 15 14 16 15 16 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Year Year c. High-income countries, consumption based 100 90 80 70 Percent 60 50 40 30 20 10 0 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Year Poor (