W O R L D B A N K W O R K I N G P A P E R N O . 1 6 2 47612 Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union Some Insights from the 2006 Life in Transition Survey Salman Zaidi Asad Alam Pradeep Mitra Ramya Sundaram THE WORLD BANK W O R L D B A N K W O R K I N G P A P E R N O . 1 6 2 Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union Some Insights from the 2006 Life in Transition Survey Salman Zaidi Asad Alam Pradeep Mitra Ramya Sundaram THE WORLD BANK Washington, D.C. Copyright © 2009 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First Printing: February 2009 printed on recycled paper 1 2 3 4 5 12 11 10 09 World Bank Working Papers are published to communicate the results of the Bank's work to the development community with the least possible delay.The manuscript of this paper therefore has not been prepared in accordance with the procedures appropriate to formally-edited texts. Some sources cited in this paper may be informal documents that are not readily available. The findings,interpretations,and conclusions expressed herein are those of the author(s) and do not necessarily reflect the views of the International Bank for Reconstruction and Development/ The World Bank and its affiliated organizations, or those of the Executive Directors of The World Bank 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 of the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/The World Bank encourages dissemination of its work and will normally grant permission promptly to reproduce portions of the work. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, Tel: 978-750-8400, Fax: 978-750-4470, www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, Fax: 202-522-2422, email: pubrights@worldbank.org. ISBN-13: 978-0-8213-7900-4 eISBN: 978-0-8213-7901-1 ISSN: 1726-5878 DOI: 10.1596/978-0-8213-7900-4 Library of Congress Cataloging-in-Publication Data Satisfaction with life and service delivery in Eastern Europe and the former Soviet Union : some insights from the 2006 life in transition survey / Salman Zaidi . . . [et al.]. p. ; cm.--(World Bank working paper, ISSN 1726-5878 ; no. 162) Includes bibliographical references. ISBN 978-0-8213-7900-4 1. Health surveys--Europe, Eastern. 2. Health surveys--Asia, Central. I. Zaidi, Salman, 1967-II. World Bank. III. Series: World Bank working paper ; no. 162. [DNLM: 1. Consumer Satisfaction--Europe, Eastern. 2. Health Services--Europe, Eastern. 3. Data Collection--Europe, Eastern. 4. Quality of Life--Europe, Eastern. 5. Socioeconomic Factors--Europe, Eastern. W 85 S253 2009] RA407.5.E85S28 2009 362.1094--dc22 2008052797 Contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix 1. Key Factors Affecting Satisfaction with Life in Eastern Europe and the Former Soviet Union . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Satisfaction with Life as a Welfare Measure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Key Factors Influencing SWL: Multivariate Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Comparisons over Time and Across Countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Concluding Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Annex: Tables and Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2. Employment, Sources of Income, and the Poor in Eastern Europe and the Former Soviet Union . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Choosing Between Alternate Survey-Based Welfare Measures . . . . . . . . . . . . . . . . . 29 How Good is the LiTS Welfare Metric?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Poverty Profile for ECA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Employment, Sources of Income, and Welfare. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Concluding Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Annex: Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 3. Satisfaction with Publicly-provided Health Services in Eastern Europe and the Former Soviet Union . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Evolution of Publicly-provided Health Services in Eastern Europe and Central Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 Utilization Rates, Satisfaction, and Prevalence of Informal Payments. . . . . . . . . . . 71 Estimation Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Key Findings and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 Health Sector Reform in the Caucuses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 Concluding Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 Annex: Tables and Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 LIST OF TABLES 1.1. Average SWL Score by Self-perceived Economic Status . . . . . . . . . . . . . . . . . . . . 7 1.2. Average SWL Score by Present and Past Self-assessed Economic Status . . . . . . . 7 iii iv Contents 1.3. Average SWL Score by Present and Past Level of Social Capital . . . . . . . . . . . . . 8 1.4. Simulated Probabilities Derived from Ordered Probit Model . . . . . . . . . . . . . . 10 1.5. Ordered Probit Results: SWL by Country Groups . . . . . . . . . . . . . . . . . . . . . . . 11 1.6. Change over Time in Average SWL Rates by Country . . . . . . . . . . . . . . . . . . . . 14 1.7. Comparing GDP and SWL Changes in Recent Years . . . . . . . . . . . . . . . . . . . . . 15 A1.1. Satisfaction with Life Question by Country. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 A1.2. Perceptions Regarding Changes over Time in Economic Situation by Country. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 A1.3. Satisfaction with Changes over Time in Living Conditions by Country. . . . . . 21 A1.4. Average SWL Score: Colleagues in 1989 Rather than School Mates as Peers . . 22 A1.5. Tendency to Feel I've Done Worse During Transition Than Others, by Level of Income Inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 A1.6. Fall in Social Capital? Decline in Reported Level of Trust in Other People . . . 24 2.1. Comparing Various Alternate Welfare Measures in the LiTS . . . . . . . . . . . . . . . 31 2.2. Subjective Assessment of Welfare . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 2.3. Ownership of Assets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 2.4. Correlation Matrices: Decile Rankings Based on Various LiTS Welfare Measures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 2.5. Asset Ownership Rates by Welfare Level using Alternate Ranking Criteria . . . 37 2.6. 2006 LiTS PCE Compared to Other Data Sources . . . . . . . . . . . . . . . . . . . . . . . 40 2.7. Overall Regional Poverty Rates from the 2006 LiTS . . . . . . . . . . . . . . . . . . . . . . 42 2.8. Overall Regional Poverty Rates from the 2006 LiTS . . . . . . . . . . . . . . . . . . . . . . 44 2.9. Distribution of the Poor by Employment Status of the Respondent. . . . . . . . . 45 2.10. Respondents Having Worked in Past 12 Months, by Age, Gender, and Education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2.11. Main Income Source by Region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 2.12. Probit Model of Likelihood of Being Poor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 2.13. Public/Private Transfers Are More Important in the CIS and EU Member States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 A2.1. Overall Poverty Rates by Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 A2.2. Sensitivity of Poverty Rates with Respect to Choice of Poverty Line . . . . . . . . 56 A2.3. Distribution of the Poor by Geographic Region . . . . . . . . . . . . . . . . . . . . . . . . . 56 A2.4. Rural Urban Disparities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 A2.5. Mean Per Capita Expenditures ($PPP per year) . . . . . . . . . . . . . . . . . . . . . . . . . 59 A2.6. Decomposition of Inequality by Geographic Region . . . . . . . . . . . . . . . . . . . . . 60 A2.7. Ratios of Selected Expenditure Percentiles in Urban and Rural Areas . . . . . . . 60 A2.8. Poverty by Age Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 A2.9. Poverty by Whether Respondent Worked or Not During Past 12 Months . . . . 61 Contents v A2.10. Poverty by Education Level of Household Head . . . . . . . . . . . . . . . . . . . . . . . . . 61 A2.11. Poverty by Household Head's Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 A2.12. Poverty by Demographic Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 A2.13. Consumption Regressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 3.1. Probit for Health Care Access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 3.2. Ordered Probit: Satisfaction with Publicly-provided Health Services. . . . . . . . 80 3.3. Prevalence of Unofficial Payments for Selected Countries . . . . . . . . . . . . . . . . . 85 3.4. Change in Health Care Access Rates for Selected Countries . . . . . . . . . . . . . . . 86 A3.1. Priorities for Additional Government Spending, By Country . . . . . . . . . . . . . . 90 A3.2. Access Rates of PPHS, By Country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 A3.3. Satisfaction with Medical Treatment in PPHS by Country . . . . . . . . . . . . . . . . 92 A3.4. Prevalence of Unofficial Payments in PPHS by Country . . . . . . . . . . . . . . . . . . 93 A3.5. Difference between General and Experienced Perception of Prevalence of Unofficial Payments in PPHS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 LIST OF FIGURES 1.1. All Things Considered, I am Satisfied with My Life Right Now. . . . . . . . . . . . . . . 3 1.2. SWL Rates by Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.3. Correlation of Satisfaction with Life with Average Incomes and Level of Inequality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.4. Satisfaction with Life Among the Youth is Generally Higher than among the Elderly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.5. Satisfaction with Life is Higher for the Healthy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.6. Satisfaction with Life is Positively Correlated with Level of Trust in People. . . . . 8 1.7. Cross-Country Comparisons: SWL and Employment Rates . . . . . . . . . . . . . . . . 16 1.8. Cross-Country Comparisons: SWL and Level of Trust in Others . . . . . . . . . . . . 17 2.1. Distribution of Normalized Expenditures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.2. Distribution of the One-question Welfare Aggregate . . . . . . . . . . . . . . . . . . . . . . 32 2.3. Distribution of Subjective Welfare Rankings by Country. . . . . . . . . . . . . . . . . . . 34 2.4. Comparing the Various Alternate Welfare Measures . . . . . . . . . . . . . . . . . . . . . . . 39 2.5. Country Welfare Rankings: National Accounts vs. Survey-based Estimates . . . . 41 2.6. Regional Variation in Poverty Rates Across the ECA Region . . . . . . . . . . . . . . . . 43 2.7. Distribution of the Poor Across the ECA Region. . . . . . . . . . . . . . . . . . . . . . . . . . 43 2.8. Respondents that Report Having Worked During Past 12 Months . . . . . . . . . . . 46 2.9. Intercountry Differences in Main Income Sources of Households . . . . . . . . . . . 48 3.1. Rates of Satisfaction with the Publicly-provided Health System By Country. . . 66 3.2. Priorities for Additional Government Spending: 2006 LiTS. . . . . . . . . . . . . . . . . 69 3.3. Utilization of Publicly-provided Health System by Country . . . . . . . . . . . . . . . . 72 vi Contents 3.4. Rates of Satisfaction By Type of Service . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 3.5. Percent of Respondents that Think that Unofficial Payments Are Needed. . . . . 74 3.6. Perceptions Regarding Unofficial Payments in Publicly-provided Health System By Country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 3.7. Negative Correlation Between Satisfaction and Prevalence of Informal Payments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 3.8. Satisfaction with Publicly-provided Health Service and Self-assessed Health Status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 3.9. General vs. Experienced Opinion of Need for Unofficial Payments . . . . . . . . . . 84 3.10. Changes in Access Rates and Prevalence of Unofficial Payments, 2001 to 2006. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 Preface T he past two decades in Eastern Europe and the former Soviet Union (ECA) have been times of tremendous change, with countries undergoing rapid transformation from centrally-planned to market-oriented economies. While poverty increased during the initial years of transition, primarily on account of the sharp economic contraction, the resurgence of economic growth in the region since 1998 has resulted in a rebound in household incomes and living standards. Little is known, however, about subjective welfare in the wake of this growth rebound, especially how people across ECA countries view their satisfaction with life as well as with the quality of services being providing by their governments. Data from the 2006 Life in Transition Survey (LiTS)--a joint initiative of the European Bank for Reconstruction and Development and the World Bank-- provides a unique opportunity to investigate the extent to which citizens of ECA countries are satisfied with their lives and with the performances of their governments, and to study key factors influencing their outlook in a systematic way across all countries of the region.1 The LiTS was carried out in 29 countries: Albania, Armenia, Azerbaijan, Belarus, Bosnia and Herzegovina, Bulgaria, Croatia, Czech Republic, Estonia, Former Yugoslav Republic of Macedonia, Georgia, Hungary, Kazakhstan, Kyrgyz Republic, Latvia, Lithuania, Moldova, Mongolia (not in ECA, but included because its an EBRD client country), Montenegro,Poland,Romania,Russia,Serbia,Slovakia,Slovenia,Tajikistan,Turkey,Ukraine, and Uzbekistan between August and October 2006. In each country, the LiTS questionnaire was administered to a nationally representative sample of 1,000 households using face- to-face interviews. The main objective of the LiTS was to assess the impact of transition on people, and so the survey questionnaire covered four main themes. First, it collected personal information on aspects of material well-being, including household expenditures, possession of con- sumer goods such as a car or mobile phone, and access to local public services and utilities. Second, the survey included measures of satisfaction and attitudes towards economic and political reforms as well as public service delivery. Third, the LiTS captured individual "histories"through transition--from around 1989 to the present, especially key events and episodes that may have influenced their attitudes towards reforms, and collected informa- tion on individuals; family background, on their employment situation, and on coping strategies during transition. Finally, the survey also attempted to capture the extent to which crime and corruption are affecting peoples' lives, and the extent to which individuals' trust in other people and in state institutions has changed over time. This volume presents the main findings of three studies by World Bank economists using data from the 2006 LiTS. Chapter 1 examines quantitative and qualitative dimensions welfare in countries of Eastern Europe and the former Soviet Union, with "satisfaction with life" being the key welfare measure used. Analysis suggests that more than 60 percent of the population reports satisfaction with life though this varies quite a bit across countries: Slovenia has the 1. For more details on the LiTS as well as the preliminary survey findings, see EBRD 2007. vii viii Preface highest rates of satisfaction and Georgia the lowest. Econometric analysis suggests a positive correlation of the overall satisfaction-with-life variable with factors such as expenditure per capita, equality of incomes, youth, working status, non-agricultural employment, non-metropolitan living, and better education. However, the analysis also highlights the important role played by subjective factors like self-assessed health status, level of trust in people, relative economic status compared to peers and own-perception of improvement in economic status over time in determining overall satisfaction levels. Chapter 2 analyzes socioeconomic characteristics of different income groups across countries, and shows how the welfare measure derived from the LiTS provides a very useful and effective means to measure household welfare and/or rank households by relative economic status, both within as well as across countries. Moreover, this welfare measure also compares favorably with welfare measures using traditional and more exten- sive household budget surveys. The chapter provides a systematic examination of the sources of household incomes,patterns of asset ownership,as well as the sectoral occupation patterns of the working poor. Wages and pensions are the primary sources of income for the bulk of the population though the poor depend more on pensions. Asset ownership-- in terms of cars, secondary residence, mobile phone, and computers--varies quite broadly, and asset inequality is quite significant across income groups both within and across coun- tries. The data also that the poor have a relative disadvantage in that they are primarily employed in low productivity agriculture, and have limited educational attainment and non-professional skills. Finally, chapter 3 focuses on three interlinked questions: (i) why are some people more likely than others to use publicly provided health services? (ii) what are some of the key influences on users' satisfaction with quality and efficiency of medical treatment received? and (iii) how does the prevalence of informal payments impact people's decision on using publicly provided health services, and upon use, the level of satisfaction with services received? Analysis shows that the elderly, the relatively better-off, and those who have con- fidence in the government are more likely to use publicly provided health services, while those with compulsory/secondary education as well as those with some tertiary education are less likely to access these services. Satisfaction with publicly provided health services in the region is quite high, though there is considerable variation evident across countries. In general, while a large majority of respondents say that unofficial payments/gifts are never needed when using publicly provided health services, in cases where users have to pay for what should essentially be "free services," this has a significant negative influence on satis- faction with service delivery. Acknowledgments T he 2006 EBRD-World Bank Life in Transition Survey (LiTS) was designed by the EBRD's Office of the Chief Economist and the World Bank's Europe and Central Asia (ECA) Region, under the general guidance of Erik Berglof (Chief Economist, EBRD) and Pradeep Mitra (Chief Economist, Europe and Central Asia Region, World Bank) and Asad Alam (Sector Manager, ECSPE, World Bank). The core task team for the project was led by Peter Sanfey (EBRD) and Salman Zaidi (World Bank), and comprised James Anderson, Pauline Grosjean, Juan Munoz, Franklin Steves, and Utku Teksoz. Field work for the survey was carried out by the global market research firm Synovate, under the direction of Savvas Kyriakides. Chapter 1 and 2 were written by Asad Alam, Pradeep Mitra, and Salman Zaidi, while Chapter 3 of this volume was written by Ramya Sundaram and Salman Zaidi. Helena Makarenko processed the report. Funding for the survey was provided by Canada, Taiper China, and the United Kingdom. In addition, the authors gratefully acknowledge support provided by the ECA Chief Economist Regional Studies Program and from the World Bank Research Support Budget. The assessment and views presented in this volume are those of the authors, and should not be attributed to the Executive Directors of the World Bank. ix CHAPTER 1 Key Factors Affecting Satisfaction with Life in Eastern Europe and the Former Soviet Union2 T here has been a resurgence of interest recently among social scientists in studying subjective measures of individual well-being, and in analyzing how peoples' sense of their personal welfare is impacted by not just their level of incomes, but also other diverse factors like health, income inequality, and employment status.3 Much research has been carried out to better understand why some people say they are satisfied with their lives, and others say they are not. While there is broad agreement in the literature about the diverse set of factors that affect individual well-being, much less consensus prevails about the relative importance of these factors, even among leading researchers. For instance, Richard Layard has argued that non-income factors like health and family circumstances impact peoples' sense of well-being more than income per se. Others have stressed that incomes play the main role in determining peoples' satisfaction with life-- Angus Deaton has noted that the map of average satisfaction levels across the world looks very similar indeed to the spatial distribution of average incomes across countries. The past two decades in Eastern Europe and the former Soviet Union (ECA) have been times of tremendous change, with countries undergoing rapid transformation from centrally-planned to market-oriented economies. While poverty increased during the initial years of transition, primarily on account of the sharp economic contraction, the resurgence of economic growth in the region since 1998 has resulted in a rebound in household incomes and living standards.4 However, little is known about subjective welfare in the wake 2. Asad Alam, Pradeep Mitra, and Salman Zaidi. 3. See, for instance, Layard (2006), Kahneman and Krueger (2006), Helliwell (2007), Clark, Frijters, and Shields (2007), Graham (2007), as well as many other papers referenced in these publications. 4. For instance, see World Bank (2005). 1 2 World Bank Working Paper of this growth rebound, especially how people across ECA countries view their satisfaction with life. Earlier work on ECA countries, using the World Values Survey, have had limited country coverage and used early data that mostly covered the first decade of transition. These earlier studies have also focused more on explaining differences in attitudes between market and socialist economies, and examining people's preferences for redistributive spending by the state and for greater income equality.5 The chapter addresses three main questions: (1) what are prevailing levels of satis- faction with life (SWL) in ECA countries, and how have they been changing over time? (2) what are the main factors that help explain SWL, and in particular what is the relative importance of income vs. non-income factors like health, family status?, and finally, (3) why are prevailing levels of SWL in ECA somewhat lower than what might be expected given relatively high income levels and good health status etc? The next section of this chapter provides an overview of satisfaction with life and its correlates in ECA countries. The results of the multivariate analysis are presented in the second section, which show that while per capita incomes and employment status are important drivers of satisfaction with life, other non-income factors such as health, relative economic status, and level of trust in other people also play a crucial role. How do average satisfaction rates in ECA countries in 2006 compare to findings of similar surveys conducted earlier in time? How about in relation to countries in other parts of the world? These questions are taken up in the third section, where we compare the LiTS findings on satisfaction with life with similar results from other surveys conducted in other parts of the world as well as in the same set of coun- tries earlier in time. Satisfaction with Life as a Welfare Measure Respondents in the LiTS were asked to what extent they agreed with the statement: "All things considered, I am satisfied with my life now," with responses coded as 1 = strongly disagree (SD), 2 = disagree (D), 3 = neither disagree nor agree (N), 4 = agree (A), and 5 = strongly agree (SA). In the LiTS sample overall, respondents that reported themselves as satisfied with their lives outnumber those that are not 3 to 2. Yet, this varies considerably across countries from a high of 8:1 in Slovenia to roughly 2:5 in Georgia (see Figure 1.1). Satisfaction rates not only vary across countries but also across groups of countries (see Figure 1.2). Most of the new member states of the European Union, which have perhaps seen the biggest political transformation in the Region, feature in the upper part of the distribution except for Hungary which is third from the bottom. Conversely, many of the countries of southeastern Europe and the south Caucuses show relatively low levels of satisfaction. Despite the clear heterogeneity in satisfaction rates observed across countries and groups of countries, there are nonetheless some similarities evident across some groups. At the 5. See, e.g. Murthi, Mamta. and Erwin Tiongson, 2008, Attitudes to Equality: The "Socialist Legacy" Revisited, Policy Research Working Paper No. 4529, The World Bank for recent examination of preferences for inequality as well as for an excellent survey of the ECA-specific literature. The study uses data from early rounds of the World Values Survey: 1990, 1995­97, and 1999­01 and covers a smaller set of 4, 15, and 17 countries in these three rounds. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 3 Figure 1.1. All Things Considered, I am Satisfied with my Life Right Now Slovenia Uzbekistan Tajikistan Belarus Estonia Kyrgyz Republic Slovakia Croatia Latvia Czech Republic Kazakhstan Lithuania Poland Russia Turkey Albania Ukraine Romania Bulgaria Montenegro Bosnia Moldova Macedonia, FYR Serbia Azerbaijan Hungary Armenia Georgia 0 10 20 30 40 50 60 70 80 90 100 Strongly Disagree Disagree Neither Agree Strongly Agree country level, the LITS data show satisfaction with life to be positively correlated with absolute incomes (PPP-adjusted), and negatively correlated with level of income inequality (Figure 1.3). One of the striking contrasts observed across most transition countries6 is the clear divide across age groups: overall satisfaction rates among the youth are considerably higher than amongst the elderly (Figure 1.4). This is not surprising given that it has been the adult population which lived through the economic decline and dislocations of the 1990s and for whom the economic transition, with its attendant uncertainties and insecurities, has been the most acute. What is striking is that even with higher unemployment rates, younger age cohorts are more positive about life satisfaction (see Figure 1.3). Similarly, overall satisfaction rates are quite low in Bosnia, Serbia, FYR of Macedonia, Georgia, and Hungary; yet those under 30 years appear to have a more positive outlook on life as compared to the rest of the population. 6. Turkey and Mongolia are useful comparators in this regard, in the sense of being "non-transition" countries, and response patterns in these two countries do not show such marked differences by age group. 4 World Bank Working Paper Figure 1.2. SWL Rates by Region All things considered, I am satisfied with my life right now 60 Agree 50 Strongly Agree 40 30 respondents 20 of % 10 0 states Other income Europe CIS-middle CIS-low member EU South-Eastern Note: "CIS-low income" includes Moldova, Armenia, Azerbaijan, Georgia, Kyrgyz Republic, Tajikistan, and Uzbekistan; "EU member states" includes Slovenia, Estonia, Slovakia, Latvia, Czech Republic, Lithuania, Poland, and Hungary; "CIS-middle" includes Belarus, Ukraine, Kazakhstan, and Russia; SEE includes Albania, Bosnia, Bulgaria, FYR of Macedonia, Montenegro, Romania, and Serbia; "Other" includes Croatia and Turkey. Other Subjective Measures of Well-being The data also suggest that the (self-reported) health status of respondents has a strong bear- ing on reported level of satisfaction with life (Figure 1.5). In addition to these quantitative measures of well being, the LiTS questionnaire also includes a number of other questions on respondents' perception of their relative eco- nomic standing, both at present and around 1989.7 The respondents were asked to what extent (on a five-point scale, ranging from 1: strongly disagree to 5: strongly agree) they agreed with statements like (a) I have done better in life than most of my high school class- mates, and (b) I have done better in life than most of the colleagues I had around 1989. Responses to these questions provide some interesting insights into the extent to which respondents' expressed level of satisfaction with life is related to their perceptions regard- ing both their current economic standing as well as how their economic standing has changed over time and in relation to their peers. These patterns of association are exam- ined in the tables below, which illustrate the "average SWL score"--the average responses 7. The specific questions were: (1) Please imagine a ten-step ladder where on the bottom, the first step, stand the poorest people and on the highest step, the tenth, stand the richest. On which step of the ten is your household today?; (2) Now imagine the same ten-step ladder around 1989, on which step was your household at that time? Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 5 Figure 1.3. Correlation of Satisfaction with Life with Average Incomes and Level of Inequality (i) Level of Per capita GDP All things considered, I am satisfied with my life right now Slovenia 70 Uzbekistan Belarus Estonia Tajikistan 60 life Kyrgyz Slovakia with Latvia Croatia Czech Kazakhstan Lithuania 50 Poland satisfaction Albania Turkey on Russia 40 neutral Ukraine Mongolia above Romania % 30 Bosnia Moldova Bulgaria Azerbaijan Hungary Macedonia Georgia Armenia 20 0 5000 10000 15000 20000 Per-capita GDP in 2005 (PPP$2000) (ii) Level of Income Inequality All things considered, I am satisfied with my life right now Slovenia 70 Uzbekistan Estonia Belarus Tajikistan 60 Slovakia life Kyrgyz Croatia with Latvia Czech KazakhstanLithuania 50 Poland satisfaction Albania Turkey on Russia 40 neutral Ukraine Mongolia above Romania % 30 Bulgaria Montenegro Bosnia Moldova Serbia Azerbaijan Hungary Armenia Macedonia Georgia 20 .3 .35 .4 .45 Gini coefficient of income inequality 6 World Bank Working Paper of the satisfaction to life Figure 1.4. Satisfaction with Life Among the Youth question, where 1 indicates is Generally Higher than Among the Elderly strong disagreement and All things considered, I am satisfied with my life right now 5 indicates strong agreement with the statement that the 50 respondent is satisfied with his/her life. 40 Table 1.1 reports aver- 30 age SWL scores by (i) self- respondents of assessed economic welfare 20 and (ii) comparisons relative Percent to peers, and reveals a num- 10 ber of interesting patterns. For instance, looking across 0 18-30 yrs 31-40 yrs 41-50 yrs 51-60 yrs 61-70 yrs 71+ yrs the table's bottom row, Age-Group of the Respondent we find average SWL score Agree Disagree increases with self-assessed level of welfare: respon- dents in the highest welfare quintile have an average SWL of 3.9 compared to Figure 1.5. Satisfaction with Life is Higher for the Healthy only 2.3 for those that place themselves in the lowest All things considered, I am satisfied with my life right now quintile. Moreover, within 60 each self-assessed welfare group, we find average SWL scores to be positively cor- 40 related with extent to which respondents of the respondents feel they've 20 done better than their high Percent school classmates (looking down each of the columns). 0 Very good Good Medium Bad Very bad Satisfaction with life thus Self-Assessed Health Status depends, it seems, not just Agree Disagree on (self-perceived absolute welfare), but rather also on how individuals seem to think they have done relative to their peers. Thus, respondents that place themselves in the highest welfare quintile and who "strongly agree" that they've done better in life than their high school classmates have an average SWL score three times as high (4.8 vs. 1.6) as those that place themselves in the lowest quintile and who "strongly disagree" with having done better than their classmates.8 8. A similar pattern is evident if instead one looks at responses to the "I have done better in life than most of the colleagues I had in 1989" statement. Please see Table A1 in Annex for these summary statistics. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 7 Table 1.1. Average SWL Score by Self-perceived Economic Status I Have Done Better in Life Than Self-assessment of Own Economic Welfare (quintile) Most of My High School Classmates Lowest 2 3 4 Highest Overall Strongly disagree 1.6 1.9 2.3 2.2 . . . 1.8 Disagree 2.3 2.7 3.1 3.0 . . . 2.7 Neither 2.6 3.0 3.4 3.7 3.7 3.2 Agree 2.7 3.5 3.8 3.9 4.0 3.6 Strongly agree 2.8 3.6 4.0 4.6 4.8 4.0 Overall 2.3 3.0 3.5 3.8 3.9 3.1 . . . Cell has fewer than 30 observations Average SWL scores reported in Table 1.2, which shows average SWL scores by self-assessed economic welfare, both (i) today, and (ii) around 1989, suggest that respondents' current satisfaction with life depend not just on current economic wel- fare, but also on how they think its changed over time (moving across the rows from left to right, average scores generally fall--i.e. we observe negative slopes across all rows except the last). Table 1.2. Average SWL Score by Present and Past Self-assessed Economic Status Self-assessment of Own Economic Self-assessment of Welfare in 1989 (quintile): Welfare (quintile) at Time of Survey Lowest 2 3 4 Highest Overall Lowest 2.5 2.4 2.4 2.1 1.9 2.4 2 3.4 3.2 2.9 2.7 2.9 3.0 3 3.6 3.7 3.6 3.2 3.2 3.5 4 3.8 4.0 4.0 3.6 3.4 3.8 Highest . . . 3.6 3.9 4.1 4.1 3.9 Overall 3.0 3.1 3.1 3.0 3.0 3.1 . . . Cell has fewer than 30 observations Figure 1.6 illustrates how SWL varies by the level of respondents' trust in people. Among those reporting "complete distrust" in people today, the proportion that are dis- satisfied with their lives outnumber those that are satisfied: by contrast, among those reporting "complete trust" in people today, those satisfied with their lives outnumber those that are dissatisfied by 3 to 1. Similarly, respondents reporting having greater trust in people today as compared to before 1989 tend, on average, to have higher SWL scores than others (as can be seen in Table 1.3, average scores in the cells below the diagonal tend to be higher than those above the shaded diagonal). 8 World Bank Working Paper Figure 1.6. Satisfaction with Life is Positively Correlated with Level of Trust in People All things considered, I am satisfied with my life right now 60 40 respondents of rcente 20 P 0 Complete distrust 2 3 4 Complete trust Level of trust in people today Agree Disagree Table 1.3. Average SWL Score by Present and Past Level of Social Capital Level of Trust in People Before 1989 Level of Trust In People Today Lowest 2 3 4 Highest Overall Lowest 2.7 2.5 2.8 2.6 2.7 2.7 2 2.8 3.2 3.1 3.0 3.0 3.0 3 2.6 3.4 3.2 3.2 3.2 3.2 4 3.5 3.2 3.4 3.4 3.2 3.3 Highest 3.5 3.1 3.7 3.4 3.4 3.4 Overall 2.8 3.0 3.2 3.1 3.1 3.1 Key Factors Influencing SWL: Multivariate Analysis In order to better examine the correlate of satisfaction with life, we use an ordered probit model to analyze respondents' expressed level of satisfaction with life according to the following model specification: yi = xi + i * We do not observe y*i directly, but rather only observe whether yi = 1, 2, 3, 4, and 5, if j < y*i < j ( j = 1, 2, 3, 4, and 5 respectively)--i.e. the expressed level of satisfaction with -1 life on a 5 point scale. Various factors that we think influence the level of satisfaction expressed by the respondent are included in the vector xi (see below). Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 9 Variable Obs Mean Std. Dev. Min Max swl 27395 3.119146 1.155076 1 5 health 27996 2.725782 1.002334 1 5 sc_now 27032 2.64542 1.232121 1 5 sc_change 23769 -1.009214 1.473212 -4 4 lnpcexp 27911 8.708414 .8421385 2.204605 12.06602 car 27993 40.81735 49.15043 0 100 secondhouse 28000 88.95 31.35176 0 100 bankaccount 27978 36.48938 48.1409 0 100 creditcard 27975 31.13137 46.3039 0 100 mobilephone 27993 62.92645 48.30104 0 100 computer 27988 27.55109 44.67794 0 100 pew 24067 2.999668 1.008523 1 5 diff 28000 -1.187571 2.355999 -9 9 unemploy 28000 .085 .2788867 0 1 transfer 28000 .0237143 .1521603 0 1 hhsize 28000 3.238929 1.840137 1 12 age 28000 46.50054 17.79848 17 97 age2 28000 2479.074 1760.618 289 9409 country 28000 14.60696 8.247696 1 29 locality 28000 1.836929 .7392072 1 3 swl: Satisfied with life, coded as 1 = strongly disagree (SD), 2 = disagree (D), 3 = neither disagree nor agree (N), 4 = agree (A), and 5 = strongly agree (SA). health: How would you assess your health, 1 = Very good, 2 = Good, 3 = Medium, 4 = Bad, 5 = Very bad. sc_now: Level of trust in people today, 1 = Complete distrust, 2 = Some distrust, 3 = Neither, 4 = Some trust, 5= Complete trust. sc_change: Difference between sc_now and sc_before (with both variables coded as above). lnpcexp: (log) Per equivalent adult (using OECD scales) annual expenditures (in PPP$). car, secondhouse, bankaccount, creditcard, mobilephone, computer: 100 = Respondent reports owning the item. pew: I have done better in life than most of my high school classmates/colleagues I had around 1989. (responses coded the same as swl). diff: Change in self-perceived decile group ranking between 1989 and present. unemploy: 1 = Not working at present and actively looked for a job at this moment, 0 otherwise. transfer: 1 = Transfers (unemployment benefits, social benefits, and/or help from charities and non-government organizations) are the most important source of livelihood for the household. hhsize: Total household size age: Age in years age2: Age in years, squared country: Country code (29 unique values for each of the 29 countries covered in the survey: Mongolia excluded) locality: 1 = Urban, 2 = Rural, 3 = Metropolitan area. 10 World Bank Working Paper Table 1.4. Simulated Probabilities Derived from Ordered Probit Model Predicted Probability of Response: Strongly Strongly Ideal Type Disagree Disagree Neither Agree Agree "Average" respondent: 0.05 0.23 0.30 0.38 0.04 Self-reported health status: Very good 0.03 0.18 0.28 0.44 0.06 Medium 0.05 0.23 0.30 0.38 0.04 Very bad 0.12 0.33 0.29 0.24 0.01 Trust in people today: Complete distrust 0.08 0.28 0.30 0.32 0.02 Neither 0.05 0.23 0.30 0.38 0.04 Complete trust 0.03 0.18 0.28 0.45 0.06 Per-capita expenditures half the sample mean 0.06 0.25 0.30 0.35 0.03 Per-capita expenditures twice the sample mean 0.05 0.22 0.29 0.40 0.04 Per-capita expenditures four times the 0.04 0.20 0.29 0.42 0.05 sample mean Done better than peers: Strongly disagree 0.26 0.40 0.22 0.11 0.00 Neither 0.05 0.23 0.30 0.38 0.04 Strongly agree 0.01 0.07 0.18 0.56 0.18 Moved from 3rd to 7th decile between 1989 0.01 0.11 0.23 0.52 0.11 to present Main income source is transfers 0.07 0.26 0.30 0.34 0.03 Person is currently unemployed 0.07 0.27 0.30 0.33 0.03 64 yr old, self-reported health status very bad, 0.77 0.20 0.03 0.00 0.00 per-capita expenditures half sample mean, down 4 decile places, strongly disagrees she/he has done better than peers; transfers main income source, currently unemployed, complete distrust in people today; complete trust in people before 1989. 28 yr old, self-reported health status very good, 0.00 0.01 0.05 0.45 0.48 employed, per-capita expenditures twice sample mean, up 4 decile places, agrees she/he has done better than peers; complete trust in people today; complete distrust in people before 1989. The results of the ordered probit model using the above set of explanatory variables/controls (Table 1.5) confirm the following hypotheses: Satisfaction with life is positively correlated with health status, with very good health status increasing the probability of life satisfaction by more than 20 percent; the role played by overall health status is particularly important in the EU country group.9 9. As noted earlier, this country group excludes Romania and Bulgaria, which are included under "Other." Table 1.5. Ordered Probit Results: SWL by Country Groups Overall EU SEE CIS_L CIS_M Other coef sd coef sd coef sd coef sd coef sd coef sd (i) Health Status: Very good 0.146*** 0.030 0.317*** 0.056 0.247*** 0.050 0.034 0.080 0.362** 0.141 0.203* 0.109 Good 0.095*** 0.019 0.096*** 0.036 0.123*** 0.038 0.232*** 0.044 0.101* 0.052 0.134* 0.074 Medium Reference Category Bad -0.246*** 0.023 -0.200*** 0.041 -0.237*** 0.049 -0.228*** 0.047 -0.347*** 0.059 -0.313*** 0.089 Very bad -0.469*** 0.039 -0.569*** 0.070 -0.306*** 0.079 -0.497*** 0.080 -0.405*** 0.124 -0.330*** 0.126 (ii) Level of trust in people Complete distrust -0.220*** 0.027 -0.216*** 0.053 -0.159*** 0.050 -0.149** 0.062 -0.221*** 0.079 -0.300*** 0.099 Some distrust -0.020 0.023 0.000 0.039 0.017 0.046 -0.027 0.058 -0.076 0.064 -0.274*** 0.092 Neither trust nor distrust Reference Category Some trust 0.146*** 0.023 0.175*** 0.038 0.106** 0.046 0.264*** 0.056 0.000 0.059 -0.022 0.093 Complete trust 0.280*** 0.039 0.386*** 0.077 0.262*** 0.093 0.315*** 0.078 0.191** 0.091 -0.075 0.167 Change in trust since 1989 0.021*** 0.007 0.030** 0.015 0.057*** 0.014 -0.033** 0.016 0.035* 0.020 0.031 0.026 (iii) Economic status: Log normalized expenditures ($) 0.093*** 0.013 0.142*** 0.026 0.082*** 0.025 0.184*** 0.027 0.009 0.035 -0.005 0.048 Household owns: A car 0.077*** 0.019 0.112*** 0.034 0.044 0.034 0.110** 0.045 0.103** 0.049 0.246*** 0.077 Second home 0.045* 0.025 0.034 0.039 0.136*** 0.052 0.068 0.079 0.008 0.077 0.124 0.085 A bank account 0.178*** 0.021 0.171*** 0.041 0.059 0.038 0.046 0.097 0.002 0.057 0.096 0.085 A credit/debit card 0.071*** 0.022 0.053 0.036 0.039 0.039 0.013 0.088 0.143** 0.060 0.048 0.085 A mobile phone -0.031 0.020 0.048 0.041 0.084** 0.043 -0.106** 0.042 0.056 0.053 0.099 0.088 A computer 0.052** 0.021 0.052 0.036 0.073* 0.039 -0.032 0.077 -0.030 0.055 0.078 0.078 (continued) Table 1.5. Ordered Probit Results: SWL by Country Groups (Continued) Overall EU SEE CIS_L CIS_M Other coef sd coef sd coef sd coef sd coef sd coef sd Done better than peers Strongly disagree -1.008*** 0.033 -0.962*** 0.060 -0.965*** 0.060 -1.085*** 0.075 -1.453*** 0.106 -0.729*** 0.105 Disagree -0.294*** 0.020 -0.366*** 0.035 -0.197*** 0.040 -0.345*** 0.044 -0.439*** 0.054 -0.149* 0.084 Neither disagree or agree Reference Category Agree 0.353*** 0.020 0.284*** 0.037 0.406*** 0.038 0.471*** 0.046 0.339*** 0.052 0.281*** 0.078 Strongly agree 0.872*** 0.036 0.910*** 0.068 0.811*** 0.071 1.081*** 0.095 0.941*** 0.088 0.628*** 0.107 Change in decile ranking: From 1989 to present: 0.121*** 0.003 0.118*** 0.007 0.129*** 0.006 0.091*** 0.007 0.125*** 0.010 0.108*** 0.013 Transfers main income source -0.141*** 0.053 -0.145 0.092 -0.298*** 0.097 -0.025 0.139 0.019 0.245 0.129 0.153 Unemployed Residence -0.238*** 0.029 -0.176*** 0.065 -0.184*** 0.052 -0.280*** 0.051 -0.262*** 0.101 -0.153 0.130 Rural Urban -0.020 0.018 -0.124*** 0.032 0.052 0.034 -0.062 0.045 0.156*** 0.047 0.071 0.071 Metropolitan areas -0.088*** 0.022 -0.073* 0.038 -0.073* 0.042 -0.189*** 0.049 -0.036 0.073 0.042 0.082 Other controls (age, age squared, household size, etc.) omitted. Pseudo-R2 0.130 0.131 0.125 0.140 0.143 0.122 Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 13 Social capital (using responses to the question on level of trust in other people as a proxy), appears to matter quite a bit--respondents that said they trusted other people were significantly more likely to be satisfied with their lives than those that did not. Overtheregionasawhole,thosewithtwicetheaveragepercapitaexpendituresare about 13 percent more likely to report being satisfied with their lives compared to those with half the average PCE. Correlation between satisfaction with life and income varies quite a bit across the various country groupings: for instance, the association between level of income and SWL is strongest in CIS low-income coun- tries, but weakest in CIS middle-income countries. Satisfactionwithlifetendstobepositivelycorrelatedwithownershipofmiscellaneous durable goods; other things being equal, a person owning a car, a mobile phone, and a computer is about 10 percent more likely to report being satisfied with his/her life compared to a person not owning any of these durable goods. Asagroup,theyoungaremuchmorelikelytobesatisfiedwithlifethantheelderly; however, controlling for the influence of other factors, there are in fact only very small differences in satisfaction levels across different age groups. Relativestatusmatters.Ahigherperceivedrelativepositionwithrespecttopeersand an improvement in the self-perceived difference in decile ranking relative to 1989 increase the likely satisfaction with life; this is true across all the country groups. Workstatusisveryimportant;theemployedareabout16percentmorelikelytobe satisfied with life than those who are unemployed; in other words, other things being equal the difference in satisfaction levels between the employed and the unemployed are greater in magnitude than those between people with half and twice the average per capita expenditures; the negative impact of unemployment status on overall satisfaction with life is found to be quite strong across all country groups. In addition, the findings of the regression analysis provide some additional insights too in areas where, a priori, the relationship between satisfaction with life and other variables is not so obvious. For instance: Wefindthatwheretransfersareasignificantsourceofincome(forexample,inthe EU and SEE groups, a person is about five times as likely to report transfers as the main source of income as in the CIS middle-income group) those reporting transfers as the households' main income source are significantly less likely to be satisfied with their lives compared to respondents whose households are not so heavily reliant on transfer incomes. For instance, in the SEE (where this difference is the starkest) those not dependent on transfers as their main income source are about 48 percent more likely to be satisfied with their lives as compared to those reporting transfers as their households' main source of income. Finally,incontrasttothefindingsofearlierstudieswhichindicatethataverageliving standards of people living in metropolitan areas are both better on average as well as have been improving faster than those living elsewhere, the results of our analysis show that, other things being equal, people living in metropolitan areas are in fact less likely to be satisfied with their lives than those living either in other urban areas or else in rural areas. However, this might simply be an artifact of the failure of the welfare measure used in the analysis (log per capita expenditures in PPP dollars) to control for cost-of-living differences across these localities. 14 World Bank Working Paper Comparisons over Time and Across Countries How do average satisfaction rates from the 2006 LiTS compare to the findings from similar surveys--for example, the World Values Survey (WVS)10--conducted earlier in the same countries? Responses to the satisfaction with life question in the latter survey are recorded on a 1­10 point scale, so need to be adjusted accordingly before comparing directly with the LiTS findings. A comparison of the adjusted WVS scores (modified to a 1­5 scale) with the LiTS for the 21 countries for which such over-time comparisons are indeed possible are presented in Table 1.6. Comparing these two sets of findings thus reveals that over the past 6­8 years, average SWL scores increased the most in Belarus, the Baltic states, and Russia, but in fact declined in some countries in the Balkans--for example, Bosnia, Serbia, and Montenegro (Table 1.6). Table 1.6. Change over Time in Average SWL Rates by Country World Values Survey 2006 Score Year LiTS Change in Score Albania 2.7 1998 3.2 0.5 Azerbaijan 3.0 1997 2.7 -0.3 Belarus 2.5 1996 3.6 1.1 Bosnia 3.0 1998 2.6 -0.4 Bulgaria 2.6 1997 2.8 0.2 Croatia 3.3 1996 3.3 0.0 Czech Republic 3.4 1998 3.4 0.0 Estonia 2.8 1996 3.5 0.7 Georgia 2.6 1996 2.5 -0.1 Hungary 3.2 1998 2.6 -0.6 Latvia 2.7 1996 3.3 0.6 Lithuania 2.8 1997 3.3 0.5 Moldova 2.2 1996 2.7 0.5 Montenegro 3.3 1996 2.7 -0.6 Poland 3.4 1997 3.3 -0.1 Romania 2.7 1998 2.9 0.2 Russia 2.5 1995 3.1 0.6 Serbia 3.0 1996 2.6 -0.4 Slovakia 3.3 1998 3.4 0.1 Slovenia 3.4 1995 3.8 0.4 Turkey 3.3 1996 3.1 -0.2 10. The World Values Survey is a worldwide investigation of socio-cultural and political change, conducted by a network of social scientist at universities all around world. For more details, see http://www.worldvaluessurvey.org Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 15 On the whole, there appears to be a fairly close conformance across the region at the country level between changes in average SWL scores and performance of the economy. For instance, countries like the Baltic states, Belarus, Moldova, Russia, and Albania, which have recorded relatively high growth rates in recent years also tend to be the ones where average SWL scores appear to have increased the most in recent years; by contrast, other countries with relatively poorly performing economies, like Bosnia, Hungary, Poland, and Turkey, also tend to be the ones with the least improvements (or in some cases actual declines) in average SWL scores (Table 1.7). Table 1.7 Comparing GDP and SWL Changes in Recent Years Change in GDP (%) Between 2000 and 2005 Change in SWL Bottom Middle Top Bottom Bosnia, Hungary, Montenegro, Serbia Azerbaijan Poland, Turkey Middle Slovenia, Czech Romania, Bulgaria, Georgia Republic, Croatia Slovakia Top -- Russia, Albania Latvia, Lithuania, Estonia, Belarus, Moldova How do average satisfaction rates in countries in the Europe and Central Asia region compare to countries with similar incomes in other regions in the world? Data from the recent Pew Global Attitudes survey reveal two interesting findings of relevance to our work: (i) in countries in Latin America and Eastern Europe for which comparable data are available over time, there is a high correlation between income growth and changes in happiness over time, (ii) Eastern European respondents appear to be less satisfied with their lives than Latin Americans.11 That individuals in transition countries tend to have lower self-reported SWL rates compared to those in non-transition countries has also been noted in other earlier studies, which have tended to attribute this to difficulties faced by people in these countries to adapt to the profound economic and social changes that have taken place over this period.12 In addition to the role played by falling incomes and rising inequality discussed above (and earlier), we offer three additional broad sets of factors that might help explain this, and examine some evidence in support of these conjectures: changes in (i) employment, (ii) trust, and (iii) rising inequality. First, unemployment rates rose and activity rates fell in many transition countries with the transitional recession. Despite the subsequent recovery of output, these indicators have 11. Stokes, Bruce: "Happiness is Increasing in Many Countries--But Why?--Rising Incomes a Big Reason, But Not the Only One" available at: http://pewglobal.org/commentary/display.php?AnalysisID=1020 12. E.g. see Peter Sanfey & Utku Teksoz "Does Transition Make You Happy?" EBRD working paper #91, April 2005. 16 World Bank Working Paper Figure 1.7. Cross-Country Comparisons: SWL and Employment Rates All things considered, I am satisfied with my life right now 70 Slovenia Uzbekistan Estonia Belarus Tajikistan 60 Slovakia life Kyrgyz Croatia with Latvia Czech Kazakhstan Lithuania 50 Poland satisfaction Turkey Albania on Russia 40 neutral Ukraine Mongolia above Romania % 30 Bulgaria Montenegro Moldova Azerbaijan Bosnia Serbia Armenia Macedonia Hungary Georgia 20 30 40 50 60 70 80 % respondents aged 65 years or under that worked during past 12 months failed to return to their pre-transition levels with slower progress in job creation.13 As illus- trated in Figure 1.7, there is clearly a strong positive correlation evident between SWL and employment rates, and the fact that employment rates in Eastern Europe continue to stagnate at fairly low overall rates may be part of the reason why, other things being equal, SWL scores are lower than in Latin America. Second, the LiTS data reveal one very interesting finding: level of trust in other people appears to have been eroded considerably during the transition period (see appendix table A8). This is not surprising given that social capital was seriously undermined during the economic crisis (World Bank 2005a). Given that SWL is clearly positively correlated with average level of trust in people (Figure 1.8), the secular decline in overall levels of trust over time in ECA may be another reason why satisfaction rates are lower than in Latin America. Third, income inequality has risen in most ECA countries during the transition. There also seems to be a fairly widespread feeling among respondents that somehow others have done better during transition than they have. Given the importance of bench- marking themselves relative to peers in determining overall SWL as identified in our earlier analysis, this in turn may be another reason why people in ECA are, on average, less satisfied than in other countries. 13. See World Bank 2005, Enhancing Job Opportunities in Eastern Europe and the former Soviet Union, Washington, DC. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 17 Figure 1.8. Cross-Country Comparisons: SWL and Level of Trust in Others All things considered, I am satisfied with my life right now Slovenia 70 Uzbekistan Estonia Belarus Tajikistan 60 life Slovakia Kyrgyz with Croatia Latvia Kazakhstan Czech Lithuania 50 Poland satisfaction Turkey Albania on Russia 40 neutral Ukraine Mongolia above Romania % 30 Bulgaria Bosnia Moldova Montenegro Macedonia Serbia Armenia Hungary Azerbaijan Georgia 20 10 20 30 40 50 % above neutral on trust in people today Concluding Observations The Life in Transition Survey provides a useful new addition to welfare measures commonly applied to ECA countries. The survey provides information on this new "satisfaction with life" measure across all ECA countries. The results show that most people in ECA report satisfaction with life. But there are large variations across ECA countries, with a highest rate of 90 percent in Slovenia and the lowest rate of 40 percent in Georgia. Our analysis confirms the importance of expected factors like income per capita, equality of incomes, youth, working status, non-agricultural employment, non-metropolitan living, and better edu- cation. However, our analysis also highlights the important role played by subjective factors like self-assessed health status, level of trust in people, relative economic status compared to peers and own perception of improvement in economic status over time are also important factors determining overall satisfaction levels. Satisfaction with life in ECA is also, not sur- prisingly, lower than the levels found in other countries. This likely reflects the transitional recessional and the social costs of the major economic transformation undertaken by most of these countries. 18 World Bank Working Paper Annex: Tables and Figures Table A1.1. Satisfaction with Life Question by Country All things considered, I am satisfied with my life now 60 Agree 50 Strongly Agree Strongly Strongly 40 Agree Disagree 30 respondents of 20 Disagree % 10 Agree 0 states Neither income income Europe Turkey CIS-low member EU CIS-middle South-Eastern Percentage of Respondents Who . . . Strongly Strongly Group/Country Disagree Disagree Neither Agree Agree Overall CIS-low income 9 21 19 41 10 100 EU member states 9 21 26 35 9 100 CIS-middle income 9 22 24 35 10 100 South-Eastern Europe 18 23 24 28 7 100 Turkey 15 16 24 32 12 100 Overall sample: 10 21 24 35 10 100 Slovenia 1 8 19 54 18 100 Belarus 2 11 20 57 10 100 Uzbekistan 3 13 16 53 15 100 Tajikistan 6 11 15 47 21 100 Estonia 4 17 14 52 13 100 Czech Republic 4 14 28 41 12 100 Slovakia 4 17 21 51 7 100 Kazakhstan 4 18 25 44 9 100 Kyrgyz Republic 5 20 16 53 6 100 Poland 6 18 27 39 11 100 Lithuania 6 22 20 40 12 100 Croatia 11 14 20 42 13 100 Latvia 8 21 17 46 9 100 Albania 8 18 30 33 11 100 Mongolia 6 21 34 32 7 100 Russia 10 22 24 33 11 100 (continued) Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 19 Table A1.1. Satisfaction with Life Question by Country (Continued) Percentage of Respondents Who . . . Strongly Strongly Group/Country Disagree Disagree Neither Agree Agree Overall Turkey 15 16 24 32 12 100 Ukraine 12 26 25 31 6 100 Romania 13 25 29 27 6 100 Bulgaria 13 32 26 23 6 100 Moldova 15 29 28 25 3 100 Montenegro 20 26 25 24 5 100 Azerbaijan 15 34 25 21 5 100 Bosnia 20 25 27 26 2 100 Macedonia 21 26 26 23 4 100 Hungary 21 26 27 22 4 100 Serbia 23 29 22 22 4 100 Armenia 20 34 20 24 2 100 Georgia 19 35 24 19 3 100 Table A1.2. Perceptions Regarding Changes over Time in Economic Situation by Country The economic situation in this country is better today than in 1989 60 Agree 50 Strongly Strongly Agree Agree Strongly 40 Disagree 30 Agree spondentser of 20 % Disagree 10 Neither 0 e mo CIS-low income inc member atests CIS-middle Europe EU South-Eastern Percentage of Respondents Who . . . Strongly Strongly Group/Country Disagree Disagree Neither Agree Agree Overall CIS-low income 14 31 15 32 8 100 CIS-middle income 15 32 16 28 9 100 EU member states 18 28 20 26 8 100 South-Eastern Europe 39 29 14 14 4 100 (continued) 20 World Bank Working Paper Table A1.2. Perceptions Regarding Changes over Time in Economic Situation by Country (Continued) Percentage of Respondents Who . . . Strongly Strongly Group / Country Disagree Disagree Neither Agree Agree Overall Overall sample: 18 30 16 26 9 100 Belarus 1 12 18 56 12 100 Estonia 3 13 17 45 22 100 Albania 8 9 12 55 16 100 Kazakhstan 5 20 13 47 15 100 Mongolia 5 23 19 45 10 100 Lithuania 7 24 16 41 12 100 Czech Republic 9 23 21 32 15 100 Uzbekistan 9 26 15 36 13 100 Azerbaijan 8 28 20 41 3 100 Latvia 13 28 15 36 9 100 Poland 16 23 20 32 10 100 Slovenia 11 29 21 28 10 100 Russia 13 29 18 29 11 100 Slovakia 12 33 17 29 9 100 Tajikistan 18 30 11 30 11 100 Armenia 23 28 11 30 7 100 Turkey 27 27 12 21 12 100 Romania 19 33 24 19 5 100 Georgia 24 36 13 23 5 100 Moldova 20 41 16 19 4 100 Kyrgyz Republic 21 48 6 22 2 100 Bulgaria 23 40 19 14 3 100 Ukraine 25 45 13 15 3 100 Croatia 42 24 18 13 3 100 Montenegro 41 35 11 11 2 100 Hungary 40 35 14 9 2 100 Serbia 41 34 15 7 2 100 Macedonia 44 38 11 5 1 100 Bosnia 53 31 11 4 1 100 Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 21 Table A1.3. Satisfaction with Changes over Time in Living Conditions by Country I have done better in life than my parents 60 Agree 50 Strongly Strongly Agree Strongly Disagree Agree 40 30 Disagree spondentser of 20 % Agree 10 Neither 0 e mo inc member atests CIS-low income CIS-middle Europe EU South-Eastern Percentage of Respondents Who . . . Strongly Strongly Group/Country Disagree Disagree Neither Agree Agree Overall CIS-middle income 5 19 23 40 13 100 EU member states 6 19 24 38 14 100 South-Eastern Europe 11 21 22 33 13 100 CIS-low income 8 31 24 30 7 100 Overall sample: 7 21 24 36 13 100 Belarus 2 9 23 53 13 100 Slovenia 2 13 23 44 19 100 Estonia 2 14 21 42 21 100 Albania 5 9 14 52 19 100 Slovakia 4 12 20 51 14 100 Romania 3 13 24 44 16 100 Lithuania 2 18 18 38 24 100 Latvia 4 18 18 42 18 100 Czech Republic 4 17 27 36 15 100 Russia 4 20 23 38 15 100 Croatia 10 15 20 34 21 100 Kazakhstan 5 20 23 41 11 100 Poland 6 21 24 35 14 100 Ukraine 7 20 23 42 8 100 Bulgaria 6 23 23 39 8 100 Tajikistan 6 25 25 32 12 100 Uzbekistan 6 28 20 37 10 100 Moldova 7 20 33 36 4 100 (continued) 22 World Bank Working Paper Table A1.3. Satisfaction with Changes over Time in Living Conditions by Country (Continued) Percentage of Respondents Who . . . Strongly Strongly Group/Country Disagree Disagree Neither Agree Agree Overall Serbia 13 22 23 30 12 100 Hungary 12 24 27 28 9 100 Macedonia 13 26 24 29 8 100 Montenegro 10 29 28 26 8 100 Turkey 18 22 25 22 13 100 Bosnia 11 27 27 30 5 100 Kyrgyz Republic 5 39 21 32 4 100 Georgia 11 36 23 24 6 100 Mongolia 11 36 30 20 3 100 Azerbaijan 12 39 30 16 3 100 Armenia 14 42 24 17 3 100 Table A1.4. Average SWL Score: Colleagues in 1989 Rather than School Mates as Peers I Have Done Better in Life Than Most Self-assessment of Own Economic Welfare (quintile): of the Colleagues I Had Around 1989 Lowest 2 3 4 Highest Overall Strongly disagree 1.6 1.9 2.3 2.2 . . . 1.8 Disagree 2.3 2.6 2.9 2.9 . . . 2.6 Neither 2.6 3.0 3.4 3.6 3.3 3.1 Agree 2.8 3.4 3.7 4.0 4.0 3.5 Strongly agree 3.5 3.6 4.2 4.2 4.9 4.0 Overall 2.3 2.9 3.4 3.8 3.9 3.0 . . . Cell has fewer than 30 observations Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 23 Table A1.5. Tendency to Feel I've Done Worse During Transition Than Others, by Level of Income Inequality Average Reported Change in Prevailing Level of Income Decile Rank Between Inequality (Gini) in Country Country 1989 and Present (based on PCEXP) Albania 0.40 0.34 Belarus -0.08 0.37 Slovenia -0.32 0.29 Czech Republic -0.37 0.29 Turkey -0.64 0.36 Kazakhstan -0.76 0.35 Poland -0.85 0.32 Slovakia -0.85 0.30 Estonia -0.87 0.34 Lithuania -1.01 0.37 Romania -1.02 0.40 Kyrgyz Republic -1.05 0.36 Moldova -1.11 0.44 Hungary -1.18 0.35 Russia -1.19 0.38 Latvia -1.19 0.38 Tajikistan -1.35 0.31 Ukraine -1.35 0.45 Uzbekistan -1.42 0.32 Armenia -1.44 0.43 Croatia -1.58 0.34 Bulgaria -1.60 0.37 Macedonia, FYR -1.93 0.34 Azerbaijan -2.09 0.38 Montenegro -2.14 0.29 Serbia -2.20 0.35 Bosnia -2.30 0.33 Georgia -2.49 0.40 Overall 1.10 0.37 Correlation coefficient: 0.16 24 World Bank Working Paper Table A1.6. Fall in Social Capital? Decline in Reported Level of Trust in Other People (a) Trust in People Before 1989 60 Some Trust Complete 50 Complete Trust Distrust 40 Complete Trust Some Distrust 30 respondents of Neither % 20 10 Some Trust 0 states income income low Europe middle South-Eastern CIS member CIS EU Percentage of Respondents Who Have . . . Complete Some Complete Group/Country Distrust Distrust Neither Some Trust Trust Overall CIS low-income 5 8 14 34 40 100 CIS middle-income 6 11 14 39 31 100 South-Eastern Europe 8 10 20 45 18 100 EU member states 6 13 29 41 11 100 Overall sample: 6 11 18 38 28 100 Mongolia 2 2 7 30 60 100 Uzbekistan 2 6 12 32 49 100 Georgia 2 7 14 45 31 100 Bulgaria 3 6 14 55 22 100 Kazakhstan 3 9 9 41 37 100 Kyrgyz Republic 7 4 5 37 47 100 Tajikistan 4 9 17 29 41 100 Lithuania 2 11 18 52 16 100 Montenegro 5 8 18 48 21 100 Hungary 4 8 29 46 14 100 Russia 6 10 13 38 33 100 Turkey 8 9 16 28 39 100 Serbia 6 10 17 48 19 100 Estonia 3 12 24 49 12 100 Slovenia 3 11 28 48 10 100 Macedonia 6 10 22 33 29 100 Moldova 7 11 17 46 20 100 Belarus 4 14 23 44 15 100 Latvia 4 14 14 56 12 100 (continued) Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 25 Table A1.6. Fall in Social Capital? Decline in Reported Level of Trust in Other People (Continued) Percentage of Respondents Who Have . . . Complete Some Complete Group/Country Distrust Distrust Neither Some Trust Trust Overall Ukraine 6 14 17 38 26 100 Azerbaijan 10 9 17 28 35 100 Armenia 6 14 20 36 24 100 Bosnia 7 12 22 41 19 100 Croatia 7 9 23 49 12 100 Poland 6 12 30 40 11 100 Slovakia 7 12 31 40 11 100 Albania 15 8 16 47 14 100 Romania 10 16 30 36 8 100 Czech Republic 5 20 38 33 4 100 (b) Trust in People Today 60 50 Complete Trust 40 Complete Some Trust Distrust 30 Complete Trust Some Trust respondents of %20 10 Neither Some Distrust 0 income income member states CIS-middle Europe EU South-Eastern CIS-low Percentage of Respondents Who Have . . . Complete Some Complete Group/Country Distrust Distrust Neither Some Trust Trust Overall CIS-middle income 20 26 17 29 8 100 CIS-low income 30 21 16 24 9 100 EU member states 19 27 27 24 3 100 South-Eastern Europe 29 22 23 23 3 100 Overall sample: 25 25 19 24 7 100 Georgia 12 22 21 36 9 100 Belarus 9 23 27 35 6 100 Estonia 8 30 22 36 4 100 Kazakhstan 14 31 12 35 8 100 (continued) 26 World Bank Working Paper Table A1.6. Fall in Social Capital? Decline in Reported Level of Trust in Other People (Continued) Percentage of Respondents Who Have . . . Complete Some Complete Group/Country Distrust Distrust Neither Some Trust Trust Overall Slovakia 13 21 30 33 4 100 Ukraine 17 29 16 33 6 100 Slovenia 11 24 37 26 2 100 Latvia 13 36 15 33 3 100 Lithuania 13 33 23 27 4 100 Russia 23 25 17 26 8 100 Czech Republic 10 32 32 22 4 100 Tajikistan 29 21 14 26 10 100 Uzbekistan 30 19 16 26 10 100 Moldova 24 25 17 28 6 100 Poland 17 29 27 24 4 100 Serbia 26 21 22 28 3 100 Croatia 21 25 28 23 3 100 Montenegro 21 26 26 24 3 100 Romania 24 21 30 22 3 100 Hungary 23 25 28 21 3 100 Kyrgyz Republic 36 29 7 21 8 100 Albania 38 18 17 21 6 100 Azerbaijan 40 17 19 16 8 100 Mongolia 31 27 19 19 5 100 Armenia 35 26 18 16 5 100 Bulgaria 33 31 15 19 3 100 Bosnia 32 24 24 19 2 100 Turkey 47 20 14 11 8 100 Macedonia 47 17 20 14 3 100 CHAPTER 2 Employment, Sources of Income, and the Poor in Eastern Europe and the Former Soviet Union14 T he past two decades in Eastern Europe and the former Soviet Union have been times of tremendous change, with countries undergoing rapid transformation from centrally-planned to market-oriented economies. While poverty increased during the initial years of transition, primarily on account of the sharp economic con- traction, the resurgence of economic growth in the region since 1998 has resulted in a rebound in household incomes and living standards. The most recent comprehensive assessment of growth, poverty and inequality in ECA was done in 2005 (see World Bank 2005a). The study documented the sharp reduction in poverty and the modera- tion of inequality that is taking place in the region. The key source for understanding these changes have been the household and labor force surveys conducted in most ECA countries. While the above-mentioned study is a valiant attempt at providing a quantitative assess- ment of income and poverty in the region, differences in structure of the various question- naires used across different countries (for example, level of disaggregation, recall period, variable coverage of sources of income, and so forth) along with differences in definitions and concepts followed (for example, commodity classification schemes followed, treatment of imputed consumption, and so on) render the task of constructing a comparable measure of welfare across countries an extremely difficult one. The 2006 LiTS provides a hitherto unprecedented opportunity to systematically examine differences in socioeconomic char- acteristics of different income groups using the same survey instrument across all ECA coun- tries. Using this new data source, this paper also provides a systematic examination of household welfare, the sources of household incomes and therefore of potential channels 14. Asad Alam, Pradeep Mitra, and Salman Zaidi. 27 28 World Bank Working Paper for affecting them, the sectoral occupation patterns of the working poor which, in turn, is a good guide to the opportunities available to them for improving their income growth and living standards, as well as of asset ownership and underlying inequalities which may con- strain future accumulation in human and physical wealth. Because our main concern here is with the survey data collected on household expen- ditures, it is worth elaborating upon the specific questions used to gather this information. Respondents in the LiTS were asked two main questions in this regard: (1) approximately how much their household spent on (i) food, beverages, and tobacco, (ii) clothing and footwear, (iii) transport and communications (including phone, mobile phone, and inter- net charges), (iv) recreation, entertainment, meals outside the home, and so forth, during the past 30 days, as well as (2) approximately how much their household spent on (v) edu- cation (including tuition, books, kindergarten expenses), (vi) health (including health insurance), (vii) furnishings (such as sheets, towels, blankets, linen), (viii) household durable goods (such as furniture, household appliances, TV, car), and (ix) other expenses (any additional expenses that the respondent would like to report) during the past 12 months. These nine subcomponents of household expenditures were converted to annual amounts and aggregated to derive a per-capita (and, using the modified OECD scales,15 an normal- ized) measure of individual welfare. In addition to the questions on monthly and annual household expenditures, the LiTS also included a number of other potentially useful questions from a welfare analysis stand- point, such as ownership of various types of assets (cars, computer, mobile phones, and so forth) as well as a question on the minimum amount of money that the household would need in order to make ends meet at the end of each month. Furthermore, the LiTS also includes a subjective welfare measure whereby respondents were asked to place themselves on a ten-step ladder (1=Poorest, 10=Richest), the response to which could potentially also be used as a welfare metric. Finally, LiTS provides a uniform module of sources of income asking respondents about the main sources of livelihood of their households. The next section provides a more detailed discussion of the various pros and cons of each of the alternate welfare measures that could potentially be constructed from the LiTS. We then use the per capita expenditures welfare metric (PCE) to derive a poverty profile for the region as a whole using $PPP2.15 and $PPP4.30 regional poverty lines using 2000 PPPs.16 However, before doing so, it is important to first assess how reliable is this welfare measure. The second section compares the estimates of private consumption per capita from LiTS with those obtained from other more traditional sources, such as the National Accounts as well as other more detailed nationally representative household surveys. The third section then presents the poverty profile for the region obtained using this welfare measure, and wherever possible also compares the extent to which the poverty estimates and profile based on the LiTS data is indeed consistent with those from traditional house- hold surveys. Finally, the fourth section examines in more detail the differences between the poor and non-poor in employment status, sector of employment, as well as main sources of income, while the fifth section provides some concluding observations. 15. These equivalence scales assign a weight of 1 to the first and 0.5 to each subsequent adult house- hold member, and a weight of 0.3 to each household member aged less than 14 years. 16. For a justification and more detailed description of these poverty lines, please see World Bank (2005). Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 29 Choosing Between Alternate Survey-Based Welfare Measures While there is general agreement on the merits of using household survey based estimates of per capita (or suitably normalized) consumption and expenditures as a summary mea- sure of living standards, there is little consensus regarding how long the survey question- naire should be to yield good estimates of household expenditures.17 Greater disaggregation in coverage of items is generally assumed to result in fuller reporting and greater accuracy. However, very detailed consumption modules are costly to administer and may crowd out other information to be collected in the survey while short questionnaires can save time and money and still deliver reasonably accurate PCE estimates. Survey questionnaires currently in use vary in length from as much as several hundred items purchased/consumed over the past one year (for example, India's National Sample Surveys) to as little as one question about the household's total expenditures over the past one month.18 Exploiting the fact that the LiTS questionnaire enables the construction of consump- tion aggregates in 29 countries using data collected in the short consumption module (approximately 10 expenditure items in total), this paper assesses the adequacy of these con- sumption aggregates compared to those obtained from more traditional consumption mod- ules in typical household budget/LSMS surveys in the same countries. While multiple assessment criteria can be used for this purpose, our primary interest is in assessing how well the short consumption module does in terms of enabling the analyst to rank households into broad welfare groups (such as quartiles/quintiles). We start first by first examining var- ious possible individual welfare metrics that can be derived from the 2006 LiTS data. The LITS has several sources that could potentially be used for this purpose: A battery of questions on total household spending on food, beverages, tobacco, transport and communications, recreation and entertainment, education, health, furnishings, and household durables--henceforth referred to as the "short con- sumption aggregate." Aquestionontheminimumamountofmoneythatthehouseholdwouldneedin order to make ends meet at the end of each month--the "one-question welfare aggregate." A question where respondents are asked to place themselves on a ten-step ladder ranging from the poorest (1) to the richest (10). Respondents are asked this ques- tion both in relation to their standing today as well as around 1989. A battery of binary-response questions (yes/no) on ownership of various assets, such as cars, computers, mobile phones, and so forth. Normalized or Per Capita Expenditures? In theory, just as a price index is used in order to make welfare comparisons across house- holds facing different cost-of-living, use of equivalence scales is a way of making comparable 17. Please see the related discussion in "Chapter 5: Consumption" by Angus Deaton and Margaret Grosh in Grosh and Glewe eds. 2000, Designing Household Survey Questionnaire for Developing Countries: Lessons from 15 years of the Living Standards Measurement Study. 18. Thus, for instance, the World Bank's Living Standards Measurement Study (LSMS) surveys tend to have shorter consumption modules (50­80 items) in comparison to typical household budget sur- veys (where, as noted above, the number of consumption items can be as high as 200­300 items). 30 World Bank Working Paper consumption aggregates of households with different demographic composition. In practice, however, does it really matter much if the welfare measure used is based on consumption per equivalent adult (ECE) rather than consumption per capita (PCE)? Rather than prejudge the issue in favor of either measure, particularly given the relatively widespread use of the latter measure in poverty analysis, we use both normalized welfare and per-capita welfare measures in our analysis. Equivalence scales are the deflators used to convert household consumption aggregates into money metric utility measures of individual welfare.19 The modified OECD equivalence scale we use is the same as those used by Eurostat to make welfare comparisons across countries of the European Union. This scale assigns a weight of one to the first person in each household, and 0.5 to each subsequent adult household member. In addition, each household member aged less than 14 years is assigned a weight of 0.3. As one would expect, applying this equivalence scale to the LiTS data results in normalized expendi- tures that are systematically higher than per-capita expenditures across all countries (Table 2.1). Not surprisingly, the difference between the two (normalized vs. per capita expenditures) is lesser in countries with relatively small households (Latvia, Lithuania, Estonia), compared to those countries where the average household size is higher (Tajikistan, Uzbekistan, Azerbaijan). The distribution of (log) normalized expenditures in each country is, in general, quite close to a log-normal distribution (Figure 2.1). One-question Welfare Aggregate (OQE) When asked: "Living in this dwelling and doing what you do, what would be the minimum amount of money that this household would need to make ends meet at the end of each month?" about three-fourth of respondents reported an amount greater than their total (normalized) expenditures. Average OQE across the LiTS sample is $4,376, about 56 per- cent higher than average ECE of $2,804 per-equivalent adult per annum (Table 2.1). In addi- tion, OQE has, in general, both higher dispersion and is "more spiked" compared to the distribution of ECE (ref. Figure 2.1 and Figure 2.2). Subjective Assessment of Welfare (SAW) The LiTS included a question: "Please imagine a ten-step ladder where on the bottom, the first step, stand the poorest people and on the highest step, the tenth, stand the richest. On which step of the ten is your household today?" Thus, unlike the various welfare measures considered so far--PCE, ECE, and OQE, which are based on a continuous monetary scale-- individuals' subjective assessment of their welfare is based on a ten-point scale, with "1" denoting the poorest and "10" the richest. When replying to this question, respondents had a tendency to rank themselves in the middle of the income distribution rather than at the 19. For a more detailed discussion, see (for instance): Deaton, A. and S. Zaidi (2002) Guidelines for Constructing Consumption Aggregates for Welfare Analysis, LSMS Working Paper 135, World Bank, Wash- ington DC. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 31 Table 2.1. Comparing Various Alternate Welfare Measures in the LiTS Expenditures One-question Household Per Equivalent Per Capita Ratio of Per Equivalent Adult Country Size Adult (ECE) (PCE) ECE/PCE Expenditures (OQE) Czech Republic 2.1 4,556 3,673 1.24 7,448 Hungary 2.1 3,455 2,774 1.25 7,516 Latvia 2.1 3,870 3,099 1.25 7,465 Lithuania 2.1 3,234 2,575 1.26 5,831 Bulgaria 2.2 2,125 1,672 1.27 4,017 Romania 2.4 2,372 1,865 1.27 4,652 Serbia 2.6 2,752 2,163 1.27 4,867 Belarus 2.1 2,203 1,728 1.27 2,847 Ukraine 2.2 2,687 2,094 1.28 3,735 Estonia 2.1 4,045 3,145 1.29 7,409 Croatia 2.4 5,260 4,052 1.30 8,098 Poland 2.4 3,642 2,803 1.30 6,132 Russia 2.2 3,108 2,372 1.31 4,531 Slovenia 2.5 6,531 4,980 1.31 7,774 Slovakia 2.5 3,577 2,708 1.32 5,901 Bosnia 2.7 2,860 2,143 1.33 4,423 Montenegro 2.9 4,173 3,095 1.35 5,821 Moldova 2.4 1,302 958 1.36 1,860 Georgia 2.9 1,315 947 1.39 1,946 Kazakhstan 3 1,951 1,386 1.41 2,800 Macedonia, FYR 3.3 2,308 1,633 1.41 3,012 Armenia 3.6 1,604 1,092 1.47 2,486 Turkey 3.6 2,784 1,891 1.47 4,683 Kyrgyz Republic 3.7 1,060 704 1.51 915 Albania 3.7 2,459 1,577 1.56 3,102 Azerbaijan 4 1,136 708 1.60 1,085 Uzbekistan 4.3 721 440 1.64 993 Tajikistan 4.8 787 466 1.69 787 Overall 2.6 2,812 2,120 1.33 4,390 tails. Thus, one problem with this measure is that it does not result in self-rankings that are uniformly distributed across decile groups (in particular, respondents are more-likely-than- average to list themselves as belonging to the 3rd, 4th, or 5th decile groups) (see Table 2.2 and Figure 2.3).20 Three groups of "poor," "middle," and "rich" were formed by group- ing together those respondents that placed themselves in (a) the first three, (b) fourth and 20. Given that the LiTS interviewed a nationally representative sample of respondents from each country, one would have expected their responses to have been uniformly distributed across the ten response classes, assuming respondents could indeed make an accurate comparison of their own welfare in relation to that of other people in the same country. Figure 2.1. Distribution of Normalized Expenditures Albania Belarus Bosnia Bulgaria Croatia Czech Republic .8 .6 .4 .2 0 Macedonia Hungary Moldova Montenegro Poland Romania .8 .6 .4 .2 0 Serbia Slovakia Slovenia Turkey Ukraine Armenia .8 .6 .4 .2 Density 0 Azerbaijan Estonia Georgia Kazakhstan Kyrgyz Republic Latvia .8 .6 .4 .2 0 4 6 8 10 4 6 8 10 Lithuania Russia Tajikistan Uzbekistan .8 .6 .4 .2 0 4 6 8 10 4 6 8 10 4 6 8 10 4 6 8 10 Log of ECE Graphs by country Source: 2006 LiTS. Graphs show distribution of (log) normalized expenditures by country, with the appropriately-scaled normal distribution (same mean and standard deviation as the data) overlaid. For each country, the distribution has been censored at the 1% and 99% level. Figure 2.2. Distribution of the One-question Welfare Aggregate Albania Belarus Bosnia Bulgaria Croatia Czech Republic 1.5 1 .5 0 Macedonia Hungary Moldova Montenegro Poland Romania 1.5 1 .5 0 Serbia Slovakia Slovenia Turkey Ukraine Armenia 1.5 1 .5 Density 0 Azerbaijan Estonia Georgia Kazakhstan Kyrgyz Republic Latvia 1.5 1 .5 0 5 10 5 10 Lithuania Russia Tajikistan Uzbekistan 1.5 1 .5 0 5 10 5 10 5 10 5 10 Log of OQE Graphs by country Source: 2006 LiTS. Graphs show distribution of (log) normalized one-question welfare aggregates by country, with the appropriately scaled normal distribution (i.e. same mean and standard deviation as the data) overlaid. For each country, the distribution has been censored at the 1% and 99% level. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 33 Table 2.2. Subjective Assessment of Welfare Percentage of Respondents Ranking Themselves in the Decile Group Country Poorest 2 3 4 5 6 7 8 9 Richest Overall Albania 8.9 7.6 12.7 18.0 31.0 10.3 5.9 3.4 1.0 1.1 100 Belarus 1.8 2.1 13.1 18.1 27.6 19.5 10.7 5.8 1.3 0.1 100 Bosnia 7.3 10.2 14.4 15.8 25.9 12.6 9.6 2.6 0.9 0.6 100 Bulgaria 13.0 15.8 19.9 15.1 17.2 7.8 3.6 4.3 2.9 0.3 100 Croatia 7.1 8.0 16.0 15.0 34.4 10.9 6.6 1.5 0.4 0.2 100 Czech Republic 1.9 8.3 12.8 22.1 25.8 13.1 7.9 5.7 1.8 0.5 100 Macedonia, FYR 8.1 9.5 16.1 16.9 34.1 8.5 4.2 1.9 0.6 0.1 100 Hungary 6.2 9.8 18.6 20.1 24.2 11.8 6.1 2.8 0.3 0.1 100 Moldova 6.3 9.6 12.5 15.9 25.1 14.2 10.5 4.7 1.3 0.0 100 Montenegro 7.3 10.3 19.4 14.8 26.6 10.8 7.2 2.2 1.0 0.3 100 Poland 6.3 9.8 17.2 18.6 21.3 11.2 7.3 5.7 1.9 0.7 100 Romania 5.7 9.1 13.3 18.8 28.6 14.2 6.1 3.0 0.9 0.3 100 Serbia 8.7 13.4 18.0 17.9 28.2 8.6 3.7 0.6 0.4 0.5 100 Slovakia 5.3 8.4 17.3 18.8 32.0 11.9 4.7 1.3 0.2 0.1 100 Slovenia 1.4 3.3 7.9 16.1 37.0 19.4 9.3 4.5 1.0 0.2 100 Turkey 15.3 16.8 16.8 14.5 20.2 7.1 5.1 2.6 1.2 0.2 100 Ukraine 8.6 12.8 22.7 21.3 19.3 8.9 4.1 1.7 0.4 0.2 100 Armenia 4.9 6.8 17.2 18.7 32.7 11.8 5.8 1.1 0.5 0.5 100 Azerbaijan 11.8 21.3 25.7 17.4 17.1 4.9 1.3 0.3 0.1 0.0 100 Estonia 3.4 7.3 21.7 21.9 31.6 8.6 4.3 0.8 0.1 0.4 100 Georgia 11.4 12.8 23.0 18.6 20.3 8.5 3.7 1.0 0.4 0.3 100 Kazakhstan 3.9 9.1 17.0 17.5 28.3 12.0 7.1 4.2 0.9 0.0 100 Kyrgyz Republic 2.9 3.5 11.0 15.8 33.2 19.7 8.0 3.7 1.7 0.6 100 Latvia 5.6 10.3 22.2 23.2 26.5 8.4 2.9 0.6 0.0 0.2 100 Lithuania 9.1 12.4 19.5 21.4 26.1 8.1 2.4 0.8 0.1 0.0 100 Russia 6.0 13.9 22.4 21.3 17.0 8.3 6.1 3.8 1.0 0.2 100 Tajikistan 1.5 7.0 14.5 21.5 31.8 13.3 5.9 4.0 0.3 0.2 100 Uzbekistan 5.1 6.3 11.1 16.6 39.6 15.4 4.3 1.1 0.2 0.3 100 All countries: 6.6 9.9 17.0 18.3 27.3 11.4 5.9 2.7 0.8 0.3 100 Source: 2006 LiTS (data are not weighted). fifth and (c) higher decile groups (comprising respectively 33, 45, and 21 percent of the population). Ownership of Assets The LiTS also collected data on household ownership of various consumer goods, such as cars, mobile phones, computers, internet access, etc (Table 2.3). In general, average ownership rates of these consumer durables are quite high, though there is a fair degree 34 World Bank Working Paper Figure 2.3. Distribution of Subjective Welfare Rankings by Country Albania Belarus Bosnia Bulgaria Croatia Czech Republic .4 .3 .2 .1 0 Macedonia Hungary Moldova Montenegro Poland Romania .4 .3 .2 .1 0 Serbia Slovakia Slovenia Turkey Ukraine Armenia .4 .3 .2 .1 Density 0 Azerbaijan Estonia Georgia Kazakhstan Kyrgyz Republic Latvia .4 .3 .2 .1 0 0 5 10 0 5 10 Lithuania Russia Tajikistan Uzbekistan .4 .3 .2 .1 0 0 5 10 0 5 10 0 5 10 0 5 10 Subjective Welfare Ranking Graphs by country Source: 2006 LiTS. X-axis denotes welfare ranking from 1 (poorest) to 10 (richest) of variation between countries--for instance, ownership of mobile phones varies from 15 percent in Uzbekistan to 90 percent in Montenegro. Overall, 64 percent of house- holds report owning mobile phones, 32 percent have cars, 28 percent own computers, while 16 percent have access to internet at home. Overall, one-in-ten respondents report owning a secondary residence. In general, asset ownership rates are quite high in wealthier countries (Croatia, Slovenia, Czech Republic), and lower in comparatively poorer Central Asia (Uzbekistan, Tajikistan, and the Kyrgyz Republic). About 30 per- cent of respondents said their household did not own any of the above-mentioned assets, with the ratio varying from a low of around 8 percent in the Czech Republic to over 70 percent in Tajikistan. Comparing the Various Welfare Measures To what extent are these various LiTS welfare measures correlated with one another? Table 2.4 provides the correlation matrix for these variables. As one would expect, corre- lation between ECE and PCE is the highest among six possible pair-wise comparisons, given these two are based on essentially the same set of variables (albeit with different weights). In general, ECE/PCE has the highest correlation with other welfare measures, while SAW has the lowest, due in part to the tendency of respondents to rank their house- hold in the middle of the income distribution ladder. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 35 Table 2.3. Ownership of Assets Respondents (percent) reporting owning the consumer good Mobile Internet Secondary Country phone Car Computer access residence Albania 88 24 10 3 7 Belarus 56 30 34 17 10 Bosnia 68 51 29 14 18 Bulgaria 59 38 19 14 9 Croatia 75 60 42 34 30 Czech Republic 89 56 51 38 19 Macedonia, FYR 69 49 28 10 8 Hungary 73 46 37 21 6 Moldova 31 24 12 7 3 Montenegro 90 58 32 22 19 Poland 69 48 45 28 7 Romania 61 33 31 19 11 Serbia 73 46 37 27 24 Slovakia 81 58 47 24 11 Slovenia 88 78 65 52 16 Turkey 73 22 15 8 8 Ukraine 67 26 27 11 8 Armenia 46 23 10 5 6 Azerbaijan 52 17 5 2 4 Estonia 82 46 47 41 15 Georgia 46 20 8 4 15 Kazakhstan 41 29 13 5 3 Kyrgyz Republic 21 24 5 1 9 Latvia 76 38 34 24 14 Lithuania 75 49 38 26 11 Russia 69 31 33 20 12 Tajikistan 17 18 4 1 4 Uzbekistan 15 20 2 1 6 All countries 64 32 28 16 10 Table 2.4. Correlation Matrices: Decile Rankings Based on Various LiTS Welfare Measures Variables ECE PCE OQE SAW ECE 1.0000 -- -- -- PCE 0.8707 1.0000 -- -- OQE 0.4530 0.4560 1.0000 -- SAW 0.2500 0.2148 0.1431 1.0000 36 World Bank Working Paper How good are these various welfare metrics in identifying the poor? One possible criterion to ascertain this is to first rank respondents into different groups--for instance, three equal-sized welfare groups within each country (the poor, middle class, and rich)--based on the distribution of the respective measures. In the case of SAW, this is done by grouping together those respondents that placed themselves in (a) the first five, (b) sixth and seventh and (c) higher decile groups (so as to form roughly three equal groups comprising respectively 33, 46, and 20 percent of the population). We then compare asset-ownership rates by income group across different welfare metrics-- to the extent that asset ownership is correlated with welfare status, we would expect a lower rate of ownership among the "poor" compared to the "rich." While around 30 percent of "rich households" (as classified by ECE) own a mobile phone, only 5 per- cent of "poor households" (again, as classified by ECE) reporting owning one. In gen- eral, all four welfare metrics do quite well in discerning the poor from the rich; however, ECE consistently does the best job in the sense of giving the sharpest gradient in asset- ownership rates across the various poor-middle-rich welfare classes (Table 2.5). This can also be seen by focusing on the 29­30 percent of respondents that report not hav- ing a mobile phone, car, computer, internet access, or a secondary residence--these respondents own none of the various asset variables on which questions were asked in the LiTS. While the so-classified "asset poor" are considerably more likely to be among the poor than the rich as classified by all other welfare metrics under consideration, the odds-ratio of probability of the asset poor being classified as poor to the probability of their being classified as rich is greater than five in the case of the ECE--higher than any other measure (Figure 2.4). How Good is the LiTS Welfare Metric? The above analysis suggests that ECE is the preferred welfare measure from LiTS, but also confirms that ECE and PCE are in fact quite highly correlated. In this section, we compare PCE from the LiTS with per capita consumption expenditures obtained from other more traditional sources, such as the National Accounts and other household surveys. Average PCE in the LiTS sample is US$ 2,049 per annum, but varies considerably across countries from $440 in Uzbekistan to $4,980 in Slovenia (Table 2.6). Across the entire LiTS sample, average PCE is about 72 percent of the region's estimated per-capita 2006 GDP. Across the 15 countries for which survey data are available in the World Bank's Europe and Central Asia Region household survey archive (henceforth ECAPOV), average PCE in the LiTS ($1,603) is also quite close to the ECAPOV average ($1,814), though this close conformance overall hides a fair amount of variation at the national level.21 As Table 2.6 shows, in 7 out of 28 countries presented (Montenegro, Moldova, Kyrgyz Republic, Ukraine, Tajikistan, Serbia, and Bosnia), the LiTS PCE is greater 21. In five cases, the difference in the LiTS and ECAPOV PCE measure is greater than 25 percent--in Armenia and Georgia, mean PCE from ECAPOV is lower than the LiTS, but higher in Belarus, Estonia, and Azerbaijan. In 3 cases--namely Armenia, Belarus, Georgia--the LiTS measure accords much better to per-capita GDP estimates from the National Accounts, but not so in the remaining two--i.e. Estonia and Azerbaijan. Satisfaction Table 2.5. Asset Ownership Rates by Welfare Level Using Alternate Ranking Criteria (1) ECE (2) PCE (3) OQE (4) SAW % of persons in with households with Poor Middle Rich Poor Middle Rich Poor Middle Rich Poor Middle Rich All Groups Internet access Life EU members 10 26 49 14 27 44 17 27 42 12 31 49 28 and S.E. Europe 8 18 32 9 20 29 12 18 30 9 20 36 19 Service CIS-low income 0 1 7 0 1 7 2 2 6 1 3 6 3 CIS-middle 2 11 28 3 11 27 3 11 27 6 14 24 14 Delivery Turkey 2 6 16 2 7 16 0 8 16 2 10 22 8 Overall 5 15 30 7 16 28 9 15 26 7 18 30 17 in Computers Eastern EU members 18 41 66 26 40 59 29 40 57 21 46 64 41 S.E. Europe 15 31 49 17 32 45 22 29 44 17 34 53 32 Europe CIS-low income 1 3 14 1 3 14 3 4 12 3 6 12 6 CIS-middle 6 27 49 9 27 47 10 27 46 15 28 45 28 and Turkey 5 13 27 5 14 26 4 16 26 7 18 34 15 the Overall 11 26 45 14 26 42 17 26 40 14 29 44 27 Former Secondary residence EU members 6 10 18 7 10 18 9 11 16 6 12 20 12 Soviet S.E. Europe 12 18 26 13 18 25 15 18 23 14 18 29 19 (continued) Union 37 38 World Table 2.5. Asset Ownership Rates by Welfare Level Using Alternate Ranking Criteria (Continued) Bank (1) ECE (2) PCE (3) OQE (4) SAW Working % of persons in households with Poor Middle Rich Poor Middle Rich Poor Middle Rich Poor Middle Rich All Groups CIS-low income 4 6 11 4 6 11 5 6 10 4 7 12 7 Paper CIS-middle 3 8 13 4 8 13 4 8 12 4 8 14 8 Turkey 2 7 15 3 7 15 4 7 14 4 9 18 8 Overall 6 11 17 7 10 17 8 11 16 7 11 19 11 Cars EU members 28 52 73 36 51 66 37 52 66 31 57 69 51 S.E. Europe 36 56 68 42 54 64 45 52 64 36 58 71 53 CIS-low income 13 22 35 14 23 33 19 23 29 12 25 36 23 CIS-middle 16 32 47 22 29 43 19 30 45 18 33 48 31 Turkey 12 20 38 13 23 34 12 24 35 12 29 45 23 Overall 23 41 57 28 40 52 30 40 51 24 45 57 40 Mobile phone EU members 51 78 93 59 74 88 61 76 87 58 80 85 74 S.E. Europe 64 82 91 70 80 88 71 79 89 66 84 90 79 CIS-low income 15 31 55 18 32 52 22 31 49 28 36 39 34 CIS-middle 29 65 86 36 63 81 37 61 81 45 65 73 60 Turkey 62 73 86 66 74 81 63 77 82 68 78 83 74 Overall 42 65 81 47 63 77 50 63 76 51 67 72 63 Note: Based on un-weighted observations. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 39 Figure 2.4. Comparing the Various Alternate Welfare Measures Percent of Group that is "Asset-Poor" 60 "Poor" "Rich" 50 40 30 20 10 0 ECE PCE OCE SAW than the average per capita 2005 GDP estimate. However, this discrepancy between data from the national accounts and household surveys is not unique to the LiTS per se--of the countries for which ECAPOV household survey data is available, the survey-based estimate exceeds mean per-capita GDP in 7 cases (Albania, Belarus, Moldova, Azerbaijan, Kyrgyz Republic, Russia, and Tajikistan) out of 15 countries for which this comparison is possible. National Accounts data indicate that household final consumption expenditures (HFCE) were, on average, about 67 percent of GDP in 2004, the latest year for which these data are available. However, this share varies quite a bit across countries, from a low of 50 percent in the Czech Republic to 93 percent in Serbia and Montenegro. While the LiTS PCE definition does not fully coincide with the HFCE concept, a comparison of the two estimates is interesting in that it shows fairly close conformance between the two (Figure 2.5, Panel A), and in fact no worse than those derived from more detailed ECAPOV household budget surveys (Figure 2.5, Panel B). What about poverty estimates based on the two respective household survey sources? Using the $2.15 and $4.30 PPP poverty lines with the LiTS PCE welfare measure suggests a poverty incidence in ECA in 2006 of about 10.5 percent for the lower poverty line and 33.6 percent for the upper poverty line (Table 2.7). This compares favorably with recent sur- vey based estimates of around 10.8 percent and 37.8 percent poverty rates derived from the ECAPOV household survey database using the same poverty lines.22 To sum, the evidence presented in this section indicates that the LiTS consumption aggregate provides a credible welfare metric with which to paint a profile of variation in living conditions across ECA, which is taken up in the next section. 22. See Alam, Sulla, and Yemtsov (2007), "Income Poverty in Eastern Europe and the former Soviet Union--An Update," The World Bank, Washington, D.C. 40 World Bank Working Paper Table 2.6. 2006 LiTS PCE Compared to Other Data Sources ECA survey archive Country Mean PCE 2006 LiTS Per-capita GDP (2005) Mean PCE Albania 1,577 1,535 1,647 Belarus 1,728 1,868 2,565 Bosnia 2,143 1,486 -- Bulgaria 1,672 2,071 1,987 Croatia 4,052 5,138 4,995 Czech Republic 3,673 6,515 -- Macedonia, FYR 1,633 1,889 -- Hungary 2,774 5,691 -- Moldova 958 429 900 Montenegro 3,095 1,369 -- Poland 2,803 5,194 2,974 Romania 1,865 2,259 -- Serbia 2,163 1,369 -- Slovakia 2,708 4,761 -- Slovenia 4,980 11,382 -- Turkey 1,891 3,390 -- Ukraine 2,094 959 -- Armenia 1,092 1,128 696 Azerbaijan 708 1,182 1,325 Estonia 3,145 5,866 4,071 Georgia 947 971 725 Kazakhstan 1,386 1,972 1,407 Kyrgyz Republic 704 319 394 Latvia 3,099 5,023 -- Lithuania 2,575 4,838 -- Russia 2,372 2,444 2,459 Tajikistan 466 237 554 Uzbekistan 440 673 516 Overall 2,049 2,842 -- (15 country average) (1,603) (2,068) (1,814) Sources: 2006 LiTS, WDI, and various household survey data sets from the ECA survey archive. Mean PCE from the ECA survey archives have been projected to 2006 using country-specific growth rates based on annual changes in household final consumption expenditures (HFCE) from the National Accounts. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 41 Figure 2.5. Country Welfare Rankings: National Accounts vs. Survey-based Estimates Panel A 2006 LITS 6000 Slovenia Croatia 4000 Czech Latvia Estonia Poland estimate(2006) Slovakia Hungary Lithuania PCE Russia Ukraine Bosnia Serbia Romania LITS 2000 Turkey Belarus Bulgaria Albania Macedonia Kazakhstan Moldova Armenia Georgia Kyrgyz Azerbaijan Tajikistan Uzbekistan 0 0 2000 4000 6000 8000 10000 National Accounts HFCE (2004) Panel B ECAPOV estimate 8000 6000 Croatia estimate PCE Estonia 4000 Poland ECAPOV Belarus Russia Bulgaria 2000 Albania AzerbaijanKazakhstan Moldova Georgia Uzbekistan TajikistanArmenia Kyrgyz 0 0 2000 4000 6000 8000 10000 National Accounts HFCE (2004) 42 World Bank Working Paper Table 2.7. Overall Regional Poverty Rates from the 2006 LiTS Squared Poverty Headcount Rate (P0) Poverty Gap (P1) Gap (P2) Poverty Line $PPP 2.15 Urban 6.3 1.7 0.8 Rural 17.4 5.5 2.7 Overall 1.5 3.2 1.5 Poverty Line $PPP4.30 Urban 25.7 8.5 4.1 Rural 46.5 19.2 10.4 Overall 33.6 12.6 6.5 Headcount Rate Poverty Line $PPP 2.15 ECAPOV 10.8 LiTS 10.5 Poverty Line $PPP4.30 ECAPOV 37.8 LiTS 33.6 Poverty Profile for ECA EU Member States and CIS Middle-income Countries Have the Lowest Poverty Rates in the Region Using the $2.15 and $4.30 PPP poverty lines with the LiTS PCE welfare measure,23 we find considerable spatial variation in poverty rates across the ECA region. Overall, the analysis shows 10.5 percent of the region's population to be below the $PPP 2.15 poverty line, while 33.6 percent were below the $PPP 4.30 poverty line. Across different subregions, poverty rates were, as expected, highest among CIS low-income countries and lowest among EU member states and CIS-middle-income countries (Figure 2.6, and Appendix Table A2.1).24 But the Majority of the Poor Live in Middle-income Countries Even though CIS low-income countries have the highest poverty rates, these countries are home to less than one-fifth the total number of poor in the region. Instead, mirroring the overall distribution of population across countries in the region, about two-thirds of the poor in the region in fact live in five middle-income countries--Turkey, Russia, Roma- nia, Poland and Kazakhstan (Figure 2.7) (which also account for about two-thirds of the region's total population). 23. These poverty estimates have been derived using PPP adjustments factors taken from the 2005 ICP Preliminary Results, December 2007 publication. 24. EU member states includes Czech Republic, Hungary, Poland, Slovakia, Slovenia, Estonia, Latvia, Lithuania; CIS middle-income countries include Belarus, Ukraine, Kazakhstan, Russia; South-East Europe includes Albania, Bosnia, FYR of Macedonia, Montenegro, Serbia; CIS low income includes Moldova, Arme- nia, Azerbaijan, Georgia, Kyrgyz Rep., Tajikistan, Uzbekistan (PPP data not available); "Other" includes Bul- garia, Croatia, Romania, Turkey. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 43 Figure 2.6. Regional Variation in Poverty Rates Across the ECA Region 80 70 60 $2.15 line (%) 50 $4.30 line Rate 40 30 Poverty 20 10 0 EU member CIS-middle South-East Other CIS-low income ECA Region states income Europe Sources: Authors' estimates based on the 2006 LiTS. $PPP data is from the 2005 International Comparisons Program Preliminary Results, December 2007. Figure 2.7. Distribution of the Poor Across the ECA Region 40 34 35 34 30 (percent) 25 poor total 20 16 15 regon's of 10 6 5 5 Share 5 0 Turkey Russia Romania Kazakhstan Poland All other countries Note: Poor = $PPP 4.30 line. Sources: Authors' estimates based on the 2006 LiTS. $PPP data is from the 2005 International Comparisons Program Preliminary Results, December 2007. 44 World Bank Working Paper Considerable Rural/Urban Disparities, Especially in the CIS Middle-income Group Roughly one-fifths of the ECA region's population now resides in metropolitan regions, while the remainder of the region's population is split roughly equally (about 40 percent each) between other urban areas and rural areas. In Romania, Armenia, and the Baltic coun- tries, more than half the urban population is in metropolitan areas, while in Russia and Kazakhstan, Serbia and Bosnia, and Poland, only about one-quarter or less reside there. Dis- parities in living conditions between urban and rural areas are generally the highest in CIS- middle-income countries, and lowest in EU member states (Table A2.4).25 Thus, depending on the particular poverty line used, about around 53­63 percent of the poor reside in rural regions, even though these areas account for less than 40 percent of the region's total pop- ulation (Table A2.3). In Most Countries, the Unemployed Have the Highest Risk of Poverty Classifying the population covered in the LiTS into three main groups based upon whether the primary respondent was (a) employed, (b) unemployed, or (c) not working (not in labor force), we find that the incidence of poverty is generally the highest among the unemployed. However, this is not necessarily true across all country groups; for instance, in CIS-middle-income countries, the non-working population face the highest poverty risk (34.7 percent) of all three subgroups, more than twice that among the employed (15.4 percent) (Table 2.8). Table 2.8. Overall Regional Poverty Rates from the 2006 LiTS Group Employed Unemployed Not working Overall EU member states 13.2 29.3 23.4 18.2 South-Eastern Europe 19.9 40.1 32.7 27.3 CIS-low income countries 64.0 75.5 72.9 69.3 CIS-middle-income countries 15.4 26.6 34.7 21.3 Other 44.8 50.2 58.6 51.8 Overall 24.8 43.8 44.9 33.6 But the Working Poor are the Largest Population Group Among the Poor However, among the poor as a group, the largest share of the poor in-fact live in households where the respondent was employed--i.e. among the "working-poor" (Table 2.9). Addi- tional information pertaining to the profile of the poor in the ECA region is provided in Annex Tables A2.1­A2.13. 25. The considerable metropolitan/rural divide in living conditions in CIS-middle countries is cor- roborated by the average satisfaction with life (SWL) score of 3.5 in metropolitan vs. 2.9 in rural areas using LiTS data; by contrast, no major differential in SWL scores is observed in other country groups (Hungary, Serbia, Slovakia, and Georgia show a similar trend as the CIS middle-income country group). Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 45 Table 2.9. Distribution of the Poor by Employment Status of the Respondent Poor (<$PPP4.30) Non-Poor (>$PPP4.30) Not Not Employed Unemployed working Overall Employed Unemployed working Overall EU member 45.4 14.6 40.1 100.0 65.2 7.8 27.0 100.0 states South-East 39.9 19.9 40.2 100.0 59.2 10.8 30.0 100.0 Europe CIS-low 45.4 18.6 36.1 100.0 55.3 13.2 31.5 100.0 income CIS-middle 59.7 7.8 32.6 100.0 78.4 5.2 16.5 100.0 income Other 40.2 8.3 51.5 100.0 53.2 8.9 38.0 100.0 Total 46.7 11.1 42.3 100.0 69.1 7.0 23.9 100.0 Employment, Sources of Income, and Welfare Labor Force Participation Remains Low As the above analysis has shown, labor market status is an important correlate of welfare. Even though average living standards in virtually all countries in the region have recovered and are now higher than their pre-transition levels, labor market conditions remain diffi- cult in many countries (see World Bank 2005a, 2005b). Employment rates--the share of the working-age population this is employed--continue to be very low, in many cases well below the so-called Lisbon target of 70 percent by the year 2010 set by the European Com- mission (EC) its member states. The LiTS data confirm the relatively low level of labor force participation in most countries in the region: overall, only about 56 percent of respondents reported having worked during the 12 month period preceding the survey. These rates were generally highest among CIS middle-income countries and lowest in South Eastern Europe (Figure 2.8). Age, Gender, and Level of Education Are Key Correlates of Work Status Table 2.10 reports the variation in employment rates by different respondent character- istics, and clearly illustrates the important role played by factors like age, gender, educa- tional background, etc. in influencing the likelihood of respondents having worked during the 12 month period preceding the interview. For instance, men are considerably (about 1.5 times) more likely than women to have worked. Similarly, age is an impor- tant correlate of work status, with the data showing a clear inverted-J shaped age- profile. However, by far the most important determinant of work status appears to be the level of education of the respondent: those with higher professional / post-graduate edu- cation level are about five times as likely to have worked as compared to those with no education. 46 World Bank Working Paper Figure 2.8. Respondents that Report Having Worked During Past 12 Months (percent) Russia Belarus Latvia Czech Republic Estonia Slovenia Ukraine Kazakhstan Bulgaria Uzbekistan Mongolia Lithuania Serbia Hungary Slovakia Kyrgyz Republic Croatia Romania Moldova Tajikistan Poland Montenegro Albania Macedonia Bosnia Georgia Turkey Azerbaijan Armenia 0 20 40 60 80 Percent of respondents 15-64 years Table 2.10. Respondents Having Worked in Past 12 Months, By Age, Gender, and Education Proportion of respondents that worked during 12 months preceding interview By Highest educational attainment Compulsory / Higher No degree / secondary / professional / By Age group no education vocational post-graduate Overall 18­30 yrs 14.5 59.0 80.0 63.2 31­40 yrs 34.9 74.8 85.9 75.7 41­50 yrs 26.3 71.1 90.7 74.6 51­60 yrs 28.3 51.4 69.5 54.6 61­70 yrs 7.1 10.5 29.6 12.9 71+ yrs 2.4 3.0 9.4 3.4 All ages 14.5 53.1 75.3 55.6 Men 66.9 Women 45.5 Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 47 Wages and Salaries Are the Primary Source of Income Wages and salaries are the main income source of about one-half of all households in the region, and pensions for about one-fourth of all households (see Table 2.11).26 Wages are rel- atively more important for the CIS middle-income countries, with 62 percent of the respon- dents citing wages as the main income source. Pensions account for the second main income source and are relatively more important for the EU countries where more than one-third of the respondents report it as the main source of income. Not surprisingly, self-employment agricultural and non-agricultural income as well as remittances from friends and relatives, both are relatively more important for the CIS low-income countries. Once pensions are taken out, transfers--through various social assistance programs--are a relatively small part of household incomes. While this broad pattern prevails across the region, inter-country dif- ferences are distinct (see Figure 2.9).27 What is also interesting is that it is only in Turkey where self-employed income has a stronger importance, perhaps reflecting the historical legacy of market enterprise and entre- preneurship in Turkey. But the small size of respondents in other subregions in transition countries citing self-employment income, whether agriculture or non-agriculture, as a pri- mary source of income suggests that efforts to promote entrepreneurship and the growth of new small businesses still has some ways to go. But for the Poorest One-third, Pensions Are the Most Important Source of Income The pattern of income for the poorest one-third is different, with a stronger reliance on pen- sions. The odds-ratio provides a measure of the likelihood of each of the group factors con- tributing to poverty relative to the other group. The use of the bottom one-third, rather than an absolute measure of poverty common across countries, provides a sense of relative poverty in each country. The data clearly suggest that pensioners, agriculture workers, and those on transfers have greater likelihood of being poor. Interestingly, income from friends and family are no more important for the poor than for the overall population (contrast with top one-third or non-poor). The sectoral pattern of employment of the bottom one-third who rely upon wages as the primary source of income suggest a clear divergence with those of the non-poor. The working poor in wage employment are more likely to be found in agriculture and other primary activities, than in the service sector where productivity and wages are likely higher (see World Bank 2008 forthcoming). By contrast, the non-poor are disproportionately 26. Respondents in the LiTS were asked about the various sources of livelihood of their households as well as to report which of these was the most important income source for their household. (1) income from wages in cash, (2) wages in kind, (3) income from self-employment, (4) sales or bartering of farm products, (5) pensions, (6) unemployment benefits, (7) investments, savings, rental of property, (8) state- provided social benefits, (9) community/privately provided social benefits, (10) help from relatives/friends in the country, (11) help from relatives/friends abroad, (12) stipend income, (13) help from charities and NGOs, and (14) other sources. 27. Annex Figure A2 illustrates the country rankings based on relative importance of different house- hold income sources. 48 World Bank Working Paper Figure 2.9. Inter-Country Differences in Main Income Sources of Households Main Source of Income of the Household Russia Belarus Kazakhstan Ukraine Slovakia Estonia Slovenia Latvia Bulgaria Czech Republic Hungary Romania Lithuania Serbia Croatia Bosnia Montenegro Poland Azerbaijan Moldova Mongolia Macedonia Armenia Tajikistan Uzbekistan Turkey Georgia Albania Kyrgyz Republic 0 20 40 60 80 100 Percent of respondents Wages SE non-agri Agriculture Pensions Friends/family Transfers Other Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 49 Table 2.11. Main Income Source by Region Percent of respondents reporting category as main income source: Self-empl. Friends/ Country Group Wages non-agri. Agriculture Pensions family Transfers Other Total (i) All households: EU member states 49.1 8.0 0.9 34.9 2.2 3.5 1.4 100.0 South-Eastern Europe 44.2 11.2 3.7 26.0 8.3 3.1 3.4 100.0 CIS-low income 35.3 19.2 9.1 18.5 10.2 2.1 5.6 100.0 CIS-middle income 62.5 4.8 1.2 26.4 3.4 0.6 1.1 100.0 Other 39.1 15.9 9.2 24.0 4.3 3.3 4.3 100.0 Overall 51.9 9.4 3.7 26.5 4.3 1.9 2.4 100.0 (ii) Poorest one-third: EU member states 31.6 3.1 0.5 54.7 2.2 6.3 1.6 100.0 South-Eastern Europe 27.6 7.2 4.7 39.2 10.1 6.9 4.4 100.0 CIS-low income 25.1 14.5 10.8 32.3 8.3 3.4 5.6 100.0 CIS-middle income 34.1 2.0 2.0 57.1 3.3 0.8 0.8 100.0 Other 22.8 15.7 13.5 32.5 6.5 5.2 3.8 100.0 Overall 30.3 6.3 5.0 48.7 4.5 3.1 2.1 100.0 Odds-ratio (ii) / (i): EU member states 0.64 0.38 0.50 1.57 0.98 1.81 1.18 1.00 South-Eastern Europe 0.62 0.64 1.26 1.51 1.21 2.19 1.29 1.00 CIS-low income 0.71 0.76 1.19 1.74 0.81 1.65 1.00 1.00 CIS-middle income 0.55 0.41 1.72 2.16 0.96 1.34 0.67 1.00 Other 0.58 0.99 1.47 1.35 1.53 1.57 0.89 1.00 Overall 0.58 0.68 1.35 1.84 1.05 1.63 0.89 1.00 employed in the higher productivity growing sectors of the economy such as transport and communications, financial intermediation, and other service sectors. A more detailed analysis of the correlates of poverty using an ordered probit of the like- lihood of being in the bottom one third of per capita expenditures (Table 2.12), suggests the following: Thelikelihoodofbeingrelativelypoorishighestforpensioners,farmers(especially in the CIS middle-income countries), and those dependent on transfers (except for Turkey). Thelikelihoodofrelativepovertyislowerfortheworkingpeople,forthoseinself- employed non-agriculture, the better educated (especially those with higher profes- sional or postgraduate degrees), and those in urban or metropolitan areas. Results of the ordered probit model of welfare status summarized in Table 2.12: 50 World Table 2.12. Probit Model of Likelihood of Being Poor Bank (1 = poorest one-third income group ranked by PCE) Working Overall EU SEE CIS_L CIS_M Other coef sd coef sd coef sd coef sd coef sd coef sd Paper Household size 0.053*** 0.005 0.025* 0.013 0.074*** 0.012 0.043*** 0.008 0.063*** 0.019 0.057*** 0.015 Education level: No degree / 0.126*** 0.038 0.108 0.091 0.178** 0.075 0.026 0.098 -0.169 0.162 0.241*** 0.071 no education Compulsory Reference category school education Secondary -0.180*** 0.025 -0.272*** 0.047 -0.365*** 0.062 -0.043 0.051 -0.229** 0.091 -0.229*** 0.066 education Professional, -0.248*** 0.024 -0.308*** 0.043 -0.215*** 0.054 -0.131** 0.058 -0.341*** 0.088 -0.432*** 0.064 vocational school Higher -0.527*** 0.030 -0.625*** 0.057 -0.480*** 0.072 -0.382*** 0.062 -0.680*** 0.097 -0.661*** 0.088 professional degree Post graduate -0.832*** 0.124 -1.140*** 0.197 -0.082 0.312 -0.455 0.327 -0.811** 0.382 -1.082*** 0.355 degree Age category: 18­30 yrs Reference category 31­40 yrs 0.017 0.027 0.097 0.061 0.040 0.060 -0.105** 0.048 0.042 0.076 0.125 0.078 41­50 yrs 0.111*** 0.027 0.339*** 0.059 0.064 0.059 -0.028 0.048 0.219*** 0.073 0.030 0.078 51­60 yrs 0.177*** 0.028 0.251*** 0.058 0.127** 0.063 0.115** 0.056 0.473*** 0.081 0.067 0.081 61­70 yrs 0.264*** 0.033 0.312*** 0.066 0.110 0.076 0.231*** 0.063 0.563*** 0.096 0.244*** 0.090 Satisfaction 71+ yrs 0.534*** 0.036 0.594*** 0.069 0.426*** 0.089 0.426*** 0.074 0.776*** 0.113 0.561*** 0.095 Worked during -0.154*** 0.020 -0.244*** 0.045 -0.208*** 0.046 -0.093** 0.036 -0.088 0.060 -0.130** 0.057 past 12 months with Main income source: Life Wages Reference category and SE non-agriculture -0.123*** 0.029 -0.371*** 0.078 -0.197*** 0.063 -0.080* 0.048 -0.212** 0.101 0.025 0.084 Agriculture 0.250*** 0.038 -0.176 0.162 0.396*** 0.092 0.165*** 0.057 0.597*** 0.157 0.229** 0.103 Service Pensions 0.565*** 0.026 0.493*** 0.053 0.422*** 0.060 0.609*** 0.053 0.700*** 0.080 0.558*** 0.071 Friends/family 0.070* 0.040 0.306** 0.149 0.052 0.079 -0.048 0.062 0.186 0.148 0.452*** 0.137 Delivery Transfers 0.727*** 0.052 0.756*** 0.099 0.723*** 0.103 0.660*** 0.114 0.549** 0.255 0.713*** 0.123 Other 0.144*** 0.053 0.207 0.146 0.154 0.109 0.065 0.082 0.273 0.242 0.223 0.144 in Place of residence: Eastern Urban areas -0.177*** 0.019 -0.035 0.035 -0.064 0.044 -0.200*** 0.042 -0.536*** 0.050 -0.203*** 0.054 Rural areas Reference category Europe Metropolitan -0.521*** 0.025 -0.319*** 0.045 -0.335*** 0.062 -0.508*** 0.049 -1.305*** 0.103 -0.588*** 0.062 Constant -0.528*** 0.038 -0.528*** 0.078 -0.576*** 0.086 -0.505*** 0.075 -0.299** 0.127 -0.573*** 0.103 and the Former Soviet Union 51 52 World Bank Working Paper Some illustrative estimates from the above model: Simulated type Predicted probability that person is poor: Random respondent: 0.33 Wage employees in capital: EU members 0.17 SEE 0.19 CIS low 0.16 CIS middle 0.03 Other 0.12 Pensioners: EU members 0.42 SEE 0.43 CIS low 0.50 CIS middle 0.49 Other 0.42 Rural Farmers: EU members 0.22 SEE 0.45 CIS low 0.39 CIS middle 0.62 Other 0.39 Respondent 61­70 years old, with no education, did not work in past 12 months, pensions main income source, lives in rural areas 0.73 Respondent 26 years old, with post-graduate degree, worked in past 12 months; wages main income source, lives in metropolitan area 0.04 Public Transfers Have Reach But Are Not an Important Source of Income Given the socialist legacy and the recent real increases in social assistance payments in most countries in ECA, transfers reach out to a significant part of the population in the CIS coun- tries and in the EU member states (see Table 2.13). But they play only a small role in the incomes of the population, including the poor. This suggests that state-provided social ben- efits appear to be largely untargeted transfers, probably because of family allowances which tend generally to be untargeted transfers, and that their levels are inadequate. However, in the relatively richer ECA countries of the EU and in SEE, transfers are still reported as the primary income source by 3­4 percent of the population. Concluding Observations Our analysis above indicates that the welfare measure derived from the LiTS provides a very useful and effective means to measure household welfare and compare both within as well as across countries. Using a per capita adult equivalent measure of household consumption, this paper develops a unique, contemporaneous profile of poverty in the ECA Region in 2006. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 53 Table 2.13. Public/Private Transfers Are More Important in the CIS and EU Member States Percent of respondents receiving transfer payment State provided Unemployment Community/privately Country Group social benefits benefits provided benefits Charities/NGOs CIS-middle 12.7 1.1 0.1 0.3 CIS-low income 12.2 0.6 0.6 0.9 EU member states 9.4 2.9 3.4 0.9 Other 9.6 0.9 0.7 0.3 South-East Europe 4.1 3.3 0.4 0.7 Overall 11.1 1.4 0.8 0.5 Percent of respondents reporting transfers as main income source Country Group Poor Middle Rich Overall South-East Europe 6.9 2.0 0.9 3.1 EU member states 6.3 2.6 1.7 3.5 Other 5.2 2.6 2.3 3.3 CIS-low income 3.4 1.7 1.3 2.1 CIS-middle income 0.8 1.0 0.1 0.6 Overall 3.1 1.7 1.0 1.9 The profile suggests a diverse region, with significant differences among countries in the inci- dence of poverty but a preponderance of the poor in the more populous middle-income countries. Quantitative measures of poverty correspond well with the population's own sub- jective view of relative income status. Not surprisingly, poverty is correlated with low educa- tional attainment, lack of skills as well as self-employment in agriculture. But, of some concern, is the finding that opportunities for the poor to upgrade their human capital and skills, or access finance and economic opportunities, may be limited by a distinct disadvan- tage in asset ownership. Even though income inequality is moderate in the region, this may be accentuated by the asset inequality. Given the analysis above, is there any role for public policy in improving the living stan- dards, especially for the poor? The diagnosis clearly reveals the importance of labor status for improving outcomes with respect to satisfaction with life. The analysis reveals three labor market groups towards which policy should be calibrated. First, are those who are working for wages, especially the poor, but whose wages may be low. In some cases, the low wages reflect the low educational and skill level of the population which may condemn such workers to low productivity employment. Public policy can help in strengthening the quality of public education, especially when it comes to the poor, pro- viding incentives for firms to invest in lifelong learning so as to create opportunities for edu- cational progression in life, and eliminating any barriers that may exist to the easy movement of factors of production towards more productive sectors or for the creation of new startups. Second, are those who are non-participants on account of retirement, and who rely pri- marily on pensions for their income and who are at risk of poverty without these. For pen- sions, it is clear that public policy--though an appropriate combination of public and private 54 World Bank Working Paper financing--needs to provide an adequate level of social insurance against working in old age. After the serious erosion and, in many cases, nonpayment of pensions during the crisis years, real increases in pensions are now targeting to provide a modicum of old age security. Gov- ernments need to ensure that pensions cover all the population at an adequate, fiscally afford- able rate. Third, are those who are unemployed and rely upon public and private benefits to sus- tain themselves. To the extent that employment rates remain low in many countries and job creation has not progressed sufficiently rapidly to absorb the new entrants in the labor force, public policies that can spur job creation, especially in higher value added jobs, is essential. These would include measures to improve the investment climate, invest in human capital, and to promote labor market flexibility. To the extent that some of the unemployed are unemployable, as may be the case because of skill obsolescence and mismatches, adequate unemployment benefits are needed to protect them from poverty and maintain minimum living standards. But public policy needs to be sufficiently discerning to prevent dependency on benefits and discourage labor market participation. Finally, public policy should also address the issue of asset inequality which could lead to an inequality of opportunity.28 For instance, lack of ownership of a car in the absence of adequate and affordable public transport system could limit the poor's ability to access bet- ter paying jobs. Similarly, lack of ownership of housing assets could limit access to finance and therefore the potential opportunities for entrepreneurship and self-employment income. Given the important role that mobile telephones and internet connectivity is playing in today's economy and helping to bring the economic divide, any significant disadvantage for the poor would also limit income growth and perpetuate inter-generational inequalities. Yet, public policy can help by improving the quality of public transport, by supporting programs for promoting housing ownership for low-income families, and ensuring competition in product markets to ensure affordable telephone and internet connectivity. 28. See, for instance, World Bank, World Development Report 2005. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 55 Annex: Tables Table A2.1. Overall Poverty Rates by Region Squared Poverty Headcount Rate(P0) Poverty Gap(P1) Gap(P2) Poverty Line = PPP$2.15 ECA Region 10.5 3.2 1.5 Urban 6.3 1.7 0.8 Rural 17.4 5.5 2.7 EU member states 2.3 0.6 0.3 Urban 2.0 0.5 0.2 Rural 2.6 0.7 0.3 South-Eastern Europe 8.1 3.3 2.2 Urban 6.0 2.9 2.1 Rural 10.6 3.9 2.2 CIS-low income 30.1 10.1 5.0 Urban 21.3 7.6 4.0 Rural 36.8 11.9 5.7 CIS-middle 3.9 1.1 0.5 Urban 2.1 0.5 0.2 Rural 7.4 2.3 1.1 Other 20.2 5.9 2.7 Urban 12.8 3.2 1.2 Rural 32.5 10.4 5.0 Poverty Line = PPP$4.30 ECA Region 33.6 12.6 6.5 Urban 25.7 8.5 4.1 Rural 46.5 19.2 10.4 EU member states 18.2 5.0 2.0 Urban 15.8 4.4 1.7 Rural 22.2 6.1 2.5 South-Eastern Europe 27.1 10.2 5.6 Urban 20.5 7.8 4.5 Rural 34.7 12.9 6.9 CIS-low income 69.2 31.1 17.7 Urban 56.0 23.8 13.3 Rural 79.1 36.6 20.9 CIS-middle 21.3 6.4 2.9 Urban 15.6 4.0 1.6 Rural 32.2 11.0 5.3 Other 52.2 21.4 11.5 Urban 42.7 15.4 7.5 Rural 68.0 31.5 18.1 56 World Bank Working Paper Table A2.2. Sensitivity of Poverty Rates with Respect to Choice of Poverty Line Poverty Incidence(P0) Change from Actual (%) Poverty Line = PPP$2.15 Actual 10.5 0.00 +5% 11.6 10.32 +10% 12.6 19.41 +20% 15.1 42.75 -5% 9.3 -11.41 -10% 8.4 -20.07 -20% 6.1 -42.04 Poverty Line = PPP$4.30 Actual 33.6 0.00 +5% 36.0 7.16 +10% 38.3 13.90 +20% 42.8 27.33 -5% 31.6 -6.09 -10% 29.2 -13.05 -20% 24.2 -27.97 Table A2.3. Distribution of the Poor by Geographic Region Poverty Distribution Distribution of Headcount Rate of the Poor Population Poverty Line = PPP$2.15 ECA Region 10.5 100.0 100.0 Urban 6.3 37.1 61.8 Rural 17.4 62.9 38.2 EU member states 2.3 100.0 100.0 Urban 2.0 56.3 62.0 Rural 2.6 43.7 38.0 South-Eastern Europe 8.1 100.0 100.0 Urban 6.0 39.2 53.5 Rural 10.6 60.8 46.5 CIS-low income 30.1 100.0 100.0 Urban 21.3 30.4 43.0 Rural 36.8 69.6 57.0 CIS-middle 3.9 100.0 100.0 Urban 2.1 35.8 65.9 Rural 7.4 64.2 34.1 (continued) Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 57 Table A2.3. Distribution of the Poor by Geographic Region (Continued ) Poverty Distribution Distribution of Headcount Rate of the Poor Population Other 20.2 100.0 100.0 Urban 12.8 39.5 62.5 Rural 32.5 60.5 37.5 Poverty Line = PPP$4.30 ECA Region 33.6 100.0 100.0 Urban 25.7 47.2 61.8 Rural 46.5 52.8 38.2 EU member states 18.2 100.0 100.0 Urban 15.8 53.6 62.0 Rural 22.2 46.4 38.0 South-Eastern Europe 27.1 100.0 100.0 Urban 20.5 40.4 53.5 Rural 34.7 59.6 46.5 CIS-low income 69.2 100.0 100.0 Urban 56.0 34.8 43.0 Rural 79.1 65.2 57.0 CIS-middle 21.3 100.0 100.0 Urban 15.6 48.4 65.9 Rural 32.2 51.6 34.1 Other 52.2 100.0 100.0 Urban 42.7 51.1 62.5 Rural 68.0 48.9 37.5 58 World Bank Working Paper Table A2.4. Rural Urban Disparities Average PCE (US$) As Ratio of National Average Metrop./ Country Metropolitan Urban Rural Overall Metropolitan Urban Rural Overall Rural Russia 6,093 2,787 2,322 3,108 1.96 0.90 0.75 1.00 2.62 Romania 3,255 1,856 1,449 2,372 1.37 0.78 0.61 1.00 2.25 Armenia 2,292 1,243 1,035 1,604 1.43 0.77 0.65 1.00 2.21 Belarus 2,869 2,568 1,314 2,203 1.30 1.17 0.60 1.00 2.18 Kazakhstan 2,947 2,130 1,365 1,951 1.51 1.09 0.70 1.00 2.16 Bulgaria 3,124 2,074 1,546 2,125 1.47 0.98 0.73 1.00 2.02 Ukraine 4,058 2,613 2,051 2,687 1.51 0.97 0.76 1.00 1.98 Lithuania 4,188 2,896 2,178 3,234 1.29 0.90 0.67 1.00 1.92 Kyrgyz Re 1,649 975 874 1,060 1.56 0.92 0.82 1.00 1.89 Moldova 1,918 1,353 1,025 1,302 1.47 1.04 0.79 1.00 1.87 Tajikistan 1,281 785 703 787 1.63 1.00 0.89 1.00 1.82 Georgia 1,828 1,165 1,014 1,315 1.39 0.89 0.77 1.00 1.80 Azerbaijan 1,549 1,155 887 1,136 1.36 1.02 0.78 1.00 1.75 Serbia 4,001 2,667 2,360 2,752 1.45 0.97 0.86 1.00 1.70 Mongolia 1,466 856 884 1,095 1.34 0.78 0.81 1.00 1.66 Albania 3,175 2,593 2,011 2,459 1.29 1.05 0.82 1.00 1.58 Bosnia 3,700 3,227 2,432 2,860 1.29 1.13 0.85 1.00 1.52 Croatia 6,423 5,602 4,247 5,260 1.22 1.07 0.81 1.00 1.51 Turkey 3,257 2,855 2,189 2,784 1.17 1.03 0.79 1.00 1.49 Poland 5,001 3,418 3,375 3,642 1.37 0.94 0.93 1.00 1.48 Hungary 4,087 3,549 2,800 3,455 1.18 1.03 0.81 1.00 1.46 Czech Rep 5,523 4,419 4,113 4,556 1.21 0.97 0.90 1.00 1.34 Montenegro 5,163 3,950 3,856 4,173 1.24 0.95 0.92 1.00 1.34 Uzbekistan 878 766 674 721 1.22 1.06 0.93 1.00 1.30 Latvia 4,335 3,820 3,408 3,870 1.12 0.99 0.88 1.00 1.27 Slovenia 7,990 6,192 6,304 6,531 1.22 0.95 0.97 1.00 1.27 Slovakia 4,269 3,636 3,376 3,577 1.19 1.02 0.94 1.00 1.26 Estonia 4,719 3,499 3,800 4,045 1.17 0.87 0.94 1.00 1.24 Macedonia, FYR 2,575 2,186 2,321 2,308 1.12 0.95 1.01 1.00 1.11 Overall 4,067 2,774 2,131 2,804 1.45 0.99 0.76 1.00 1.91 Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 59 Table A2.5. Mean Per Capita Expenditures ($PPP per year) Mean per capita Mean per capita expenditure expenditure ECA Region 3,939 CIS-low income 1,990 Urban 4,546 Urban 2,534 Rural 2,955 Rural 1,581 Lowest quintile 1,001 Lowest quintile 944 2 1,958 2 1,913 3 2,940 3 2,892 4 4,427 4 4,304 Highest quintile 9,372 Highest quintile 8,359 EU member states 4,426 CIS-middle 4,892 Urban 4,693 Urban 5,555 Rural 3,991 Rural 3,610 Lowest quintile 1,156 Lowest quintile 1,080 2 1,978 2 1,988 3 2,968 3 2,947 4 4,445 4 4,467 Highest quintile 8,664 Highest quintile 9,823 South-Eastern Europe 3,899 Other 2,726 Urban 4,379 Urban 3,172 Rural 3,346 Rural 1,982 Lowest quintile 952 Lowest quintile 976 2 1,977 2 1,923 3 2,936 3 2,921 4 4,423 4 4,304 Highest quintile 8,597 Highest quintile 8,559 60 World Bank Working Paper Table A2.6. Decomposition of Inequality by Geographic Region GE(0) GE(1) GE(2) Overall inequality Overall ECA Region 31.4 31.5 46.4 EU member states 19.7 20.1 25.3 South-East Europe 24.5 23.4 30.5 CIS-low income 27.9 28.4 40.7 CIS-middle income 27.7 29.3 43.9 Other 29.4 29.6 40.6 Within group inequality Overall ECA Region 31.4 31.5 46.4 Between group inequality Overall ECA Region 4.5 4.1 3.8 Between group inequality as % of overall inequality Overall ECA Region 14.4 12.9 8.2 Table A2.7. Ratios of Selected Expenditure Percentiles in Urban and Rural Areas p10 p25 p50 p75 p90 ECA Region 1.59 1.60 1.50 1.52 1.49 EU member states 1.07 1.15 1.16 1.17 1.20 South-Eastern Europe 1.40 1.39 1.33 1.35 1.28 CIS-low income 1.19 1.34 1.45 1.60 1.74 CIS-middle 1.53 1.44 1.51 1.51 1.53 Other 1.57 1.63 1.66 1.53 1.68 Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 61 Table A2.8. Poverty by Age Groups Poverty Headcount Rate Distribution of the Poor Distribution of Population 0­5 yrs 50.8 10.4 6.9 6­14 45.1 16.1 12.0 15­19 34.0 8.1 8.0 20­24 27.8 6.1 7.4 25­29 29.1 7.0 8.1 30­34 31.6 7.0 7.5 35­39 30.0 6.8 7.6 40­44 26.6 5.9 7.5 45­49 26.2 5.7 7.3 50­54 27.3 5.3 6.6 55­59 27.8 5.1 6.2 60­64 31.6 3.9 4.2 65+ yrs 38.9 12.3 10.6 Overall 33.6 100.0 100.0 Table A2.9. Poverty by Whether Respondent Worked or Not During Past 12 Months Poverty Distribution Distribution Headcount Rate of the Poor of Population Poverty Line = $PPP4.30 Yes 25.7 44.2 57.9 No 44.5 55.8 42.1 Overall 33.6 100.0 100.0 Table A2.10. Poverty by Education Level of Household Head Poverty Distribution Distribution Poverty Line = $PPP4.30 Headcount Rate of the Poor of Population Highest Educational Attainment No degree/no education 69.5 14.1 6.8 Compulsory school education 52.1 29.0 18.7 Secondary 35.6 26.2 24.8 Professional/vocational school 24.0 21.6 30.2 Higher professional degree 16.4 9.1 18.6 Post-graduate degree 3.7 0.1 0.9 Overall 33.6 100.0 100.0 62 World Bank Working Paper Table A2.11. Poverty by Household Head's Gender Poverty Distribution Distribution Headcount Rate of the Poor of Population Poverty Line = $PPP4.30 Male 35.5 77.2 73.1 Female 28.4 22.8 26.9 Overall 33.6 100.0 100.0 Table A2.12. Poverty by Demographic Composition Poverty Distribution Distribution Poverty Line = $PPP4.30 Headcount Rate of the Poor of Population Number of children 0­6 years old no children 27.3 59.2 72.9 1 43.2 23.6 18.4 2 57.2 10.8 6.3 3 or more children 90.8 6.5 2.4 Household size 1 18.0 5.9 11.0 2 22.0 14.8 22.7 3 24.2 16.6 23.0 4 31.6 21.1 22.4 5 55.0 16.7 10.2 6 71.7 10.5 4.9 7 or more 82.9 14.4 5.9 Overall 33.6 100.0 100.0 Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 63 Table A2.13. Consumption Regressions Urban Rural coef se coef se Household characteristics Log of household size -0.255*** 0.03 -0.321*** 0.04 Log of household size squared -0.066*** 0.02 -0.026 0.02 Share of children 0­6 (dropped) (dropped) Share of children 7­16 0.388*** 0.05 0.334*** 0.06 Share of male adults 0.698*** 0.05 0.575*** 0.06 Share of female adults 0.658*** 0.05 0.564*** 0.06 Share of Elderly (=60) 0.416*** 0.06 0.320*** 0.06 Country Group EU member states (dropped) (dropped) South-Eastern Europe 0.028* 0.02 -0.027 0.02 CIS-low income -0.626*** 0.02 -0.787*** 0.02 CIS-middle -0.110*** 0.02 -0.456*** 0.02 Other -0.115*** 0.02 -0.342*** 0.02 Characteristics of household head Log of household head's age -0.314*** 0.02 -0.239*** 0.03 Gender Male (dropped) (dropped) Female -0.069*** 0.01 -0.070*** 0.02 Education of the household head No degree/no education (dropped) (dropped) Compulsory school education 0.098*** 0.03 0.076*** 0.02 Secondary 0.344*** 0.03 0.263*** 0.03 Professional/vocational school 0.346*** 0.03 0.282*** 0.03 Higher professional degree 0.570*** 0.03 0.488*** 0.03 Post-graduate degree 0.672*** 0.05 0.707*** 0.11 Employment status of the household head Yes (dropped) (dropped) No -0.170*** 0.01 -0.116*** 0.01 _cons 8.990*** 0.08 8.801*** 0.11 Number of observations 15,633 11,276 Adjusted R2 0.325 0.355 Note: 0.01--***; 0.05--**; 0.1--*; CHAPTER 3 Satisfaction with Publicly provided Health Services in Eastern Europe and the Former Soviet Union29 I n this paper we explore citizens' satisfaction with publicly provided health services (PPHS) in the Eastern Europe and Central Asia region. In particular, we focus our analysis on three inter-linked questions: (a) why are some people more likely than others to use PPHS; (b) what are some of the key influences on users' satisfaction with quality and efficiency of medical treatment received; and (c) how does the prevalence of informal payments impact people's decision on using PPHS, and upon use, the level of satisfaction with services received? The 2006 LiTS provides us with a unique opportunity to pose these questions. In addi- tion to eliciting respondents' perceptions of the quality and efficiency of PPHS, the survey explores priorities for public policies and investment, attitudes to a market economy and democracy, as well as living standards (including expenditure aggregates and sources of livelihood), and demographic characteristics. Thus we are able to assess the impact of `objective' variables such as gender, expenditure, age and education level, as well as `sub- jective' variables such as the level of trust in government and police, satisfaction with life, and perception of corruption, on satisfaction with PPHS. Countries in the Europe and Central Asia region inherited a health care system, called the "Semashko" model after the Soviet statesman who envisaged it, that was state owned and financed through general taxation. Although the system delivered quick improve- ments in population health when first implemented, the absence of market forces (or alternate mechanisms) to ensure provider and administrator accountability for health out- comes led to hospital-dominated networks that adapted very slowly, if at all, to changing disease patterns and medical innovations. The fall of the Soviet Union in 1991 and the 29. Ramya Sundaram and Salman Zaidi. 65 66 World Bank Working Paper resulting macroeconomic contraction in this region led to severe shortfalls in health sector budgets. Countries began to undertake health sector reforms as they moved along the tran- sition path, partly necessitated by the budget cuts, and partly in keeping with the other fun- damental social and economic changes occurring in the region. Strong economic growth in recent years, the fruition of early reform efforts and the continuation of further reforms have resulted in a health sector that is extremely dynamic and diverse across the region, featur- ing a pluralism of health systems. Against this backdrop of transition and reforms, how sat- isfied are the citizens of ECA with publicly provided health services in 2006? Overall, we find that 45 percent of all respondents report themselves as being satisfied with the services they received from the publicly provided health system. Reported satis- faction rates vary considerably by country, with the "net satisfaction rate"--the difference between the share of satisfied and dissatisfied respondents--generally a little higher among EU member states than among middle-income CIS countries (Figure 3.1). A first glance at country ranking suggests that a combination of (a) economic growth, and (b) health sector reform, have an impact on satisfaction. Most EU member states in this region have enjoyed relative political stability through the transition, and, with the exception of Romania, began recovering from recessions in the second half of the 1990s. They have also successfully instituted radical reform of their health sectors. With the exception of Romania, EU member states are at or near the top with regards to satisfaction with PPHS. Figure 3.1. Rates of Satisfaction with the Publicly-provided Health System, By Country Slovenia Czech Republic Croatia Georgia Slovakia Armenia Lithuania Estonia Latvia Bulgaria Hungary Azerbaijan Bosnia Poland Belarus Turkey Mongolia Montenegro Macedonia Uzbekistan Russia Serbia Kyrgyz Republic Kazakhstan Romania Moldova Tajikistan Ukraine Albania 0 20 40 60 80 100 Percent of respondents Very dissatisfied Dissatisfied Neither Satisfied Very satisfied Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 67 Georgia and Armenia, two low-income CIS countries, also do very well, with over 60 percent of respondents being satisfied with the care received. Armenia has seen broad based economic growth, with per capita GDP increasing in double digits in the recent past. In addition, both Georgia and Armenia have instituted effective reforms in provider payment mechanisms (Bonilla-Chacin, Murrugarra, and Temourov 2005) and have experienced a drop in informal payments (see the discussion on informal payments in a later section of this chapter). In contrast, the Kyrgyz republic, another low-income CIS country which has insti- tuted far reaching health sector reform, but which has experienced tepid growth, ranks low in satisfaction. The scenario is reversed in Azerbaijan, which has experienced spectacular, but narrow based economic growth, and where the health sector reform has not been substan- tial. Azerbaijan stands at the middle of the rankings according to the "net satisfaction rate." The middle-income CIS countries are clustered at the bottom of the satisfaction rank- ings, despite spending more per capita on health services than the low-income CIS countries. Health sector reform in these countries has been incremental and piecemeal with medical practices continuing to deviate considerably from internationally established evidence-based medicine. What are the individual factors that influence satisfaction with publicly provided health services? To determine this, we employ an estimation strategy that accounts for selection bias--it takes into account the fact that responses to satisfaction with PPHS are only obtained from the subsample of respondents that chose to access the service during the 12 months preceding the survey. Therefore, we jointly estimate factors that influence whether a respondent chooses to use PPHS, and upon use, his/her satisfaction with services received. Finally, we compare our results with a health utilization survey that was conducted in 2001 in 8 countries (Balabanova and others 2004). One of the key findings of this analysis relates to the perception that unofficial pay- ments are needed to obtain services. It is well documented that, with the decrease in pub- lic funding of the health care system, informal payments have emerged as a fundamental aspect of health care financing in many ECA countries (Lewis 2000; Balabanova 2007; Allin, Davaki, and Mossialos 2006; and several others). The study of the effects of informal payments is complicated by the existence, particularly in the CIS countries, of strong tra- ditions of innocent gift-giving as an expression of gratitude, usually after service is pro- vided. The LiTS does not separate informal payments from gifts in its questionnaire; thus the results associated with informal payments that are presented in this paper should be interpreted with some caution in light of this omission. We find that having to pay for essentially "free" services has a significant negative influence on satisfaction with PPHS. Our empirical model indicates that, other things being equal, respondents who say unofficial payments are often necessary are about 1.4 times more likely to report being dissatisfied with service delivery in the publicly provided health system compared to those who say that such payments are never needed. A com- parison of the 2006 ECA LiTS survey to the 2001 survey referred to in Balabanova and others 2004 (for details, see a later section of this chapter) suggests that there has been a dramatic fall in the prevalence of informal payments in both Georgia and Armenia between 2001 and 2006, while there is an increase in Ukraine and Russia. This could go some distance in explaining why satisfaction rates are high in Georgia and Armenia, and low in Russia and Ukraine, although the total (public and private) PPP adjusted per capita health spending in 2004 is lower both in Georgia ($137) and Armenia ($211) when 68 World Bank Working Paper compared to Russia ($546) and Ukraine ($382). Similarly, a comparison of the LiTS to Lewis (2000) suggests that the incidence of informal payments have increased in Albania between 1996 and 2006; Albania ranks the lowest among all countries in terms of satis- faction with PPHS. The second key finding of this analysis relates to accessing PPHS. We find that those who have confidence in the government (as proxied by trust in government and the police) are more likely to use PPHS--other things being equal, a respondent that says he/she trusts the government and the police is about 7.5 percent more likely than a person who does not to have used PPHS in the past 12 months. Political scientists have theorized that a positive experience with public services leads to satisfaction and trust in the government (for instance, see Bratton 2007; Bouckaert and Van De Walle 2003). While we find a positive correlation of 0.097 between satisfaction with PPHS and trust in government and police, we additionally find that trust in government and police has a positive and significant effect on accessing PPHS. The results have broad policy implications. Rapid and broad based economic growth accompanied by sensible health sector reform increases user satisfaction with PPHS. A two-pronged approach to health sector reform is suggested, particularly in middle and low- income CIS countries. Firstly, in addition to improving the actual quality of service pro- vided by improving primary care facilities and encouraging the practice of evidence-based medicine, the reform effort should address the complex set of circumstances that underlie the prevalence of informal payments. Providing citizens' with means to hold providers directly accountable for quality of service, and for health outcomes, should remove some of the incentives for these payments. Secondly, a good communication strategy should be a key component of any effective health reform effort. Take the case of a reform effort to improve the quality of a historically poorly provided public service. Simply taking steps to improve the quality of the service is likely to be insuf- ficient: the government must also actively seek to change the poor perception of the service in citizens' minds. For instance, if people have typically experienced poor service when they go to a public hospital, they are less likely to want to go back. They might delay or avoid seeking service until they are very sick. Even a significantly improved public hospital would find it challenging to provide good treatment in the later stages of a disease. Instead, by con- currently trying to improve both the actual quality, as well as the perception of the quality of publicly provided health services, the government might persuade people to seek service early, ensuring better treatment outcomes than otherwise. Effective communications should also be aimed at boosting awareness that the gov- ernment is committed to delivering free, and good quality, basic health services, and that citizens should hold PPHS accountable for such results. A striking example of the need to clearly communicate the government's health policy comes from Georgia. By law, all chil- dren 0­3 are entitled to free publicly provided health services in Georgia. A large number of parents seem apparently unaware of this policy as disclosed by this focus group partic- ipant in Georgia in 2001: "My friend told me that in their polyclinic children receive ser- vices for free . . . I do not know if there is any age limitation. I think this is an initiative of their polyclinic." (Belli, Gotsadze, and Shahriari 2004). By encouraging and providing citi- zens with the means to seek redress if substandard treatment is provided, or if they are asked for unofficial payments, the government can co-opt the patient to be an active player in the reform process. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 69 Communication strategies that simply focus on an anti-corruption message, but that do not provide users with alternate means of holding providers accountable are not likely to succeed. Vian and Burak (2006) find that there is no difference in moral beliefs between people in Albania who intend to make informal payments and those who do not--the main difference is that people who do not intend to make informal payments are more likely to report that they have connections with medical personnel, which may be substi- tuting for informal payments. Well functioning PPHS are perceived as important by most ECA citizens. When asked about their first priority for extra government spending, 40 percent of LiTS respondents chose health care (Figure 3.2). This equals the sum of respondents who identify education and housing (the second and third categories) as the top priority, 27 and 13 percent, respectively. In terms of country groupings, healthcare was picked by the highest share of respondents in the EU member states (44 percent) and the lowest in South-East Europe and Turkey (33 and 29 percent). The remainder of the paper is laid out as follows: Figure 3.2. Priorities for Additional Government the next section provides Spending: 2006 LiTS some background on the evolution of PPHS in ECA First Priority countries; section two de- scribes the LiTS data used in the analysis along with sum- Other mary tables and graphs of Housing key variables. In section three, the empirical strat- Health Pensions egy is described, while the results of the estimation are Environment presented in section four. Section five probes into the Infrastructure Education strong performance of some low-income CIS countries in terms of satisfaction with PFHS, while section seven concludes. Evolution of Publicly Provided Health Services in Eastern Europe and Central Asia Health care delivery in the ECA countries was centrally managed and financed during the era of communism, and the system sought to provide universal care that was free at the point of access. When first implemented, the `Semashko' model led to quick improvements in population health through introduction of practices that prevented the spread of com- municable diseases, and through investment in infrastructure and in training physicians. Until the 1960s, reported measures of life expectancy in this region were comparable to those of Western Europe (Balabanova 2007). In addition to the free, comprehensive health 70 World Bank Working Paper care system, factors such as high levels of education and adequate access to water and san- itation enabled countries to achieve better reported health outcomes than other countries with similar income levels (Bonilla-Chacin, Murrugarra, and Temourov 2005). The absence of market forces started becoming evident however, with health care bud- gets allocated predominantly to hospitals rather than to primary care facilities, and accord- ing to historic patterns and fixed norms of number of physicians and beds rather than on the changing health needs of the population. This led to a hospital dominated network of extensive infrastructure and poorly paid personnel that adapted very slowly to changes in disease patterns and innovations in medical technology. The greater failure of the Soviet era health care system was due to the lack of accountability of providers and administra- tors for quality of care provided and for health outcomes. There were no mechanisms by which sound clinical practices were rewarded or widely disseminated. With the intensifi- cation of the cold war, and the increasing isolation of the Soviet Union, there was very lit- tle interaction between specialists behind the iron curtain and those in the western world. Unable to benefit from the new medical breakthroughs in the rest of the world, health care systems in this region moved further away from evidence based medicine. The lack of direct accountability also gave rise to alternate mechanisms through which patients could influence the quality and outcome of health services. Even during the com- munist era, those with connections--such as the party or military elite--or those willing to provide private monetary incentives, could always ensure superior facilities. Jumping the queue, or expediting service through the use of under-the-table payments, was not uncommon. The large upheavals that accompanied the break-up of the Soviet Union, and the severe economic downturn in the region during the 1990s, led to drastic cutbacks in gov- ernment spending on the health sector, with further deterioration in the quality and dis- tribution of health services. Many countries, particularly among the CIS, experienced decreases in reported measures of life expectancy and increases in infant and child mortality during the 1990s.30 There has been a recovery in economic growth in the region more recently, accompa- nied by varying degrees of health sector reforms in particular economies. Health financing reform has focused on moving from funding health services through general revenues to some form of social insurance system. While the EU member countries have achieved some success in raising money for national health insurance through payroll taxes (with short- falls being met by transfers from general revenue), low-income CIS countries face chal- lenges due to the narrow employment base given the significance of non-cash as well as informal activities in their economy. Thus health financing has become increasingly reliant on out-of-pocket payments in these countries (Bonilla-Chacin, Murrugarra, and Temourov 2005; Lewis 2000). Rationalizing the excess human and physical capacity invested in the health system has been met with limited success, particularly among middle-income CIS countries. For instance, maternal in-patient capacity in Russia has not declined in response to the large 30. Reported life expectancy in Belarus, Kazakhstan, Russia, Ukraine, and Uzbekistan continues to be lower now than in 1990. In addition, in Kazakhstan, both infant and child mortality is currently higher than in 1990. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 71 decline in birth rates, and increasing numbers of pregnant women are hospitalized to pre- serve excess capacity (Danishevski and others 2006). Some low-income CIS countries such as Georgia, Armenia, the Kyrgyz republic and Moldova are further along the reform trajectory than others, likely necessitated by the almost total collapse in health sector funding in these countries. Despite this, Bonilla-Chacin, Murrugarra, and Temourov (2005) confirm that there still are more beds and health facilities in these countries than in the EU. In Georgia an interviewed hospital provider had this to say: "We are too many compared to the workload, and our remuneration depends on how many patients we treat. Thus, doctors have an internal agreement, whereby we `rotate': 1 week one doctor serves all patients, and the following week another one comes in" (Belli, Gotsadze, and Shahriari 2004). Stretching the ever shrinking set of resources across the large networks that exist has led to a reduction in the effectiveness of services (World Bank 2005a). At the same time, the spread of communicable diseases, such as HIV/AIDS and tuberculosis, are beginning to pose even greater challenges to the outdated hospital-dominated networks (World Bank 2005b). Life expectancy is currently falling in the former Soviet Union, with this region being only one of two regions in the world (the other being sub-Saharan Africa) where life expectancy is currently declining (Balabanova 2007). Against this backdrop of changing disease patterns and macroeconomic instability, the LiTS elicits citizen's satisfaction with the publicly provided health services in their country. Utilization Rates, Satisfaction, and Prevalence of Informal Payments The EBRD-World Bank Life in Transition Survey (LiTS), conducted in September 2006, probes the relationship between living standards and satisfaction with life in 28 Eastern Europe and Central Asian countries, and Mongolia. A sample of 1,000 individuals was interviewed in each country, making a total of 29,000 respondents across ECA. The analysis in this paper focuses on the attitudes and values section of the LiTS questionnaire, particularly on responses to three sets of questions, which respondents were asked with reference to eight different public services.31 First, respondents were asked: "In your opinion, how often is it necessary for people like you to have to make informal payments/gifts in these situations." One of the eight situations listed was "Receive medical treatment in the public health system." Respondents could choose among five options: 1: Never, 2: Seldom, 3: Sometimes, 4: Usually, and 5: Always. Respondents were next asked: "During the past 12 months, have you personally used these services?" A follow-up question was addressed to those who had used the service during the previous 12 months: "How satisfied were you with the quality and the effi- ciency of the service/interaction?" Responses to this third question were coded using a progressive five-point scale; 1: Very dissatisfied, 2: Dissatisfied, 3: Indifferent, 4: Satisfied, and 5: Very satisfied. 31. These included (i) the road police, (ii) official documents (e.g., passport, visa, birth or marriage certificate, etc), (iii) police (other than road police) (iv) civil courts (v) public health system (vi) public education (tertiary and vocational), (vii) unemployment benefits, and (viii) social security benefits. 72 World Bank Working Paper Figure 3.3. Utilization of Publicly-provided Health System, By Country Received treatment in the publicly provided health system during past 12 months Albania Kazakhstan Hungary Croatia Czech Republic Turkey Azerbaijan Latvia Tajikistan Ukraine Lithuania Macedonia Russia Estonia Romania Moldova Serbia Bulgaria Georgia Montenegro Uzbekistan Bosnia Poland Kyrgyz Republic Armenia Slovenia Mongolia Belarus Slovakia 0 20 40 60 80 Percent of respondents Utilization of the Public Services in ECA Countries The number of respondents reporting having had an interaction with different services in the last 12 months varies quite considerably by type of service. Of the various services cov- ered, the largest number of affirmative responses is for health, with slightly more than half of those surveyed having had an interaction with PPHS during the 12 month period pre- ceding the date of the interview. The response rate for other services is much smaller32: the service with the second largest response rate is "interacting with the authorities granting official documents (e.g., passport, visa, birth or marriage certificate, etc.)," where a little more than one fifth of those surveyed responded "yes." Utilization rates of PPHS varied quite considerably by country, from a high of around two-thirds of all respondents in Albania to one-third only in Slovakia33 (Figure 3.3). While Slovenia and Slovakia have some private provision of health, the lower access rates for 32. The scope of enquiry into the public education system was restricted to tertiary and vocational education--only 12% of those surveyed reported an interaction with the vocational and tertiary public education system in ECA. 33. It is important to note that the LiTS survey did not enquire in detail about health seeking behav- ior. Therefore we do not have information on the health status of those we accessed publicly provided health services, or details about those who needed to use the system, but did not do so for various reasons. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 73 countries such as Belarus, Armenia and the Kyrgyz Republic probably reflect the barriers to access that out of pocket payments represent. Satisfaction with Publicly-provided Health Services Compared to Other Services and Across Countries Of all services covered in the LiTS, the public education and publicly provided health sys- tems received the most favorable ratings (Figure 3.4), with 49 percent and 45 percent of all respondents reporting themselves as being satisfied with the education and medical treat- ment they received (24 and 20 percent respectively reported being dissatisfied with services received). Figure 3.4. Rates of Satisfaction, By Type of Service Public education Publiclly provided health system Request for documents Social security benef Road police Courts Unemployment benefits Other police 0 10 20 30 40 50 60 70 80 90 100 Very unsatisfied Unsatisfied Indifferent Satisfied Very satisfied Reported satisfaction rates with publicly provided health services vary quite consider- ably by country. The EU member countries typically have higher "net satisfaction rates"-- the difference between the share of satisfied and dissatisfied respondents--when compared to the middle-income CIS countries (Ukraine, Kazakhstan, Russia). The low-income CIS countries (Georgia, Armenia, Azerbaijan) also perform a little better than the middle- income CIS countries. Albania has the lowest net satisfaction rate. Perceptions Regarding Prevalence of Unofficial Payments/Gifts Turning to the third key variable of interest: when LiTS respondents were asked how often it is necessary for people to have to make unofficial payments/gifts when using public ser- vices, a large majority said that such payments are never needed--however for publicly provided health services, the share reporting such payments to be usually/always needed was notably higher than for other services (Figure 3.5). There is considerable variation at the country level, from a low of less than 10 percent in Estonia, Slovenia and Georgia to around 48 percent in Albania (Figure 3.6). 74 World Bank Working Paper Figure 3.5. Percent of Respondents that Think that Unofficial Payments Are Needed Publiclly provided health system Road police Public education Request for documents Police Courts Social security benef Unemployment benefits 0 10 20 30 40 50 60 70 80 90 100 Never needed Seldom Sometimes Usually Always needed Figure 3.6. Perceptions Regarding Unofficial Payments in Publicly-provided Health System, By Country Estonia Slovenia Georgia Czech Republic Kazakhstan Turkey Belarus Latvia Croatia Poland Mongolia Serbia Montenegro Armenia Bosnia Slovakia Macedonia Lithuania Bulgaria Russia Kyrgyz Republic Azerbaijan Romania Moldova Hungary Uzbekistan Tajikistan Ukraine Albania 0 10 20 30 40 50 Percent of respondents Always needed Often needed Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 75 Finally, on comparing country rankings of satisfaction with publicly provided health services with prevalence of unofficial payments, the data indicate that prevalence of unofficial payments adversely impacts the level of satisfaction with the service (Figure 3.7)--this apparent link is subjected to more rigorous examination later on in this paper. Figure 3.7. Negative Correlation Between Satisfaction and Prevalence of Informal Payments Satisfaction with Publicly Provided Health System 70 Slovenia Croatia Georgia Armenia Estonia Czech Bulgaria 60 Slovakia Latvia Lithuania Azerbaijan satisfaction on Turkey Macedonia Hungary 50 Montenegro Kyrgyz Mongolia BelarusBosnia Uzbekistan neutral Serbia Romania Poland Moldova above Russia % Albania 40 Kazakhstan Tajikistan Ukraine 30 0 20 40 60 % saying informal payments usually/always required when using publicly provided Estimation Strategy In our estimation, we take into account the fact that responses to satisfaction with PPHS is obtained only from that subsample of individuals who choose to access the service, that is, we correct for the selection bias. While people don't completely control whether they fall ill or not, once ill, they do make some choice about whether to use PPHS. This is con- firmed by many studies that research health utilization behavior, which find that not all individuals who should seek treatment do so (Balabanova and others 2004; World Bank 2005a). Next, with regard to the level of satisfaction with PPHS, the ordered nature of the dependent variable suggests the use of an ordered probit or logit model. However, models with 76 World Bank Working Paper ordered dependent variables that correct for sample selection are practically non-existent, and standard tools are not readily available to estimate such models. We proceed following a two- step "Heckman ordered probit" procedure. We first use a binary probit model to study factors that influence whether or not an individual accesses PPHS during the past 12 months. We compute a correction factor, the inverse mills ratio, from this estimation. At the second stage, we use an ordered probit model to analyze the level of satisfaction with the PPHS, incorporating the correction factor calculated in the first step. In other words, the first equation determines sample selection, which we write as follows: z*i = wi + u i We do not observe z*, just zi = 1 if z*> 0, and zi = 0 otherwise; that is, we observe whether i i the respondent interacts with publicly provided health services (zi = 1) or doesn't (z1 = 0). Factors that we believe influence this decision, including variables such as per-capita expenditure, gender, self-assessed health status, etc, are included in the vector wi. Next, the equation that determines satisfaction with the public service is writ- ten as: y*i = xi + i We do not observe y*.i When zi = 1, we only observe whether yi = 1, 2, 3, 4, and 5 if j -1< y*i < j, (j = 1, 2, 3, 4, and 5)--i.e. when a survey participant interacts with the publicly pro- vided health service, we observe the level of satisfaction with the service received. Factors that influence the level of satisfaction are included in the vector xi. These include individ- ual-level characteristics such as per-capita expenditure, gender, age, general satisfaction with life, etc, in addition to country-level characteristics such as per capita GDP, growth rate, etc. We can reformulate the model as: Selection equation: Prob zi = 1 = wi ( ) ( ) Prob zi = 0 = 1- wi ( ) ( ) Main equation: Prob yij = 1 = j - xi - j - xi ( ) ( ) ( -1 ) yij is observed only if zi = 1, (ui, i) bivariate normal with correlation . Key Findings and Results We find that people do choose whether or not to use PPHS--we can reject the hypothesis that the selection and the main equations are independent at the 1 percent level. The vari- ables used in the analysis are summarized below; the results of the selection equation are reported in Table 3.1, and the main equation in Table 3.2. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 77 Variable Obs Mean Std. Dev. Min Max access_health 28999 .516 .500 0 1 trust 28999 .200 .400 0 1 logpcexp 28909 7.62 .833 .98 10.25 female 28999 .529 .499 0 1 over_65 28999 .155 .362 0 1 health 28995 2.79 .937 1 5 educ 28991 2.16 .494 1 3 locality 28999 1.76 .765 1 3 country 28999 18.8 7.89 1 29 access_health: Used PPHS during past 12 month: 1 = Yes, 0 = No lpgpcexp: (log) Per equivalent adult (using OECD scales) annual expenditures trust Respondent has some or complete trust in the police and in the government/ cabinet of ministers female: 1 = Female respondent, 0 otherwise over_65: Respondent aged over 65 years health: Self-assessed health status is 1 = "Very good", 2 = "Good", 3 = "Medium", 4 = "Bad", 5 = "Very bad" educ: 1 = "No education/no degree"; 2 = "Compulsory/secondary education"; 3 = "Higher/ post-graduate" locality: 1 = Rural; 2 = Urban; 3 = Metropolitan country: Country code (29 unique codes for each of the 29 countries covered in the survey) Factors Influencing Health Care Access We find that relatively better-off persons (as measured by log annual expenditures per equivalent adult) are more likely to access PPHS than those who are poor. This confirms that monetary costs have some bearing on access to services, and belies the entitlement to universal access expressly stated in many constitutions across this region. Females are also much more likely to use the service than males. This effect likely captures the routine use of PPHS by women for child birth and during pregnancy. Similarly, elderly respondents (those aged over 65 years) are also much more likely to have used PPHS compared to the rest of the surveyed population, a finding that is consistent with Balabanova and others (2004). The individual attributes that decreases health care access include self assessed good health and level of education--we find that people with compulsory/secondary education as well as people with some tertiary education are less likely to access PPHS com- pared to those with no education (note that the effect of compulsory/secondary education-- is not statistically significant at the 10 percent level).34 This in turn may be because those 34. Balabanova and others (2004) find that, when ill, people with lower education consult profession- als less often than people with higher education. However, a crucial difference between their findings and ours is that they are able to control for differences in morbidity levels. If average morbidity rates among those with higher education are significantly lower than among those with less education, there is no con- tradiction necessarily between this and our finding that, other things being equal, the likelihood of having accessed the health service during the past 12 months is negatively associated with the level of education of the respondent. 78 World Bank Working Paper Table 3.1. Probit for Health Care Access Overall sample Coefficient s.d. Individual-level variables: Log normalized expenditures 0.112*** 0.028 Trust in government 0.092** 0.045 Female 0.164*** 0.039 Over 65 years age 0.239*** 0.079 Female and over 65 years of age -0.276*** 0.093 Health status Very good -0.279*** 0.062 Good -0.158*** 0.042 Medium Reference category Bad 0.158*** 0.054 Very bad 0.287*** 0.087 Locality Urban -0.073* 0.041 Rural Reference category Metropolitan -0.105** 0.049 Education level No education/no degree Reference category Compulsory/secondary education -0.092 0.073 Higher professional/post-graduate 0.018 0.087 Other controls: Country-level dummies . . . not shown here . . . Pseudo-R2 0.0268 Source: Staff calculations based on 2006 LiTS data. [Note: .01 - ***; .05 - **; .1 - *;] with more education tend to adopt lifestyles that are healthier than those with less/no edu- cation, and are therefore likely to suffer lower morbidities. Finally, we find that respon- dents living in rural and in metropolitan areas are somewhat more likely to use PPHS than those living in urban areas, though the effect is not statistically significant at the 1 percent level. Trust in Government and Police We find that people who trust the government/police are significantly more likely to access PPHS, compared to those who do not trust the government or police. Given that the LiTS is a cross-sectional survey, one cannot determine whether greater trust in government leads to greater willingness to access public services, or whether this correlation is driven by other unobserved factors. Nevertheless, this finding seems to lend credence to the idea that there exists a social contract between the government and citizens; a functioning government, Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 79 and the experience of the rule of law in everyday lives goes hand-in-hand with a greater willingness to use public services. Thus policies targeting reform of specific public sectors may be handicapped in that they do not try to reform `the general culture' of government. Concurrently reforming public service delivery, along with the improvement in the performance of the police, and other institutions, could lead to an improvement in citizens' overall perception of the func- tioning of the government. If this has the effect of encouraging more people to use pub- licly provided health services, then they would directly experience the improved quality of these services, leading to a positive feedback loop to fall in place. Factors Influencing Satisfaction with Publicly-provided Health Services Respondents who reported having used PPHS during the past 12 months were asked how satisfied they were with the quality and efficiency of the medical treatment received. Their responses, coded as follows: 1=Strongly Disagree (SD), 2=Disagree (D), 3= Neither disagree or agree (N), 4=Agree (A), and 5=Strongly Agree (SA), are summarized below, while the results of the second-stage estimation, after including the correction factor calculated from step one of the estimation procedure, are reported in Table 3.2. The first point to note is that the coefficient of the inverse mills ratio correction factor is significant, indi- Level of Percent Cum. cating that we can safety reject the Satisfaction Freq. (weighted) (weighted) hypothesis that the selection and SD 1,536 11.3 11.3 main equations are independent. D 2,615 18.7 30.0 The negative coefficient implies that N 3,049 24.5 54.5 unobservedcharacteristicsofrespon- A 6,174 38.3 92.8 dents that affect the likelihood of SA 1,333 7.2 100.0 their having used PPHS during the Total 14,707 100.0 100.0 past 12 months are inversely related to the level of satisfaction with the quality and efficiency of service. As shown above, the variables at the individual level that enhance satisfaction with PPHS include (i) living in urban areas, (ii) self assessed good health, and (iii) satisfaction with life in general (individuals who are more satisfied with life are also more satisfied with PPHS). Balabanova 2007 reports that, as inequalities increased in the health care delivery sys- tem in the post-transition period, the quality of service available in urban centers were vastly superior to service in rural areas. This could well account for the increased satisfac- tion of urban residents. Self-assessed health status is seen as a fairly good indicator of gen- eral health and morbidity. This is corroborated by the finding, above, that those with good self-assessed health status access PPHS less frequently. Once ill, they probably also recover faster, leading to greater satisfaction with treatment received. The reason for the positive association between satisfaction with life in general and sat- isfaction with publicly provided health services is likely due to the personality of the respon- dent (Figure 3.8). Psychologists report that "measures of temperament and personality typically account for much more of the variance of reported life satisfaction than do life cir- cumstances" (Kahneman and Krueger 2006). Thus a respondent who is temperamentally 80 World Bank Working Paper Table 3.2. Ordered Probit: Satisfaction with Publicly-provided Health Services Overall sample Coefficient s.d. Inverse Mills Ratio (from selection equation) -3.074*** 0.231 Individual-level variables: Log normalized expenditures -0.190*** 0.021 Female -0.167*** 0.025 Age group 19­30 years 0.021 0.027 31­40 years Reference category 41­50 years 0.097*** 0.028 51­60 years 0.086*** 0.031 61­70 years -0.028 0.035 71+ years 0.000 0.041 Health status Very good 0.605*** 0.059 Good 0.482*** 0.032 Medium Reference category Bad -0.248*** 0.033 Very bad -0.632*** 0.060 Satisfied with life Strongly disagree -0.285*** 0.033 Disagree -0.072*** 0.026 Neither disagree nor agree Reference category Agree 0.215*** 0.024 Strongly agree 0.386*** 0.036 How often are unofficial payments needed Never 0.127*** 0.026 Seldom 0.027 0.029 Sometimes Reference category Usually -0.080*** 0.029 Often -0.134*** 0.030 Locality Urban 0.060** 0.023 Rural Reference category Metropolitan 0.039 0.031 Education level No education/no degree Reference category Compulsory/secondary education 0.075* 0.042 Higher professional/post-graduate -0.169*** 0.046 (continued) Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 81 Table 3.2. Ordered Probit: Satisfaction with Publicly-provided Health Services (Continued) Overall sample Coefficient s.d. Country-level variables: Per capita GDP 2005 (US$ `000) 0.245*** 0.023 Growth rate of per capita GDP (2004 to 2005) 0.056*** 0.007 Other controls: Country-level dummies . . . not shown here . . . Pseudo-R2 0.0297 Note: .01 - ***; .05 - **; .1 - *. Source: Staff calculations based on 2006 LiTS data. Figure 3.8. Satisfaction with Publicly-provided Health Service and Self-assessed Health Status Satisfaction with publicly provided health system by self-assessed health status Very good Good Medium Bad Very bad 35 40 45 50 55 Percent of satisfied respondents more satisfied with life is also likely to be more satisfied with a given quality of health ser- vice. Kahneman and Krueger additionally cite a couple of experiments that suggest that those who are satisfied with life in general may be less susceptible to morbidities35. It may simply be that those who are more satisfied with life fall ill less frequently and recover more easily. The health outcome for any illness episode, given a fixed quality of treatment, may 35. In one study, subjects were exposed to a cold virus, and their symptoms were monitored. Those who reported a higher level of life satisfaction at baseline were less likely to come down with a cold, and quicker to recover if they became sick. Another study subjected individuals to a controlled wound, which was then monitored. The study found that subjects who were more satisfied with their lives healed quicker than others. 82 World Bank Working Paper be better for those with a higher baseline satisfaction with life, leading to increased levels of satisfaction with the health service received. Similarly, other things being equal, respondents aged 41­60 years also tend to report higher levels of satisfaction with PPHS compared to the 31­40 year-old reference category (differences for other age groups are not statistically significant). By contrast, however, higher living standards, as proxied by per capita expenditures, as well as higher profes- sional/post-graduate educational attainment tend to be negatively associated with satisfac- tion with PPHS--i.e. the richer the respondent and the higher their education level, the less likely they are to be satisfied with PPHS. This likely indicates that such individuals tend to judge the quality and efficiency of PPHS using more exacting standards as compared to the poor/less well-educated. In addition, women also tend to express lower satisfaction than men. Women routinely use health services during pregnancy and child-birth. A likely explanation for women's lower satisfaction could be associated with the type of prenatal and child delivery care they receive. Balabanova 2007 reports that, particularly in maternal and child health in Russia, there is a "culture of over-medicalization tolerating ineffective and wasteful clinical practices dependent on mothers' ability to pay informally rather than need, administrative incentives to retain underused capacity, and user disempow- erment." Similarly, although services for pregnant women are ostensibly free in Georgia, informal payments are routinely demanded. One patient who was interviewed in Tbilisi in 2001 reports "For the delivery, we had to pay the doctor 200 Gel . . . We agreed the price with the doctor. They also told us that as long as we arrange for their private services (their guaranteed assistance during delivery) several month prior to delivery, the public coverage does not work, and we are not eligible for free services." (Belli et al, 2004). Finally, turning to variables at the country-level, we find both per capita GDP and the growth rate of per capita GDP to have significant positive effects on the satisfaction with PPHS. There are many ways through which macroeconomic well-being affects satisfaction with PPHS--it enables increased spending on health services, either by the government or by private households, it fosters optimism about the future, and so on. Once again, due to the cross-sectional nature of the LiTS survey, it is not possible for us to disentangle the means by which economic well-being affects satisfaction with PPHS, we merely note the strong and significant correlation. The Impact of Unofficial Payments Unofficial payments, as the name implies, are payments that do not go through official channels, and that are often made for what are meant to be free services. The reasons why they exist are myriad, and have been well documented (Lewis 2000; Belli, Gotsadze, and Shahriari 2004; Balabanova 2007; Bonilla-Chacin 2005; Allin, Davaki, and Mossialos 2006). At the same time, it serves us well to differentiate between three types of "unofficial" pay- ments, which have all typically been lumped together in most data, including in the LiTS. Unofficial payments could be (a) gifts, which are voluntarily given, (b) payments for sup- plies (such as gauze, bed clothes, medicines) that are not budgeted or paid for by the gov- ernment, or (c) payments that are required by medical personnel or administrators to Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 83 provide or expedite what should be free services. Only those unofficial payments that fall under category (c) can truly be called corruption. Given these different motivations, varied strategies are required to decrease the inci- dence of unofficial payments. Gift-giving is a strong and innocent cultural tradition to express gratitude to medical healers that has existed in the ECA region, particularly in CIS countries, for many generations. Gifts are often in-kind, and are more prevalent in rural areas within close-knit communities than in more impersonal urban areas. While these characteristics could be used in surveys to try and distinguish between voluntarily given gifts and unofficial payments that are demanded, there is no reason to wish to decrease or elim- inate the practice of giving gifts (such a campaign would be as meaningless and ineffective as trying to eliminate the practice of tipping for restaurant service in the United States). Unofficial payments that are used to purchase essential materials (as categorized under (b) above) do not go either to medical personnel or to administrators. They exist because of declining revenues but little rationalization of existing infrastructure resulting in inad- equate medical equipment, drugs and supplies. Health care reforms that address these issues, including identifying and eliminating funding gaps that ensure that all materials needed for a given health service are indeed available should suffice to decrease such types of unofficial payments. The third type of payment, which are requested by medical personnel or administra- tors, for access to (or to expedite) what should essentially be free service, is indeed a form of corruption. Some of the factors that contribute to the existence of such payments include underpaid or unpaid doctors, as well as financing and administration of hospitals and facilities that make them unresponsive to the evolving medical needs of the popula- tion. Reforms that address these deficiencies, while complex and difficult to implement, would likely lead to some decline in unofficial payments that are used to influence health outcomes. Simply running anti-corruption campaigns without instituting mechanisms that provide effective accountability for health outcomes to users of the PPHS are likely to be ineffective. As Vian and Burak (2006) find in Albania, there is no difference in the moral beliefs of people who intend to make unofficial payments the next time they access the PPHS and those who do not intend to do so. On the other hand, people who do not intend to make unofficial payments report other means by which they can influence health out- comes, such as connections with medical personnel, etc. Measuring unofficial payments, particularly those that indicate corruption, are diffi- cult, due to the very nature of the transaction. It is not always straightforward to distinguish between formal and informal payments, or between payments that are demanded and that are given voluntarily out of gratitude. Another confounding factor has been the prevalence of unofficial payments for some types of services (inpatient services in hospitals) as opposed to other services (outpatient physicians visits); and in some geographic areas (urban) as opposed to others (rural). Thus, depending on the sample of respondents, and on the actual question fielded, one could get widely varying estimates of unofficial payments. One advantage of the LiTS has been the general nature of the question: "In your opin- ion, how often is it necessary for people like you to have to make informal payments/gifts when receiving medical treatment in the public health system." Responses were obtained both from those who used the system in the recent past (last 12 months) and those who did not. A comparison of general perception (the responses from those who have not used PPHS in the last 12 months) with the opinion of experienced users is revealing. It appears 84 World Bank Working Paper Figure 3.9. General vs. Experienced Opinion of Need for Unofficial Payments Respondents that think unofficial payments are needed in publicly provided health system Estonia Slovenia Kazakhstan Georgia Turkey Czech Republic Mongolia Croatia Latvia Montenegro Serbia Belarus Slovakia Poland Bosnia Bulgaria Armenia ECA REGION Lithuania Macedonia Russia Azerbaijan Hungary Tajikistan Romania Ukraine Uzbekistan Kyrgyz Republic Moldova Albania 0 20 40 60 Percent of respondents Did not use Used in past 12 months that general perception consistently underestimates the prevalence of unofficial payments (Figure 3.9). Our analysis indicates that the perception that unofficial payments are necessary for access is an important factor causing dissatisfaction with PPHS. For instance, dissatisfaction with health care is highest in Albania, with 48 percent of respondents being dissatisfied or very dis- satisfied with the service interaction. Albania also has the largest percentage of people (48 per- cent), who believe that unofficial payments are usually or always needed to access PPHS. Our empirical model indicates that, other things being equal, respondents who say unofficial payments are often necessary are about 1.4 times more likely to report being dis- satisfied with PPHS compared to those who say that such payments are never needed. As reasoned above, one of the motivations for unofficial payments has been to ensure higher quality of service than otherwise available. Thus, one might expect to find that, in some cases at least, unofficial payments go hand-in-hand with higher satisfaction with the ser- vice interaction. If this effect exists at all, it is likely very small, and is overwhelmed by the amount of dissatisfaction generated by having to pay for free services. Bratton, 2007, finds that the perception that officials are corrupt decreases citizen satisfaction with services in African countries, and this is in agreement with the findings in this analysis. We found two other sources that report the incidence of unofficial payments in a limited set of ECA countries in different time periods (Lewis 2000; Balabanova and others 2004). Table 3.3 summarizes the prevalence of unofficial payments from these two sources, and from the experienced users in LiTS. Comparing changes in access rates, satisfaction level, Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 85 and reported prevalence of unofficial Table 3.3. Prevalence of Unofficial Payments payments derived from the LiTS with for Selected Countries those from earlier studies provides Prevalence of unofficial another means of analyzing the payments in PPHS complex interaction between these 2001: three sets of variables. Balabanova 2006: Although the usual caveats Country Lewis36 et al37 ECA-LiTS38 apply regarding the need for caution Georgia 65 12 when comparing the findings of dif- Kazakhstan 38 12 ferent surveys, these three sources Armenia 91 (1999) 56 30 provide a rough indication at least Moldova 70 (1999) 46 48 of changes in prevalence of unoffi- Kyrgyzstan 75 (1996) 42 46 cial payments in these 15 coun- Belarus 8 24 tries between the late 1990s and Ukraine 27 44 2006. Comparing the Lewis findings Russia 74 (1997) 19 37 with the LiTS, we see that the inci- Azerbaijan 78 (1995) -- 40 dence of unofficial payments (or in Poland 78 (1998) -- 25 the case of the LiTS of the opinion Tajikistan 66 (1999) -- 44 that unofficial payments are neces- Slovak sary to access PPHS) declined in Republic 60 (1999) -- 25 all countries except Albania and Latvia 31 (2000) -- 17 Bulgaria between the late 1990s and Albania 22 (1996) -- 55 2006. Comparing the findings of Bulgaria 21 (1997) -- 30 Balabanova and others (2004) with the LiTS, we see that the percent- age of people who had to pay informally/make a gift declined in Georgia (dramatically so), Armenia, and Kazakhstan, between 2001 and 2006. This percentage has remained fairly constant in Moldova and Kyrgyzstan, while it has, by contrast, increased in Belarus, Russia, and Ukraine. Evidence that unofficial payments act as barriers to access emerges when one compares access rates in the LiTS to that in Balabanova et al study. Firstly, Balabanova and others (2004) find that, among all respondents who report an illness that justified seeking attention, about 21 percent did not do so. The reason for not seeking care cited by 78 percent of respondents was the lack of money to pay for treatment. Comparing access rates from the two data sources by country, we find that access rates have not changed by much in Kyrgyzstan, Moldova and Ukraine, but have increased in Armenia, Kazakhstan, and in Georgia (quite dramatically in the case of the latter); declined in Russia, and almost halved in Belarus (Table 3.4). An important finding emerging from comparison of access rates and prevalence of unofficial payments between the Balabanova et al (2004) study and the LiTS is highlighted in Figure 3.10, which contrasts the extent to which the two data sources indicate that the 36. Per cent of patients that reported making informal payments (the year for each observation is included in brackets) 37. Respondents that reported paying informally/making a gift during their most recent consulta- tions; data is from surveys of adults aged 18 years and older conducted in Autumn 2001. 38. Of the respondents who accessed the public health system in the last 12 months, percent that say that unofficial payments/gifts are usually / always necessary. 86 World Bank Working Paper prevalence of unofficial payments Table 3.4. Change in Health Care Access Rates went down over this period, with for Selected Countries changes in the share of the popula- Access Rates Among the tion that reported using publicly Population Aged 18 and Over provided health care facilities dur- 2001: ing the past 12 months. Balabanova 2006: As the figure clearly shows, the Country et al39 ECA-LiTS40 Difference*** eight countries for which this com- Georgia 24 46 +22 parison is possible fall into three Kazakhstan 50 64 +14 main groups. In the first group, com- Armenia 30 39 +9 prising Georgia, Kazakhstan, and Moldova 53 50 -3 Armenia, declines in the reported Ukraine 58 54 -4 prevalence of unofficial payments Kyrgyzstan 44 40 -4 during this period was accompanied Russia 66 52 -14 by significant increases in the share Belarus 66 37 -29 of the population utilizing public health care facilities. In the second group, which includes Moldova, Ukraine, and the Kyrgyz Republic, access rates did not change by very much during this period, and neither did the prevalence of unofficial payments (with the exception of Ukraine). Finally, in the third group, with Russia and Belarus, the increase in prevalence of unofficial payments between 2001 and 2006 was accompanied by significant declines in the share of the population accessing PPHS. These findings underline the barrier that unof- Figure 3.10. Changes in Access Rates and Prevalence of Unofficial Payments, 2001 to 2006 Georgia Kazakhstan Armenia Moldova Ukraine Kyrgyzstan Russia Belarus -40 -30 -20 -10 0 10 20 30 40 50 60 Increase: Access rates Decline: prevalence of unofficial payments Source: Balabanova and others (2004) for 2001, LiTS data for 2006. 39. Survey conducted in autumn 2001. Access rate refers to consulting a health care professional (pub- lic or private); (access rates have been translated from a graph, so the numbers are approximate). 40. Survey conducted in autumn 2006. Access rate refers to accessing the public health system. Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 87 ficial fees impose upon access to health service, with access rates falling as unofficial fees rise. They also provide strong corroborative evidence to our earlier reported findings that the perception of the necessity of unofficial payments negatively impacts satisfaction levels. Finally, given the strong negative influence of unofficial payments on satisfaction with PPHS, one may plausibly be able to argue that the LiTS question indeed captures unofficial payments that act as barriers to access rather than unofficial payments that are given as gifts. Health Sector Reform in the Caucuses In this section we probe a little into the strong performance of Georgia and Armenia, and to a smaller extent Azerbaijan, on satisfaction with PPHS. All three countries suffered catastrophic economic decline with the fall of the Soviet Union--real GDP per capita shrank by 19.2 percent on average per year in Georgia, by 16.8 percent per year in Azerbai- jan, and by 8.4 percent per year in Armenia, from 1991 to 1995.41 This economic collapse led to sharply shrinking budgets, particularly for the health sector. The expenditure on PPHS as a percent of GDP decreased dramatically in Georgia, from about 3 percent in 1991 to a little above 0 percent in 1994. Available data for the latter half of the decade show that health out- comes stagnated (and in some instances actually worsened) during this period.42 Spending recovered to about 1 percent in 1995, and has remained at that level ever since. While the decline in spending was more gradual in Armenia, spending has hovered between 1 and 2 percent of GDP in the last decade. Health sector expenditure in Azerbaijan fluctuated in the early 1990s, before settling down to between 2 and 3 percent of GDP (Bonilla-Chacin, Murrugarra, and Temourov 2005; UNICEF TransMONEE Database 2006). This collapse in public expenditure led to dramatic increases in out of pocket payments-- Bonilla-Chacin, Murrugarra, and Temourov (2005) report that out-of-pocket private spend- ing (both formal and informal) in Georgia has been estimated to range between 66 and 87 percent of total health spending, while household expenditure on health plus donor con- tributions account for between 63 and 69 percent of total health spending in Armenia. They also document the drastic decline in health care utilization, particularly among the poor, as a result of these increasing costs of access. Most countries in the low-income CIS embarked on health care reforms--with the focus being to realign facilities and medical personnel to give more emphasis to the pri- mary health care system. In addition, the reforms have tried to raise additional revenues for the health sector. All low-income CIS countries have implemented some cost-sharing arrangements by charging fees for some health services. In 1995, the Government of Georgia formally removed entitlement to free health care from its constitution. Public insurance coverage was limited to services included in a basic benefit package (BBP). A social insurance contribution was imposed on formal employment, with shortfalls being met by transfers from general taxation (Belli, Gotsadze, and Shahriari 2004). Many low-income CIS countries have also tried to rationalize costs. Substantial progress has been made in decreasing the number of hospital beds, although the numbers still exceed those available in the EU. There is some mixed progress in rationalization of 41. World Development Indicators 2007. 42. Public Information Document: Georgia-Primary Health Care Development Project, World Bank, Washington DC, January 2002. 88 World Bank Working Paper health care staff, with Georgia and Moldova achieving significant reductions in the num- ber of health sector workers. Georgia and Armenia have shown some progress in privatizing hospitals. The most dramatic evidence of changes in health care practice have been visible in Georgia and Armenia, with much less reform in Azerbaijan (Bonilla-Chacin, Murrugarra, and Temourov 2005). Combined with these general reforms in the health sector, Oxfam and other inter- national agencies have supported a particular innovation in the Caucasus (Armenia, Georgia and Azerbaijan), aimed at mitigating the situations of the rural poor--community health insurance schemes. Ensuring equity of access to primary health care has been an explicit objective of the scheme. Balabanova 2007 reports that Oxfam's schemes have contributed to improved access to and quality of care, through rehabilitation of local health posts, train- ing of nurses, and subsidy of running costs. She further reports that these schemes have grown to be major providers of health care to rural communities, to the point where they have assumed significant responsibility for public sector provision. Real GDP per capita in Georgia, Armenia and Azerbaijan began to grow, more mod- estly in the late 1990s but quite dramatically since 2000. Real GDP per capita growth aver- aged 12.6 percent per year from 2001 to 2005 in both Armenia and Azerbaijan, and by 8.4 percent per year in Georgia over the same period. We posit that the reforms in the health sector undertaken in Georgia and Armenia, and the innovations in community health insurance schemes, combined with the transformational and broad based growth occurring in both countries, have led to recovery in the quality of PPHS from an extremely low base in the early 1990s. These factors help explain, to some extent, the significant rise in access rates and decline in prevalence of unofficial payments in countries like Georgia and Armenia. While Azerbaijan has equally strong economic performance, it has not pro- gressed as far in health sector reform. Our empirical model provides some support for this; we find that the growth rate of per capita GDP is one of the few macroeconomic variables that have a significant positive effect on the satisfaction with PPHS. Concluding Observations In this paper we investigate satisfaction with publicly provided health services (PPHS) in 28 ECA countries and Mongolia using data from the Life in Transition survey conducted in these countries in the fall of 2006. The countries in this region inherited a health system that was state owned and financed through general taxation, and that was largely charac- terized by over capacity, both in terms of human as well as physical factors, at the time of the fall of the Soviet Union. The system was largely unsustainable, particularly given the economic decline in the region in the 1990s. Most countries undertook reform of the health sector, with mixed results. The EU member countries have the most positive outcome, with (i) mixed financing of health, through insurance and general taxation, (ii) rationalization of health sector capacity, (3) and more recently, solid economic growth, that has enabled them to provide stable financing for the health sector. Thus, among ECA countries, many EU member are at or near the top with regards to satisfaction with PPHS. A surprising finding from the LiTS has been the levels of satisfaction in a few low-income CIS countries, particularly Georgia and Armenia. Some discussions of these findings are Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 89 presented in section six above. Finally, many middle-income CIS countries (Ukraine, Kazakhstan, Russia) are among the lower ranked in terms of citizens' satisfaction with PPHS. We use a two-step Heckman probit to estimate the effects of various factors on (i) why some people select to use PPHS (ii) upon use, the level of satisfaction derived from the service provided. We find that factors that have a significant positive effect on accessing health care include (a) being relatively better off, (b) female, (c) elderly, and (d) having trust in government and police. The factors that have a significant negative effect on accessing health care are (a) self-assessed good health, and (b) level of educa- tion (those with higher levels access the health care system less than those with lower levels of education.) Factors that have a significant positive effect on satisfaction with service received include (a) per capita GDP (b) growth rate of per capita GDP (c) living in urban areas, (d) self assessed good health, (e) satisfaction with life in general (individuals who are more satisfied with life are also more satisfied with PPHS), and (f) being between 41­60 years of age. Factors that have a negative, and significant, effect on satisfaction include (a) the per- ception that unofficial payments are needed to access service, (b) higher living standards, (c) higher professional/post graduate education, and (d) being female. In terms of broad policy implications, the reduction/elimination in informal payments for health service is the most urgent reform needed in this region--particularly among the CIS countries. The primary reason for the existence of these informal payments seems to be to influence availability and quality of services received through the PPHS as there are no other means available to hold providers and administrators accountable. Thus, the estab- lishment of mechanisms that enable provider accountability is a first step towards the elim- ination of informal payments. Sustained and long-term engagement in the reform process is needed to address this and several other complex factors that give rise to these fees. This paper finds that trust in government and police go hand-in-hand with increased access of PPHS. Thus, changing the culture of government is as important as undertaking reforms in the health care system. A well coordinated communications strategy that is aimed at informing health sector users of their entitlements, and of emphasizing government com- mitment to reform, could influence citizens' perception of what they should expect from health care services, and could co-opt them into partners in enabling change. 90 World Bank Working Paper Annex: Tables and Figures Table A3.1. Priorities for Additional Government Spending, By Country 1. First priority: Percentage of Respondents Choosing: Group/Country Education Healthcare Housing Pensions Environment Infrastructure Overall EU member states 22 44 10 17 2 4 100 South-Eastern Europe 32 33 8 16 3 8 100 CIS-low income 27 41 9 16 3 4 100 CIS-middle 21 42 19 13 4 2 100 Total 27 40 13 13 3 3 100 Albania 24 23 9 21 5 18 100 Belarus 28 32 21 13 2 4 100 Bosnia 39 27 8 18 1 6 100 Bulgaria 23 53 3 17 2 4 100 Croatia 35 32 8 19 2 4 100 Czech Republic 29 41 11 13 5 2 100 Macedonia 39 33 7 14 2 4 100 Hungary 16 53 8 13 4 6 100 Moldova 24 45 8 14 2 6 100 Montenegro 27 35 11 13 3 10 100 Poland 21 41 11 19 1 6 100 Romania 18 42 15 21 2 2 100 Serbia 28 41 7 12 4 8 100 Slovakia 31 50 5 9 2 3 100 Slovenia 32 37 13 12 4 2 100 Turkey 58 29 4 5 1 3 100 Ukraine 21 49 13 11 3 3 100 Armenia 20 48 6 21 2 2 100 Azerbaijan 29 41 12 15 1 3 100 Estonia 30 41 4 15 4 5 100 Georgia 24 35 11 23 2 5 100 Kazakhstan 22 44 12 13 7 3 100 Kyrgyz Republic 32 37 13 9 2 7 100 Latvia 32 36 6 18 2 5 100 Lithuania 24 49 10 14 0 2 100 Mongolia 40 28 7 9 5 11 100 Russia 21 40 21 13 4 1 100 Tajikistan 40 29 9 14 6 2 100 Uzbekistan 24 44 8 16 4 4 100 Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 91 Table A3.2. Access Rates of PPHS, By Country Percent of population who ______ the public health care system in the last 12 months Country Accessed Did not access Albania 65.7 34.3 Belarus 37.3 62.7 Bosnia 42.1 57.9 Bulgaria 48.4 51.6 Croatia 61.1 38.9 Czech Republic 59.5 40.5 Macedonia, FYR 51.9 48.1 Hungary 61.3 38.7 Moldova 49.4 50.6 Montenegro 45.3 54.7 Poland 41.3 58.7 Romania 51.4 48.6 Serbia 49.1 50.9 Slovak Republic 33.4 66.6 Slovenia 38.3 61.6 Turkey 58.0 42.0 Ukraine 54.3 45.7 Armenia 38.5 61.5 Azerbaijan 56.9 43.1 Estonia 51.6 48.4 Georgia 45.7 54.3 Kazakhstan 64.0 36.0 Kyrgyzstan 39.9 60.1 Latvia 56.5 43.5 Lithuania 54.4 45.6 Mongolia 37.8 62.2 Russia 51.7 48.3 Tajikistan 55.4 44.6 Uzbekistan 44.7 55.3 92 World Bank Working Paper Table A3.3. Satisfaction with Medical Treatment in PPHS by Country 50 Satisfied Very Very satisfied satisfied Very 40 unsatisfied 30 Unsatisfied respondents 20 Satisfied of % 10 Indifferent 0 South- member Eastern Europe CIS-low income states income EU CIS-middle Percentage of respondents that are: Very Very Group/Country Unsatisfied Unsatisfied Indifferent Satisfied Satisfied Overall EU member states 10 15 24 43 8 100 South-Eastern Europe 14 17 19 40 11 100 CIS-low income 13 20 17 43 7 100 CIS-middle income 10 23 28 36 5 100 Total 11 19 24 38 7 100 Slovenia 3 6 26 46 20 100 Czech Republic 3 7 30 48 11 100 Croatia 7 8 20 45 20 100 Georgia 4 11 22 54 10 100 Slovakia 8 6 29 39 18 100 Lithuania 4 16 23 45 11 100 Armenia 9 17 11 53 11 100 Estonia 5 23 12 50 10 100 Latvia 6 20 19 48 9 100 Hungary 7 16 27 42 8 100 Bulgaria 9 21 11 48 12 100 Belarus 3 24 23 43 6 100 Poland 9 14 31 42 4 100 Bosnia 13 14 24 39 11 100 Azerbaijan 13 19 13 47 8 100 Turkey 17 11 21 37 14 100 Montenegro 12 20 19 36 13 100 Mongolia 12 17 22 42 6 100 (continued) Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 93 Table A3.3. Satisfaction with Medical Treatment in PPHS by Country (Continued) Percentage of Respondents that are: Very Very Group/Country Unsatisfied Unsatisfied Indifferent Satisfied Satisfied Overal Russia 10 21 28 37 5 100 Uzbekistan 13 21 18 42 6 100 Kazakhstan 9 22 31 34 5 100 Macedonia 17 16 16 45 5 100 Serbia 16 20 19 37 9 100 Kyrgyz Republic 16 23 11 44 6 100 Moldova 13 24 21 37 6 100 Romania 19 19 16 39 6 100 Tajikistan 17 21 25 33 4 100 Ukraine 10 28 28 31 3 100 Albania 21 27 12 38 2 100 Table A3.4. Prevalence of Unofficial Payments in PPHS by Country 15 Usually Always Always 10 Usually respondents Never of 5 % Sometimes 0 Seldom South- member Eastern Europe CIS-low income states income EU CIS-middle Percentage of respondents that say unofficial payments are needed: Group/Country Never Seldom Sometimes Usually Always Total EU member states 51 11 18 12 8 100 South-Eastern Europe 55 9 15 11 10 100 CIS-middle income 39 15 20 15 12 100 CIS-low income 37 16 18 14 15 100 Overall sample: 42 14 19 14 11 100 Estonia 75 11 11 2 1 100 Slovenia 73 9 12 5 1 100 (continued) 94 World Bank Working Paper Table A3.4. Prevalence of Unofficial Payments in PPHS by Country (Continued) Percentage of respondents that say unofficial payments are needed: Group/Country Never Seldom Sometimes Usually Always Total Georgia 59 21 12 6 2 100 Kazakhstan 63 14 13 5 5 100 Belarus 59 17 12 8 4 100 Turkey 66 9 13 7 5 100 Czech Republic 53 18 19 8 2 100 Croatia 65 9 13 11 3 100 Latvia 60 12 15 8 5 100 Poland 57 11 17 10 5 100 Serbia 61 8 14 9 8 100 Mongolia 47 16 21 11 6 100 Bosnia 57 10 14 9 10 100 Armenia 45 20 16 11 8 100 Montenegro 51 11 20 9 9 100 Slovakia 46 14 19 13 8 100 Macedonia 44 13 20 11 13 100 Russia 40 15 20 14 10 100 Lithuania 35 14 27 16 8 100 Bulgaria 41 10 25 13 11 100 Kyrgyz Republic 41 13 17 18 11 100 Romania 47 9 14 16 14 100 Azerbaijan 39 19 12 12 18 100 Moldova 35 16 17 19 13 100 Hungary 39 10 18 19 13 100 Tajikistan 24 21 21 18 15 100 Uzbekistan 34 12 21 14 19 100 Ukraine 25 13 22 20 20 100 Albania 26 8 18 18 30 100 Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union 95 Table A3.5. Difference between General and Experienced Perception of Prevalence of Unofficial Payments in PPHS Percentage of respondents that say unofficial payments are needed (Accessed: used PPHS in last 12 months; Not accessed: Did not use) Never/Seldom Usually/Always Country Accessed Not Accessed Accessed Not Accessed Albania 27.9 46.8 54.9 33.5 Belarus 59.6 85.5 23.7 5.0 Bosnia 52.0 77.9 29.5 10.7 Bulgaria 40.7 59.9 29.6 19.4 Croatia 68.1 82.0 16.5 9.0 Czech Republic 62.9 82.7 13.5 4.7 Macedonia, FYR 42.4 73.0 31.9 13.6 Hungary 37.4 68.3 43.6 15.5 Moldova 26.5 75.7 48.5 15.0 Montenegro 56.7 65.7 19.8 16.7 Poland 48.5 81.6 25.3 7.3 Romania 38.3 74.2 44.4 16.0 Serbia 59.4 78.0 22.6 11.9 Slovakia 54.0 63.7 24.9 18.9 Slovenia 72.8 87.4 9.6 4.8 Turkey 73.5 77.6 12.4 10.2 Ukraine 31.7 46.8 44.5 33.7 Armenia 42.5 79.4 30.2 11.2 Azerbaijan 45.3 73.9 40.2 17.1 Estonia 85.2 87.3 4.7 1.8 Georgia 74.6 84.9 11.9 4.3 Kazakhstan 74.6 80.7 11.6 9.1 Kyrgyzstan 26.0 72.8 46.1 16.8 Latvia 64.3 81.7 16.5 8.4 Lithuania 33.7 67.9 32.0 13.2 Mongolia 53.2 68.4 16.0 17.0 Russia 35.4 76.7 36.9 11.0 Tajikistan 30.2 64.1 44.0 20.0 Uzbekistan 31.7 57.6 45.3 23.1 Bibliography Allin, S., K. Davaki, and E. Mossialos. 2006. "Paying for `free' health care: the conundrum of informal payments in post-communist Europe." Global corruption report 2006. Berlin: Transparency International. Balabanova, D. 2007. "Health Sector Reform and Equity in Transition." Processed. http:// www.wits.ac.za/chp/kn/Balabanova.pdf Balabanova, D., M. McKee, J. Pomerleau, R. Rose, and C. Haerpfer. 2004. "Health Service Utilization in the Former Soviet Union: Evidence from Eight Countries." Health Services Research Supplement Part 2, 39(6):1927­49. Belli, P., G. Gotsadze, and H. Shahriari, 2004. "Out-of-pocket and informal payments in health sector: evidence from Georgia." Health Policy 70:109­23. Bonilla-Chacin M. E., E. Murrugarra, and M. Temourov. 2005. "Health Care during Tran- sition and Health Systems Reform: Evidence from the Poorest CIS countries." Social Policy and Administration 39(4):381­408. Bouckaert G., and S. Van de Walle, 2003. "Quality of Public Service Delivery and Trust in Government." In A. Salimen, ed., Governing Networks: EGPA Yearbook. Amsterdam: IOS Press. Bratton, M. 2007. "Are you being served? Popular Satisfaction with Health and Education Services in Africa." http://www.afrobarometer.org/papers/AfropaperNo65.pdf Danishevski, K., D. Balabanova, M. McKee, and J. Parkhurst, 2006. "Delivering babies in a time of transition in Tula, Russia." Health Policy Plan 21(3):195­205. EBRD (European Bank for Reconstruction and Development). 2007. Life in Transition, a Survey of People's Experience and Attitudes. London. Falkingham, J. 2004 "Poverty, Out-of-Pocket Payments and Access to Health Care: Evidence from Tajikistan." Social Science Medicine 58:247­58. 97 98 World Bank Working Paper Kahneman D, and A. Krueger, 2006 "Developments in the Measurement of Subjective Well-Being." Journal of Economic Perspectives 20(1):3­24. Lewis, M. 2000. "Who is paying for health care in Eastern Europe and Central Asia?" The World Bank, Washington, D.C. Vian, T, and L. Burak. 2006. "Beliefs about informal payments in Albania." Health Policy and Planning 21(5):392­401. World Bank. 2005a. Growth, Poverty, and Inequality: Eastern Europe and the Former Soviet Union. Washington, D.C. ------. 2005b. MDGs in Europe and Central Asia: Performance and Prospects. Washington, D.C. Eco-Audit Environmental Benefits Statement The World Bank is committed to preserving Endangered Forests and natural resources. We print World Bank Working Papers and Country Studies on 100 percent postconsumer recy- cled paper, processed chlorine free. The World Bank has formally agreed to follow the rec- ommended standards for paper usage set by Green Press Initiative--a nonprofit program supporting publishers in using fiber that is not sourced from Endangered Forests. For more information, visit www.greenpressinitiative.org. In 2008, the printing of these books on recycled paper saved the following: Trees* Solid Waste Water Net Greenhouse Gases Total Energy 355 16,663 129,550 31,256 247 mil. *40 in height and 6- Pounds Gallons Pounds CO2 Equivalent BTUs 8" in diameter Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union is part of the World Bank Working Paper series. These papers are published to communicate the results of the Bank's ongoing research and to stimulate public discussion. The past two decades in Eastern Europe and the former Soviet Union have been times of tremendous change, with countries undergoing rapid transformation from centrally-planned to market-oriented economies. While poverty increased during the initial years of transition, primarily on account of the sharp economic contraction, the resurgence of economic growth in the region since 1998 has resulted in a rebound in household incomes and living standards. Data from the 2006 Life in Transition Survey (LiTS)--a joint initiative of the European Bank for Reconstruction and Development and the World Bank-- provides a unique opportunity to investigate the extent to which citizens of ECA countries are satisfied with their lives and with the performances of their governments, and to study key factors influencing their outlook in a systematic way across all countries of the region. The main objective of the LiTS was to assess the impact of transition on people, covering four main themes. First, it collected personal information on aspects of material well-being, including household expenditures, possession of consumer goods such as a car or mobile phone, and access to local public services and utilities. Second, the survey included measures of satisfaction and attitudes towards economic and political reforms as well as public service delivery. Third, the LiTS captured individual "histories"--key events and episodes that may have influenced their attitudes towards reforms, and information on family background, employment, and coping strategies. Finally, the survey also attempted to capture the extent to which crime and corruption are affecting peoples' lives, and the extent to which individuals' trust in other people and in state institutions has changed over time. World Bank Working Papers are available individually or on standing order. Also available online through the World Bank e-Library (www.worldbank.org/elibrary). ISBN 978-0-8213-7900-4 THE WORLD BANK 1818 H Street, NW Washington, DC 20433 USA Telephone: 202 473-1000 Internet: www.worldbank.org SKU 17900 E-mail: feedback@worldbank.org