Report No. 25343-BIH Bosnia and Herzegovina Poverty Assessment (In Two Volumes) Volume I: Main Report November 21, 2003 Poverty Reduction and Economic Management Unit Europe and Central Asia Region Document of the World Bank TABLE OF CONTENTS VOLUME I ACKNOWLEDGEMENTS EXECUTIVE SUMMARY .......................................................................................................................................... i PART I POVERTY PROFILE . .................................................................................................................... 1 IMEASURINGPOVERTYINBIH . ............................................................................................................ A.Background: Why a Poverty Assessment?........................................................................... 1 1 B.Data for Poverty Analysis ............................................................................................... 2 4 D.Poverty Numbers............................................................................................................. C.Living Standards and Poverty Lines ............................................................................... 6 E.Inequality......................................................................................................................... 7 F.Depthand Severity of Poverty......................................................................................... 8 G.Dimensionsof Poverty............................................................................................................... 9 (i) EducationPoverty: Legacy of the Past?................................................................................ 12 (ii) Rights...................................................................................................................... Human 12 H.(iii) Health. Nutrition. Environment and Sanitation ................................................................. 15 Conclusions............................................................................................................................... 17 I1 VULNERABILITY TO POVERTY: RISKSAND PERCEPTIONS . ...................................................... 19 A. Vulnerability to Poverty as Exposure to Risks......................................................................... 19 (i)Dangerously Low Consumption ........................................................................................... 19 (ii)Employment Risk ................................................................................................................ 21 (iii) Risks....................................................................................................................... 23 Health B.(iv)Corruption, Political, Governance Risks ............................................................................ 24 Vulnerability to Poverty as Failureto Cope with Risks............................................................ 24 (i) toLaborMarket......................................................................................................... Access 25 (ii) andAccesstoCredit................................................................................................. Assets 27 (iii) toSafetyNetsandFamilyhformalNetworks....................................................... Access 27 29 D.Data C.(iv)Mobility and Access to Transportation................................................................................ and Trends in Poverty Perceptions................................................................................... 29 Conclusions............................................................................................................................... 31 I11 CHARACTERISTICSOF THE POOR? . .................................................................................... 33 A.Poverty Profile for BiH:Who Are at Risk?Who Are the Poor? ........................................... B. Children versus the Elderly....................................................................................................... 33 37 D.Women's Work and Poverty..................................................................................................... C. Work and Poverty..................................................................................................................... 38 39 E. Locality-Based Variations of PovertyRiskinBiH................................................................. -43 F.GeographicalConcentrationof Poverty.................................................................................... 44 H.FromPoverty Ratesto PovertyCorrelates............................................................................... G. Education and Poverty............................................................................................................. 45 I.Conclusions............................................................................................................................... 45 47 PARTI1 POVERTY AND PUBLIC POLICIES . ................................................................................. 49 IV INITIAL CONDITIONS AND STRUCTURALREFORMS . ....................................................... A. War and Reconstruction.......................................................................................................... 49 49 C. Conclusions and Policy Implications........................................................................................ B. Structural Reforms: HardLessons........................................................................................... 51 54 V GROWTH. EMPLOYMENT AND THE BUSINESSENVIRONMENT . ...................................... 57 A.MacroPolicies and Growth..................................................................................................... 57 B.Employment Generation........................................................................................................... 59 C. Business Environment............................................................................................................... 64 D.Conclusions andPolicy Implications........................................................................................ 66 V I HEALTH AND POVERTYINBIH . ................................................................................................. A.Introduction............................................................................................................................... 67 67 B.Population's Health Status........................................................................................................ 69 D. HealthService Utilization......................................................................................................... C.Health Care Provision ............................................................................................................... 71 74 E. Barriers to Access...................................................................................................................... 76 F.HealthCareFinances: Public Expenditures.............................................................................. 78 H. Conclusions andPolicy Implications........................................................................................ G Health Care Finances: PrivateExpenditures............................................................................. . 79 82 VI1 EDUCATIONAND POVERTY INBIH . ........................................................................................ A.Overview................................................................................................................................... 85 85 B. Educational Status of the Population inBiH............................................................................ 87 C Reasons for Low School Enrollment ........................................................................................ 90 D.ExpenditureinEducation.......................................................................................................... . E.Conclusions andPolicy Implications........................................................................................ 91 95 VI11 TARGETING. EFFICIENCY AND EFFECTIVENESSOF SAFETY NETS . .......................... A.Background............................................................................................................................... 99 99 B. HouseholdPerspective............................................................................................................ 101 C. Pension System and Poverty................................................................................................... 102 D.Targeting of SocialAssistance................................................................................................ 106 E.Unemployment Benefits.......................................................................................................... F. Conclusions andPolicy Implications...................................................................................... 110 111 IX CHALLENGESAHEAD. . ............................................................................................................... A. Understanding Causes of Poverty to FightIt ......................................................................... 115 115 B.How Far Can Economic GrowthReduce Poverty inthe Future?.......................................... C. Setting Priorities for Actions .................................................................................................. 117 122 D.Monitoringof PovertyReduction: Status and Recommendations.......................................... 126 REFERENCES ...................................................................................................................................... 129 TABLES Table A. Material Poverty Incidence, Depthand Severity by Entities I Table B. Poverty and Number of Children (below 14) inHousehold 111 ... Table C. Poverty and Number of Earners inHouseholds with Children iv Table D. Concentrationof the Poor inBiHAcross Poor and Non-Poor Communities V Table E. Poverty and Education of HouseholdHead V Table 1.1 Bosnia and Herzegovina: Poverty Incidence (Percent) 6 Table 2.1 Comparison of Labor Market Outcomes, Bosnia and Herzegovina and Other Balkan Countries 21 Table 2.2 MainProblem inRelation to the Provision of Public Services (Most Frequent Answers) 29 Table 3.1 Poverty Profile for BiH2001 34 Table 3.2 Poverty and Number of Earners in Households with Children 38 Table 3.3 Poverty Indices and Composition by Location and Entity 43 Table 3.4 Poverty Indices and Composition by Location and Entity 44 Table 3.5 The "Net" Impact of Various RiskFactors on Probability of Becoming Poor 46 Table 5.1 Bosnia and Herzegovina: MainEconomic Indicators, 1997-2003 58 Table 5.2 Men: Labor Market Outcome by School Attainment (Prime Age Workers, 18-35 Years Old), PercentageShares 64 Table 5.3 Women: Labor Market Outcome by School Attainment ( PrimeAge Workers, 18-35 Years Old), PercentageShares 64 Table 6.1 Self-perceivedHealth Status. Percentagesof Individuals 70 Table 6.2 Percentagesof InsuredPopulation by Age and Income Category 72 Table 6.3 Utilizationof Health Care Services Across Insurance Coverage, Health Status and Regions - Individuals Percentages 75 Table 6.4 Expenditure on Transportationto Access a Public or Private Health Care Facilities across Entities and Locality Types (KMpervisit) 78 Table 6.5 Out-of-pocket Expenditure on Health Care by Type of Services (KM) 80 Table 6.6 Out-of-Pocket Health Expenditure inPublic Facilitiesby Insurance Status (KM) 81 Table 7.1 Net Enrollment Rates for the Poor and Non-Poor 88 Table 7.2 Comparisonbetween Survey Data and Official Data on Total Students Enrolled inthe School Year 2000/2001 89 Table 7.3 Reasons for not Continuing Education (Secondary School Age children with Primary School Diploma only) 90 Table 7.4 Public Expenditure Per Student Relative to Per Capita GDP, 1999 (Percent) 91 Table 7.5 Incidence of Public Expenditure inEducation by Consumption Quintile and Type of School (Shares of Total and Spending in KM),2001 92 Table 7.6 Incidence of Public Expenditure inEducation by Consumption Quintile and Type of School (Shares of Total and Spending inKM), 2001 93 Table 7.7 Private Annual Expenditure in Educationby ItemLine, School Level, inKM 94 Table 8.1 Key Features of the Pension System inBiH 101 Table 8.2 The Role of Public Transfers in BiHfor PopulationWelfare: an Overview 101 Table 9.1 Simple Poverty Projections 117 Table 9.2 FullPovertyProjections: Effect of Changesinthe Mean Consumption and its Distribution on Poverty in 5 and 10years 119 Table 9.3 Types of Economic Growth and Poverty Reduction 120 FIGURES Figure A. Trends in Gender Differences for Labor Force Participation Rates inTransition Economies iv Figure B. BiH:Economic Collapse, Growth andEmployment vi Figure C. Simulations of Poverty reductionProspects ix Figure 1.1 Inequality in Selected Transition Economies (End 1990s- 2001) 8 Figure 1.2 Dimensions of Deprivation Among the Consumption Poor 11 Figure 2.1 Distribution of the Populationby the Level of Per Capita Consumption 20 Figure3.1 Trends in Gender Differences for Labor Force Participation Rates InTransitionEconomies 39 Figure 3.2 PanelA) Rates of Employment by Consumption Percentiles (Panel B) Ratio of Women's to Men's Employment Rates and Absolute Gap between Women and Men's Employment Rates by Percentiles 40 Figure 3.3 Unemployment Rate for Men and Women Age 15-59 by Consumption Per Capita Percentiles (Panel A) and by Age (Panel B) 41 Figure 5.1 BiH:Growth Rate Forecasts:InitialTargets and Revisions 59 Figure 5.2 BiH:Economic Collapse, Growth andEmployment 59 Figure5.3 Non-Agricultural Self-Employment as Percent of Total Non-Agricultural Employment 61 Figure 6.1 Health insurance Coverage by Income Quintiles across Entities 72 Figure 6.2 Proportion of Health Insured among the Paid Employed by Occupation 73 Figure 6.3 People that Mentioned an Economic Reasonfor not Seeking Care when Needed by Expenditure Quintile 77 Figure7.1 Educational Attainment by Income Groups, Cohort 16-25 Years of Age 87 Figure 7.2 Concentration Curves of Public Expenditure inEducation by School Levels 93 Figure 7.3 Percentageof Students MakingPayments for School Repairs and Maintenance 94 Figure 8.1 (A) Incidence of Benefits by Ex-Post Consumption Quintiles 104 (B)Incidenceof Benefitsby Ex Ante Consumption Quintiles 105 Figure8.2 International Comparisons of Social Assistance Efficiency and Effectiveness 109 BOXES BOX I LSMS Survey: Basic Consistency Checks 3 BOX I1 Measuring Welfare UsingLSMS 6 BOX I11 How Do This Study's FindingsCompare With Previous Analysis? 9 BOX IV MillenniumDevelopment Goals and Non-Income Dimensionsof Poverty inBiH 10 BOX V Gender Violence in Zenica: Findings of a Qualitative Study 13 B O X VI The Plight of Roma Returnees 14 Box VI1 Environmental Problems as a Time Bomb 15 BOX VI11 Informal Employment inBiHand Poverty: Cure or Disease? 26 Box IX The Invisible Face of the Poor: Roma inBiH 36 BOX X Female Headed Households inTraditional Societies 42 BOX XI Uneven Distribution of International Presenceand Regional Inequalities 50 BOX XI1 Costs of Postponing Difficult Reforms: The Case of Bulgaria 51 BOX XI11 Business is Difficult inBiH 65 Box XIV Mental Disorders InfluenceJob Opportunities, Accessibility of Health Care and Economic Development 70 BOX XV Recent Upswing inReturnee Numbers 122 ACKNOWLEDGEMENTS The report consists of two volumes. The first volume contains the main Poverty Assessment report produced by the World Bank team, integrating various data sources and inputs from local counterparts. The second volume provides an in-depth analysis of the LSMS survey, and i s a joint product of World Bank staff, and the staff of the Republika Srpska Statistical Institute (RSIS), the Federation of Bosnia and Herzegovina Statistical Institute (FIS), and the Bosnia and Herzegovina State Agency for Statistics (BHAS) (please refer to the second volume for details). Ruslan Yemtsov led the World Bank team producing this report, and was the key author of the report, with crucial contributions by Kinnon Scott (poverty profile analysis), and by Maria Eugenia Bonilla Chacin (health and education). The World Bank's team also included: Paulette Caste1 (social safety nets); Milan Vodopivec and Jacob Tomse (labor market and labor market policies); Lejla Somun (IBHI),Michael Lokshin, Tomas Mroz (University of North Carolina) and Patricia Poggi (gender and poverty); Rosanna Polastri (macro projections); Marjana Karahasanovic and Ozren Tosic (health sector). The team i s indebtedto the World Bank project and sector teams which commented on the draft version of the document and provided substantive suggestions, in particular to Nedim Jaganjac, Betty Hanan and Loraine Hawkins (health analysis), to Jim Stevens, Peter Darvas and Nick Burnett (analysis of education), to Reeta Klees (environmental risks), and to Eva Molnar and Seema Manghee (infrastructure and poverty), and to Steven Holtzman (refugee issues). Special thanks to Phillip 0'Keefe for help in consolidating comments received. Together with Kendra Gregson he also made valuable suggestions for the social safety net analysis. Peter Bocock provided outstandingly efficient and thorough help in editing the document. Irina Smirnov has provided support from the field. Samra Basic has provided very efficient support for the organization of the mission work and Tanja Bijev for the translations. Helena Makarenko has undertaken the processing of the document. The team i s indebted to many government officials from BiHfor their help throughout the preparation process. Various comments were received during the presentation of first results to the key stakeholders (Summer 2002), especially from Hasan Zolic, Marijan Vlaho, Slavka Popovic (BHAS), Slavko Sobot, Bogdana Radic (RSIS), Dervis Durdevic, Vedran Milisav and Munira Zahiragic (FIS). The team also held numerous consultations with the PRSP working groups in the period September 2002-August 2003. Special thanks to Zlatko Hurtic, who helped inlinkingthe report to the PRSP process. The team also benefitedfrom discussions with various international organizations, includingUNICEF, UNDP, IMF,and DfID. Given this collaborative approach the report was prepared over an extended period of time between March 2002 and August 2003 under the guidance of the World Bank Country Director for BiH Orsalia Kalantzopoulos, who took over this responsibility from Christiaan Poortman. Daniela Gressani and Asad Alam (Poverty and Gender Sector Managers) oversaw the process of preparing the report. N. Simon Gray (Lead Country Officer) provided encouragement and invaluable insights into the on-going dialogue with the Government during the entire period of report production. The peer reviewers were Martin Rama, Jesko Hentshel, and Phillip O'Keefe, who contributed substantively to the report's analysis and presentation throughout the preparation process. Andy Mason helped the team in mainstreaminggender analysis into the analytical work. The team i s also grateful to the Norwegian Trust Fund which supported the analysis of gender aspects of poverty. EXECUTIVESUMMARY Poverty Incidence 1. This report presents the most comprehensive analysis of poverty in Bosnia- Herzegovina (BiH) to date.' Calculations based on data collected in the 2001 Living Standards Measurement Survey (LSMS) suggest that poverty defined as lack of adequate consumption i s quite substantial, affecting every fifth citizen. Poverty i s present in both Entities and the number of poor i s split almost equally between them (Table A). Table A. Material poverty incidence, depth and severity by Entities Poverty Rate Poverty Depth Poverty Share of BiHPoor (%) Severitv FBiH 16 3 1 52 RS 25 7 2 48 BiH 20 5 2 100 Source: BiHLSMS,2001. See this report, Vol. I,Chapter I11for details. Notes: Poverty depth measures how badly off the poor are -how far below the poverty line their consumption levels are. Poverty seventy is a measure closely related to the poverty gap but giving those further away from the poverty line-the poorest-a higher `weight' in aggregationthan those closer to the poverty line. 2. Poverty i s substantial compared to some neighboring countries.2 Nevertheless, and despite the country's post-conflict situation and its low officially measured level of GDP, there i s no extreme poverty, as shown by the indicators of poverty depth and severity in Table A. 3. Inequality in material well-being also appears to be quite moderate by international standards (see Volume Iof this report, Chapter 111, Section E and Figure 3.1). Non-Income Dimensions of Poverty 4. The non-income dimensions of poverty are also generally not extreme in BiH: decent housing, access to water and sanitation, basic health care and education are all within the reach of the majority of the poor. This i s a very positive-and surprising-outcome in a country devastated by civil war. It can be traced to the successful implementation of the internationally endorsed reconstruction program, to highly developed and sophisticated informal safety net arrangements, and to effective private coping strategies. 'The second volume of this report thoroughly assesses the quality of data and methods used to analyze poverty. It shows that the report adopts a methodology that provides robust and accurate results. BiHdoes not have the critical data needed to produce internationally comparable poverty statistics. Nevertheless, on the basis of national poverty lines derived with the same methodology, poverty in BiHexceeds the rate in Croatia (8 percent in 1999) and FYR Macedonia (16 percent in 1999),but i s much lower than in Kosovo (50 percent in 2000). For reference, GDPper capita in BiH is estimated at around USD 1,200, while that in neighboring countries for the corresponding year was as follows: FYRMacedonia, around USD 1,800; Croatia, around USD 4,500; Kosovo, around USD 850 (Data are from the World Bank WDI[2002]). 5. At the same time, the analysis based on LSMS data finds sizeable deprivation in non- income dimensions of poverty. Thus, > > An astonishing 27 percent of adults inBiHcan be considered "education poor"3; > Health poverty affects 16 percent of the population4; Housing conditions for 11 percent of the population are below the poverty > standard;j Finally, and most importantly, 29 percent can be called "rights-poor", defined as belonging to households without established property rights for their housing. Deprivation of wider human rights, in terms of physical safety from violence and crime, i s also a significant problem. 6. There i s some good news on this front. Multiple deprivation across a number of dimensions i s not widespread; in fact, not a single individual interviewed for the LSMS i s simultaneously poor in any three of the above listed dimensions. But this fact also has negative connotations: since poverty in different dimensions affects different people, many BiH citizens are poor at least in one dimension: 72 percent of all adults in BiH are poor at least in one key aspect of well-being, i.e. with respect to material consumption, education, health, housing, employment or property rights. 7. Poverty in each dimension may look low only when BiH i s compared to other countries with similarly low GDPs; it i s too high compared to neighboring economies (poverty in BiH i s higher than in all countries of the former SFRY for which data are available), and compared to the population's expectations and aspirations. Hence, poverty reduction needs to be the top priority for the Government. The data on the different aspects of poverty also make it clear that an effective poverty reduction strategy has to be multidimensional. Vulnerability to Poverty I s Widespread and TakesMany Forms 8. The analysis finds that a sizable proportion of the population faces the risk of falling into poverty. Thus the percentage of the population classified as poor on the basis of current consumption levels i s only a fraction of those in BiHwho must worry about, and struggle to avoid, becoming poor at some point in the relatively near future. The report estimates that a typical household in BiH (couple with two children with employed head of the household) faces a substantial 10percent risk of falling into poverty in any given year. 9. The analysis identifies risk factors that may thrust a non-poor householdinto poverty: k Precarious earning sources, because the predominant living arrangement features > a single-earner household; Widespread health risks, including uninsured financial risks associated with becoming ill,and lingering post-conflict risks (minefields, violence, high accident rates); With unfinished primary education. Having serious physical limitations due to illness, being disabled or unable to work. Living in a partly destroyed building, or without running water and proper sanitation 11 > > Violation of human rights, discrimination and corruption; > Limited geographical mobility; Limited access to formal safety nets and reliance on personalized informal > arrangements; A worn-out stock of householdassets and limitedaccess to credit. 10.Among these risk factors the report strongly emphasizes the importance of non- income vulnerability with respect to health, human rights and limited mobility. The fight against poverty needs to include a wide spectrum of policies to tackle vulnerability-related issues. 11.Poverty risks are not shared equally. Vulnerability is much higher than average for particular groups, including: those living in rural areas, internally displaced persons (IDPs) and refugees, households with poor education, and women. The poor, and especially the very poor, are characterized by a cumulation o f risk factors. Vulnerability Makes Many People Feel Poor 12.Widespread vulnerability explains why many people in BiHreport feeling poor when asked about their own assessment of their living standards. The gap between perceptions and reality suggests that people also lack information with which to assess their circumstances. Regular monitoring of poverty and living standards would help to improve the quality o f information. Similarly, better and more regular communication from the Government about the nature, objectives and progress of reforms would offer a more realistic understanding of the reform path. WhoAre the Poor? 13. The report analyzes the profile of the poor in detail. It finds that poverty in BiH has five distinct characteristics: (i) Inno other country of the former SFRY for which recent data are available do children have as high poverty risk as in BiH. Poverty in Bosnia and Herzegovina has a young face: around a third of all poor individuals are aged under 18. And 2/3 of the poor live in households with children (Table B). Table B. Poverty and Number of Children (below 14) inHousehold Poverty Rate Poverty Depth Poverty Share of BiHPoor Severity No children 12 3 1 31 1child 20 4 1 22 2 children 27 6 2 27 3 and more children 56 16 6 20 Total 20 5 2 100 Source: LSMS BiH, 2001. See Vol. 1 Chapter I11for details. Notes: Poverty depth measures how badly off the poor are -how far below the poverty line their consumption levels are. Poverty seventy i s a measure closely related to the poverty gap but giving those further away from the poverty line-the poorest-a higher `weight' in aggregation than those closer to the poverty line. ... 111 (ii)Thoughjobless households makeupahighproportion of the poor, most poor households include someone who does actually work, usually a single earner. Inthese households, the income of a single eamer i s either insufficient to pull a household out of poverty, or leaves it vulnerable to the risk of job loss (Table C). Table C. Poverty and Numberof EarnersinHouseholdswith Children Percent Poor Poverty Depth Poverty Share of All Poor Severity No employed household member 49 13 5 31 One informally employed earner 36 9 3 17 One formally employed earner, owner 23 5 2 39 or farmer Two or more earners 17 3 1 13 Source: LSMS BiH, 2001. Notes: Poverty depth measures how badly off the poor are -how far below the poverty line their consumption levels are. Poverty severity is a measure closely related to the poverty gap but giving those further away from the poverty line-the poorest-a higher 'weight' in aggregationthan those closer to the poverty line. (iii)Female participation inthe labor market is lower inBiHthan inany other transition economy (Figure A). FigureA Trends in Gender Differences for Labor Force Participation Rates inTransition Economies 1 00 0 80 0 - 000 g9 040 2 0 20 ow Source: World Bank Gender DatabasecitedinPaci (2002).See Vol. IChapter111, Section Dof this report. While women have gained ground in the area of education, they still participate much less frequently in the labor market than men. The economy of BiH thus loses vital human capital. The analysis shows that female participation i s especially low among the poor. Thus this is a poverty issue as well as a gender issue. (iv) Poverty i s strongly concentrated in particular communities, which suffer from economic dislocation and social cleavages that combine to reduce social capital in iv these communities. Often these communities also have a high concentration of IDPs and refugees (Table D). Table D. Concentration of the Poor inBiHAcross Poor and Non-Poor Communities. Poverty rate and composition Community poverty rate, Average Share of Share of compared to national mean poverty rate IDPshefugees BiH Share of (%) inpopulation population BiH poor Less than $5 of national poverty rate 1 10 47 4 Between $5 and average poverty rate 13 20 17 11 Between average and twice poverty rate 29 17 14 21 More than twice national poverty rate 58 27 21 63 National 20 16 100 100 Source: LSMS BiH, 2001. See Vol. IChapter 111for details. (v) The majority of the poor live in households where the household head has only primary education or less, suggestingthat a significant number of working age adults have inadequate skills (Table E). Table E. Poverty and Education of Household Head Poverty Rate Poverty Depth Severity Share of BiHPoor None or Unfinished Primary 28 9 3 12 Primary 26 6 2 47 Secondary General 10 2 1 7 Secondary Vocational 18 4 1 32 Junior College 10 2 1 2 University 2 0 0 1 Total 20 5 2 100 Source: LSMS BiH, 2001. See Vol. I,Chapter I11for details. 14. Poverty in childhood can permanently affect children's future. It can undermine the integrity and stability o f their families, their health, their schooling, and their likelihood of either participating in the labor force or becoming economically and socially marginalized. The report examines in detail the barriers the poor face in educating their children and providing them with adequate child care. 15. The report's analysis o f the poverty profile concludes with the observation that the poor in BiH are made up of two groups. Poverty in the first group (a minority among the poor) i s due to a combination of causes and factors that push households into poverty. The second group (a majority among the poor) shares many characteristics with a typical household in BiH and i s in poverty often dues to a single factor or cause, sometimes unique to a particular household. This latter group has more chances to benefit from economic growth and employment generation, while the former group may be so deep in the poverty that may stay poor even despite some improvement in its living standards generated by growth. 16. Thus, policies to reduce poverty have to be multifaceted and include general economic policies to prevent currently non-poor falling into poverty in the future, specific policies to help the poor to benefit from growth, and targeted policies to support those who are likely to be left behind. Causes of Poverty in BiH 17. Poverty i s caused by limited opportunities, lack of security and powerlessness. With respect to opportunities,economic growth in BiH has been significant since 1995, but has had little impact on employment since 1998 (Figure B). Labor market participation i s still below the level needed to have a decisive impact on poverty. Figure B BiH: EconomicCollapse, Growthand Employment _ _ - - -- .-.- -- __- -____ - -- - - or, 1990 1991 1992 1593 1994 I595 1996 1997 1998 1999 2" 2031 ............. ..... ...-.... ~ I-Formal Employment, 1990=1M) 4-Percapita GDPincurrent US$, 1990=100 ~ Sources:IMF, World Bank, See Chapter V. 18. Meanwhile, private consumption far exceeds what might be expected from the level of output in the economy. Thus far, highlevels of consumption have been sustained thanks to large inflows of international aid. But this situation cannot continue forever. The poverty rate could rise sharply in the future in the absence of major progress in achieving sustainable growth. Trying to sustain growth rates sufficient to improve the living standards of the poor represents a major challenge. And if inequality were to increase, the benefits of growth accruing to the poor would be reduced. 19. The relatively low level of labor market participation in BiHstems from two mutually reinforcing problems - limited demand for the skills and assets of the population, and the inadequacy of supply of slulls and assets by the population. Demand i s limited by the lack of dynamism in highly localized, constricted and fragmented labor markets, together with an inadequate business environment and in particular low levels of private sector investment. The supply of adequate skills i s limitedby problems with human capital formation (including vi quality of education, the prevalence of a range of health risks) and the failure to match people with new jobs. Both these aspects of sluggish labor market performance militate against opportunities throughout the economy, but their adverse consequences bear especially heavily on the prospects o fpoor people to find goodjobs at decent wages. 20. Poor performance in generating new employment also reflects worryingly slow progress in establishing BiH as a "single economic space.'' This fundamentally political problem i s ultimately responsible for reduced opportunities, including opportunities for poor people. Factors Limiting Opportunitiesand Security: Health Care and Education Provision 21. Effective health and education services can substantially improve both opportunities and security for the population as a whole and for poor people in particular. In BiH, despite success in rebuilding health and education infrastructure, a large gap still remains between the delivery o f health and education services and the needs of a modern market economy. This gap has led to uneven allocation of scarce resources in general, and growing problems of access, especially for the poor. Lack of progress in improving the quality and efficiency of services has been accompanied by an increasing tendency to shift costs to users - which in turnleads to the exclusion of the poor from these services. 22. Health and education provision remain largely dominated by the public sector. It i s supply-driven, divorced from modern labor market requirements, and offers few incentives for achieving much needed efficiency. Despite high overall spending, serious problems of access persist - ranging from pre-school education that i s unavailable to poor children (who need it most to offset their relatively deprived backgrounds) to low enrollment rates in secondary education, and from the prohibitively high cost o f medical treatment for most Bosnians to a widespread perception of uncertainty and exposure to risks related to catastrophic illnesses. 23. In health care, substantial public and donor resources support broad coverage and reasonable standards. However, inefficient treatment protocols, duplication o f services, and neglect of primary health care make the system not only one o f the most expensive in the ECA region in terms of its share of GDP, but also one that provides inadequate services to the poor, especially to those who are not insured. The high costs of the system also crowd out resources that could otherwise be used for measures to support poverty reduction more effectively (for a detailed discussion, see Chapter VI). 24. The education system i s even less efficient in terms of its impact. It provides largely irrelevant skills, and it i s the poor who face the most serious long term consequences of its inadequacies (see Chapter VII). vii Other Sources of Insecurity, and the Role of Safety Nets 25. Physical security remains an issue (though it i s not as acute a problem as it was immediately after the conflict). This i s a predictable consequence of social disorganization, combined with erosion of the authority o f the State and traditional institutions of social control. These factors have contributed to an increase in the incidence of crime, accidents, and disease. The report demonstrates that physical insecurity affects all groups of society and creates a pervasive environment of uncertainty that undermines confidence. 26. Economic security is also a problem. It has many dimensions, including uncertainty regarding basic property rights (particularly with respect to housing property rights among those displaced by the war); corruption; and failure to adequately protect the population against economic and non-economic risks, such as disability and old age. 27. In the area of social protection, substantial increases in efficiency could be achieved by creating a better designed and targeted safety net. Substantial reallocation of spending from untargeted "categorical" benefits, supported by appropriate incentives to reduce dependency on transfers, i s needed (see Chapter VIII). Features of Powerlessness and Voicelessness 28. Finally, powerlessness and voicelessnessimpede the capacity o f poor people to break out of a vicious circle o f impoverishment. Powerlessness and voicelessness in BiH reflect the unsatisfactory nature of institutional relations and connections between people (poor and non-poor), between poor people and government, and between poor people and civil society. Poor people are generally excluded from the institutions that make the decisions and allocate the resources that affect their lives. Further, in an effectively decentralized system of resource allocation, powerlessness in BiHi s primarily linked to the lack of social capital and weak community mobilization in poor areas, along with disruption of social capital in communities affected by the war. Meanwhile, violation of human rights, including violation of the basic right to life during the years of conflict, together with persistent physical violence (including crime, recurrent armed attacks and violence against women) i s a troubling element in the mix of non-income dimensions of poverty. How to Reduce Poverty: Selectivity, Stability and Growth 29. Poverty reduction depends on attacking its key causes: limited opportunities, lack of security and powerlessness. As the report shows, these are all unintended consequences of past or current policies in BiH. Achieving better outcomes will require major change. 30. A key message of this report is the need to change the focus of policy making with respect to poverty, based on a review o f the assumptions underlying it. Current policies have implicitly assumed that the majority of the population i s poor, and therefore that any action that produces benefits for the population at large helps with poverty reduction. This assumption i s exemplified by the 1998 UNDP Human Development Report which claimed that two thirds of the population was poor. It needs to be reassessedin light of the data on VI11 ... poverty presented in this report, with a view to making use of more selective, targeted approaches where appropriate. 31. Nevertheless, the report emphasizes that the majority of the poor still share many characteristics with a typical household. This argues for the retention of aspects of a "universalist" approach, albeit with policies that differ from the current ones. For example, if even a typical householdhas a 10percent riskof falling into poverty inany particular year, then it i s important to make sure that every household has access to social protection mechanisms. Given that a majority o f the population considers itself poor, universalism would also be more acceptable than targeting from a political perspective. It will thus be important to give careful consideration to the trade-offs between universalism and selectivity. 32. A core policy goal that will be important for successful poverty reduction will be maintaining a stable macroeconomic environment-especially prudent monetary policy and responsible fiscal management. Such an environment will help to reduce the risk of falling into poverty among the near-poor. It will also help BiH to meet the major economic challenges the country faces over the coming five years. N o scenario, even the most optimistic, suggests that private consumption growth rates will exceed 3 percent a year; under more realistic assumptions, the rate i s likely to hover around 1percent p.a. There are no major gains to be realized quickly from foreign trade reform, and prospects for FDI inflows are not encouraging. Meanwhile, international assistance i s on the decline. All these factors underscore the importance of unwavering commitment to maintaining macro stability. 33. Policies that promote an appropriate future growth path will also be o f central importance. The report reviews possible sectoral growth paths and find that different growth scenarios offer dramatically different benefits to the poor (Figure C). Investment in agriculture and manufacturing, together with continued efforts to rehabilitate infrastructure, are most likely to benefit the poor. The benefits from broadly shared growth will be substantial, but poverty will most likely remain an ongoing concern. Figure C Simulations of Poverty ReductionProspects 1- With StatusGoPolicies tWith-go-hchgrowth A WithFro-PoorGrowtQ 25 1 ... ..- . .. - L h p> 20 a-< -----------. d&i -5 I S - -I 10- B2 5 - A Poverty now Poverty in 10years Source: Simulations basedon LSMS 2001, See Vol. IChapter IXfor figures and details. ix How to Reduce Poverty: Supporting the Implementation of BiH PRSP 34. BiH is in the process of completing and start the implementation of its national Poverty Reduction Strategy Paper (PRSP). Reform commitment i s comprehensively expressed in the second draft of the PRSP, which was publishedby the authorities for a third round consultation at the end o f June 2003. The PRSP lists the following four policy objectives conducive to accelerated poverty reduction: 1. Achieve financially sustainable growth that would rapidly enable to country to reach and surpass its pre-war production capacity. 2. Ensure social stability by reducing poverty and unemployment. 3. Create the conditions for full membership in the EUbefore the end o f the decade; and 4. Create the conditions for greater participation of BHinthe global economy. 35. In order to achieve these development objectives, the PRSP identifies a set of policy priorities including: (i)maintaining macroeconomic stability; (ii)implementing a fiscal reform program; and (iii) establishing sustainable systems of social welfare and services. The analysis of this report strongly supports the multi-pronged approach proposed in the PRSP. Our analysis shows that it i s impossible to make a significant impact on poverty in each sector in isolation. 36. In addition to supporting the PRSP's priorities, this report underscores a number of areas that will be of special importance for successful implementation of BiH's poverty reduction strategy. As already noted, it draws attention to the need to assess options for adopting a targeted as opposed to a purely universalist approach to policy design, and to the need for macroeconomic stability. It emphasizes the importance o f ensuring mobility and adequate infrastructure in order to strengthen the business environment. I t suggests that the issue of gender i s central to the problem of poverty in BiH, and that gender considerations will need to permeate many elements o f the action plan to implement the strategy. In this connection, one of the key criteria for the success of poverty reduction policies will be improving labor market outcomes for women, especially including progress inraising female labor force participation rates. Finally, it offers a detailed analysis and action proposals for social sector reform, as discussed below. How to Reduce Poverty: Agendafor Reform in the Social Sectors 37. The sections that follow summarize findings and reform proposals with respect to health care, education and social safety nets in BiH. 38. Health care reform needs to take account o f the following issues: a. There i s are many uninsured people inboth Entities. Most o f them are poor. b. Utilization of health services is subject to differential access between the insured and the uninsured. c. Thus, there are problems with access and equity in a system that i s supposed to assure universal coverage. X d. The poor allocate a greater proportion of their consumption to health care than the rich, and thus are more exposed to financial consequences of health risks. e. The prevalence of large out-of-pocket expenditures negatively affects access to health services. f. The existing system is expensive andrequires substantial public resources. 39. Based on this diagnosis as developed in the report, health sector reform should include as priorities moving towards puttingthe system on a more solid financial footing and making access more equitable. 40. Specifically in the area o f finances, the large proportion of the population engaged in informal activities makes it impossible for a payroll based health insurance system to cover all citizens. Policy makers should revise the mechanisms that currently exist to cover those groups who currently do not fall into any category of the insured, and should enforce contributions from those who can and should pay for them. For instance, more efforts should be made to minimize non-payment of health contribution by firms and institutions in the formal sector. In the case o f agricultural workers and the self-employed, ways need to be found to develop arrangements to permit them to contribute to and benefit from the system. Finally, approaches need to be developed to take into account the capacity o f the budget to cover the currently uninsured. 41. Improving access will also involve addressing the issue of the large out-of-pocket expenses that can result from the fragmentation of the system and the lack of portability of health insurance. Great progress has been made toward reducing fragmentation with the creation of the Federal Solidarity Fundin the FBiH, and the Inter-Entity agreement that allow persons who live in one Entity but work in another to receive health services in their place o f residence. However, much remains to be done - first, to ensure complete portability of health insurance across health jurisdictions, and second, to lessen the problems created b y small and unsustainable pools o f risk. 42. More generally, the governance of the system as a whole needs to be improved. The authorities in BiH are already allocating substantial resources to health care. Enhancing equity will require them to use these resources more effectively. This implies improving coordination among health jurisdictions of the kind recently achieved in the reform o f the health insurance, and urgent action to ensure a more efficient mix of levels of care, facilities, and staff. Better coordination with the private sector i s also essential. Private facilities are providing an important share o f health care provision, and the prevalence of large out-of- pocket payments suggests that many households are willing to pay for health care in both the private and public sectors. The public sector might then be able to withdraw from the provision of some o f these services to the better-off and use the resources saved in assuring a more equitable access to health care for poor households. 43. The education sectorpolicy agenda needs to respond to the following concerns: a. The extremely low proportion of children of the relevant age attending pre- school. xi b. The fact that primary enrollment rates are still lower than they were before the war. c. The low proportion of children continuing on to secondary school and the marked differences in secondary school enrollment across income levels. d. The large proportion of secondary school children enrolled in vocational or technical schools instead of in general academic schools (which only about 20 percent of students attend). e. The highly regressive distribution of public expenditure in secondary and tertiary education, with most of public funds financing education for the rich. 44. Policy options to promote a pro-poor orientation of the education system include: Pre-school education: It i s unrealistic to plan to dramatically increase publicly- subsidized pre-schools. The reduction of the primary school entrance age to six years throughout BiH, combined with targeted remedial programs, such as core subject tutoring in the first year or two of primary education, are more sustainable options. Primary and Secondary Education: Policy makers should work towards universal primary school enrollment and attendance. To ensure that the poor have equal access to secondary education, policy-makers should focus on two main constraints that households face: economic barriers to access and the lack of relevance of (and, hence, interest in) education. With respect to economic barriers, the provision of subsidized textbooks and education materials for poor students could be increased (on the basis of a family means test ). With respect to improving relevance, more broad-based skills that can be used in a wider range of jobs such as computer literacy, foreign languages, team work, problem solving, communication, and learning ability are competencies that need to be emphasized in arapidly changing market economy. Inthe latter years of secondary vocational education, specialization should focus on a few broad-based cross-occupational families (services, manufacturing skills, management, etc.) rather than a proliferation of highly specific occupations, which should to be the responsibility of private training providers, public post secondary institutions, or employers themselves. Higher Education: The current politically-driven fragmentation in higher education financing and organization i s inefficient and effectively prevents qualitative improvements in programs. Important steps towards reducing inefficiency and providing an equitable distribution of resources include: (a) establishing Entity-level management of higher education funding on the basis o f a per-student funding formula; (b) consolidating autonomous university faculties into modem universities; (c) converting some existing universities/faculties into community college-like institutions that offer short-term programs more relevant to the current labor market; and (d) given the regressive incidence o f public spending in higher education, gradually increasing fees and eliminating currently xii non-means tested fee-exemptions to students on the basis o f high scores on entrance exams. Education Finance: Finally, education authorities should ensure a more progressive distribution of public resources earmarked for education, especially those going to secondary and higher education. It i s important that geographically disadvantaged jurisdictions do not depend solely on their own resources to finance education in their localities. The need for some revenue redistribution in favor of education should be taken into account in the context of assigning tax revenue to various levels o f governance in BiH, including decisions on the assignment o f the VAT tax currently under consideration. 45. Social safety net reform needs to address the following problems: a. Insufficient resources are allocated to specific poverty targeted benefits aimed at the very poor, as opposed to non-targeted "categorical" benefits; b. Very limited attention and resources are devoted to programs that provide insurance or coping assistance to mitigate shocks (temporary assistance, unemployment benefits). 46. Targeted programs command an inadequate share of overall social protection expenditures, while substantial resources are misallocated to non-poor beneficiaries in politically driven social programs (such as veterans' benefits). This report advocates a serious reallocation o f resources to address more adequately the needs of the poor. 47. The report finds that locality of residence i s strongly associated with poverty. Hence, programs specifically directed to poor communities are the most efficient way to improve targeting, and need to be incorporated into the delivery mechanism o f some poverty reduction programs. 48. Benefit provisions inherited from the past, which feature multiple types o f benefits managed by different agencies (with programs that often duplicate, or even conflict with each other) need to be re-evaluated. The process of reforming them should rely on improved governance of the system, better coordination and specific targeting. 49. There are fiscal and capacity-related limits on the provision of safety nets that would cover all citizens in BiH. On the fiscal side, expansion of social benefits would require compensatory cuts in other public spending. With respect to capacity, the current system i s not particularly efficient in delivering benefits. Operating within these limits, the following agenda i s proposed: Develop and legislate a new Social Protection Strategy in both Entities with an emphasis on: (i)determining a realistic core set of entitlements, with clearly identified financing sources; (ii)ensuring that Entity budgets finance one or more core "poverty benefits" equitably within each Entity; (iii)creating incentives for adequate allocations for social protection by local governments; (iv) redefining Xlll ... institutional responsibilities on the basis of a functional review of social protection arrangements. To support this strategy, develop a set of projections focused on the fiscal and social implications of the on-going reforms of the pension system, veteran's programs and other social transfers. Reform the cash transfer programs with the following priority goals: (i) thatensure protection for disabled veterans and survivor families i s better focused on the most disabled, while reducing the share of public spending on veterans' programs; (ii) improve the impact and efficiency of the social safety net for the poorest, inparticular social welfare benefits and services; (iii) improve the efficiency and equity of pension and disability programs; and (iv) ensure that social policy reforms are done on the basis of better information, both within the system and to the public, and that their impact on poverty and social welfare i s monitored effectively and timely. Monitoring and Evaluation 50. One of the driving principles of the PRSP approach i s a focus on results, specifically including the selection of appropriate monitoring indicators against which progress can be measured. The ability to produce statistics needed for design, implementation and evaluation of economic and social policy in BH has been damaged by the war and transition. Three institutions are formally responsible for statistics in the country, two entity level Institutes and a State level agency. The statistical offices have limited capacity to produce the range of statistics necessary for policy-makers. The area that has been especially hard hit by the war i s that of data collection at the household level. The lack of reliable data not only hampers efforts to develop and monitor policy, but contributes to significant lack of transparency in the operations of public social policy institutions. 51. With support from the Bank and other donors, the Statistical Institutes have made major progress in developing a new representative household survey, based on the Living Standards Measurement Survey (LSMS) model. The joint poverty profile analysis conducted as part of the preparation o f this report has been a critical input to PRSP development. As a follow-up of the survey, first panel round of the LSMS has been completed in 2002, and field work for the second round o f the panel initiated in late 2003. While providing data for compiling monitorable indicators in some important dimensions of living standards (such as access to health and education, unemployment rate, housing conditions etc.), the panel survey does not include data on material welfare as measured by the consumption of households. It therefore cannot be used for poverty monitoring. 52. This report proposes a framework to for monitoring poverty. For monitoring and evaluation to be effective, the Government needs to improve coordination among the national and Entity-based statistical institutes and other data collectors with a view to creating countrywide statistical data. Critical steps in this effort will include launching regular household budget and labor force surveys. 53. Using the household budget survey data (conducted in BiH in 2003-2004 with the support of the Italian Government) for poverty monitoring will ensure a continuity in data xiv collection and tracing the progress of the PRSP implementation. It will require a special effort to make newly collected data fully comparable to the baseline poverty data from LSMS. The Statistical Agency and Institutes will play a key role in this effort assisted by local researchers, NGOs and international partners, such as the World Bank, UNDP and DFID. To ensure a broaduse of the newlycollecteddata, the program of capacity buildingi s now currently being implemented in close coordination between the above mentioned agencies. xv PART I.POVERTY PROFILE. A solid understanding of the magnitude and depth of poverty is essentialfor designing poverty reduction policies and actions and for setting realistic targets for public action. This Part presents findings on the extent of poverty in BiH on the basis of the first representative household survey data: the Living Standard Measurement Study Survey, conducted at the end of 2001. 1. MEASURINGPOVERTYINBIH The chapter starts with a description of the survey, and discusses its basic results with respect to welfare levels and non-income dimensions of well-being. Despite thefact that inequality is moderate and there is no measurable extremepoverty, poverty in BiH is significant, affecting one citizen in five. The analysis of non-income dimensions of poverty confirms the diagnosis based on material deprivation: poverty in BiH is a major public policy concern. A. Background: Why a Poverty Assessment? 1.1 A Poverty Assessment6 for Bosnia and Herzegovina (BiH) is long overdue. Until recently there were no reliable data about poverty inBiH. Yet poverty in has been known to be a serious issue: the country emerged from the devastating war of 1992-95 with drastically lower living standards and major disruptions across society. Out of 4.4 million of the prewar population, over 250,000 were killed or registered as missing. Around half of the population were forced out of their homes. Out of the 900,000 people employed in the civilian economy in 1991 only about 72,000 were able to keep their jobs by the end of the ~onflict.~Assets - housing, automobiles, farm equipment, householdpossessions- were lost or destroyed. Bank deposits were eroded or frozen. A generation lost critical years of schooling. Access to health services was interrupted, and social protection programs were disrupted. War damaged the very core of community- and family-based social networks. Poverty Assessments have been conducted by the staff o f the World Bank for its member countries since the 1980s, and are now among the most comprehensive and influential of its analytical instruments. B y synthesizing information on living standards from various sources, Poverty Assessments: provide details for each country on the number and characteristics of the poor and the causes of poverty; link the observed poverty profile to public policies, expenditures and institutions; and make proposals to the Government about how to reduce poverty. Their intended audiences have typically been the Board o f Directors of the World Bank and the Government of the country. Over time, and especially after the initiation o f the Poverty Reduction Strategy Papers (PRSP) process, the way in which Poverty Assessments are done has changed. Poverty Assessments are now conducted by the Bank in partnership with countries and donors and are intended to provide a critical input for country-owned PRSP processes. BiHDemobilization andReintegration Project, World Bank. 1 1.2 Since the war critical infrastructure has been rebuilt under the 1996-1999 Priority Reconstruction and Recovery Program (PRRP), endorsed by the BiH Government and the donor community. Economic growth has been rapid; real output in 2001 was 180 percent of its 1995 level. Nevertheless, per capita GDP - measured at current exchange rates - was only around USD 1,270 by 2002, just over half its level in 1991. Important structural reforms still remain to be completed,* while foreign assistance i s declining. 1.3 To address the challenges of sustaining the reform process, the BiH Council of Ministers adopted in 2001 the Economic Development Strategy.' Simultaneously, work was initiated on drafting the Poverty Reduction Strategy Paper (PRSP), and the Coordination Board for Economic Development was established to develop a PRSP in an open, participatory process. The Government's strategy, with its focus on achieving social and economic sustainability by creating a favorable environment for entrepreneurship, provided a key policy context for the PRSP. 1.4 This study i s fully aligned with the Government priorities - which include building more reliable poverty information in the overall context of working to achieve social sustainability - and directly contributes to reducing "the paucity of poverty data" identified in the World Bank CAS (2000) as an impediment to effective policy formulation. The PRSP and the Government's Economic Development Strategy recognize the problems created by inadequate data on the well being of the population. The three statistical organizations in the country (the State Agency for Statistics for BiH and the statistical agencies for its two constituent Entities - the Republika Srpska Institute of Statistics-RSIS, and the Federation of Bosnia and Herzegovina Institute of Statistics-FIS) conducted the LSMS survey on which this study i s based, and engaged in the concurrent process of capacity building for poverty monitoring and analysis. 1.5 The information contained in this report i s the result of a collaborative effort on the part of the Government of BiH and the World Bank, and offers a baseline against which future progress in addressing poverty can bejudged and evaluated. B. Datafor Poverty Analysis 1.6 Understanding poverty starts with the collection of comprehensive household-level (micro) data on the living standards of the population. In 1999, the three statistical organizations began work on the design of a Living Standards Measurement Study (LSMS) Survey, and carried it out in late 2001 for a large sample o f 5,400 households: 2,400 in the Republika Srpska (RS) and 3000 in the Federation (FBiH)." 1.7 The LSMS survey collected information from households and individuals in the country on their income and consumption levels, access to and use of public services, The year 2000 saw a marked slow-down in economic reform (IMF,Press Release December, 22,2000). For the period up to 2004, entitled "Entrepreneurial Society". loThe fieldwork for the LSMS survey was carried out between 26 September and 23 November 2001 as ajoint project of the three statistical organizations with financial and technical support from DfID,UNDP,the Japanese Government, and the World Bank. More details are provided in Volume 11, Chapter 1. 2 economic activities, housing, and other important characteristics. The survey sampling frame provided representative data *for BiH as a whole, and for each of the two Entities (RS and FBiH). The survey coverage and accuracy are unprecedented inBiH(see Box I). BOX I.LSMS Survey: Basic Consistency Checks Comparison of survey data with other reliable sources of information i s an important validity check on the survey results. While making such comparisons, one has to recognize that survey data differ from other types of information sources. It i s misleading to say, for example, that the LSMS found that 38.2 percent of households in BiH have cars; a survey can never provide such a precise estimate. In fact what the LSMS found was that car ownership in the sample was 38.2%; generalizing to the population (see Chapter 1 in Volume I1for details), we infer based on this result that between 35 and 42 percent of households in BiH have cars (or 38.2 percent +3.6%.p.). Estimates from the LSMS are reported here accompanied by a confidence interval (k) around the mean (95% statistical confidence). Bearing this in mind, LSMS estimates o f key socio-economic characteristics match well with the non-survey administrative records. For example: o The number o f registered employed and unemployed obtained from the LSMS i s consistent with administrative sources: Total registered employment in FBiH at the and of 2001 was 407,000; the LSMS data give an estimate of 375,000 G5,000* "oficially " employed. In the RS, registered employment was 220,000; the corresponding number from the LSMS is 260,000 G3,OOO. Similarly, by the end of 2001 there were 424,000 registered unemployed in BiH; according to the LSMS survey there were around 450,000 244,000 registered at the labor oficer. o The average pension reported in the LSMS and the estimated total number o f public transfer recipients match data from the PensionFundand other sources: The average pension reported in the survey (by area) matches the Pension Fund (PF) data (taking into account the clearance in 2001 of pension arrears from previous years). The number of pension and veterans' benefit recipients in the LSMS matchesfairly well with the oficial data. In both Entities, only slightly fewer respondents declared receiving pensions and veteran benefits than PF and budget data suggest: in RS 245,30OGI,200 recipients in a survey against 261,833 in the oficial data; in FBiH 380,200 G2,900 in the survey vs.417,612 according to the oficial data. o The estimate o f the number o f persons displaced by the war in the LSMS is consistent with administrative data: According to UNHCR by the end of 2001 there remained 443 thousand displaced person (DPs) in BiH. In the LSMS survey, 399,000240,000 declared themselves as DPs. Further, according to UNHCR statistics of cumulative minority returns in 1996-2001 (returns of refugees and DPs to their original place of residence in an area with a different dominant ethnic majority) there were 287,227 such cases. Extrapolating the LSMS results, gives around 270,000G4,000 persons with a returnee status, with a split between RS and FBiH closely mirroring the oficial numbers Comparison of household consumption from the LSMS with National Accounts (NA) data i s difficult, owing to incomplete NA data. Per capita annual personal private consumption (defined consistently with NA principles) i s estimated at 3,050+140 KM (Convertible Marks) on the basis o f the LSMS. NA data total (which include not only personal, but also general government) annual per capita consumption for 2001 can be estimated at 3,100 KM (using the same population numbers). On this basis, the LSMS data would yield an effectively much larger number. The most likely explanation is significant underestimation by the NA of the informal economy, which i s captured by the LSMS consumption data. Indirectly the presence of a large informal economy i s confirmed by the LSMS data on income: respondents did not declare sources of income averaging 40 percent o f their reported cash expenditures. There is in fact a broad consensus that the existing NA and other personal income data significantly underestimate the level o f GDP, and revisions are inprogress with substantial technical assistance. Finally, regarding price levels, the LSMS confirmed that there has been a convergence in prices between Entities: inNovember 2001 the food price level was only four percent lower inthe RS compared to FBiH. Sources: LSMS 2001, LDB 2002, UNHCR, State Agency for Statistics for BiH, the RS Institute of Statistics-, and the FBiHInstitute of Statistics. Volume I1 of this report. Nore: * The margins of error shown for the L S M S data are 95 percent confidence interval for means adjusted for stratification and intra-cluster correlation 3 1.8 While the LSMS remains the main data source on poverty available to date inBiH, other sources o f data were also used inthis report.l1 C. Living Standards and Poverty Lines 1.9 A central feature of this study i s establishing a baseline for measuring of living standards and poverty in BiH. Modern views of poverty focus not only on material deprivation but also on deprivation in many other dimensions. Although this report treats poverty as multidimensional (see Section F), many of its non-material aspects-such as the psychological pain of being poor, low achievements in education and health, vulnerability to shocks, and a sense of powerlessness-are in fact closely linked to material poverty. This is because material resources12 are critical instruments with which to tackle unexpected occurrences, to achieve better health and education outcomes, and to empower individuals to escape from social marginalization. When such material resources are missing, households are often forced to devise. coping strategies-such as neglecting spending on health-which lead to greater future vulnerability and deprivation. 1.10 To examine material poverty, one needs to measure material well being and establish a poverty line. The methodological and practical aspects o f this are discussed in Chapters 2-4 of Volume 11. This Chapter only reviews basic concepts and results. 1.11 The existence of a well-defined poverty line helps to increase public awareness of the particular circumstances of the poor and how they are changing over time. BiH inherited from SFRY a food subsistence basket for a standardfamily offour members developed more than a decade ago by the Statistical Institute of the then Federal State. This food basket i s still used sometimes as a poverty line in BiH (e.g. by the UNDP Early Warning System). Analysis undertaken for this report, however, revealed that in its current form and application it is an inaccurate measure of poverty. First, it includes only food items, when non-food goods and services are clearly also essential for even basic living standards. Second, its composition by food items i s outdated. Finally, in its current form it i s applied in an arbitrary way to all households regardless of their size and composition. Living arrangements vary greatly in BiH, and to assume that different households have the same needs as the standard famiZy offour does not make much sense.13 1.12 The derivation of the new poverty line was undertaken in two stages. At both stages the concern was to reflect accurately country-specific actual circumstances in BiH at the end of 2001. First, a minimum food basket was chosen to obtain a food poverty line. Then, to obtain the full poverty line, an amount was added to the food poverty line to allow for essential UNICEF Multiple indicator cluster survey (MICS) on health, nutrition and education, the World Bank anti- corruption diagnostics, household and enterprise level surveys, household level data from a qualitative survey of local institutions, qualitative studies (a Social Assessment and the Voices of the Poor study), and the results o f a regular attitudes survey conducted by PrismResearch for the UNDPEarly Warning System (results published on a quarterly basis for 2000-2). Qualitative research was also conducted by IBHIspecifically for this study, inorder to gain understanding of perceptions o f gender issues among the poor. '*Public as well as private. l3The LSMS found that only nine percent of the BiHpopulation live in such "typical" households. 4 nonfood consumption. Volume 11, Chapter 3 of this report provides details on both steps of this analysis. 1.13 The value of the new minimum food basket in 2001 was found to be KM 760 (per person per year in average BiHpri~es).'~ This value can be considered as an extreme poverty line, as it represents a bare minimumneeded for physical survival. 1.14 The full poverty line, including both food and non-food components, was derived by estimating the level of total consumption (including non-food items) o f households whose total food consumption i s equal to the extreme poverty line.15 Households with food consumption at this level have a total consumption level that amounts in average prices to 2,198 KM per person per year. This level i s used as a general or "absolute" poverty line in this report. It has a very clear meaning: a household with consumption below 2,198 KMper person faces absolute deprivation either in essential food or essential non-food needs, or inboth. 1.15 In addition to the absolute poverty line, Section D below also uses relative poverty lines. A relative poverty line defines poverty compared to a "typical" national living standard. EUcountries publish poverty figures utilizing relative poverty lines, and countries that aspire to join the EU wish to publish poverty statistics based on relative lines for comparability. However, as discussed in Volume 11, when the calculation of a poverty rate and its monitoring over time are the primary goals, as in BiH, an absolute poverty line set in an objective and scientific manner, has to be used. 1.16 It was recognized from the beginning that consumption was a more accurate indicator of material well-being than income, the latter being subject to large under-reporting. The definition of welfare in BiH i s highlighted in Box 11. Those households whose consumption levels are below the value of the poverty line are considered poor and those whose consumption exceeds the poverty line are considered non-poor. 1.17 The poverty line developed in this way i s a cut-off for consumption levels as measured in the LSMS, and it is not applicable to any other data source. Different types of data may require a re-definition of the poverty line to produce comparable results. In some countries the poverty line serves as a reference income level to which state transfers are linked. Those with incomes below the poverty line are entitled to various publicly provided benefits. Functions of critical importance for poverty reduction policy and action (monitoring and targeting) make the exact value of the poverty line very important, and may require a re-definition o f the BiH poverty line in the future; the LSMS consumption poverty line cannot be used for these purposes.16 l4Itis informative to compare this figure to the official minimumfood basket. As of November 2001 the cost of the official minimumfood basket was equal to 1413.6 KMper capita per year inFBiHand to 1337.2 KMper capita per year inRS. The new minimumfood basket based on LSMS data amounted to 780 KM inFBiHand to 735 KMinRS, - i.e. approximately only half the value o f the official minimumfood basket. l5Itis noteworthy that food costs represent only 34.5 percent of the general poverty line, confirming that non-food costs constitute an essential component o f consumption, even for the poor. l6For example, if one uses income instead of consumption for the LSMS, one has to set the poverty line at a lower level to account for underreporting o f income; such an income poverty line was estimated to be 418 KMper 5 BOX 11. MeasuringWelfare UsingLSMS. This Poverty Assessmentrelies on measuring material well-being, and follows three principles: 1. using household consumption, which is the most reliable indicator of living standards in the BiH context, collecting data that take into account seasonality, constructing a comprehensive consumption measure that includes imputed values for consumption in kind (own production, gifts, housing), and using local market prices for valuation; it also excludes expenditures not directly contributing to higher well being (health care expenditures, household business spending, investment); 2. to account for regional price differences the annual consumption aggregate of each household is adjusted to average BiHprices; 3. per capita measures are used as a baseline; a per adult equivalent scales were found to be unsupported by household level data and are used only to check for the robustness of the main conclusions in Chapter 6 of Volume 11. Source: Volume 11, Chapters 2,3 and4. D. Poverty Numbers 1.18 Table 1.1 below shows the incidence of poverty on the basis of the absolute poverty line calculated for this report and two relative poverty lines. The data show that poverty i s quite substantial if measured by the former and the higher of the latter, affecting up to 20 percent of the population. The table does not report extreme poverty rates (based on food poverty line only), as these were found to be zero in the LSMS data. Relative (50% Relative (66% Absolute of the median) of the median) (baseline) Head Count of Poverty, percent of population 6.9 16.7 19.5 Statistical precision range percentage points k2.3 r3.6 k3.6 1.19 Zntemational comparisons using relative poverty lines. The two commonly used relative poverty lines shown in Table 1.1yield poverty estimates of between seven percent and 17 percent (Table 1.1). How do these figures compares with those for other transition countries? BiH i s in the middle of the range, with relative poverty higher than in the Czech Republic, Hungary, Slovenia and Croatia, on a par with Romania and Albania, but lower than inPoland, Bulgaria, Macedonia, the Baltic States andthe countries of the CIS.17 capita per year at average BiHprices - almost five times below the general consumption poverty line. Note that this line will produce accurate overall poverty incidence given the current overall response behavior of the population, but will not be able to reveal the true characteristics of the poor, as all those who are unwilling to report their income will figure as poor. This i s why consumption i s a superior indicator of poverty. l7Using the World Bank study "Making Transition Work for Everyone: Poverty and Inequality in Europe and 6 1.20 Intemational comparisons of absolute poverty. Comparing poverty incidence using national absolute poverty lines derived with the same methodology i s a useful consistency check. From this perspective it i s noteworthy that the incidence of absolute poverty in BiH significantly exceeds absolute poverty rates in Croatia (8 percent in 1999) and Macedonia (16 percent in 1999), two countries o f the former Yugoslavia for which a similar methodology was applied to survey data.18 These findings suggest that measured poverty in BiH i s in line with expectations, as its level of GDP per capita i s below that of the other countries in this comparison (Macedonia's GDP per capita was around USD 1800, while that of Croatia was around USD 4,500). 1.21 The World Bank in Making Transition Workfor Everyone: Poverty and Inequality in Europe and Central Asia (2000) developed two special internationally comparable poverty lines for Eastern and Central European transition economies ($2.15 and $4.30 per person per day at purchasingpower parity). In applying these lines to a particular country it i s critical to convert national currencies to dollars using a purchasing power exchange rate - a statistical index reflecting the price level in a country compared to international prices - not the current exchange rate. There are no PPP data available for Bosnia and Herzegovina, making it impossible to estimate poverty based on international poverty line." E. Znequality 1.22 Inequality directly affects the share of income or consumption available to those at the bottom of the distribution, and i s therefore a critical determinant of the extent of poverty in an economy. The higher the level of inequality, the more widespread and deeper poverty will be at a given mean income level. Thus it i s important to measure how unequally consumption and income are distributed in the population. Figure 1.1below reports a key inequality statistic,20 the Gini coefficient, for BiH and several other transition economies. As inequality measures can be quite sensitive to the definition of a welfare measure, every attempt was made to use definitions inthe international comparisons that are as comparable as possible.2' Central Asia", 2001. l8 Data for Croatia are from the World Bank report Croatia: Poverty Assessment; data for Macedonia are from the national PRSP. Inboth cases poverty lines are based on minimumfood baskets and adjustments for non-food consumption, derived from actual consumption patterns. Note, however, that the resulting absolute poverty lines may (and should) in fact represent different standards of living, as poverty lines tend to be higher in real terms, the higher the level of economic development (see Ravallion, 2001 for details). One might consider taking as an approximation an average ratio of PPPs to current exchange rates from countries in the former Yugoslavia for which such data are available (FYR Macedonia, Slovenia and Croatia). But the values differ widely: from 219 percent inMacedonia (meaning that one could buy more than two times more inthe country exchanging a dollar into the national currency than using this dollar outside the country) to 135-150 in Croatia and Slovenia. This example shows that using a market exchange rate to estimate the international poverty line is totally erroneous. I t also shows that one needs the exact PPP data to do this estimation with sufficient precision. *'Since inequality can be defined indifferent ways, there are many statistics that measure it. The Gini coefficient i s perhaps the best-known inequality statistic. It ranges between 0 (perfect equality) and 1(complete inequality).The Gini i s most sensitive to inequality inthe middle o f the distribution. Nevertheless, the data come from different surveys and from different years. Hence, the resulting inequality statistics may not always be perfectly comparable. 7 1.23 It is noteworthy that across all countries in South Eastern Europe andthe CIS measures of inequality based on incomes exceed those based on consumption. This gap i s particularly large for BiH, Bulgaria and Tajikistan. Bearing in mind the problem of pervasive under- reporting o f income, the consumption-based indices are clearly likely to be more reliable; they are also of more interest in the context of this report, because it relies on consumption-based measures. Figure 1.1 suggests that consumption based inequality in BiH i s in line with other countries in the same region: it i s lower than in Macedonia and Croatia, and very similar to Bulgaria and Hungary. On a broader regional scale, inequality levels are definitely below those observed in CIS countries. Figure 1.1Inequality inSelected Transition Economies (end 1990s- 2001). hicom-bascd 0 Consurqtion-based Sources: For BiH: staff estimates based on BiH LSMS 2001. For Macedonia: recent estimate from 2000 data, obtained in the PRSP working group. For Bulgaria: Poverty Assessment (2002).For other countries the World Bank study "Making Transition Work for Everyone: Poverty and Inequality in Europe and Central Asia", Annex D. F. Depth and Severity of Poverty 1.24 So far we have used only the simplest and most common measure of poverty: the headcount index. This measure, however, does not tell us whether the poor are only slightly below the poverty line or whether their consumption falls substantially short of it. The headcount measure also does not reveal whether all the poor are about equally poor or whether some are very poor and others just below the poverty line. 1.25 To highlight the complex distributional aspects of poverty, we measure the depth (measured as the poverty shortfall or the poverty gap) and the severity of poverty.22Poverty depth measures how badly off the poor are - how far below the poverty line their consumption levels are. For BiH, the poverty gap, based on LSMS data and our baseline definition of absolute poverty i s equal to 4.6 percent (kl.0). A corresponding measure of an average shortfall of 24 percent means that mean consumption of the poor falls 24 percent short of the poverty line. Poverty severity i s a measure closely related to the poverty gap but giving 22For further discussion, see Chapter 5 inVolume I1of this report. 8 those further away from the poverty line-the poorest-a higher `weight' in aggregation than those closer to the poverty line. Its level in BiH i s found to be 1.6 percent (k0.5). These data suggest that the depth and severity o f poverty are not extreme, consistent with the absence of extreme (food) poverty and moderate inequality. BOX 111. How DoThis Study's FindingsCompare With Previous Analysis? Prior to the LSMS, there were no consistent, representative and comprehensive household welfare data for the country as a whole. Earlier surveys did not allow generalization of results or did not collect detailed information on material living conditions. Reliance on reported income is a common feature o f these data. The resulting poverty estimates deviate substantially from estimates based on a detailed household consumption survey such as the LSMS. As a result, previous estimates o f poverty inBiH are known to vary widely: The UNDP's 1998 Human Development Report was the first comprehensive analysis o f well-being of the citizens of the country as it recovered from war. There was no micro data, and the report used various official sources to assess the extent and structure of poverty in 1997. Each group with average recorded incomes below a rough cut-off value equivalent to the minimum pension was considered poor, - a very imprecise assessment dictated by the lack of better data. This method of assessment yielded a poverty level amounting to around two- thirds of the population. Unfortunately despite being outdated and having no solid backing, this number i s the most often quoted estimate o f poverty inBiH. The World Bank's 1999 Poverty Note was based on the first micro dataset from a survey (conducted by CEIT) comparable across the Entities. Unfortunately, CEIT collected information on welfare using a very short questionnaire, which made the assessment o f poverty very approximate: it estimated that around a third of the country's households were poor (below the relative poverty line). The UNDP's Early Warning System Reports 2000-2002 are based on special quarterly on-going small-scale surveys. These reports indicated that many households declare themselves as living on very low incomes. In a baseline survey conducted on May 2000, 17 percent of households had declared incomes below KM 200 per month per household, which was considered as a rough poverty cut off; by the end of 2001 the number of people inthe lowest income group decreasedslightly, to 14percent. Sources: UNDP 1998 Human Development Report "Poverty and Inequality in Bosnia and Herzegovina: The Legacy of the War", World Bank, December 1999. See also M. Bisogno and S. Chong "Bosnia and Herzegovina: Labor Market Study". Mimeo. World Bank, 2000 Bosnia and Herzegovina."Consultations with the Poor". Poverty group, PREM,World Bank and "Voices of the Poor" for World Development Report2000/01; "Early Warning System", UNDP October-December2000, andquarterly, 2001. G. Dimensions of Poverty 1.26 Different aspects of poverty - income and non-income - interact and reinforce each other in ways that often exacerbate the deprivation that poor people face. Poor health status and low educational achievement not only decrease well-being, but also limit people's income- earning potential. Lack of voice and participation feeds exclusion, and can bias the distribution of public spending against the poor. 1.27 Recent advances in understanding the causes and consequences of poverty have led to the concept of social exclusion, which, according to the definition of the 1998 Eurostat Task Force on Social Exclusion and Poverty Statistics, i s : ... a dynamic process.. which ends up " with persistent multiple (deprivation) disadvantages. Individuals, households and spatial units can be excluded from access to resources like employment, health, education, social or 9 political life." (Eurostat Task Force 1998). The internationally endorsed Millennium Development Goals (MDGs) also draw on this concept (see Box IV). 1.28 Defining poverty as multidimensional, however, also raises the question of how to measure these different dimensions. There are no strict and agreed standards suitable for all countries. This Chapter uses the LSMS data and other sources of information to obtain individual or household-level measures of deprivation inthe following dimensions: > Education and Literacy P Social Inclusion andHumanRights P Health, NutritionandSanitation 1.29 In addition, Chapter 2 focuses on vulnerability to poverty and subjective perceptions of welfare (or lack of it) as special dimensions of poverty. BOX IV. MillenniumDevelopmentGoals andNon-IncomeDimensionsof PovertyinBiH In September 2000, 147 heads of State and Government - and 191 nations in total - adopted the Millennium Declaration; discussions were held to develop a comprehensive set of indicators, or Millennium Development Goals (MDGs). The MDGs are examples of a multidimensional approach to poverty. Income poverty i s listed first, but each o f the goals addressesan aspect of poverty which is important initself, and because it interacts with other aspects of poverty. These eight global goals - each to be achieved by 2015 - include: Reducing poverty and hunger Achieving universal primary education Promoting gender equality Reducing under-five mortality Reducing maternal mortality Reversing the spread of HIV/AIDS and other diseases Ensuring environmental sustainability Developing global development partnerships for aid, trade, and debt relief These goals represent widely accepted international development objectives. The globally defined MDGs, however, are goals and targets that are to be adapted and integrated into each country context. Using LSMS data and the development priorities as reflected in the BiHPRSP, the following set of indicators to track poverty in its many dimensions was developed: 1.Consumption poverty (defined in Section C of this Chapter). 2. Education poverty, differentiating between extreme (illiteracy) and basic education poverty (defined as being above 15 years old, but not in school and having no education or only incomplete primary schooling). 3. Health and nutrition poverty spans a range of indicators from the ones describing status (individuals who are suffering a major physical ailment precluding their normal independent functioning, disabled and unable to work), to process (inadequate, poor nutrition) and access to services (individuals unable to use medical services which they need). 4. Housing poverty (which i s defined in a broad sense: households not connected to piped water and using latrines, households living ina partly destroyed building or one unsuitable as a dwelling). 6. Rights poverty is a broad notion and can be measured with an array of indicators, such as incidence of crime, corruption, use o f political rights, data on discrimination and lack o f protection for basic property rights (in this narrow latter sense the indicator i s defined as a share of households with uncertain housing status, e.g. illegal and temporary occupants, households without documents confirming their ownership o f housing). Each of these indicators can be further disaggregated by gender to track the gender equality goal. Source: MillenniumDeclarationon `Development andPoverty Eradication', PRSPfor BiH and LSMS. 1.30 Identifying new dimensions o f poverty i s important in the context of understanding the profile of vulnerable groups because: 10 > > Households that are not income poor may be poor inother dimensions; Same householdsmay suffer from multipledeprivations, thus constituting "the hard core" poor. 1.31 The application of this multidimensional concept of poverty to LSMS data (see Box IV for definitions) shows that substantial proportions of people in BiH suffer from non-income aspects of poverty: >> An astonishing 27 percent of adults inBiHcan beconsidered"education poor"; Healthpoverty affects 16 percent of the population; P Housingconditions for 11percent of population are below the poverty standard; P Finally, and most importantly, 29 percent of the population belong to households without establishedproperty rights for their housing. 1.32 The data also suggest, however, that deprivation in several dimensions is rare. Infact, not a single individual interviewed for the LSMS i s simultaneously poor in any three of the listed dimensions. This i s strong confirmation of the absence of extreme poverty in the country. But as poverty consists of non-overlapping or only partially overlapping groups, the majority of the population i s poor at least in one dimension. As many as 72 percent of all adults in BiH are poor in terms of one of the following: material consumption, education, health, employment, housing or property rights. 1.33 Figure 1.2 illustrates the point by focusing on the consumption poor. The Figure reports the share of the consumption poor in BiH suffering from a particular kind of deprivation. The longer the bars, the more widespreadthe problem among the consumption poor. Figure 1.2 Dimensionsof DeprivationAmongthe ConsumptionPoor 0% 10% 20% 30% 40% 50% 60% 70% 80% Source: BiH LSMS 2001. Poor individuals are definedaccordingto our baseline:per capitaconsumptionand absolute povertyline. 1.34 Figure 1.2 suggests that (other than with respect to schooling) only relatively small minorities of the consumption poor also suffer from other forms of deprivation. The majority 11 of the consumption poor in BiH are literate, socially active, and have a diet that assures that their basic needs are met. But as subsequent sections will demonstrate, specific groups of the population - women, children, the elderly, rural populations, IDPs and refugees, and minority returnees - face higher risks of deprivation in different dimensions of poverty. The analysis will also show serious problems in each dimension. Most of these are related to outcomes that are difficult to measure at the household level (such as environment), but are nevertheless as important for well being as deprivation of consumption. (i) Education Poverty:Legacy of the Past? 1.35 As much as five percent of the adult population of BiH is illiterate, one of the highest illiteracy rates in transition economies.23 When defined as having difficulty reading a text, illiteracy, as shown on Figure 1.2, reaches 10 percent among the poor (and 8 percent among all adults). Not surprisingly, as illiteracy i s a legacy of the past, it i s concentrated among the oldest age cohorts. Within these cohorts, however, females and the poor are particularly prominent: a staggering 59 percent of pensionable age females from poor households are illiterate or read with difficulty. Less than one percent of the population under age 50 i s illiterate; in fact, prevalence i s so low that any comparison by population categories i s precluded. I s education poverty simply a legacy of the past? 1.36 The LSMS found that at least 5 percent of primary school age children are not enrolled in school. Thus, poverty ineducation is not only a legacy of the past. This issue is analyzedin Chapter VII, which will show that much of the inequality in accessing secondary and tertiary education in BiHreflects problems in accessing adequate primary education. As Figure 1.2 has illustrated, lack of education i s a major issue for the consumption poor, about two-thirds o f whom have completed primary schooling or less. (ii) Human Rights 1.37 The war in BiH meant a denial to the population of the most basic human right - the right to life. Close to six percent of the pre-war population of BiH were killed or reported missing.24As documented in numerous UNHCR reports, security concern and widespread violations of the human rights of ethnic minorities remained real problems until very recently.25 1.38 Public concern has recently been rising about criminal and other forms of violence - including robberies, armed attacks, and sexual assaults. The UNDP Early Warning System surveys reveal that every month between 15 and 20 thousand people in BiH are exposed to 23 According to the LSMS, among the CEE countries for which data is available, only Albania and Kosovo have higher illiteracy rates among adults (WDI, 2001 and Kosovo: Poverty Assessment 2001). 24 Including also the increase in mortality rates, the casualties were 269,800 inhabitants Demographic Changes in Bosnia- Herzegovina from 1991 to 1998. See in the Intemational Forum of Bosnia,Document no. 1, Return of Displaced Persons and Refugees as a Conditionfor the Survival of Bosnia-Herzegovina, Sarajevo,January 1999. 25 For example, UNHCR (2000) reported that during one month of 2000 there were 50 reported incidents of ethnically motivated violence, including throwing of hand grenades, arson of newly constructed returnee houses, shootings of automatic rifles at returnees, and desecration of graveyards. Recently the number of such incidences has been reduced (UNHCR (2002)). 12 physical violence or attacks on their property. Combining data from different surveys,26 the Poverty Assessment team confirmed that the incidence of violence in BiH i s high: as much as 15 percent o f the population lives in households whose members were victims of serious crimes involving property or violence over the last three years.27 All income groups experienced similar levels of violent crime, but property crime disproportionately affected the rich and the poor, with the middle quintiles experiencing much lower levels of victimization. This U-shapepatternholds true for both urban andrural areas. 1.39 Despite being subject to similar risks of violent crime, the rich and the poor have very different responses with respect to seeking help. While the rich report almost 80 percent o f property crime incidents to the police, the reverse i s true for the poor, who report only 20 percent of cases. The most frequent explanation offered i s the belief that reporting a crime will not help to recover the property, which in turn highlights another dimension of poverty in BiH: the lack of legal protection available to poor people. 1.40 Household violence against women i s a major problem across the world, and i s a critical concern with respect to women's human rights and health. It typically goes hand in hand with poverty. A study on gender violence in BiH strongly suggests that it disproportionately affects women from poor households and i s linked with other aspects of deprivation in a post-conflict environment (see Box V). BOX V. GenderViolence in Zenica: Findingsof a Qualitative Study Medica, an NGO operating in Zenica, conducted a series of surveys and in-depth interviews with 542 women in the Zenica municipality, followed by a study in2000 in Banja Luka and Sarajevo. I t looked at different forms of assaults and coercive behavior, including physical, sexual, and psychological attacks, as well as financial manipulations against a partner in an intimate relationship. I t found that domestic violence i s not typically experienced as an isolated incident, but as an on-going cycle of abuse. Overall a quarter of the women surveyed reported that they suffered from systematic violence in their family or in their relationship, and one half reported knowing other women who suffer similarly. Half of the women who report being beaten say they are beaten daily. O f particular concern is the fact that the young women reported being subject to violence as often as did other age groups, and that their experience of violence usually began before marriage. Domestic violence is not caused by alcohol or drug abuse. Rather, both alcoholism and domestic violence are symptoms of serious social dysfunctions caused by a common set o f factors, such as poverty, unemployment and war trauma. The qualitative study o f poverty and gender conducted by IBHI noticed an increase in violence within the family after the war. The victims are mostly women, and less often children. Education of women and their partners i s the strongest violence-reducing factor (more educated women tend to be more active in terminating relationship in which they are subject to violence; they also have higher status in a relationship). The study did not discover any differences in the pattern of violence across Entities. Source: To Live Without Violence. Second Look 2. Final Report, Zenica, Medica, Infoteca 1999, and Violence Against Women. Women to Women -Udruzene-Medica. BanjaLuka, Zenica, Sarajevo, 2000. : QualitativeSurvey of Gender and Poverty. IBHI, 2002 26 Anti-Corruption Diagnostic survey contained detailed block on question o f crime victimization. This dataset was supplemented with imputed data on consumption for each household. The imputation used variables which are similar across datasets: demographic characteristics, education, type o f location, employment and detailed list of consumer durables; it estimates household consumption as a function o f these variables inthe LSMS data; obtained parameters were used to predict consumption for each household in the anti-corruption survey dataset. For discussion of methodology see: Poverty and Social Developments in Peru, 1994-97. The World Bank 1999. 27Unfortunately, the Anti-Corruption Diagnostic survey is not comparable to internationally conducted surveys on this issue- see UNICRIPublications on International Crime Victim Surveys, www.unicri.it. 13 1.41 The displacement of the population during the war created a major rights problem in the form of uncertainty over property rights in general, and rights in housing in particular. This i s a very substantial aspect of measurable deprivation in BiH. While matters have improved greatly since a few years ago when the proportion was twice as high- a clear sign of the success o f reconstruction efforts - about 30 percent of the poor are still in temporary occupation of somebody else's property28or have uncertain tenancy status. The percentage for IDPs i s an alarmingly high78 percent. 1.42 UNHCR (1999) demonstrated that the process of establishing property rights put onerous demands on claimants until relatively recently, making IDP and refugee households headed by single females particularly ~ u l n e r a b l eThe . ~ ~ situation has improved recently. 30 Discrimination and denial of rights affect ethnic minorities that are traditionally regarded as second class citizens, such as the Roma (see Box VI). 1.43 Finally, qualitative surveys conductedin BiHhave emphasized the lack of prospects for youth and the limited nature of social and cultural life. These factors are often seen as reasons why very young people turn to drug or alcohol abuse, or perceive leaving the country as their only hope for the future. The UNDP Human Development Report on Youth identified cultural deprivation as a serious policy issue inBiH. BOX VI. The PlightofRoma Returnees. Many of the Roma of BiH were displaced during the ethnic war. Like other IDPs they usually settled in houses abandoned by their owners, who had fled to other areas. When the owners claim their property back, the Roma, like other IDPs, must go back to their homes, where they often find their own properties either destroyed, mined or occupied by other IDPs displaced from other areas. In the north-eastern Bosnian city of Bijeljina (RS), a vast majority o f the local Romani community left within days in early 1992, either out o f fear or as a result o f expulsion. During the war, 168 Romani houses were destroyed, even though there were no armed conflicts in Bijeljina. Ten years later, with around 1200 Roma back in Bijeljina, only 5 1Romani houses have been returned to their pre-war owners, and more than 700 houses belonging to Romani families are still illegally occupied by IDPs. Huso Beganovic, a 46-year-old Romani man from Bijeljina, tells his story: "My house was occupied by a 5-member Serb family. We asked them to let us in, even in a part of it, but they refused. In August 1999, we received a Court decision on the return of our property. However, when we came home, the Serb refugees did not want to move out. Iwent to the municipal office of the Ministry for Refugees and Displaced Persons, asking for help. The officers either refused me, or promised help in a couple of months. After two and a half years o f such life, the Serb refugees let us finally move into the first floor of the house. A t that time, my brother Huska and his family returned to Bijeljina, and as they were not able to move into their own house, Ireceived them in. So, we were 24 persons altogether, living in30 square meters". Source: The EuropeanRoma Rights Center (ERRC),Roma Rights ,nr 2, 2002. see: http://errc.org . 28 See Box VI for examples o f conflicts arising in the process of restitution. 29 In the FBiH,proceedings over property or inheritance rights had to be initiated in the municipality of pre- conflict address, raising potential security concerns for single women without male support. Inaddition, proving one's property rights i s an emotionally painful matter for spouses o f missing people, who need to declare them as deceasedin order to inherit the right to their pension or to pre-conflict property. 30 Laws facilitating the return of property were passedinboth Entities in 1998, and implementation began in earnest in 2000. The process was more efficient in the FbiH,where property had been reinstated in 32% of claims. InRS the property repossessionprocess is slower; although the percentage of casesruledon was 34%, e only 15% o f claims have successfully ended inproperty reinstatement. Data are from UNDPNational Human Development Report 2002. UNHCR reported that by 31 January 2002, 66,552 families (5 1percent of all claimants) had recovered their property inFBiH,and 36,901 (32 percent) had done so in RS. 14 (iii) Health, Nutrition, Environment, Sanitation, and Housing 1.44 Poverty can be seen as deprivation of basic capabilities such as the ability to live in decent housing, to have a healthy environment, to avoid undernourishment, and to escape unnecessary morbidity and premature mortality. The LSMS confirms that hunger i s not a concern in BiH: all households were found to be able to afford the minimum food basket. But the suffering during the war may have left deep public health wounds. Deprivation of food in childhood in particular i s proven to have long term consequences for development, imposing irreversible losses. Research has demonstrated that malnourished children are more likely to get sick and suffer from chronic diseases later in life, have lower learning achievement and be less productive as adults. 1.45 Responding to such concerns, UNICEF carried out in 1999 the first large scale representative survey done in BiH, the Multiple Indicator Cluster Survey (MICS). It was estimated that a sizeable 11percent of children below the age of 5 had slow growth disorders, although only 2 percent o f children had severe disorders. The MICS revealed that the real problem was in fact not famine but too much of the wrong kindof nutrition, leading to obesity. Based on the MICS, 21 percent o f those aged 12-23 months were found to be overweight, with 5 percent classified as obese. These data suggest that BiH faces a problem of nutrition quality, rather than quantity. 1.46 Chapter VI provides a detailed discussion of health status and access to health care. It shows that while overall qualified medical help i s within reach for the vast majority o f people inBiH,the poor have difficulty inaffordingpropertreatment, suffering from a vicious circle of illhealth,lowproductivityandinabilitytoadequatelyaddresstheirailments. 1.47 Even in the pre-war period, pollution and an unhealthy physical environment were major problems in BiH. The war created a number of new risks that directly or indirectly jeopardized health status, including inadequate waste disposal, an estimated 500 tons of outdated medical drugs, and the devastation of forests (see Box VII). Box VII. Environmentalproblemsas a time bomb A qualitative study on Gender and Poverty discovered that ecological problems were perceived as deeply alarming by a focus group in a site located in Prijedor. Concern focused on deposits of out of date medicaments and remedies from a nearby hospital. "We have up there, an atomic bomb, - a deposition of medicaments. Why did I mention this problem - because the deposition of medicaments has been existing for 5-6 years, those medicaments are out of date, placed in iron barrels, next to the Hospital, and up to date, only media have been informing about that ... these barrels have already begun to leak, so that the outflow of poisons will happen there.. .the sewage system, possibilities to provide the water for other countries and not to mention everything that refers to this." I t seems that local authorities (inspection services, decontamination services etc.) are inactive in dealing with this problem. Source: Qualitative Survey of Gender and Poverty. IBHI,2002. 15 1.48 Inadequate waste collection and disposal capacity i s a critical environmental and health issue, and affects the poor most heavily. The area coveredby collection services i s only about 60 percent in larger municipalities in BiH and much lower in more rural localities; not surprisingly, the poorest are more often cut off from service or find themselves unable to pay for it. Significant quantities o f waste are being discarded in unofficial sites such as unauthorized dumps, roadsides, small village dumps, rivers, and mines, posing a direct risk to public health. 1.49 Runoff and leachate infiltration from dumpsites are potential hazards in areas of BiH where groundwater aquifers provide sources of water supply. The most serious problems caused by such dumpsites include deterioration o f groundwater quality, destruction o f vegetative cover, and proliferation o f insects and rodents. 1.50 The problem o f improper water management was already acute in BiH before the war; and the water and sanitation system suffered severe damage during the war. Nevertheless, by1999, according to the results from the MICS survey, 98 percent of the population had access to "improved drinkingwater sources (mechanically protected from contaminants)". The LSMS confirmed that a large proportion of the population has access to piped water and sanitation systems. 1.51 Finally, with respect to housing conditions, only around 10 percent of the population suffer from "housing poverty" as defined in Box IV. In some respects, the situation i s particularly good. For example, not a single household reported problems with electricity supply. This shows that some elements of basic infrastructure are now being brought to minimally acceptable standard, and that the PRRP goals (World Bank, 1997) were largely achieved, 0 Rehabilitation of over 30 percent o f the housing stock inFBiHand 15 percent inRS; 0 Most electricity networks and lines are in good condition; 0 The production of electrical energy reaching over 80 percent o f the pre-war level; 0 Water supply reaching 90 percent o f the pre-war level. 1.52 After the war, over 400,000 people benefited from housing assistance provided by the PRRP. But where problems exist with respect to both quantity and quality of housing, they are mainly suffered by the poor. For example, while about 40 percent of the population as a whole lives in congested accommodation with more than 2 persons per room, the comparable figure for the poor i s 60 percent. Many Bosnians complain about the low quality of water, poor sewage systems, and their continuous leakage.32A substantial fraction o f the poor - almost one-third - have to carry water to their homes, and 40 percent are still using latrines. In rural 31Bosnia and Herzegovina: 1996-1998 Lessons and Accomplishments. Review of the Priority Reconstruction Programand Looking Ahead Towards Sustainable Economic Development. 1999 Donor Conference. 32Ina Study onLocal Level Institutions (The World Bank, 2002). The study also mentions that poor quality water caused a recent hepatitis outbreaks in Bihac and Tesanj. Actual data reported inthe PRSP and the national Environment Action Plan (NEAP)confirm these fears and concerns. I t is estimated that only 35 percent of total needs in potable water are covered. I t i s estimated that only 55 percent of the population is supplied with water which is regularly tested. And microbiological analysis of 10,496 samples conducted in2000 in public health laboratories in BiHfound that 18 percent were not suitable for human consumption. 16 areas, this rate i s as high as 60 percent among the poor, but i s only 20 percent among the rural non-poor. These findings do not, however, indicate a regionally atypical level of deprivation; the some of the findings noted here are in fact similar to observed conditions in much richer, neighboring Croatia.33 H. Conclusions 1.53 The development of an effective poverty reduction strategy requires a realistic assessment o f poverty. The LSMS i s the first dataset that offers such an assessment. It estimates material poverty as affecting around 19.5 percent of the population. The fact that this number i s much lower than some other less reliable but better known estimates listedin Box I11 should not be taken to mean that poverty i s not a serious problem in BiH. In fact, poverty in BiHi s a major public policy concern for at least three reasons. 1.54 First, absolute poverty directly affects every fifth citizen. This estimate relates to material poverty, but deprivation in non-income dimensions of well being, such as health or education, i s also significant. When poverty i s defined as deprivation in at least one o f these dimensions, it affects the majority o f the population. In a country that has experienced much higher living standards, nobody i s willing to accept the current extent of poverty as a long term proposition, making poverty reduction an urgent public policy priority. 1.55 Second, current relatively tolerable levels of consumption may not be sustainable in the long run. Decisive steps towards self-reliance have yet to be made, and the exceptional levels of international assistance received to date are on the decline. Given this outlook, today's twenty percent consumption poverty rate looks worrisome; if nothing i s done to address it, it may increase in the future. 1.56 Third, as repeatedly emphasized in this chapter, many people in BiH face real deprivation in one or more non-material basic needs, leading them to fall below an acceptable level of well being. To address these many dimensions of deprivation, a poverty reduction strategy needs to deliver progress in, e.g., health, education, human rights, housing, environment, social inclusion. This means that it has to be multi-sectoral innature. 33 In Croatia a quarter of the populationand half of the poor live in accommodations with morethan2 personsper room; one fifth of the poor and around5 percent of the total populationdo not haverunningwater indoorsor on their properties. - see Croatia:Economic Vulnerability and Welfare Study. World Bank2001. 17 2. VULNERABILITY TO POVERTY RISKSAND PERCEPTIONS Many people in BiH, even if not currently defined as poor, recognize that they are vulnerable to poverty. Exogenous events-a lost job, a delay in pension or wage payments, an illness or a bad hawest-could easily push them into poverty. This Chapter discusses vulnerability to poverty in BiH. In addition to the 20 percent of the population currently defined as consumption poor, a comparablefraction of the population is at risk of becoming poor in the fiture, because they are either exposed to risks or do not have sufJicient resources to cope with them. As a result, many people sufSer from an overwhelming perception of vulnerability. Vulnerability is a major concem because it forces many people to adopt risk-mitigating strategies that condemn them to low-retum and low-income activities. 2.1 Like poverty itself, vulnerability i s a multi-faceted concept. For example, in the context of income, vulnerability implies the risk that a household or an individual may experience an episode of income (consumption) poverty over time. But vulnerability exists in many other dimensions as well. For a child, it may be expressed in terms of the risk of being pulled out of school. More generally, it may involve vulnerability to being the victim of a natural disaster, o f crime, of arbitrary use of power by the State, including corruption or violence. 2.2 Indicators of vulnerability reflect both the extent of exposure to shocks and the ability to cope with them. Unfortunately the BiHLSMS could not draw on the panel data, needed to develop dynamic measures of exposure to risks at the household level. This chapter therefore relies on three indirect measures. First, vulnerability i s measured by exposure to economic and non-economic shocks. Second, vulnerability i s assessedby measuring the ability to cope with shocks. Finally, vulnerability i s measured based on perceptions of the population about it. Each of these methods has advantages and disadvantages, but together they provide a helpful picture of the extent and different forms of vulnerability A. Vulnerability to Poverty as Exposure to Economic and Non-Economic Risks 2.3 The extent o f exposure to risk depends on the degree of volatility of the economic and social environment in which a household operates, and reflects many factors. But whatever the extent of volatility, those with low levels o f consumption are particularly vulnerable. In determining vulnerability to poverty, this indicator needs to be complemented by analysis o f 19 exposure to key causal factors associated with risk: unemployment, health and violation of rights. (i) Dangerously Low Consumption 2.4 Poverty i s not a simple `yes-no' status. Measures of poverty depth and severity show that a poor household may be more or less poor depending on its level of consumption. And a household may be just above the poverty line, thus classified as non-poor, but face the imminent risk of falling into poverty with even a small fluctuation in its internal or external circumstances. 2.5 To illustrate this point, Figure 2.1 shows the distribution of the population in BiH by the level of real per capita consumption. The figure plots the level of current consumption (vertical axis) for cumulative percentages of the population ranked from the poorest to the richest (horizontal axis). The interesting point about the Figure i s the flatness of the consumption profile: most people who are neither extremely rich nor extremely poor have broadly similar levels of consumption. The distribution i s therefore quite compressed in the middle, puttingmany people close to the poverty line. Figure2.1 Distributionof the Populationby the Levelof Per CapitaConsumption. 1 4.9771111 , .I .2 I I .3 I .4 I .5 I .6 I .7 I .8 .9 I 1I Per capita consumption and poverty line +/-50% Share of populat. poor->rich Note: The figure represents weighted percentages of the population ranked from the poorest to the richest (from left to right), and the corresponding level of annual per capita consumption (in logs). The figure also reports the level corresponding to the general national poverty line (horizontal line in the middle) and poverty lines higher and lower than the general line by 50 percent (upper and lower lines). Source: LSMS 2001data set. 2.6 Because so many people in BiHseem to be "bunched" just above the poverty line, the graph also draws a poverty line that i s 50 percent above the general absolute poverty line to capture the extent of this bunching.34 This higher poverty line yields a total incidence o f poverty of half o f the population. This implies that around 30 percent of people in Bosnia 34 Taking 50 percent in this case is effectively arbitrary, but is inline with common practice inother countries. It is regularly usedboth for illustrative purposes and for statistical robustness checks (see World Bank (2003)- PRSP Source Book). 20 and Herzegovina could be considered near poor, and therefore vulnerable to economic shocks that could pushthem easily to poverty. (ii) Employment Risk 2.7 Analysis of vulnerability to poverty in other countries suggests that the loss of employment i s the leading and the most significant factor affecting household well being. 35 Opinion polls show that the Bosnian public i s overwhelmingly preoccupied with unemployment, which i s rated as one of the most important issues determining how people vote. 2.8 The most direct measure of the risk of employment loss is the unemployment rate. For someone who i s employed now, a high unemployment rate today implies that it would be difficult to find a new employment incase the current one terminates. Table 2.1 reports data on unemployment from the LSMS based on two different definitions-( 1) the "economic" definition used in labor force surveys (i.e. being without work, willing to work, actively looking for a job and ready to take it) and (2) the "administrative" definition, (i.e. those registering themselves as ~nemployed)~~-fora number of countries in transition. The comparison shows a relatively high rate of economically defined unemployment (a more accurate gauge than the "administrative" definition) in Bosnia as measured by the LSMS: over 16 percent of the labor force in BiHi s unemployed-although the percentagei s broadly in line with those for Bulgaria, Croatia and Yugoslavia, and substantially lower than that for Macedonia. Unemployment rate -LFS 16.4 16.9 15.1 30.5 7.7 12.6 Unemployment rate - registered 38.0 (FBiH) unemployment 40.1 (RS) 17.9 22.4 54.1 10.8 26.8 35 "Making Transition Work for Everyone: Poverty and Inequality in Europe and Central Asia", World Bank 2001. 36The differences between these two measures and their interpretation in the context of BiH are fully discussed inthe Labor Market InPostwar Bosnia And Herzegovina: World Bank (2002), and inHumanDevelopment Report. Bosnia and Herzegovina. UNDP2002. 21 2.10 Itis important to notice that unemployment in BiHis mostly long-term, as revealed in the Labor Market This is a very important policy issue, discussed further in Chapters 111 and V, but here the concern i s about exposure to employment loss. How prevalent i s job separation, especially forced separation due to mass lay-offs, enterprise closures etc.? 2.11 LSMS data suggest that about 20 percent of all employed persons voluntarily or involuntarily changed their jobs or moved out of employment during 2000-01, of which only one-sixth were subject to mass layoffs or redundancies. Overall, around 15 percent of all households were affected byjob transitions in these two years, and only 3 percent by layoffs. Job losses are only slightly higher among poor households. 2.12 These results are consistent with the official employment data and the analysis of the job mobility in the Labor Market study. The effects of privatization and restructuring on employment by the end o f 2001 were minimal. The main source of the relatively high unemployment rate inBiHi s therefore not ajob destruction process by itself. 2.13 Contrary to widespread belief, unemployment risks do not affect the majority of the population. The popular perceptions of employment as a major problem reflect a much broader range of issues, ranging from poor labor market access for new entrants, to poor worlung conditions and low pay. 2.14 Nevertheless, exposure to the risk of job loss may increase in the future. More intensive employment restructuring will likely raise separation rates with more workers losing their current jobs. This may have significant consequences for welfare, as BiH i s typically made up o f one-earner households. Over 38 percent of the population depends on a single earner (usually the household head). If a household's single earner loses his or herjob and cannot find another, the consequences for the household could be catastrophic. (iii) Health Risks. 2.15 Chapter VI discusses in detail the health status of the population in BiH: it notes that close to one third of LSMS respondents report a chronic health condition, and four percent are seriously handicapped. As the analysis of non-income dimensions of poverty in Chapter I suggested, around 16 percent of the population can be classified as being health poor. More importantly, Chapter V I reveals very significant health expenditure risks for a typical BiH citizen, especially related to catastrophic illnesses. The proliferation of out-of-pocket payments (legal and illegal) has rendered access to basic health difficult for many poor families. A regular hospitalization costs on average over 300 KM, but any serious operation require thousands of KM in out-of-pocket expenses. Thus health risks in BiH substantially take the form of uninsured financial risks. 37Two thirds of thejobless in BiHare long term unemployed (i.e., they have not worked for a year or more). Labor Market InPostwar Bosnia And Herzegovina. World Bank 2002. 22 2.16 Chapter VI also lists a number of risk factors affecting the population's health status, among which lifestyle factors, (high prevalence of smoking, unhealthy diets, etc.) seem to play an important role. Health is also affected by hightraffic accident rates, described below, and by relatively highprevalence of violence (as discussed in Chapter I). 2.17 Traffic accidents appear to be a major risk factor. According to a report prepared for EU-Phare in November 1999, the 1997 fatality rate of around 18.6 deaths per 10,000 vehicles was around nine times higher than the fatality rate of 1.5 to 2.0 deaths per 10,000 vehicles in industrialized countries such as the United Kingdom. There i s evidence that traffic safety i s worsening and that the rate of injuries remains high or i s increasing. These numbers highlight problems that affect all population groups in society, but that pose the greatest risk to those who do not have the resources to pay for their consequences in terms of their health care (and other) costs. 2.18 Another health-related hazard affecting the population i s the wartime legacy represented by landmines. An estimated one million mines were laid during the war across 20,000 minefields. Up to date only 5% of the minefields have been demined.38Imprecisely marked or unmarked mine fields, unknown locations of mine fields, absence of records, mines laid by inappropriately trained persons, and movement of mines caused by weather and erosion are extremely serious problems. As a result, up to the end of 1999, 598 persons were wounded by mines, and 167 persons died; mine deaths and injuries are still reported every year. Mines not only kill, they often act as a major deterrent to return (abandoned houses are often mined), to agricultural development, and to mobility. Landmines have rendered approximately 15 percent of farmland inaccessible. Returnees, especially those returning to rural areas, are the group most exposed to the risk of mines. 2.19 Health risks therefore affect many people in BiH, and constitute one of the key dimensions of vulnerability. These risks take many forms, ranging from the heavy financial burden of illnesses to direct threats to life. (iv) Corruption and Political GovernanceRisks 2.20 Corruption tends to exacerbate risks stemming from other sources, because it increases the financial burden facing poor people or limits their opportunities to adopt particular coping strategies. Strong perceptions exist in BiH about rampant corruption in the public sector. The empirical evidence on c~rruption,however, suggests that its direct ~ ~ impact on household welfare i s limited,40and that most of its harmful effects take the form of 38Data from the BiHNational Environment Action Plan, 2002. 39The Anti-Corruption Diagnostic Study: World Bank 2000, already referred to in the analysis of violence (Chapter I). 40 The amounts reported are negligible, with median values of around 10 KM,and a lion's share of bribes paid by households are for traffic police. The UNDPHuman Development Report 2002 explains why, despite the relatively limited burden o f bribes on households, corruption is often listed as one of the two or three top public concerns inBiH. 23 seriously undermining the business environment. This aspect of corruption will be analyzed inChapter IV. 2.21 Meanwhile, governance risks broadly defined, mainly involving very poor implementation of laws, discrimination and infringements of the rule of law by political forces, are very significant.. Consultations with the poor in BiH 41revealedthat the poor are excluded by many institutions and that their rights are neglected. The findings highlighted many instances of governance malfunctions and discrimination by local authorities. Vulnerable groups, especially minority returnees, risk falling victim to the arbitrary use of power by the State or to discrimination with little recourse to the law. 2.22 The courts are known, even by their own admission, to be weak, deficient and, in many instances, subject to political direction.42The enforcement of laws, contracts, and court decisions i s exceptionally poor, reflecting the inadequacies of the legal system. Both Entities have new laws (the Law on Judicial and Prosecutorial Service (Federation), the Law on Prosecutors (RS) and the Law on Courts and Court Services (RS)) that are designed to de- politicize the judiciary. 2.23 Significant problems remain. Legislation i s fragmented among many levels of authority in the country, creating inconsistencies and contradictions. This state of affairs undermines the principle of equality before the law, and creates insecurity in people's personal and professional lives. Meanwhile, the web of public administration i s reportedly daunting and oppressive to all those who come in contact with the State. Political parties in power are still exercising crucial influence on the financing of the judiciary, on public administration and on official appointments. 2.24 Thus, the population rightly sees contacts with the State as sources of frustration and potential risk, rather than as opportunities to assert rights. This means that instead of reducing the risk of poverty and deprivation, current institutional and governance structures can generate new risks of their own. B. Vulnerability to Poverty as a Failure to Cope with Risks 2.25 The following factors may render a household more or less vulnerable to shocks that can lead to poverty and d e p r i ~ a t i o n : ~ ~ P Access to labor markets. For many households, access to alternative employment opportunities, both formal and informal, i s critical for income diversification. 41The World Bank's Social Assessment and the Voices of the Poor Study (1999) are closely connected participatory assessmentsof poverty, the former conductedas an independent study and the latter as a companionpieceto the World Development Report2000/01. 42InternationalCrisis Group reports. Denied Justice: Individuals Lost in a Legal Maze, February 2000; Rule of Law in Public Administration: Confusion and Discrimination in a Post-Communist Bureaucracy, December 1999; and Rule Over Law: Obstacles to the Development of an Independent Judiciary in Bosnia and Herzegovina, July 1999; and World Bank'sAnti-Corruption Diagnostic, March2001. 43As presentedindetail inNon-IncomeDimensionsOf PovertyInECA World Bank 2003. 24 > Access to credit, saving instruments and ownership of liquid assets. A household's physical assets are a measure o f its capacity to self-insure. Physical assets can be sold to compensate for temporary loss of income, or to cover unexpected expenses. Access to credit (formal or informal) can help households to cope with a health shock or a temporary loss of income. > Participation in formal safety nets, including health insurance, and links to informal/family networks. A household's vulnerability i s reduced if it i s entitled to publicly provided transfers. But if transfers are unreliable, they only increase the uncertainty that households already face. Family-based networks, remittances, or transfers in cash or in-kind from friends and relatives are important elements o f support for many, but are evidently extremely selective and determined by individual families' circumstances. > Mobility. Geographical mobility, based on access to transportation, helps households to cope with risks through better access to labor markets, to credit, to family networks, and to safety nets. 2.26 If they function well, formal mechanisms for risk mitigation, usually associated with public sector interventions, generally offer universal access to all in need. But they also tend to be expensive and can distort incentives. Informal support mechanisms help to mobilize additional resources, but are also not free from incentive problems and operate in a very inequitable manner: only those considered deserving can count on support. The analysis in subsequent subsections will show that risk management strategies in BiH are overwhelmingly informal. As a result, mechanisms for coping with vulnerability and risk are highly selective, with some groups of the population almost completely left out. 2.27 General factors influencing the current living standards of a household could also protect its long-term consumption levels against adverse shocks. Better education and health status help households to manage risk better, to have more income diversification opportunities, and to adapt to changes. Empowerment and good governance are factors that reduce the arbitrary use o f power and vulnerability to adverse change in policies. Nevertheless, although conceptually vulnerability i s multidimensional, in reality much of it i s tied to material vulnerability, whether resulting from a drop in incomes, unexpected expenses, illness or natural disasters. This i s because material resources and incomes are a critical instrument with which to tackle unexpected occurrences. The discussion that follows will therefore focus on material aspects. (i) Access to Labor Market 2.28 For many households access to alternative employment opportunities i s critical to income diversification and reduced vulnerability. This i s especially true, for example, for rural households whose main source of income i s tied to agricultural use of their land, and thus potentially subject to severe external shocks. In the context of BiH the ability to find 25 alternative employment i s critical, since, as already noted, one-earner households are the norm. 2.29 Employment opportunities are usually more diverse, and often more equitable, when labor markets function well. But, the analysis in the Labor Market Study 44 shows that the access to formal sector jobs i s very limited. The informal sector, on the other hand, seems to be an important source of flexibility (see Box VIII). The Labor Market Study estimates that about a third o f all jobs in BiH are in the informal sector. Thus, informal labor does provide an important way to cope with risk, but this i s not always an appropriate long-term strategy. In addition, many sources of information show multiple violation of workers' rights and significant instability in many types of informal e m p l ~ y m e n t . ~ ~ BOX VIII. InformalemploymentinBiHandpoverty:cure or disease? In the past few years the informal sector (or economy) in countries in transition has increasingly become the focus of research, public policy and the media. The informal sector is important because it provides a considerable source of income and employment in countries where formal employment opportunities and social security are limited. However `informal' activities are heavily stigmatized, and `entrepreneurial' activity is still somewhat associated with a `dishonest and criminal way of making money'. The term `informal sector' has been used to describe an extremely wide spectrum o f activities which do not have much in common, including tax evasion, corruption, money laundering, organized crime, bribery, subsistence farming, barter, petty trade, and the stealing of State property. Chapter Iin Volume I1provides an example o f how this concept of informal activities can be operationalized to analyze informal employment, in BiH LSMS data. We define as informally employed those who are engaged in productive income generating activities, but are not covered by the usual type o f obligations and benefits associated with a formal labor contract: Le., pension fund contributions, health and disability insurance. Several observations can be made about informal sector employment on this basis, as follows: 0 The majority of informal employment is in agriculture. 0 The majority of private sector employment is informal. 0 Youth is more likely to be informally employed. 0 Lower education seems to be associated with informal employment. 0 Some regions have especially high shares of informal employment. Analysis reveals that the informal sector in BiH is sizeable, but not extraordinarily large. Several CIS countries for which comparable data are available have larger informal sectors. As elsewhere in the region, the informal sector i s very heterogeneous in BiH. Survival type activities clearly predominate, as half of all workers having their primary jobs in the informal sector are employed in agriculture. At the same time, some formal sector employees are engaged in various informal activities as secondary occupations. I t i s therefore counterproductive and even dangerous to use a broad term `informal sector' in policy discussions, rather than specific definitions of particular activities as the relevant policy implications differ. But it i s important to recognize the informal sector's role as a coping strategy o f the poor. Sources; Bernabe.Labor basedon LSMS. see also Barberia. Johnson, and Kaufmann 1997. 44Labor Market InPostwar Bosnia And Herzegovina. World Bank 2002. 45See Qualitative Study o f Gender and Poverty. IBHI2002. 26 (ii) Assets and Access to Credit 2.30 Cash savings in national currency were wiped out several times in recent history by episodes o f high inflation in the late years of SFRY. Cash holdings in foreign currency were significant, and the exchange of old DMs has demonstrated that there i s more cash in the hands of the population than had previously been estimated. However, it i s unclear how these holdings are distributed among the population. 2.31 Unlike cash, physical assets can be more reliably measured in a survey. The LSMS survey permits construction of a very detailed measure and valuation of household assets. The emerging picture i s cause for concern: this form of coping with risks i s seriously underdevelo ed in BiH. The value of all movable property averages just over 2400 KMper household. 4BThisstock of wealth i s subject to significant depreciation, as the median age of a durable i s 10 years. Thus, few resources can be mobilized by selling physical assets. Establishing a cutoff of 1000 KM as an "asset poverty line", we find that 40 percent of the population have assets below this level. This rate of asset poverty is particularly high for IDPs, over 70 percent of whom have assets valued at less than 1000 KM. Inclusion of the market value47o f owner occupied housing in total assets increases the market value o f assets belonging to an average household to 40,000 KM, but further increases the difference between "haves" and "have not^"^^. Usinga cutoff of 10,000KMfor the total value o f assets belonging to a household as a notional "security" threshold, LSMS data suggest that 24 percent o f the population i s below this cut-off point. Among the poor this percentage i s significantly higher (42 percent); and among poor IDPs it reached a staggering 80 percent. 2.32 Borrowing can be a very efficient strategy to cope with risks, but there are serious problems with respect to access to credit in BiH. Real interest rates remains high, and the market for credit i s underdeveloped. According to the LSMS, nearly 60 percent of poor individuals wanted to borrow money during the past year, but did not do so because they felt that they would have been unable to find a willing lender. According to LSMS results, fewer than 10 percent of people aged 22-35 borrowed money during the 12 months prior to their interviews; the comparable figure for individuals aged 36-59 i s only 16 percent. A nearly identical fraction of individuals in the latter age group in poor households and in non-poor households borrowed money (17 percent versus 16 percent), but the mean amount borrowed by the poor was only a third of that borrowed by the non-poor (970 KMversus 2800 KM). (iii) Access to Safety Nets, and Family/Znformal Networks 2.33 This sub-section discusses the role of both formal andinformal support mechanisms inaddressing poverty andhelpinghouseholds to manage risks. 46Based on respondents' valuation of market price of their belongings. 47See Chapter Iin Volume I1for explanation o f derivation; depreciation i s assumed to be 5 per cent per year in constructing the value of the housing stock. 48Using the term from the Voices of the Poor in BiH,World Bank 2000. 27 2.34 The war completely changed the extent and depth of coping mechanisms. In particular, it revived traditional practices, such as garden cultivation, preserving winter food, exchange of labor for construction and agriculture, the prevalence of multigenerational households, etc. It also forced many previously self-sustaining households to rely on humanitarian assistance, private and public help, and non-market economic a c t i v i t i e ~ . ~ ~ Immediately after the conflict, many households were the recipients o f much international attention and aid, which gave them a sense of security. With the withdrawal of much international support, however, many programs implemented by local NGOs have been discontinued owing to fundingcuts. 2.35 Public transfers provide some protection against poverty for many: in 2001, over 40 percent o f the population lived in households which were receiving at least one type of benefit. Total public spending on transfers constituted 15 percent of GDP in the consolidated BiH budget, but most of these transfers, as the analysis in the PEIR has shown, are "categorical" (pensions, veterans benefit), and not targeted to those in greatest need. Until very recently most transfers were themselves a source of unpredictability with significant arrears. As Chapter VI11 will show, many people fall through the cracks in the system. Meanwhile, Chapter VI shows that a significant share of population i s also not covered by the health insurance system, and, even for those covered, benefits do not provide sufficient resources to offset the financial risks associated with health costs. 2.36 At least as many households rely on various informal coping strategies, including remittances, migration, worlung multiple jobs, own production of food, merging households, absorption of the most vulnerable into extended families, informal borrowing, and non- payment of public service bills. 2.37 Macroeconomic data provide some indication o f the importance o f remittances and assistance from relatives abroad. A recent Central Bank of BiH report estimates the amount of private transfers from abroad at 2.1 bln. KMin 2001 (22 percent o f GDP), a figure which ' far exceeds previous estimates . Close to half a million refugees and emigrants from BiH have found permanent jobs abroad,50 helping their family members at home in various ways. Many Bosnians are still leaving the countr even years after the end o f the war: an estimated 70 thousand left between 1997 and 2000. Bisogno and Chong based on CElT data (World Development, 2002) outline the paramount importance o f remittances for poverty reduction inBiH. 2.38 Unfortunately, it i s impossible to determine the importance o f remittances for poverty from LSMS data. But the data do show that a substantial fraction of the population (7 percent of all individuals in the LSMS) has no apparent source of livelihood: no agricultural land, earnings, public transfers, or social assistance. Only a third of them are poor. Thus, one may 49 See Local Level Institutions Study for a detailed discussion. 50UNDPHumanDevelopment Report 2002. 51 Inthe 1999 Social Assessment,nearly 10 percent of households interviewed mentionremittances from the diaspora as a source of income (see World Bank 1999a, p. 12). But since this survey focused on the poor, it is impossible to generalize from these data. The UNDP's Early Warning System notes that "more people receive financial support from cousins [relatives] living abroad than are helped in various ways by the State" (see UNDP2001, p. 36). 28 assume that some of these people are recipients of private transfers, such as remittances, that permit them to escape poverty. Clearly, however, remittances are spread across both non- poor and poor alike. 2.39 Informal networks are essential to weathering hard times and navigating highly personalized and politicized social institutions in BiH. But as the Local Level Institutions Study documented, such networks very rarely extend beyond the boundaries of local communities. Moreover, there are sharp divisions even within communities, by ethnicity, b y rural or urban background, by refugeeheturnee status versus local population etc. The non- poor usually manage to cross such boundaries with their extended families, but the poor are often locked inside local poverty traps.52 The poor and near-poor suffer from exclusion from effective networks because for the most part, these networks require basic resources to partake in activities, an ability to reciprocate, and as a result are mostly horizontal, with limitedpotential to lift someone out of poverty or to prevent further impoverishment. (iv) Mobility and Access to Transportation 2.40 One of the key constraints on returning to normal life was the destruction of the country's transportation networks during the war. Reconstruction efforts have successfully restored key roads, but large unmet demand remains and imbalances persist. Lack of funding has contributed to the continued deterioration and delayed maintenance of the main road network, even as traffic has increased considerably. This increasing traffic on narrow main roads has a negative impact on road safety, as evidenced by the problem of accidents discussed earlier. As a result, there i s widespread dissatisfaction with the quality of and access to the road network and public transportation. Table 2.2 i s based on responses from a large scale qualitative survey conducted for the Local Level Institutions study. The quality of roads easily tops the list as a perceivedproblem. Public secunty Low quality (16.7), Corruption(11.4),Non-existence(8 1) Socialassistance Non-existence(34 8), Low quality (18.5), Corruption (6.8) Public transport Low quality (34.4), Highpnces (16.0), Non-existence(7.4) Water supply and sewage Low quality (36.1), Poor condition of infrastructure(16 l), Non-existence(8.9) Healthcare Low quality (22.4), Highpnces(20.4), Corruption (12.6) Education Low quality (16.0), Corruption(8.4), Highpnces(7.0) Source: Local Level InstitutionsStudy, The World Bank (2002). 2.41 Close to 40 percent o f households in BiH (and 22 percent o f the poor) own cars. Maintaining roads of adequate quality provides the mobility needed to maintain social networks, access job markets in major cities and shopping in cheaper sources. But the 52 The Social Capital Study found major disfunctionalities in local level institutions (municipalities, community organization and neighborhoods) owing to the war. I t is thus not surprising that, in the Voices of the Poor and the Anticorruption Study, the lowest level of trust for formal institutions was in municipalities and neighborhood units, along with criticisms o f their lack of transparency and accountability. A Social Assessment of Bosnia and Herzegovina, World Bank. 1999. 29 majority has to rely on public transportation which in many areas turns out to be the lowest priority o f local authorities. Private alternatives have proved to be expensive, as shown in Chapter VI with respect to the cost of transportation to health care facilities. Such costs can represent a barrier to access to health services. These costs are disproportionately borne by the poor and rural dwellers. C. Data and Trends in Poverty Perceptions 2.42 Surveys where respondents rate their economic status as "poor" or "rich" often suggest a positive correlation with individual or household incomes, but this correlation i s generally not strong. The gap between objective conditions and perceptions i s not unusual, and a large body of literature in economics and psychology has sought to explain it. This subsection briefly outlines some of the contributing factors involved, including the level of inequality (not a major factor in BiH, where consumption inequality i s relatively low), unemployment and other forms of social exclusion, exposure to risks, and past trauma. 2.43 A number of studies have found that the unemployed have lower self-rated welfare.53 Those who became unemployed tended to say that they were worse off, even controlling for the loss of income. The relatively high level of unemployment in BiH may thus exaggerate perceptions of poverty. 2.44 Risk plays a very important role as a factor directly influencing perceived welfare or lack o f it. People usually feel significantly poorer when a shock occurs. The literature on the subject suggests that the welfare loss as a result of shocks arises from some combination of lower expected future incomes and increased uncertainty and perceived insecurity arising from exposure to uninsuredrisk. The widespread feeling of uncertainty and insecurity across society in BiHmay help to explain elevated self-assessments of poverty. 2.45 These factors may help to explain why qualitative studies conducted in BiH consistently find that the majority of respondents indicate that they consider themselves to be poor and destitute.54We find similar self-assessments elsewhere in the region.55 2.46 When considering subjective evaluation data, the reported level i s o f less significance than the trend. The UNDP Early Warning System Reports 2000-2002 have hinted at some improvements in people's perceptions of their own living standards: respondents assessing their status as one o f bare survival or well below average declined from 47 percent to 28 percent. 2.47 Inaddition to the general factors noted above, there are at least five additional reasons for the differing objective and subjective assessments of poverty in BiH: (i)the common "reference point" for many citizens i s the pre-war period, and current living standards still fall short of this benchmark;56 (ii) welfare state in BiH i s still based on definitions of the j3See Winkelmannand Winkelmann(1998). j4Bosnia and Herzegovina UNDPHuman Development Report (2002) 55For the most recent example, see Poverty Assessment for Bulgaria, World Bank, 2002, p.15. 56The war made such a large part of the 1990s so unreal and incomprehensible that pre-war conditions came to 30 basic rights associated with belonging to "vulnerable" categories, leading many people to call themselves poor so as not lose some of their benefits; (iii) the design o f some qualitative surveys calls for a sample that purposefully over-represents the poor; (iv) subjective assessments reflect a broader conception of welfare than income (or consumption) poverty as measured by the LSMS; (v) perceptions are also shaped by higher expectations. 2.48 High subjective assessments of poverty may partially reflect worries about vulnerability to future shocks, but they also influence the behavior o f households today. A widespread perception of poverty - even where it i s not consistent with objective measures - may foster risk-averse economic behavior that can help to perpetuate poverty, low levels of political participation, lack of trust in institutions, and non-participation in other areas of social and political life. The gap between perceptions and reality suggests that people lack information with which to assess their circumstances. Regular monitoring o f poverty and living standards would help improve the quality of information on the welfare of the country and its citizens, and help enhance the realism of their perceptions. Similarly, better and more regular communication by the Government about the nature, objectives and progress of reforms would offer a more realistic understanding of the reform path and its prospects.57 D. Conclusions 2.49 The widespread vulnerability and the limited capacity to cope with risk that characterize a sizable fraction o f the population implies that the 19 percent classified as poor on the basis of current consumption comprise only a fraction of those in BiHwho must worry about, and struggle to avoid, falling into poverty. Risk i s often concentrated in particular groups, that are especially vulnerable. Groups identified as prone to be consumption poor also face increased vulnerability in other dimensions. 2.50 Factors also exist that can mitigate vulnerability. For example, community based networks, where they are operational, offer shelter against poverty. Informal coping mechanisms seems to be particularly well developed in communities where social capital was not damaged by the war or where it has been rebuilt. But it will not be sufficient to rely on informal networks to reduce vulnerability to poverty in BiH. These arrangements offer only limitedprospects, especially for the very poor. And informal networks can hardly + be regarded as sustainable solutions. Thus public policies are needed to address systemic causes of widespread vulnerability to poverty. 2.5 1 Finally, the apparent contradiction between overwhelming subjective perceptions of poverty and lower assessments of poverty measured with objective standards probably reflects two strands o f opinion that contribute to the former. The first i s "elite" opinion - the extremely bleak perceptions about poverty in the media and academia, which are lagging reality. The second i s that o f "popular" perceptions of poverty, which may reflect an be considered as the last "normal" period in the economic history of the nation. 57 For example, the process of consultations with civil society undertaken as a partof the PRSP. 31 overwhelming feeling of vulnerability to unpredictable changes in policies, economic risk (especially with regard to employment prospects), health risks and the uncertainties bred by a post-conflict environment. 32 3. CHARACTERISTICS OFTHE POOR Knowledge about the characteristics of the poor helps to determine which policies can best reach them. In addition, knowledge about which groups have the highest and lowest poverty rates helps analysts and policymakers to understand why people are poor. This chapter examines both questions, starting with an overview of key characteristics. Subsequent sections assess each of the characteristics in tum. The analysis includes comparisons with other countries, showing that poverty in BiH has several very distinct characteristics. Poor households in BiH are predominantly those who have children, are headed by working-age adults, rely on a single eamer, are concentrated regionally in specific" mixed" urban-rural or rural communities, and are usually characterized by low educational attainment of both parents and children. Poverty is higher in RS than in FBiH. The Chapter concludes with a presentation of the analytically derived net efsects of each of these factors on poverty, arguing that all of them matter empirically and calling for a multi-pronged approach to combating poverty. A. Poverty Profilefor BiH: WhoAre at Risk? WhoAre the Poor? 3.1 Existing studies on poverty in Bosnia and Herzegovina suggest that poverty i s concentrated among distinct and easily identifiable groups within the population: the unemployed, the displaced (IDPs) and refugees, the disabled, etc. In addition, non-poor members of these same groups are usually seen as having the greatest risk of falling into poverty. I s this true? To what extent i s poverty concentrated in specific, easily identifiable groups? 3.2 In order to answer this question we compare poverty indices (incidence, depth and severity of poverty) across socio-economic grou s, and we compute the share of each group in the total number of the poor in the ~ountry.~`Table 3.1 presents the key findings, using the definition of poverty described in Chapter I(which i s based on consumption and a national general poverty line); each of the panels A-K provides data based on subdivision into mutually exclusive sub-categories. Panels A-J are household based, while panels Iand K report key group findings for individuals. Detailed results by Entities can be found in Chapter V o f Volume 11. 3.3 The highest poverty rate in Table 3.1 i s shown in panel F, column 1: over half (56 percent) of the population living in households with 3 and more children are poor. In general, households with children seem to represent a disproportionately high share of all poor people (69 percent, panel F, column 4), while households with elderly members, or '*These indices were identified in Chapter 1F, and are defined mathematically in Volume 11, Chapter I1E. 33 headed by the elderly, seem to be close to the average poverty rate (panel G, column 1) and constitute a minority among the poor (panel G,column 4). Detailed poverty figures reported in Volume 11, Chapter V show that the elevated poverty risk for households with children is confirmed for all possible poverty measurement methods. - Table 3.1 Povertxprofile for BiH 2001 Characteristics Poverty rate, Poverty gap, Poverty severity, Confidence Share Of the BiH interval for poverty (personalandhousehold) percent percent percent poor, percent m+n A Entity . FbiH 16 3 1 52 +4%p. RS 25 7 2 48 +7%p. B. Location Urban 14 3 1 18 +3%p. Mixed 24 6 2 38 +7%p. Rural 20 5 2 44 &%p. C. War Displacement Status Placeof residenceunaffectedby war 19 5 2 46 +5%p. Were displaced duringthe war 12 3 1 18 +3%p. Returnees(DPs andrefugees) 17 4 1 7 +7%p. Remain Refugeeor DP 34 8 3 29 +8%p. D.Educationof theHousehold Head None 28 9 3 12 +lO%p. Primary 26 6 2 41 +6%p. Secondary 10 2 1 7 +3%p. SecondaryVocational 18 4 1 32 +4%p. Junior College 10 2 1 2 *5%p. University 2 0 0 1 +l%p. E. Number of EmployedAdultsina Household Nobody i s employed 25 6 2 37 +5%p. 1personi s employed 18 4 1 35 +4%p. Two and morepersonsare employed 16 4 1 28 *6%p. F.Number of Children(below 14) Nobody in the householdis achild 12 3 1 31 +3%p. 1child 20 4 1 22 *4%p. 2 children 27 6 2 27 +6%p. 3 and more children 56 16 6 20 ~ 1 4 % ~ . G. Number of Elderly (above workingage) Nobody is an elderly person 20 4 2 61 +4%p. 1elderly person 18 5 2 21 +5%p. >=2 elderly person 19 5 2 18 &%p. H.Gender of theHouseholdhead Male 21 5 2 90 +4%p. Female 13 3 1 10 +3%p. J. Age ofthe Householdhead Under 25 14 3 1 0 +23%p. Between26 and49 21 5 2 56 +4%p. Between50 andretirement 18 5 2 19 +6%p. Retirementage 18 4 2 25 +4%p. I.IndividualemploymentStatus(forindividualsabove15) Economicallyinactive 19 5 2 55 *4%p. Unemployed(ILO) 28 9 3 13 53%~. Informalsector worker 16 5 2 14 *5%p. Formalsector worker 12 3 1 19 +5%p. K.Individualeducation (forindividualsabove 15) None 23 6 2 12 *6%p. Primary 23 6 2 56 +5%p. Secondary 11 2 1 8 +2%p. SecondaryVocational 15 3 1 22 +3%p. Junior College 8 2 1 1 +3%p. University 3 0 0 0 +2%p. ; Total 19.5 2 100 SI%-. Source: LSMS 2001 primary data. Only covers current householdmembers. Poverty is basedon baselineconsumption aggregate per capita with adjustment for spatial price variation, and general national poverty line. Employment status i s defined according to the ILO criteria (see Volume 11Chapter Ifor details of the definition). The upper boundfor workinghetirement age is definedat 55 for females and 60 for males; the lower bound i s 15. For an explanationof confidence intervalsreported (95 percent statistical confidence) see Box Iand Chapter 1of Volume 11, standarderrors are corrected for stratified sampledesign. 34 3.4 Refugees and displaced persons have a high poverty rate (panel C, column 1); 34 percent of them fall below the poverty line. At the same time, however, a not inconsiderable 19 percent o f those who were not displaced by the war are also poor. According to LSMS data, by the end o f 2001 most of those who fled the conflict remained displaced: 16 percent of the entire country's population were still displaced within the country or were in the process of ret~rning;~'among those who fled the war abroad an estimated 50 percent have settled permanently abroad, and the other 50 percent remain potential returneese6' Interestingly, however, panel C column 4 shows that nearly half of all poor people classified by displacement status are those whose place of residence was unaffected by the war, with refugees and displaced persons accounting for less than one-third o f the total. 3.5 Another clear message from Table 3.1 i s the strong relationship between employment status and poverty. As panels E and Isuggest, households in which no-one works, and individuals who are unemployed, have high poverty rates. As panel I, column 1, shows, 28 percent o f all the unemployed are poor. But the data also show that poverty i s not only a consequence o f unemployment or inability to work. In fact, as shown in panel E, column 4, close to two-thirds of all poor people (63 percent ) live in households where someone works. Finally, poverty falls with the number of earners in the household, generally reflecting whether the spouse of the household head i s employed. The income of a single earner inBiHi s often insufficient to pull a household out o f poverty. 3.6 Poverty varies across regions of BiH and by type of settlement, with "mixed" ruralhrban and rural areas having higher poverty rates than urban areas (panel B, column 1). But although poverty rates vary by location type, poor individuals live in all areas of BiH- urban, rural and mixed. The RS population has a substantially higher poverty rate, depth and severity than that o f the FbiH (panel A). This difference i s in line with, but somewhat less pronounced than, previous findings.61It should be noted, however, that the FbiH accounts for a slightly higher share (52 percent) of all the poor and accounts for nearly half o f all BiH poor (panel A, column 4). 3.7 Finally, education plays an important role as a correlate of poverty. This i s not unique to BiH. But the extent of the effect of lack of education on poverty in BiH i s very substantial; households where the household head has only primary education or less account for 59 percent o f BiH's poor (panel D, column 4), suggesting that a significant number of poor working age adults have inadequate skills (a maximum of eight years of schooling). 3.8 It should be noted that the poverty profile discussed here and based on the LSMS excludes those who are not covered in the sampling frame, especially including institutionalized or unregistered population groups. IDPs in collective centers (around 1000 individuals according to the PRSP) are an example of a particularly poor group that i s too 59This i s a greater share than in any other country intransition. See: UNHCR, The State of the World's Refugees Seeking Asylum, 1995, and UNHCR, Populations of Concern to UNHCR - 1997Statistical Overview, 1998. UNHCR, Statistics package, Sarajevo, 01.07.2001 and 28.02.2002 61Poverty and Inequality in BiH: The Legacy of War, The World Bank (1999) found large regional differences inlivingstandards: the poverty rate inthe Federation(FBiH) in 1997was 22 percent compared to a very high 52 percent in the Republika Srpska (RS). 35 small to be included in a survey. The Roma are another example of a group that i s inadequately covered (Box E).Thus the poverty profile needs to be interpreted with caution, because the LSMS can only provide data for groups, large enough to be statistically captured ina general population survey. 3.9 In addition, the diagnosis of poverty depends on how it is measured. Chapter VI in Volume I1 therefore evaluates how robust key findings are to different measurement assumptions, It finds that the same five key features of the poverty profile are also robust correlates o f poverty: presence of children, employment o f household head and presence of secondary earners, regional effects, displacement status, and level of education. Subsequent sections will explore each of these factors in detail. Box IX. The Invisible Faceof the Poor: RomainBiH According to the UNHCR the estimated Roma population in BiH i s between 40,000 and 50,000, i.e. between 1and 2 percent of the population. This estimate is, however, very far from being exact , as it is based on the old Census data, which were approximate at best, as many Roma were known to identify themselves as members o f one of the dominant ethnic groups. " IknowI'mRom, - saidFatimainasurvey inZenica conducted by Medica, - but they didn't call us that before, we were called Muslims before. `What is your nationality?' -`Muslim'. That's what we were on our birthcertificates and how everyone called us." According to Slavko Kukic in his study "The position o f Roma in Bosnia and Herzegovia," Roma are one of the largest ethnic minorities in BiH (1999). However, this author argues that Roma have been virtually ignored and excluded from every aspect of life: education, job opportunities, health care. In the past years, more organizations, both local and international, have begun to work with Roma. Medica - an NGO from Zenica focusing on women's issues - conducted a special survey o f Roma women in this municipality, interviewing 112 Romani women sampled on the basis of information from the local authorities and Roma NGOs. The results are striking. None of the households interviewed included a women over 50 years old - as the respondents explained, Romani women "have a lower than average age of death'. Contrary to the common perception that all Roma fled the war, % of the Roma respondents never left Zenica. Only four percent were employed (receiving some regular salary and registered as employed), but 30 percent worked in marketplaces or sold goods on the streets, and an additional fifteen percent of respondents lived by gathering aluminum and other metals from garbage and landfills. For many, however, these jobs paid too little to provide enough cash to buy food, let alone children's clothes and shoes. Less than half of the Romani interviewed had health insurance. Forty percent o f women had never attended school, and only half of those who attended managed to complete primary or secondary school. Unfortunately, the survey did not find large differences between mother's education and the their children's school attendance. Every fourth child of school age had never gone to school, and an additional 12 percent of respondents could not remember what education their children had. Finally, 17 percent had no running water at home (as opposed to 1percent on average in the Zenica municipality), 9 percent had no electricity (zero on average according to municipality statistics), and 18 percent had a least part of their home destroyed. Sources: How We Live. Second Look 4. Medica, Infoteka, Zenica, 2001 Kukic, Slavo "The position of Roma in Bosnia and Herzegovina. Center for Minority Rights: Sarajevo. Memisevic, Fadila "Roma of Bosnia and Herzegovina: Documentation of the Bosnian Section of the Society for Threatened Peoples." Sarajevo. 36 B. Children versusthe elderly 3.10 Throughout the world the presence o f children and elderly people in a household has a significant impact on its welfare. Poverty has a long-established relationship to the life- cycle in Western countries, with a higher risk in childhood and in old age. Children and the elderly are about 25 percent more likely to be poor than the average person in most EU countries; each group constitutes well over a quarter of the total number of poor in these countries. Transition economies in the countries of former SFRY have a similar age profile of poverty.62 3.11 As already noted, Table 3.1 highlights the elevated poverty rate for households with two or more children. The LSMS data also indicate inter-Entity differences with respect to this variable. Children inRS are at a particular disadvantage: one half of children below the age of seven live in poor families, as compared to a still high one-third of the children in FBiH.With respect to the elderly, the data show that only one-quarter of all poor people are of retirement age, and these households headed by retirees have an essentially average poverty rate (18 percent). 3.12 A poverty profile b demographic characteristics i s always sensitive to changes in measurement assumptions? Chapter VI of Volume I1 emphasize that regardless o f the poverty line, welfare measure or equivalence scale (within a reasonable range) used, larger families with children have the highest povertv risk. 3.13 One possible explanation o f the high risk of poverty for households with children in BiH might be that poor households are larger and having higher dependency rates than average households. A simple comparison can help check this hypothesis. An average household with children in BiH has 4.6 members, of which on average 1.8 are children, 2.7 are adults of working age and 0.2 are elderly. Poor households are slightly larger, and have an average of 2.1 children, 2.9 adults and 0.2 elderly members per household, with very little inter-Entity variation. Clearly, however, poor and non-poor households are not substantially different with respect to size or dependency; in particular, the number o f children in poor households i s only very slightly larger than that in "average" households. (A similar pattern i s observed in Croatia, where there are 1.6 children in an average household versus 1.8 children in poor households, living on average with 2.5 adults (2.7 adults in poor households). By contrast, Kosovo and Albania have larger households with more children. A typical poor Kosovar or Albanian household consists of 7-8 members made up of one elderly person, 3 adults and 3-4 children, while the typical non poor household includes only 1-2 children).64 62 We use approaches that are commonly applied to such international comparisons. The most comprehensive dataset on poverty for transition economies available to date has been developed for the World Bank study "Making Transition Work for Everyone: Poverty and Inequality inEurope and Central Asia"; similar statistics for OECD countries are cited by Klugman and Mickelwright (2001). 63 Our baseline measures use per capita consumption as a welfare indicator, i.e. we assume unitary elasticity of needs with respect to household size. If greater scale economies are assumed, small households move down the expenditure distribution and large households move up. As the elderly tend to live in smaller households and children inlarger ones, this adjustment affects the age profile of poverty. 64 Albania: Poverty Assessment, The World Bank 2003; Kosovo:Poverty Assessment, The World Bank, 2002; 37 3.14 The data raise a significant issue. Poor households in BiH have on average slightly more children than ones non-poor, but the difference i s small. Interestingly, poor households also have more working age adults than non-poor households, as shown above. Thus the question of why households with large numbers of children are poor can be reformulated as a question that asks why adults living with children are not generating enough income to pull their families out of poverty. The answer must be that work i s too poorly paid, or that there are not enough adults working in the household, or both. 3.15 The multivariate analysis of poverty risks conducted in Volume I1 (Chapter VII) confirms that demographic factors cannot explain why households with children are poorer. C. Work and Poverty 3.16 The number of eamers matters for poverty. Table 3.2 reproduces the analysis presented in Table 3.1 but provides selective data for large households with children. It looks at poverty rates by the number (and type) of earners inhouseholds with children. Each row represent as group of households with either no earner, a single earner in the informal sector, a single earner in the formal sector, or two or more earners. The Table shows significant differences in poverty rates, and in the depth and severity o f poverty, by number and type of earners. Table 3.2 also shows that the majority of poor households with children - 56 percent - are single-eamer households. Table 3.2 Poverty and Number of EarnersinHouseholdswith Children Percent Poor Poverty Depth Severity Share of All Poor Nobody has any employment 49 13 5 31 One informally employedearner 36 9 3 17 One formally employed, owner or farmer 23 5 2 39 Two and more eamers 17 3 1 13 Source: LSMS BiH,2001. Only for households with children and three and more adults. 3.17 Table 3.2 also shows that the type of employment mattersfor poverty. Informal sector work does reduce household poverty rates considerably compared to those for jobless households, but it offers a much smaller reduction than formal employment. Poverty rates vary considerably by sectors of employment of the household: agriculture and construction have the highest poverty rates, while financial services, public administration, transport and communication are associated with the lowest. 3.18 The unemployed have a higher poverty rate than the employed or inactive. But because the unemployed are not a large group relative to others (Table 3.1, panels E and I, column 4) , they do not represent a major fraction of all poor individuals. Thus, the study rejects a common perception that in BiH poverty i s fundamentally the result of unemployment. The unemployed (actively searching for work) represent only thirteen percent of all poor adults (see Table 3.1, Panel I). Less than 40 percent of all poor live in households where no one works (Table 3.1, Panel E), among the poor population in families Croatia: Economic Vulnerability and Welfare Study.World Bank 2001. 38 with children the share of jobless households is even lower (Table 3.2). On the other hand, over 60 percent of all the poor live in families where one or more people are employed, thus comprising a broad category of the workingpoor. D. Women's work andpoverty in BiH 3.19 A particular feature of poor households in BiH(the very low prevalence of secondary earners) can be explained by very low female labor force participation rates. Figure 3.1 shows that BiH stands out among all countries in transition with respect to the ratio of female-to-male participation rates. Nowhere else in the region do only 35 percent o f females aged between 16 and 64 do any work. Only in Tajikistan and Kosovo do recent surveys (not reported on the graph) reveal similarly low activity rates for women (33 and 30 percent correspondingly -Paci (2002)). Figure 3.1 Trends inGender Differences for Labor Force Participation Rates inTransition Economies i w 0 80 0 60 gI Y 0 40 P 0 PO ow Source: Staff calculations based on Paci (2002). Data are from official sources and do not correspond to any o f the surveys. 3.20 Why are we concerned about the low participation rates of females in BiH as a poverty issue? There are two reasons. First, the transition and the war have profoundly affected the real value of the earnings of primary workers; the fall in real wages means that one wage earner alone may not be able to support a family. Second, the transition process and the devastation brought about by the war have led to a shrinking government budget and a sharp reduction inthe real value of public expenditures to support family formation. 3.21 Figures 3.2 show some aspects o f the relationship between the wealth o f a household, ranked according to consumption (by percentiles, from the poorest on the left to the richest 39 on the right), and gender difference in employment among worlung age adults. Panel 7 (A) presents observed employment rates by gender and consumption percentile, while panel (B) reports the absolute malelfemale gap in percentage employment rates (lower line) and the ratio of the female employment rate to that of males (upperline). Figure 3.2 (Panel A) Rates of employment by consumption percentiles. (Panel B) Ratio of women's to men's Employment ratesand Absolute Gap between women and men's Employment rates by percentiles. (for menand women age 15-59) (A) .75 __--- -__---- __-- __*- .5 Q ._ - 0- 2 ,'Females .25 , Female-MaleLFP abullutegap , , 0 , II Consumptionper50 100 0 I 100 I capita percentiles Consumptionper50 capita percentiles Source: Lokshinand Mroz (2002)usingBiHLSMS 2001. 3.22 Panel A of Figure 3.2 shows the extent to which the poor in BiH, both men and women, are locked outside productive employment. Among the poorest, 30 percent of working age men work, but the percentage for women-only 10 percent-is far worse. Panel B shows two things. First, the absolute gap in employment rates i s only slightly higher among the poor compared to the rich (Figure 3.2, Panel A). Second, however, the relative position of women i s profoundly different. For women from well off households, employment rates are only 25 percent below that of men. For poor women, employment rates are only about one-third of men's (Figure 3.2, Panel B). 3.23 This difference does not stem simply from women's unwillingness to work. Joblessness among women, and specifically among women from poor families, i s often a result of unemployment, as Figure 3.3 demonstrates. Unemployment rates reach staggering 50 percent for young women from poor households and the female-male differential is highest for the poorest (Figure 3.3, Panel A). And for almost all age groups unemployment rates among females are higher (Figure 3.3, Panel B). And this in cannot be traced to the legacy of the past or outdated cultural norms, as the most affected groups are the youth and primeworking aged individuals. 3.24 Analysis of the gender wage gap in favor of men as a disincentive for females to work (Labor Market Study and Lokshin and Mroz) reveals that it i s in the same range as in other 40 countries in transition and therefore in itself cannot explain wide differences in activity rates between men and women in BiH compared to other countries demonstrated in Figure 3.1. Thus, other dimensions of employment disincentives (access to child care, or household economies) need to be considered in order to explain the outcome. In particular two factors analyzed elsewhere in this report are important: the complete breakdown o f the early child care system (Chapter VII) and an apparent absence of any economies of scale in household consumption (Volume 11). Figure 3.3 Unemployment rate for menand women age 15-59by consumption per capita percentiles (Panel A) and by age (Panel B). v -6 n E E .3 .-C 5n a e 0 0I Consumption per50 100 capita percentiles Source: Lokshin and Mroz (2002) using BiH LSMS 2001. 3.25 Both of these factors shift the balance in favor o f home activities for women, especially for those from poor households that are unable to afford costly substitutes for female labor (such as household appliances or private child care arrangements). This leads to the perpetuation of a traditional division of labor between men and women. It i s interesting in this context to examine in detail cases when this traditional division breaks down, for example, in female headed households. 3.26 One of the most unexpected findings of the profile i s the lower incidence of poverty infemale headedhouseholds (Table 3.1, Panel H).65 conclusions about the lower poverty The of female headed households in BiH i s robust to modification in the main measurement assumptions.66Traditionally in BiHfemale headed households have been treated as a part of 65The poverty analysis for the report had to be carried out at the households level; hence there is no information inthe household survey about how consumption is allocated within households. Given this constraint, gender differences inpoverty incidence can only be evaluated by comparing poverty incidence between female headed and other households. 66This i s really an unusual result, as in most countries this advantage under the per capita measurement assumption starts disappearing very fast even with small adjustments for economies o f scale inconsumption. The critical values o f economies o f scale adjustments that make female headed households poorer than male headedones range from as low as 0.95 (close to the per capita scale) in Bulgaria to 0.8 inHungary (Lanjouw et al. 1998). InBiH, based on staff estimates using LSMS (2001) the critical value (at which female households as a group becomes poorer than male headed) of the economies of scale adjustment parameter i s as low as 0.55, 41 the population that i s at risk (see Box X), requiring explanation of why LSMS contradicts these perceptions. BOX X Femaleheadedhouseholdsintraditional societies In the traditional sense generally accepted in many Balkan countries, a female-headed household is "incomplete" in a number of ways. The absence of a man as "head of the family" deprives the household of male protection. Protection i s understood to imply the financial and existential security that a man provides for the household. The man takes care of the existential needs of his wife and children and protects them from actual or potential attacks on their physical, mental and moral integrity. He protects, preserves and ensures the family's standing in society, since it i s through his public activities that the family's social involvement takes place. The lack of a "head of the family", that i s of a man, renders the household vulnerable socially. I t i s not unusual for women to have an adult male in the family to ensure that nothing unpleasant happens to them (the assumption i s that a man would protect them as well as the interests of the household). This may relate to the perception of the status of the female-headed household in its immediate environment (including the risk o f marginalization, discrimination, being ignored or neglected by the authorities, the local community, or the extended family -based on the knowledge that this i s a household without protection). Thus most of the available studies in BiH on poverty assume that female headed households are particularly vulnerable to poverty. Basedon Dr.Jasna BakSiC-Mufti6 FemaleHeadedHouseholdsInBiH. UNDPSarajevo2002. 3.27 show a mixed picture with respect to the well- being of female headed household^.^^ Poverty Assessments and regional reports on poverty produced by the World Bank Normally female headed households account for a little more than one-quarter of the population in Central and Eastern Europe. According to available data obtained from the BiH LSMS, women head 25 percent of households in BiH, and 16 percent of the population live in such households, thus broadly approximately the regional average. 3.28 In the majority of transition countries, single parent households have a poverty risk that i s greater than average; single mother make up a significant percent of female headed households in some countries. In BiH, single mothers live in a substantial 10 percent o f all households, but only 2 percent of the population lives in a household headed by a single mother. This figure, (contrary to expectations given the impact o f the war) i s lower than the average for Central European countries (3 percent of the population), and suggests that the traditional family may be integrating those left without husbands into larger extended families. Most female heads of household in BiH (83 percent) are widows; near 20 percent of them are of pre-retirement age and over 60 percent are o f retirement age; many are war widows, and half remain refugees or IDPs. 3.29 The explanation for why female headed households are not poorer than male headed households, as revealedby LSMS, lies in the unique characteristics of families inBiHheaded by women and in the special nature of the safety net arrangements in BiHnoted in Chapter 11. As these households are perceived as "deserving" poor, they benefit from relatively generous which is outside a reasonable degree of economies of scale in a household from transition economy (see Volume 11,Chapter I11for discussion about a plausible range of such adjustment in the context of BiH). 67Female-headed households sometimes tend to be poorer than male-headed households, and some poverty assessments have found that this finding persists even after controlling for age, education, household size, and location. Inseveral countries in Central and Eastern Europe, however, female headed households on average fare significantly better than male headed households (Hungary, Lithuania, Moldova, Macedonia, Slovenia). See Paci (2002) 42 support from both the State (seventy percent of such households receive State transfers) and the community, support that compensates for their lack of economic resources and brings most o f them above the poverty line. E. Locality-Based Variationof Poverty Risks in BiH 3.30 Table 3.1 suggests that fewer than 20 percent of poor households live in urban municipalities, and that poverty i s widely distributed among smaller "mixed' (or semi-urban) and rural communities across the country. During the war, the combined effect of the destruction o f many rural settlements, forcible expulsions, and the policy of ethnic cleansing drove the population out of rural areas towards urban areas where they found refuge. Therefore today many former rural settlers live in urban or "mixed" rural-urban areas, and many rural communities are devastated. The relatively low poverty rate for rural areas shown inTable 3.1 therefore conceals large welfare losses among former rural residents. 3.31 Agriculture also provides a source of livelihood for many urban dwellers. As much as 37 percent of the urban population use land and 40 percent own livestock. Over half of the population of poor households are involved in some form of agriculture and around 40 percent o f the non-poor are directly involved in agriculture. Agricultural production i s still far from levels comparable to 1990. Massive population displacements have adversely affected the production of key agricultural goods, particularly livestock. The recovery in agriculture has been slow and yields of most of the main crops have remained low compared to pre-war levels. Thus, it i s not surprising to find an elevated poverty risk for those who are engaged in agriculture. 3.32 Unfortunately, the official definition o f urban and rural areas adopted for the BiH LSMS i s very problematic. To overcome this problem, we simplify the classification used in the LSMS by comparing urban to non-urban municipalities by Entity. The results are presented in Table 3.3 below, which shows that the higher poverty rate for RS (Table 3.1) i s a non-urban phenomenon, Le., it i s a feature of localities classified as either rural or mixed. (see Volume I1Chapter V and VI for further details ). Table 3.3 Poverty Indices and Composition by Location and Entity Type of Poverty Oxfd" Poverty COnfidence Poverty Confidence Percent of Percent of Location Headcount in^^^^^ poverty rate for G~~ poverty gap interval for Severity population the poor RS Urban 12 k5.4 0.02 kO.01 0.007 k0.004 7 5 Non-urban 27 k8.8 0.07 k0.03 0.028 kO.011 31 45 FBiH Urban 15 k3.2 0.03 LO.01 0.012 k0.004 18 14 Non-urban 16 k4.9 0.03 kO.01 0.012 kO.005 43 36 BiH Total 20 k3.6 0.05 LO.01 0.017 *0.005 100.00 100.00 Source: BiH LSMS,2001, baseline definitionof poverty, 95% confidenceintervalsfor means with correction for stratification. Non- urbani s acombination of ruraland mixed municipalities as definedby the sampling frame. 43 F. GeographicalConcentrationof Poverty in BiH 3.33 Poverty i s strongly geoeraphically concentrated in particular communities, which suffer from a combination of economic dislocation and social cleavages that dramatically reduce social capital of these communities. Often these communities also have a high concentration of IDPs and refugees. 3.34 To quantify the extent of this problem, we need to go at the lowest level of disaggregration possible using LSMS data. We divide the population into groups of local enumeration areas (blocks in the city surroundings and villages in rural areas), thereby obtaining the closest approximation possible to the actual distribution of real communities. The results, presented in Table 3.4, show that the most vulnerable communities (those with a poverty rate twice the national average) account for a modest 21 percent of the population but 63 percent of all the poor, while the least vulnerable communities (those with less than half the national povert rate) represent a substantial 47 percent of the population and only 1 percent o f all the poor.6 l 3.35 Thus vulnerability to poverty has a strong association with locality-specific conditions, such as the health of the local economy. Poverty in the poorer localities cannot be fully explained by individual household characteristics such as level of education, age, economic activity or demographic variables. Poverty i s clustered in a spatially-related sense, having a strong "neighborhood" or interaction effect and providing evidence of the existence of strongly concentrated pockets of poverty. This has important implications for setting the priorities for a national poverty reduction strategy, as discussed in Chapter IX. Table 3.4 Poverty Indices and Compositionby Location and Entity Community poverty rate, Share of Share of relative to national mean Average BiH poor Share of poverty rate IDPshefugees BiH inpopulation population Less than ?hof national poverty rate 1 10 47 4 Between ?4and average poverty rate 13 20 17 11 Between average and twice poverty rate 29 17 14 21 More than two times national Bovertv rate 58 27 21 63 National 20 16 100 100 Source: LSMS BiH 2001. Communities are defined as groups of enumeration areas (a total of 450 in LSMS, see Volume I Chapter I). 3.36 As the LSMS did not collect community level information, we lack data to explain why certain communities have such high poverty rates and shares. In particular, the absence of data on infrastructure availability and social services provision substantially reduces the content of the analysis and will need to be addressed in subsequent rounds of LSMS. The findings of the Local Level Institutions Study, however, suggest that these communities Vulnerable groups are defined as having double the national poverty average, while the non-vulnerable are defined as having half the national averagepoverty rate. As a more robust alternative to this simplified analysis, we also ran a regression with our standard set of socioeconomic factors plus a set of dummy variables representing local areas; the results show that these narrow area identifiers explainconsumption levels even after controlling for other factors such as education, employment, etc. 44 suffer from erosion of social capital (a very strong factor mitigating against poverty elsewhere), in turn reflecting a number of social cleavages (between local and displaced populations, between returnees and IDPs, between individuals with urban and rural backgrounds, etc.), all of which seriously diminish the extent of trust and mutual support among households. As Table 3.4 suggests, IDPs and refugees represent 27 percent o f the population inpoor communities as opposed to 10 percent in non-poor communities. E. Educationand Poverty in BiH 3.37 Table 3.1 shows a very strong and significant link between poverty and lack of education. Those living in households whose head has only a primary education or less are about three times more likely to live in poverty as those with general secondary or junior college education. They also represent the largest group among the poor. These results are also robust to measurement assumptions: for all distributions, individuals with higher levels of education also have higher per capita consumption (see Volume 11, Chapter VI). 3.38 The multivariate analysis of poverty risks (Section Hbelow and Volume 11, Chapter VII) strongly confirms the role of education as a determinant of poverty. Thus, well-designed education policies are critical for an effective poverty reduction strategy. Chapter I1 demonstrated that poor education i s not simply a legacy of the past. It i s true that many poor household heads received their education under the former social system, but the connection between limited education and poverty remains strong even among young adults Those who completed only primary education constitute a staggering 50 percent of all the poor in the 25 to 35 year old age bracket. 3.39 An important issue related to education is the question of the direction of causality. Are people poor in BiH because they lack education, or worse education i s just a consequence of poverty, and the poor lag behind in educational attainment as well as in other aspects of well-being? The relationship between education and poverty i s definitely a relationship of the cause and consequence. For some of the adults grown up in the former system, their lack o f education in the past i s often a direct cause o f low wages and poverty today. Poverty today also has consequences for the future generations. Detailed sectoral analysis (see Chapter VII) shows that education in BiH suffers from very uneven quality and i s largely irrelevant to market demand. Better quality o f education as a rule i s more expensive and i s affordable generally for the better-off families. Poor people cannot afford the types of education that guarantee higher returns on the labor market, limiting their children's' potential to earn for a better life. H. From Poverty Rates to Poverty Correlates 3.40 Comparisons o f poverty rates for different population groups are informative. But they may be imprecise, especially when there i s a strong correlation between risk factors. For example, the unemployed are more likely to be poor and to have less education. What i s the net impact o f each of these variables, controlling for other factors? Since poverty i s a multi-faceted phenomenon, where many characteristics appear to interact, it i s important to try to isolate the impact of each factor controlling for the impact of others. To this end, 45 Chapter VI1 in Volume I1 presents a multivariate analysis of poverty. It confirms that practically all the characteristics of poverty identified by simple cross-tabulations are robust. 3.41 Initiating the analysis, we divide the population of BiH into subgroups defined by easily observed attributes, and assess how concentrated poverty i s in subgroups with especially high poverty i n ~ i d e n c e .We ~ ~ use attributes that are most closely and robustly correlated with poverty: the education of the household head, the number of children in the household, the number of gainfully employed household members, whether the household remains in an IDPhefugee status, and the age group of the head of household (working age or retired). The final subdivision i s therefore a full cross-combination of all these factors, representing seventy distinct subgroups. 3.42 Defining "vulnerable" groups as those with double-the-average headcount poverty rates" and "non-vulnerable" groups as those with poverty rates 50 percent below the national average, we find that in 2001 sixteen percent of the population belonged to some of the 22 vulnerable subgroups, which accounted for 43 percent of all the poor inBiH. Non-vulnerable groups on the other hand accounted for 30 percent of the population and a sizeable 11percent of all the poor. This level of concentration of poverty is quite high, but it does not ensure that vulnerable subgroups based on key observable characteristics of households "capture" the absolute majority of the poor. Household level vulnerability attributes blur compared to the local area effects, discussed in section F above. 3.43 The next stage i s to use multivariate analysis to estimate the "net" effect of each of the five factors after controlling for the impact of other factors included in the model. The full set of results is presented in Chapter VI1in Volume 11. Table 3.5 below summarizes the findings by taking a "base" householdand showing the cumulative impact of various poverty risk factors. The Table shows that a typical "base" household in BiH has already non- negligible risks of poverty (11.6 percent), which rises with each additional "disadvantage", as shown in the respective rows. Education i s easily the most important variable; having only primary education (as opposed to completed secondary) practically doubles the poverty risk (by adding 11percentage points), all other things being equal for a base household. Table 3.5 The "net" impact of various risk factors on probability of becoming poor RiskFactors Prob. of being poor Partial effect on (Estimates) Poverty Risks Base household: 0.116 +1 member 0-5 yrs 0.163 +5 P.P Head: complete primary 0.277 +11 p.p Live in non-urbanarea 0.313 +4 P.P No one works inthe HH +1 0.384 +7 P.P elderly member 0.432 +5 P.P Remain IDPshefugees 0.515 +8 P.P inRepublika Srpska 0.583 +7 P.P Note: A base household is a couple with 2 children (one under 5, one aged 6-14) and one elderly grandparent with the head having secondary education and working in the formal sector in FBiH. Poverty is based on the general poverty line expressed in terms of per capita annual consumption.Poverty risk is estimatedbasedon multivariate analysis of per capita consumption 69For a basic discussion on the theoretical underpinnings o f this approach see `Poverty and Income Distribution InA High-GrowthEconomy. Chile: 1987-1995'. The World Bank. Report No 16377 CH. 70Using our baseline definition of poverty based on per capita consumption and a general absolute poverty line. 46 3.44 The most critical message from these simulations, however, i s that there i s no single key cause of poverty. It i s a combination of misfortunes that makes people poor. Therefore a strategy to fight poverty must be multifaceted, not only because the phenomenon of poverty i s multidimensional, but also because the causes of even material poverty cannot be reduced to a narrow set of problems. I. Conclusions 3.45 The key conclusion arising from this Chapter's analysis i s that poverty affects all a wide range o f major groups in BiH. Nevertheless, poverty has a number of distinctive features. 3.46 Concem for children i s a preoccupation of many of the poor, since two-thirds of them live in household with children. At the same time, most the poor in BiH are people of working age and a majority of them live in households where someone works. The analysis of poverty rates by characteristics shows that low labor income and the predominance of single-earner households are to blame for poverty inhouseholds with children. 3.47 Ifsingle-earner householdsdominateinBiH,thisismostlybecausewomenare significantly less likely to work than men. Inactivity and unemployment are especially significant problems for women from poor households. And these problems persist even though women tend to be at least equal to men in terms of education, and despite a relatively modest gender wage gap. Persistently large gender differences in labor market activity rates are driven by the collapse of child care options, by the lack o f available substitutes to own- household production of services and goods, and by overall anemic job creation in the country. Thus the key economic cause of material deprivation in BiH i s lack of productive employment. The problem, according to this report, i s that there are not enough people working in poor households. And inparticular, there are not enough women working. 3.48 While individual socioeconomic characteristics are a useful and important way to identify the poor, it i s chiefly a combination of disadvantages that makes a household poor. 3.49 There i s evidence o f locational poverty "traps", involving significant concentration of the poor in communities where poverty rates are substantially above the national average. This factor needs to be explored in detail once community level LSMS data become available. One of the evident characteristics of poor communities i s a breakdown of social capital and informal networks that are essential for coping with poverty inthe BiHcontext. 47 48 PART 11.POVERTY AND PUBLIC POLICIES. Part I discussed various aspects of the dimensions and characteristics of poverty. Part II reviews national conditions and public policies which directly or indirectly afSect people's lives, including macroeconomic, employment, health, education, and safety net policies. It finds that though the consequences of war have been important determinant of poverty, defective policies have been responsible for a close link between poor material well-being and deprivation in non-material dimensions. This general failure to ensure that public policies work for the poor has also had repercussions for the population at large. Each chapter in Part II focuses on a particular sector and draws policy conclusions and recommendations based on the detailed analysis. Part II concludes with an assessment of the key challenges and policy prioritiesfor poverty reduction in BiH. 4. INITIAL CONDITIONSAND STRUCTURAL REFORMS The analysis of the poverty risk can help us not only to better classify drivers of poverty, but also to understand the role of policies and institutions. We start by a quick recapitulation of the consequencesof the war on decision making process in BiH, and review key institutional features of policy making in BiH. These are shown to generate problems which inhibit successful implementation of reforms, and which ultimately hurt thepoor. A. War and Reconstruction 4.1 This section briefly reviews how the war impoverishedthe population through loss of life, displacement and migration, and through the devastation of infrastructure and production facilities. It suggests that the role of local conditions as determinants of poverty, as already identified, can be traced to differential disruption of society and its economic base duringthe war andto uneven subsequent reconstructionefforts (Box XI). 4.2 BiH was one of the lower-income republics of the former SFRY. Nevertheless, in 1991, its GDP was estimated at USD 10.6 billion, or about USD 2,230 per capita. It had a highly educated labor force and a large industrial base dominated by the production of sophisticated machinery. Unlike other centrally planned economies, Bosnia and Herzegovina had a relatively open and market oriented economy - over half o f its exports were directed to Western markets. In addition, the economy was relatively diversified: 51 percent of its GDP originated in industry, miningand forestry while some 10percent came from agriculture. All of these advantages placed the country in a position that, given appropriate policies, could have ensured a smooth and successful transition to a market economy. But the 1992-95 war fundamentally changed the country's economic structure and prospects. 49 4.3 The country's highly vertically integrated pre-war production structure and the extensive pre-war trade network between member states of the former SFRY magnified the damaging effect of the war on the economy. Physical destruction was heavily concentrated in factories and industrial premises. Government assessments of the overall costs of war damage range from US$50 to US$70 billion, while estimates of the direct damage to productive capacity of the country run from about US$15 to US$20 billion.71 4.4 Although estimates on the value of physical destruction and lost GDP vary, all assessments agree that the magnitude of the latter was greater for Bosnia and Herzegovina than for any other transition country; even those that also suffered a war in parallel with transition. By the end of the war, output had fallen to just 10 - 30 per cent o f the pre-war level. GDP collapsed to a level of less than USD 500 per capita. The war disrupted trade channels and halted nascent economic reforms, while industrial production recorded a fall of more than 90 percent inreal terms. 4.5 Post-war per capita incomes have recovered from USD 456 at end-1995 to USD 1,080 in 2000. It i s estimated, however, that at least 30 percent of the latter is the result of international assistance; comprehensive restoration of the economic and social fabric will take a long time to complete. 4.6 The level of destruction of the economy initially led political leaders and the donor community to focus their attention on physical reconstruction over the first four-five years, postponing and, to some extent, ignoring the need for the kinds o f deeper reforms which have been ongoing in many other countries throughout Eastern and Central Europe. The multitude of donor funding opportunities and the chaotic post-conflict environment also contributed to reinforcing old structures and attitudes - eventually supporting an oversized public sector, an inadequate institutional framework for private business development, and a fragmented economic space. BOX XI. Unevendistribution of.internationalpresenceand regionalinequalities The influx of cash from the expenditures of international organizations and presence o f many foreigners in BiH has been large. I t has been estimated that Sarajevo alone hosts 15,000 foreign citizens, working in international NGOs, international organizations, embassies and aid organizations o f various countries. Their spending is estimated at 60 million DM per month for living expenses alone, or 720 million DM per year. Thus, if we estimate that Sarajevo today has approximately 400,000 citizens, this creates a monthly influx of 150DM for each Sarajevo resident, or 42 percent of the average monthly salary.72I t was estimated that, with 55,000 foreigners working in BiH, their spending amounted to 2.5 billion DMper year. This is almost 60 DM monthly per BiHresident. This impact was quite heavily concentrated incertain areas. Source: IBHI(2001) '*ICG,Why 71See Bisogno and Chong (1999) . will no one invesr in Bosnia-Herzegovina?, Report No. 64,21 April 1999,p. 5 50 B. StructuralReforms: Hard Lessons 4.7 Subsequently, with most basic reconstruction completed, the donors, in close cooperation with the authorities, have shifted their efforts from reconstruction to fundamental structural reforms critical to the emergence of a market based economy. These reforms are regarded not only as the building blocks of a single economic space in BiH, but also as prerequisites for association with the European Union. 4.8 The circumstances outlined above are just a few of the consequences of war and reconstruction, but each of them has tended to tear apart the communities that make up BiH. The war altered the economic map of the country, as physical destruction was heavily concentrated in factories and other industrial sites; it i s less obvious how the subsequent reconstruction efforts have affected welfare in different parts of the country. And both war and reconstruction have been accompanied by persistent limitations on the mobility of the population, "locking in" many poor people inimpoverished communities. BOX XII. CostsofPostponingDifficultReforms: The Case of Bulgaria. For a long time Bulgaria avoided taking decisive steps in structural reforms, postponing the most difficult decisions. This was frequently justified by reference to the extremely high unemployment rate. The imbalances grew out o f control, leading to a full fledged macroeconomic crisis in 1997. Since this crisis, macroeconomic policies were brought under strict control with the currency board regime. A number of bold steps have been undertaken to achieve long awaited progress in reforms. But despite very impressive performance, including large FDI inflows and deep restructuring, unemployment continued to rise, reaching 19 percent of the labor force by 2001. Why has unemployment continued to grow despite positive signs in the economy? The recently completed Poverty Assessment for Bulgaria finds an explanation for this paradox in the long-delayed restructuring. As a result o f postponement, all o f the employment reallocation in the economy is happening simultaneously once reforms have begun in earnest. The new demand from growing firms has yet to absorb enormous redundancies from labor shedding enterprises. Meanwhile, deep structural imbalances between labor market demand and supply, which were hidden by suppressed labor adjustment, have emerged. The Bulgarian example shows that the consequencesof postponing key restructuring steps out of fear o f high open unemployment may, ironically, lead to even higher unemployment. Source: Bulgaria Poverty Assessment. The World Bank (2002). 4.9 Implementation of structural reforms needed to orient the economy towards sustainable growth has been slow. Little has been done to improve the legislative environment or to eliminate administrative barriers to investment. It currently appears that it will be hard to accelerate or even to maintain recent average growth rates, given the context of declining aid flows and lagging economic and legal reform, particularly in the area o f privatization (see Box XII). Meanwhile, important economic imbalances still exist between the two Entities. 4.10 The process o f weighing difficult policy trade-offs and acting on them i s often complicated by BiH's fragmented post-war governance arrangements and weak social and political consensus. As a result, enterprise privatization has only recently begun after long 51 delays. Discretionary administrative intervention in the economy limits private sector development and encourages corruption. Slow progress in transition reforms has contributed to a weak private sector response. 4.11 Until 1999, BiH found it difficult to agree on any common national interest that would serve as a blueprint for development. The failure to define a vision and transform it into a common policy resulted in short-sighted politicking and further fragmentation of power. 4.12 This institutional fragmentation perpetuates the main ethnic cleavages of post-war Bosnian society, and helps to explain why, in the post-war period, these cleavages often take legal and statutory form. The best illustration of this problem i s the spatial and ethnic fragmentation of social insurance,73 which leads to various types o f discrimination and explains why minority returnees often are not affiliated with the health insurance or pension fund of their place of return.74Fragmentation has other consequences in terms of lost opportunities. 4.13 Tradepolicies offer an example of how lack of capacity for implementation of good decisions intended to overcome fragmentation turns them into forces that do nothing to promote growth. On paper, average tariff rates in BiH (6.8 percent) compare favorably with average ad valorem rates in the region (Albania. 12 percent; Croatia, 12.4 percent; FRY, 13.4 percent; and Macedonia, 16.5 percent). 75 But in fact, BiHrelies heavily on import duties as a source o f tax revenues and on various ad hoc surcharge duties, for example, on agricultural products, to respond to the demands of interest groups. In effect, it i s not the official trade regime that limits the ability of business to fully exploit the potential of the international division o f labor, but rather the lack o f clear rules and their enforcement. 4.14 Although the Law on Customs policy i s based on international best practices, BiH exporters and importers encounter various problems with how the system works in practice. The cost of clearances i s not trivial, organization i s cumbersome, and corruption in a country with 400 illegal border crossing points i s omnipresent and distorts the incentives incorporated inthe legal f r a m e ~ o r k . ~ ~ 4.15 A recently completedregional trade study demonstrates the failure to capitalize on the very favorable trade regime that BiH enjoys with the EU.77 The EU Stabilization and Accession process offers trade preferences to the countries of the Western Balkans. Institutional bottlenecks and inability to enforce a minimum level o f inspections and certification consistently across Entities have prevented the country from benefiting from these preferences. For example, not a single company in BiH has a EU veterinary export number; and state veterinary service was established as late as in March 2001 but has not yet 73Each Entity and, in the Federation, each canton has its own social insurance system. See Stubbs, Paul and Gregson, Kendra (eds.) 1998; OSCE -Human Right Department 1999a. 74See UNHCR 1999a; UNHCR 1999d; UNHCR 2000b. 75Trade Policy and Institutions in South Eastern Europe. The World Bank 2002. 76Hark Mesinovic, Zlatko Hurtic, Mersida Suceska "Foreign Trade Institutions and Policies. Country Studi. Bosnia and Herzegoniva" In: Trade Policy and Institutions in South Eastern Europe. The World Bank 2002. 77Trade Policy and Institutions in South EasternEurope. The World Bank 2002. 52 become operational. As a result, not a single ton of animal products i s shipped for export into EU countries. BiH has also failed to accede to relevant international agreements on standards; consequently, BiH institutions' certificates are not recognized. On the one hand, this makes exporting very difficult; on the other, it creates opportunities for imports of cheap low quality products which harm domestic producers. 78 4.16 In addition to these institutional problems, BiHbusinesses are severely hampered in making full use o f preferential arrangements by the obsolete technology and high labor costs (relative to the level of productivity) compared to other exporters to the EU.79Thus, the advantages o f the liberal external trade environment have not been fully exploited-problems which have been exacerbated by supply constraints rooted in a weak private sector environment. 4.17 Infrastructure policy. The transport system provides several examples of the coordination failures and other difficulties facing infrastructure sectors in BiH. There i s no coordination o f road safety activities between the two Entities, and only slow progress in moving beyond addressing basic transportation bottlenecks resulting from the war. Despite the success of the emergency transport reconstruction program, 8o a significant part of the main road network i s still in poor condition. Roads are maintained by separate road administrations for the FBiH and the RS. At the State level, an organization with responsibilities for road maintenance and development has been established (the Bosnia and Herzegovina Road Infrastructure Public Corporation (BRIC)). BRIC was established in February 2000, but its operations have been considerably slowed by conflicting interests. BiHis still excluded from a very important TIR network that greatly facilitates international truck movements. 4.18 The roots of the problem of poor infrastructure lie deep in the policy making process inthe post-conflict economy. Communications, transportation, andpower have been for too long sources of revenues for the governments and political parties controlling them, and were split based on ethnicity. Therefore there was little interest to proceed with their restructuring and privatization, and to open these sectors to competition. 4.19 This diagnosis of fragmentation i s specifically emphasized by sociological studies, such as a recently completed study o f social capital and local level institutions. The study find roots of fragmentation in the traditional parochialism of Bosnian society.81 Mutual resentments and prejudices between the urban and rural populations are another example of a traditional parochial attitude, which has been exacerbated by the population movements of the war and post-war period. The study noted that urban residents complain that municipal 78 Mesinovic H.,Hurtic Z. and Susecka M. Op. cit. 79 As documented in IMF(2002). This program was implemented through the Emergency Transport Reconstruction Project (ETRP) and its successor, the Second Emergency Transport Reconstruction Project (SETRP). The primary objective of the ETRPand SETRP was to remove critical infrastructure bottlenecks inorder to permit the resumptionof operation of the country's major transport network. 81 Many social scientists consider parochialism to be a central characteristic o f all Balkan societies, and some have been interested inthe way Yugoslav federalism and self-management during the Communist period reinforced this traditional feature. See Local Level Institutions Study. The World Bank 2002 for references. 53 bodies are monopolized by people coming from surrounding villages, while villagers and inhabitants of peripheral suburbs say that municipal structures are controlled by people from the city center. Parochialism influences the way Bosnian citizens conceive of political representation and citizenship. Many survey participants and other informants complain that the taxes they pay are not directly used in their community, or that enterprises settled there are employing people from outside. In BiH, trust among people living in the same place (bonding social capital) takes precedence over trust towards people from outside (bridging social capital), affecting a range of civic values and confidence in formal institutions (linking social capital). 4.20 The situation depicted above needs to be tempered by acknowledgement of a strong countervailing factor to fragmentation: that of an emerging sense of national unity and common national interest. This often takes the form of a common economic interest. For example, freedom of movement across the Inter-Entity Boundary Line (or across territory controlled by another ethnic group within FBiH) has improved greatly in recent years (especially since the establishment of a common license plate policy in 1998) and has been a first step towards closer economic integration. 4.21 This tendency has helped to initiate a nation-wide consultative process leading to a national economic development strategy. It has also laid a strong foundation for the country's Poverty Reduction Strategy Paper. Especially since 2000-01, many laws and regulations that are crucial for the establishment of a single economic space have been passed. 4.22 One of the key elements of this single economic space i s the establishment of a Central Bank and a Currency Board arrangement. The Central Bank i s judged to be at the core of Bosnia and Herzegovina's success in restoring growth and stabilizing the economy,82 reducing inflation to single digits and restoring confidence in the currency. The peg of the national currency (KM) to the Deutsche Mark since July of 1997 and subsequently, to the Euro, has been crucial in providing policy credibility and currency confidence. C. Conclusions and Policy Implications 4.23 This chapter has suggested that two types of factors related to poverty in BiH have been at play in recent years: those which confront most other economies in Central and Eastern Europe - related only tangentially to the war and post-war politics - and those which are unique to the post-conflict environment in BiH. The war is the primary but not the sole cause of poverty. There i s a strong interaction between the two types of factors: the post- conflict environment and the legacies of war make general transition problems more daunting, while transition challenges tend to perpetuate many o f the consequences of war. This has led to limited progress with the structural reforms needed to ensure that the poor benefit from opportunities opened by growth and reconstruction. Very often the policies adopted has been overly cautious, leading to mounting delays in restructuring the economy and the public sector. ** SeeIMF2002 54 4.24 The key message from this analysis i s that there i s a price to pay for delaying reforms, and this price i s often paid by the poor. The chapter strongly supports the focus of the national PRSP on measures that would promote a single economic space. One of the key elements essential to the credibility of a single economic space remains the common Central Bank and Currency Board arrangements. 55 56 5. GROWTH,EMPLOYEMENTAND THE BUSINESS ENVIRONMENT Macroeconomic policies directly and indirectly affect living standards, and the extent of poverty and inequality. This Chapter assesses macroeconomic peij+ormance and its links to employment. It shows that the key problems in translating growth into poverty reduction are related to poor labor market peij+ormance. The latter in tum can be traced to a poor business environment. The private sector can deliver more for the poor than it currently does in BiH. A. Macro Policies and Growth 5.1 Economic trends since the end of the war have been strongly linked to external assistance. Following the signing of the Peace Agreement there was an initial spurt o f reconstruction and strong growth. Economic strategy was based on strict adherence to a Currency Board arrangement, prudent fiscal management and access to large-scale external assistance on concessional terms. This phase was followed by a second period characterized by the consolidation of stabilization, moderately high growth rates and important initial reforms. By 2000, however, the economy seemed to have entered a period of sharp deceleration of growth. After leveling off at 10 percent in 1998 and 1999, real GDP growth dropped to about 5 percent in 2000. IMF staff estimates suggest that growth slowed further to around 2.3 percent in 2001, with the FBiH's GDP slowing to 4 percent in 2001 from 7 percent the previous year and the RS's officially measured GDP shrinlung by 1.9 percent in both years. 5.2 In line with large inflows of aid and remittances that covered huge foreign trade deficits, consumption was constantly well above GDP over the whole period. There are still no National Accounts data for BiH brealung down GDP by its components, but rough estimates suggest that the growth rates of private consumption were comparable to overall economic growth, remaining well in excess (in terms of GDP share) of levels observed in other countries in the region. This i s a very important structural peculiarity of the BiH economy that helps to explain the context for poverty outcomes. 5.3 The key macroeconomic problem facing BiH i s that the aid-financed post-conflict boom lost momentum before output recovered to its pre-war levels. While the country has made substantial progress in recovering from the war, reform efforts to date have not been bold enough to place the economy on a sustainable growth path. Six years into the reconstruction program, the objective of moving from reconstruction and recovery to sustainable growth remains a major challenge. 57 5.4 Private investment and foreign direct investment have been inhibited by concerns over political risks, the hostile business environment, and infrastructure bottlenecks. The export sector in particular has not played a major role in the recovery process and i s still at the early stage of revitalization and adjustment to new markets and demands (Table 5.1). - Table 5.1 Bosnia and Herzegovina: Main Economic Indicators, 1997-2003 1/ - - 1997 1998 1999 2000 2001 2002 2003 Est. Proj. Proj. RealGDP growth (annual average) 2/ 36.6 9.9 9.9 4.5 2.3 2.3 4.1 Federation 36.2 8.3 10.4 7 4 3.2 5.6 RepublikaSrpska 37.9 15.8 11.3 -1.9 -1.9 0 0.5 Fiscalbalance,Federation (Inpercentof Entity GDP, commitmentbasis,Consolidated GeneralGovemment) Revenue ... 47.7 48.7 48.2 46.6 46.9 47.6 Expenditure ... 51.7 53.2 53.1 48.5 49.5 41.5 Balance ... -4.1 -4.5 -4.9 -1.8 -2.5 0.1 Fiscalbalance, RepublikaSrpska (Inpercent of Entity GDP, commitmentbasis, Consolidated GeneralGovemment) Revenue ... 39.9 43.7 43.5 43.2 43 43.3 Expenditure ... 45.2 50.4 51.7 47.5 43.8 43.1 Balance ... -5.3 -6.8 -8.2 -4.3 -0.8 0.1 (In millions of US.dollars;unless otherwise indicated) External current account balance (excluding official -1,482 -873 -975 -971 -1,044 -1,015 -948 transfer) As apercentageof GDP -43.3 -20.9 -20.8 -21.6 -23.1 -21.3 -18.7 Exports 575 702 744 903 1,002 1,165 1,388 Sources: Dataprovidedby the authorities, and IMFstaff estimates 1/ Datarefer to the entire country, unless otherwiseindicated. 21 Startingin 1997inthe Federation,andin 2000 inthe RS, GDPi s estimatedbasedon 1993 SNA methodology. 5.5 This fall-off inhoped-for levels of growth comparedto original targets requires serious rethinking o f the achievements and challenges o f reform. It also calls for realism in setting targets for poverty reduction. Figure 5.1 compares a number of forecasts (including those of the I-PRSP) and actual outcomes. 58 Figure 5.1 BiH: GrowthRate Forecasts: Initial Targets and Revisions 20 18 - Actual and latest forecast 16 14 +1999 Donors conference report -- 12 'i10 I-PRSPforecast 8 A IMFStatus Quo Scenario 6 4 2 0 1998 1999 Zoo0 2001 2002 2003 2004 2005 Sources: I-PRSP, IMF(2002) and World Bank (2002). B. Employment Generation 5.6 The strong economic growth experienced through 1999 did not translate into a comparable growth in employment. Following an initial rebound during 1995-1997, the recovery of employment to pre-war levels has remained unfulfilled, as the real economy has not been able to generate morejobs (Figure 5.2). Figure 5.2 BiH. EconomicCollapse,Growthand Employment - . - - .. . - - .. - - - ---- I I - I 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2- 2001 I-Formal Employment, 1990=100 CPercapita GDPin currentUS$, 1990=1001 Sources: IMF, World Bank, Dataon employment are for formal sector only (see Labor Market Study). 59 5.7 Inthe absence of data on the informal sector, the Figure only reports the evolution of formal sector employment. Now employment in the formal (registered) sector in Bosnia and Herzegovina i s substantially below its pre-war level. In 1997, it was 575 thousand, or only 59 percent o f the comparable 1991 levels3. Since 1997, the growth of formal employment has been anemic: in FBiH, it has been slowly increasing so that in 2001 it exceeded its 1997 level by 9 percent; formal sector employment in RS actually fell after 1998. In 2001, 407 thousand workers were employed inthe FBIH,and 220 thousand in the RS. 5.8 The LSMS estimate of total employment (including informal sector jobs) in 2001 i s 1 million, which i s 24 thousand more than formal sector employment in 1991. It i s likely that the informal economy has increased in the postwar period, but undoubtedly it was also significant in 1991. We can therefore conclude that although the reduction of formal employment exaggerates the reduction in overall employment, the latter must nonetheless be very significant. 5.9 Such poor performance can be explained partly by poor initial condition^,'^ and partly by conflict, but the key source of the problem has been the unimpressive record of private sector employment growth. By the end of 2000 half o f all those employed in the economy - including informal sector workers, unpaid family workers, etc. - were in the public sector, a share higher than that in almost all countries intransition. 5.10 The benefits of recovery have been accruing mostly to those employed in the formal economy, dominated by the formal public sector. And, as the World Bank Labor Market Study (2003) has shown, job mobility into and out of this sector has been very slow. Comparing the profiles of public sector workers based on pension fund records, the study finds that mostjobs are taken by the same people all over the studied period (1991-99). 5.11 Duringthe 1997-99 period of very rapid growth, job flow intensity in larger Bosnian enterprises (those with more than 10 workers) was very modest. The overall job creation rate was a mere 4.2 percent, the job destruction rate was 5.3 percent, and measures o f enterprise restructuring were also quite low - the gross job reallocation rate was 9.6 percent and the excessjob reallocation rate was 7.8 percent. The rates were even smaller for enterprises with more than 100 workers, putting Bosnia behind other transition economies in job flow dynamics. 5.12 Small and medium enterprises were the key source of new jobs. According to the Labor Market Study, enterprises with less than 25 workers were the only ones that created new jobs in the period 1997-99; other formal sector firms were either not creating new jobs, or were net destroyers o f jobs. In terms of the role of small and medium enterprises in total 83 Table 2.3 and 4.7 in Labor Market InThe Postwar Bosnia And Herzegovina: How To IncreaseWorker Mobility And Job Creation?The World Bank (2002),based on StatisticalYearbook 2001, FederalOffice of Statistics, Sarajevo; StatisticalYearbook of the Republic of BiH 1992,FederalOffice of Statistics, Sarajevo; Republicof Srpska Instituteof Statistics, BanjaLuka, 2000. Informalsector estimatesare World Bank staff estimates basedon BiH LSMS (2001). 84 These included the high levels of labor hoarding that were prevalent before and immediately after the war (largely politically driven), and an unfavorable inherited industrial structure compared to other transition economies of Central Europe, with a disproportionate share o f heavy industry, defense, and mining. 60 employment, BiH i s similar to other countries in transition (with around 15 percent of wage employment inenterprises with less than 10 workers). Figure 5.3 Non-agricultural self employment as percentof total non-agriculturalemployment 12% 9% 6 8 i 3% Slovakia Romania Croatia BM Slovenia Hungary Poland Czech Republic Sources: Non-agricultural self-employment:for BiH LSMS 2001, for other countries Yearbook of Labor Statistics, ILO, 1998, for Croatia -Economic Vulnerability and Welfare Study.World Bank 2000. 5.13 A second type of employment with a very important role inproviding jobs to workers in transition economies is self-employment. In terms of non-agricultural self-employment, BiH lags behind several CEE transition economies, as shown in Figure 5.3, suggesting that the potential of this sector i s still underdeveloped. And we shall see that there are important barriers inhibiting the development of self-employment in BiH. But this factor in itself cannot explain a huge difference in employment rates between BiH and other countries, nor its weak employment generation record over the last several years; rather it reflects the generally poor state of the business climate inBiH. 5.14 The tax regime i s a limiting factor, but not a particularly strong one, impacting on employment creation. The total contribution rate calculated on gross wages (including payroll tax) i s 48.4 percent in FBiH, and 48.5 percent in RS. With respect to employers' contributions for social security and taxes, for each KM 100 of take-home pay, the employer has to bear an additional cost of KM71.6 in FBiH,and KM77.6 in RS. These numbers are comparable to those in other transition economies. 5.15 Wage determination remains inflexible and not conducive tojob creation. The way the collective bargaining process i s organized makes it very centralized and insider-biased, with heavy influence exerted by State structures. Detailed wage distribution analysis conducted in the Labor Market Study found that wages are still quite compressed, and that there i s an important "bunching" of workers around the minimumwage, indicating that it i s a bindingconstraint inwage setting. 5.16 Wage setting mechanisms have contributed to very slow employment creation and limited distribution of the benefits of growth. Despite strong growth o f wages in recent years, earnings per worker are still below what i s needed for a family with children. Nominal average net wages roughly doubled between 1997 and 2001. But the average wage o f 400 61 KMper month will only bring 1200 KMper person per year for a family of four, which is insufficient to raise it above the poverty line. 5.17 Could wages increase further, reducing or even eliminating poverty for those employed? Several indicators suggest that wages expressed in dollar terms in BiH already exceed the levels among its main competitor^.^^ Thus raising wages further would undermine export growth as a key source for economic development and will not be a viable option for poverty reduction. It will also left out significant number of poor that are currently in the jobless households. Poverty reduction therefore depends on more employment and risingproductivity of existingjobs. 5.18 To understand how poverty i s related to economic growth and productivity, we examine the changes in poverty rates across employment sectors over time. Infact, at any given point in time, a high poverty rate associated with one sector of employment need not mean that working in that sector i s a cause of poverty. Infact, the opposite may be true: the sector may be offering income earning opportunities in difficult times for the most unprivileged. 5.19 While the LSMS and CEIT surveys are not comparable, we can roughly compare the sectoral poverty rates in 1997 and 2O0lg6 to seek some insights into the stability of the poverty profile. One of the most striking results o f this rough comparison i s a significant change in sectoral poverty rates. Bisogno and Chong (2002) find that public enterprise employees (and their families) have the highest poverty rates of all employment groups and that they constitute more than 50 percent of all the poor. This striking result in not confirmed inthe LSMS data, which indicate that public sector employees have the lowest poverty rates. But on the other hand, we do find that the position of workers in state owned enterprises has not improved. 5.20 There were severe rigidities in employment legislation until labor market legislation was revised in the fall of 2000 (for example, there were large firing costs under Art.143 in the Federation and Art. 53.a in RS o f their respective Labor Codes). These rigidities, coupled with the paternalistic attitudes o f managers, militated against dynamic adjustment and job creation. One of the results was a propagation o f the phenomenon o f "waitlisted" workers (kept on employers' payrolls without actually working and at only a minimum ~alary).'~The adaptation o f new legislation reduced many barriers to more dynamic job creation, but it i s still too early to assess its impact on workers. 85See Labor Market Study. 86As already noted, the CEIT survey was basedon sampling. The welfare information is strictly non- comparable because CEIT collected responses based on income, while LSMS data are based on consumption. However, the work on other countries (for example, Croatia [2001]), where both income and consumption data were available for the same household in the survey, provides some justification for our comparisons, as it found remarkable similarities between rankings o f households by consumption and by income. 87According to the UNDP 1998 Human Development Report there were 115,000 waitlisted workers inBosnia and Herzegovina in 1998; by mid-2001 7 percent of the labor force inFBiHand 12 percent in RS were waitlisted. B y the end of 2001 the law (1992) governing workers' waitlisting was abolished, but the practice of labor hoarding has continued in modified form according to the most recent accounts (Labor Market Study). The LSMS study found that only 35 thousand workers reported being waitlisted at the end of 2001, mostly in 62 5.21 Changes in the core labor market legislation in the Fall of 2000 - the adoption of the Law on Changes and Additions to Law on Labor in FBiH,and o f the Labor Code of the RS - introduced modem employment legislation and did away with the notion of waiting lists.88 Current Bosnian employment protection legislation, as well as regulations governing fixed- term contracts, i s quite comparable to - and in some instances more liberal than - the legislation and regulations in Slovenia and developed market economies, and thus does not represent a barrier to labor reallocation. 5.22 In sum, while the pre-2000 legislation imposed onerous costs on employers when laying off workers, new legislation in both Entities represents a major improvement. It aligns Bosnianjob security legislation with European norms, and i s unlikely to impede mobility and job creation. 5.23 The informal sector emerged in the past as a way to avoid the inefficiencies generated by the pre-reform formal labor market and, to a lesser extent, as a way to avoid taxation. Growth o f the informal sector introduced labor market segmentation, exacerbated in BiHby the displacement experienced by many workers: displaced people often engage in particular kinds of jobs which are not attractive to other groups, and unemployment rates are higher among IDPs than among the rest of the population. Segmentation of the labor market and discriminatory practices become especially evident when focusing on groups which tend to be severely disadvantaged in finding employment, such as women, former combatants and young people directly affectedby war.89 5.24 Thus, poor employment generation primarily reflects very slow restructuring of the large (mostly public) enterprises, absence of new entry by individual entrepreneurs, and violation of rights imbedded in the legal framework. But limited human capital also imposes constraints. 5.25 Tables 5.2 and 5.3, based on LSMS data, suggest the extent to which higher levels of education are associated with a higher probability of being employed, especially in the public sector. RS. Official Gazette o f the FBiH,No. 32/00, and Official Gazette o f the RS, No. 38/00. 89Due to outdated skills and difficulties in adapting to civilian life, former combatants are generally disadvantaged in the tight post-war labor markets in transition countries; in BiH, however, this group enjoys a number of privileges on the labor market. Nevertheless, former combatants and young people affected by war still experience high levels o f stress and anxiety in their daily lives together with typical signs of trauma, including depression, listlessness, chronic fatigue, anguish, psychological disabilities, and recurrent recollections o f traumatic incidents. These symptoms are commonly associated with difficulties in reinsertion into the social fabric in general, and the labor market inparticular. 63 Table 5.2 M e n : Labor market outcome by school attainment (prime age workers, 18-35 years old), percentage shares Secondaryl Junior Primary Secondary Vocational College University Total Out o f labor force 20.4 12.5 15.3 1.2 5.0 15.3 Discouraged worker* 14.7 8.8 13.3 11.0 15.4 12.4 Student 0.5 24.7 1.2 0.8 1.2 7.3 Unemployed 19.6 12.7 15.9 9.2 10.2 15.7 Informal sector worker 31.2 12.1 24.7 3.1 8.6 22.2 Formal sector worker 13.7 29.1 29.6 74.8 59.6 27.1 Source BiH LSMS 2001 Nore: Percentages add to a 100 vertically. *Willing and ready to work, but without work andnot actively searchingfor employment. Secondaryl Junior Primary Secondary Vocational College University Total Out of labor force 42.9 12.4 28.2 7.1 5.3 27.4 Discouraged worker* 29.1 11.3 22.4 5.0 10.8 20.8 Student 1.3 37.4 2.2 9.1 2.5 12.0 Unemployed 8.6 8.4 14.1 14.2 8.7 10.7 Informal sector worker 13.9 9.4 10.9 4.5 19.0 11.4 Formal sector worker 4.3 21.2 22.3 60.1 53.8 17.6 1 5.26 It i s interesting to note the differences between those who graduated from a gymnasium (general secondary) and those who graduated from a vocational school. In the case of men, those with vocational secondary diplomas, as opposed to general secondary degrees, have a slightly higher tendency to be out of the labor market, to be discouraged or unemployed, and a much higher tendency to be working in the informal sector. Inthe case of women, those who have vocational as opposed to general secondary degrees also have a much higher tendency to be out of the labor market or discouraged, but their participation in the informal sector i s roughly the same as those with general degrees. Similar patterns appear for the population older than 36 years of age (results not shown). Higher education i s again associated with higher rates of labor market participation, especially for women, and higher participation in formal sector work. C. Business Environment 5.27 As already noted, notwithstanding some recent progress with structural reforms, the business environment in BiH remains weak and unable to attract desperately needed private investment. Several studies agree that overly complex and costly government structures and regulations, coupled with a lack of transparency, unpredictability and high costs have led many prospective investors to consider the risk of doing business in BiHas too high." 90 The World Bank and European Commission's Bosnia and Herzegovina: Review of the Priority Reconstruction Program and Looking Ahead Toward Sustainable Economic Development, May 1999; The Wall Street Journal's Central European Economic Review, November 1999; The World Bank Group's World Business Environment Survey, Spring 2000; Foreign Investment Advisory Service (FIAS) report Bosnia and 64 5.28 A Wall Street Journal survey of leading economists ranked BiH'sinvestment climate 19th among the 27 countries covered (between Armenia and Moldova); 18th for business ethics (just ahead of Russia and Ukraine); 21st for integration into the world economy (behind Georgia and Armenia); and 21st for rule of law (behind Ukraine and Azerbaijan). BiH's own assessment9' ranked BiH lowest among 59 countries on most measures of competitiveness, such as openness, governance, finance, infrastructure, technology, management, labor and civil institutions. The study ranks BiH at the bottom on nearly 80 percent of the 184indicators used. 5.29 BiHranks lowest among countries in ECA in terms of FDIinflows over 1996-1999. More recently, according to IMF balance of payments statistics, FDI has significantly increased, as privatization has begun to take hold and the banking sector has started to attract foreign investors. Slovenia has surfaced as the primary foreign investor in BiH, and traditional ties with Germany and Austria are increasing. However, continued problems associated with business entry and operations, as well as persistent protection for state enterprises, point to a frail system in deep need of reform. Box XV captures the frustration and cynicism that permeates the domestic and foreign business community in BiH. 5.30 Some progress has been achieved recently in creating the infrastructure necessary for privatizing large state enterprises. The FBiH has made significant progress in the area of bank privatization. Inboth Entities, "payments bureaus" have been replaced by a commercial bank-based payments and clearing system. Bank regulation and supervision have been strengthened over the past year, although procedures in place continue to fall short of best practice in some areas. The RS still lags behindFBIH, however. 1 BOX XIII. Businessis difficultinBiH. A foreign investor in BiH recently portrayed the current situation as follows: "Yes, there are new laws. But they are superimposed on old Yugoslav laws that have not been repealed. The bureaucracy doesn't know much about the new laws. They believe in the old laws and regulations. And there are State laws, Entity laws, Cantonal laws, municipal laws-who knows how many laws there are. Businesses are at the mercy of the bureaucracy. And the courts do nothing to protect property or enforce contracts. Corruption is everywhere. The whole system i s a mess." Source: World Bank (2002), Business Environment Adjustment Credit. Project Appraisal Document. 5.31 Once investors have surmounted the administrative maze which they need to master in order to begin operations, they are then faced with complying with thousands of regulation^^^ that are often inconsistent and outdated, massive reporting requirements, and inspections by numerous inspectorates with unclear and often overlapping mandates. This provides a fertile ground for rent-seeking behavior by Entity- and local-level inspectorates. Herzegovina: Commercial Legal Framework and Administrative Barriers to Investment, March 2001; International Crisis Group's (ICG) report Why No One Will Invest in Bosnia and Herzegovina, 1999; ICG's draft report Still Not Openfor Business: Bosnia's Precarious Economy, July 2001. 91 Bosnia and Herzegovina Competitive Report, May 2001, a study by a team o f BiHeconomic experts on the basis of the methodology of the World Economic Forum for 1999, was the first of its kind since the country roclaimed independence from the former Yugoslav Federation in 1992. 82 BiHBusinessEnvironment Adjustment Credit Project Appraisal Document reports that inMarch2001 a company must comply with 11,000 regulations. 65 Domestic investors estimated that as much as 20 percent of administrative time i s spent on compliance issues; one foreign investor reported that the ratio of administrative costs to production costs was an alarming 1:1. 5.32 Bosnian firms are under pressure to adapt to the changing circumstances associated with transition, but face formidable barriers in doing so. With national resources drained by the war and the reluctance of foreigners to invest in Bosnia, there has clearly been a dearth of resources available for investment. But there have been other limitations as well, not the least of which i s the lack of investment opportunities themselves, due to a non-conducive climate for the growth of businesses. The key message from the Labor Market Study i s that low quality o f government services, late payment of bills, and lack of trust in the economy are all interlinked: an inefficient and corrupt bureaucracy i s unable to establish and maintain financial discipline. And as companies, too, fail to pay their bills on time mistrust grows not only toward the state but also toward other players in the economy. The result i s a vicious cycle of mistrust that not only drives up transaction costs but also prevents companies from growing. 5.33 Trust is important for any firm, but is especially important for the small businesses that form a key engine of employment generation in transition economies. Their small size implies lack o f diversification and consequently higher business risk. Generating trust, promoting cooperation, and reducing both risk and transaction costs in the economy and society, all support employment generation. What i s missingmost in BiHi s a culture of trust that creates expectations of appropriate behavior-not only among narrow groups of people who know one another but also across the business community at large. D. Conclusions and Policy Implications 5.34 Post-war employment outcomes in BiHhave not kept pace with economic growth. In addition, the economy has seen only limited reallocation of labor resources. Economic revival has resulted in simply keeping the previously employed where they were before the war. Little has been achieved in facilitating labor reallocation to more productive firms and occupations. As a result the quality of employment deteriorated, while the quantity remained below the necessary threshold. 5.35 BiH also faces depressed labor demand, which leads to low wages and diminishes people's interest in participating in the labor market. A significant process of painstaking institutional reform i s a key condition for achieving better employment outcomes and generating better opportunities for the poor. It should be noted, however, that some of the measures needed to complete the transition in BiH (especially, dealing with public sector overstaffing), and hence to promotejob creation in the long run, will lead to large short term job losses. 5.36 But the poor also have to have adequate human capital to be able to benefit from improved opportunities. Thus the next three chapters review policies in the areas of health, education and safety nets-all of which are not only important in their own right, but are also key factors for improving employment outcomes. 66 6. HEALTHAND POVERTYINBOSNIAAND HERZEGOVINA. According to oflcial data, BiH is on track to meet the Millennium Development Goals in both health and education by 2015. In education, primary enrollment rates are close to 100 percent; in health, infant mortality is declining toward OECD levels. Reconstruction efforts and massive international assistance in both sectors were successjkl in restoring overall physical capacity to deliver services to the population. However, closer examination reveals that, under the surface, the picture is less positive. Gaps in access to both education and health services are real, particularlyfor vulnerable groups, including thepoor. This Chapter focuses on the health sector. It identifies significant inequities between the poor and the non-poor in accessing health services. In addition, the analysis shows that the existing health insurance system does not offer the average citizen enough protection against risks of catastrophic illnesses. Even those who are insured typically have to incur considerable expenses on health care - enough to be one of the shocks that can move a household intopoverty. A. Introduction: Overview of the Health Sector 6.1 Before the conflict, the health system in BiH resembled that o f many other Eastern European countries, offering easy access to comprehensive health care services for the whole population. The system was hospital based, however, relying heavily on specialized polyclinics and only marginally on primary health care. The sector was under increasing cost pressures, a problem that was exacerbated by the war and the difficult process of transition to a market economy, duringwhich public resources abruptly diminished but no adjustment was made to health infrastructure and staff levels. 6.2 The war took a heavy toll on the health of the population. During the conflict an estimated 250,000 people died and more than 175,000 were injured (World Bank, 2000). The health system was also a casualty of war. An estimated 30-40 percent o f all hospitals were damaged, and around 30% o f health care practitioners were lost to emigration or death (Jakuboswsky & Jaganjac, 2001). Meanwhile, the large percentage of displaced people and the high numbers of sufferers from psychological trauma in the aftermath of war have imposed new demands on the health care system (see Box XIV). 6.3 Psychological traumas resulting from the war, such as anxiety, depression, post- traumatic stress disorder, sleeplessness, behavioral disorders, obsessive memories, irrational fear and the inability to form emotional bonds, cannot always be satisfactorily quantified, but are expected to continue to be a health challenge for the next generation. To some degree, 67 figures showing an increase in substance abuse, suicides, domestic violence, juvenile delinquency and divorce (see UNDP 2002) are a reflection of these traumas. 6.4 After the Dayton Peace Agreement, responsibility for the health sector was delegated to the two sub-national Entities, the Republika Srpska (RS) and the Federation of BiH (FBiH). The latter in turn delegated part of this responsibility to its 10 cantons. After the signing of the Agreement, the Entities approved separate health legislation. The cantons are accountable for the finance, management and provision of all levels of care (with the exception of catastrophic tertiary health services and other expensive vertical services which are financed by the Federal Health Insurance Fund), and each also has its own health insurance fund. At the Federation level, health i s a responsibility o f the Federal Ministry of Health (with little authority), the Federal Health Insurance Fund and the Federal Public Health Institute. Responsibility for health matters at this level i s limited to functions that cannot be executed at the cantonal level. In the Republika Srpska health care i s mainly a responsibility of the central government through three institutions: the Ministry of Health and Social Welfare, the Institute of Public Health, and the Health Insurance Fund. Local governments are responsible for primary health care. Finally, the Brcko District, with an estimated population of 90,000, was created in the year 2000. This district has its own obligation to organize, finance and deliver health care services-as expressed in the Law on Health Care o f the Brcko District of Bosnia and Herzegovina. Additionally, until very recently (introduction of the Federal Solidarity Fund and the abolition o f the national/ethnic Health Insurance funds), the FBiHhealth system was polarized along ethnic lines. 6.5 The fragmentation and lack of coordination between the different "sub-systems'' of the health sector has created many inefficiencies and equity problems. These include: (i) highcosts of provision owing to failure to achieve economies of scale; (ii) of portability lack of insurance coverage for people traveling across Entities or FBiH cantons; (iii) marked inequalities in the health care provision in the FBiH produced by the varying economic strengths of the cantons and their heavy dependence on cantonal resources to finance health care; and (iv) the large percentage of the population, especially the poor, not covered by health insurance. Both Entities are trying to solve some o f these problems. For instance, FBiHrecently established a Federal Solidarity Fundas part of the Federal Health Insurance Fund (FHIF). Currently, 8% of the cantonal payroll health tax revenues, along with equivalent transfers from the Federal budget, are assigned to the FHIF for the Federal Solidarity Fund, whose main objective i s to finance potentially catastrophic tertiary health services and selected expensive services such as hemodialysis. 6.6 In the post-conflict years many efforts have been made to reform the health system and to help it recover from both the damages of war and the inefficiencies inherited from the pre-war era. The last several years have seen the beginning of wide-ranging health sector reforms, covering financing, costing, and organization, and management o f health care services. Today, taking reform in the social sectors as a whole, the health sector i s the furthest aheadinthe process. However, muchremains to be done. 6.7 This Chapter will analyze access and equity issues in the health sector in BiH, using information collected in the BiH LSMS 2001. The discussion i s intended to help decision 68 makers by providing information on: health status and insurance provision; health care utilization; barriers to access to services; and public and private expenditure on health. B. The Population's Health Status 6.8 Available information suggests that significant progress in improving the health status o f the population has been made since the conflict. Life expectancy for men in 1990 was 69 and for women 74; by the year 2000 they were 71 and 75 respectively (World Bank, 2002b). Immunization rates are also high: in 1999 they were more than 90% for BCG, DPT3 and Polio3, and 83% for measles (UNICEF, 2002). In 1991, infant mortality was 14.6 per 1000 live births; by 1998 it had fallen to 11.7 per thousand. T o put these numbers in perspective, BiH's infant mortality rates are still more than twice those prevalent in Western Europe but are similar to those of the former Yugoslav republics. These data should be interpreted carefully, however; for example, the infant mortality rate could be higher owing to underreporting and problems with data collection.. 6.9 Despite these improvements, some pre-war health problems still persist. Before the conflict lifestyle-related illnesses represented an unusually high burden o f disease in BiH. In 1990, more than 10% o f the DALYs (Disability Adjusted Life Years) lost were caused by tobacco-related illnesses and 6% to alcohol-related illnesses; worldwide only 6% of DALYs lost were caused by these two problems combined (Murray & Lopez, 1996). Although similar post-conflict data are not available, information from the LSMS offer indications o f the prevalence o f harmful health behavior in the population. Almost half of all males older than 17 years reported smoking at least once in the last week; this i s the case in both Entities, regardless of the income level of the individual. In the case of adult women, more than 22% reported smoking. Unfortunately, the LSMS cannot prove direct evidence of the prevalence of other harmful lifestyle behaviors, but qualitative data (World Bank, 2002c) suggest an increase in alcohol abuse among unemployed men, and in drug use among the young. Indirect information from the LSMS shows that the level of alcohol consumption per capita i s higher in households that include unemployedmen aged 22 to 35. 6.10 The LSMS survey collected detailed self-reported health assessments by respondents. Four measures of self-perceived health status were constructed. These measures are standard and try to quantify the level o f perceived physical health status of the population. Table 6.1 indicates that these perceptions vary across gender, entity and, not surprisingly, age groups. 93 The first measure of health status indicates limitations on basic activities of daily living. This measure includes all persons older than 17 years of age who reported having some difficulties eating, dressing, bathing or using a toilet by themselves. This indicator could suggest the extent o f major disabilities in the population at large, reflecting not only old age 93 They also vary by income (poverty) groupings; self perceptions o f health are known to be correlated with the income and education o f the individual, but the relationship i s not direct or simple. On the one hand, people with higher wages are more able to afford a disruption in their activities and this make them more likely to report as limiting their activities due to illness. On the other hand, people with lower wages are likely to be involvedin more physically demanding activities with higher risks o f accidents and could be more prone to suffer from trauma- related limitation o f their activities. At the same time they are the ones who cannot afford being sick for protractedperiods. This creates certain bias in self-reporting the health status and thus the data by poverty are not used here. 69 but also consequences of the war. The second measure, of intermediate limitations, includes all persons older than 17 who had frequent limitations on activities such as moving a table or groceries, walking uphill, wallung 100 meters, and bending, lifting or stooping. The third measure indicates whether or not the person missed at least one day o f usual activity in the last 4 weeks owing to a health problem. The fourth measure of health status i s a self-reported variable indicating whether or not the person -of any age- has a chronic condition.94 Box XIV. Mental disorders influence job opportunities, accessibility of health care and economic development The term posttraumatic stress disorder (PTSD) and its widespread use came into being to describe the suffering o f the population in BiH during the war. These scars take a very long time to heal; because of psychiatric morbidity, traumatized populations are more prone to chronic medical illnesses such as cardiovascular diseases which result in disproportionate rates of disability and premature deaths. Research under the World Bank-supported Post-conflict Mental Health Project showed that in post-conflict societies, poor mental health reducesjob opportunities for affected individuals and hampers the development of human and social capital. Moreover, in these societies, mental disorders are widespread and may represent a major obstacle to economic development. To investigate how psychological and mental conditions may influence individuals' decisions to participate in the labor market, the recently completed Labor Market Study analyzed the health status of workers by labor market status using the LSMS data. All population groups emerged troubled by mental problems. The incidence of chronic mental problems among the working age population i s staggering - around 10 percent. Many also said they suffered from insomnia and low energy. Interestingly, there are no significant differences between the poor and the non-poor, or the employed and the unemployed. This means that these factors affect the population at large and are not yet associated with social exclusion. Source: K. Scott (forthcoming); Labor Market InThe Postwar Bosnia And Herzegovina: How To IncreaseWorker Mobility And Job Creation?World Bank 2002 and ICR for War Victims RehabilitationProject (preparedby NedimJaganjac). Table 6.1 Self-Derceived health status. Percentages of individuals. L u ajor limitations bntermediate limitations bnactivity IChronic Disease" FBiH kS FBiH k S FBiH iRS FBiH Male 17-24 0.1 1.2 0.2 1.4 5.6 4.9 2.5 4.0 25-49 2.0 2.4 9.7 10.2 16.7 14.3 14.4 11.8 50-69 4.7 4.3 36.2 26.2 30.3 28.2 46.3 43.5 70 and over 18.0 12.9 65.2 40.7 53.5 44.9 68.4 64.5 Total Male 4.0 3.6 21.4 15.8 22.5 19.2 21.9 18.6 Female 17-24 0.8 1.5 2.0 6.2 5.6 3.3 3.1 25-49 0.5 1.4 12.7 8.2 14.1 15.5 17.3 14.4 50-69 6.4 5.6 46.0 35.1 40.5 34.9 59.0 55.9 potal Female 14.5 3.9 130.9 20.8 b7.0 23.9 130.7 24.8 J Source: BiHLSMS 2001. Note:* For chronic disease these cells include children 14 to 17 years of age. 94These measures overlap. For instance, an individual could report a chronic disease and also be included among those having major and/or minor limitations in daily activities as a consequence of the chronic disease. 70 6.11 LSMS data show that the health status of the population in BiH i s poor. Twenty-two percent of all adults in BiH report limitations on intermediate activities of daily living, 23 percent report having missed at least a day of their normal activities owing to health problems, and 24 percent of the entire population reports having a chronic condition. The average number of days lost in the last month due to a health condition i s quite high; nearly 10 days. On a more positive side, only about 4 percent of adults reported severe limitations, mainly reflecting old age. This percentage does not significantly vary across Entities or by gender once age i s taken into account. C. Health Insurance Coverage 6.12 The health care system in both Entities follows the social insurance model whereby employers and employees contribute to publicly managed funds that finance the bulk of health care provision. This model of health financing relies heavily on payroll taxation and cannot ensure universal access in BiH owing to a number of factors including the extent of unregistered unemployment, the number of people not participating in the labor market, and the scale of the informal economy. Also, most formal employment in BiH i s in the public sector, which effectively means that health insurance revenues are underwritten by the general tax base. 6.13 Current schemes cover working people and their dependants, plus other special groups such as pensioners, war veterans and war invalids, agricultural workers and others. Some of these groups are exempt from co-payments. Exempt groups in the FBiHinclude people under 20 and over 64 years of age, pregnant women, disabled war veterans, IDPs, returnees and some retirees and pensioners. Similarly in the RS, children, pregnant women, and the elderly are exempt from co-payments. 6.14 Even though the insurance schemes are meant to cover a large proportion of the population, some "fall through the cracks" of the systems and remain uninsured, while others do not pay insurance contribution and thus do not receive benefits. The non-payment rate in the FBiHi s 17%; it i s 35% in the RS. Official data from the Entities report a similar non- insurance rate for FBiH, but a lower one for the RS where officially only 25% of the population lacks health insurance. Table 6.2 shows the insurance coverage in each Entity by age group and broad income categories. Table 6.2 indicates that: (i)insurance coverage across all age groups i s higher in the FBiHthan inthe RS; (ii) arelargedifferenceininsurancecoveragebetweenthepoorandthenon-poor:in there the RS, close to half of the poor are uninsured while the comparable figure for the FbiHi s under 30%; (iii) 14-24agegroup, especiallythosefrompoorhouseholds, havethelowestrateof the coverage: this group i s most likely to "fall between the cracks" of the system as many young people are unemployed, or employed inthe informal sector. 71 Table 6.2 Percentages of insured population by age and income category I I RS FBiH I Total on-poor 80.0 87.9 85.5 60.0 80.4 73.3 65.9 85.1 78.4 70.6 85.1 79.4 62.1 83.5 75.5 P t a lnon-poor 1 68.0 84.7 I 78.8 oor 60.7 74.2 68.3 44.1 62.1 53.7 52.3 71.7 62.5 56.3 83.7 67.2 70 and over 60.0 74.8 65.9 Totalpoor 54.6 72.8 63.9 Source: BiH LSMS 2001. 6.15 The inequalities in insurance coverage can also be observed in Figure 6.1. In both Entities, people in the bottom quintile of the income distribution are less likely to be insured than those in higher quintiles. There are more people insured in the FBiHthan in the RS at all income levels. The inequalities in insurance coverage are also smaller in the FbiH,where. the difference in coverage between the top and bottom quintiles i s about 10 percentage points, while the gap in the RS i s about 13.5 percentage points. Figure 6.1 Health insurance coverage by income quintiles across Entities 0.9 0.8 32 0.7 0.6 n.5 0.5 n E 2 0.4 0.3 c O 0.2 0.1 0 1 2 3 4 5 Income Quintiles Source: BiHLSMS2001 72 6.16 The fact that a significant proportion of the population i s not covered by health insurance highlights three major issues for policymakers. First, as already noted, some population groups do not fall into any of the categories of insured as defined in the legislation. Second, about a third of all jobs in BiH are in the informal sector, making it very difficult for the government to enforce the payment of health insurance contributions for them. Third, formal sector employers are not paying the contributions for their workers, especially "wait listed workers" (see Chapter V). Inthe LSMS survey only 39% of people in this category reported receiving health insurance from their jobs. While the "wait list" has been formally dropped, a residual group in this category still exists-and under the 2000 Labor Law, there i s no longer a legal obligation for employers to pay health insurance for wait listed workers. 6.17 At the first glance, one may think that groups most likely to be uninsured by the system are the unemployed and those not participating in the labor market. But legislation in both Entities guarantees health insurance for the registered unemployed. Nevertheless, only 64% of those actually unemployed 95 are registered (in the RS only about half)96 and even among those registered only 84% have health insurance. Working age people not participating in the labor market can easily "fall through the cracks of the system", as they do not receive any labor income. However, this last category includes some people who are eligible for insurance, such as pensioners, disabled, and those in households where another family member has insurance that covers them. More than 90% of pensioners in BiH have health insurance (but the figure i s lower-only 80%-in the RS). Among the disabled, three out of four have insurance. Among non-labor market participants who are not disabled or pensioners, only 67% have insurance coverage, less than 60% of them inthe RS. 6.18 The large proportion of jobs in the informal sector and the non-payment o f contributions by some firms in the formal sector also creates problems with the health insurance coverage. Figure 6.2 shows the proportion of workers who received health insurance coverage from their work. Unskilled, agricultural, and service workers are the least likely to receive health insurance from their work. These occupations are also the most likely to be part o f the informal sector. Farmers, unskilled and service workers represent about 60% of all uninsured workers. While interpreting data from Figure 6.2 it i s important to remember that work i s not the only way to obtain insurance coverage. Some individuals are covered through other members of their families that are insured, others receive insurance because they belong to other categories of insured (pensioners, families o f fallen soldiers, etc). 95The registered unemployed include not only the "real unemployed', but also people who are engaged in informal economic activities. 96A possible reason why there are so many unregistered unemployed is that many are unfamiliar with the regulations. This i s especially the case among returnees and expatriates who fail to register before the 30 day deadline with the Unemployment Bureau (UNHCR, 2001). 73 Figure 6.2 Proportion of health insured among the paid employed by occupation 1 0.8 0.6 0.4 0.2 Military ManagerP Prol&Sct Technic Clerical ServWrk AgricWrk Skilllab Machoper UnskillL I ElInsured 0 Insuredfrom work Source: BiHLSMS 2001 6.19 Agricultural workers show the largest discrepancy between the total number of insured and those who receive insurance from work. Some of those who do not receive health insurance from work (18% of them) obtain it through registration at the Employment Bureau. Others may have voluntarily paid their insurance contribution but did not consider it a work-based benefit, and thus do not report it as such. But taking all insured agricultural workers together (whether or not they receive insurance from work) they, along with the unskilled, are the group with the lowest insurance coverage9'-even though the health insurance laws of the RS and the FBiHspecifically include agricultural workers as a category to be covered by compulsory health insurance. 6.20 Failure to pay insurance contributions i s thus one of the driving forces behind the large percentage of uninsured in the total population. By Federation law, insurance coverage i s withdrawn if the person or institution responsible for making the contributions fails to do so. In the RS the law states that those who fail to make the payments will not have their "health booklet" certified, and without a certified booklet the person will not be considered insured (UNHCR, 2001). D. Health Service Utilization 6.21 Qualitative studies suggest that illness i s one of the major risks poor families face in BiH (The World Bank, 2002~). Illness can significantly decrease income by reducing the economic activities of the sick person or of the person who takes care o f him or her. If the family does not have insurance, it faces two extra risks: not obtaining needed care, or incurring catastrophic health expenses that could drive the household deeper into poverty. The risk of catastrophic health expenditure also affects many families with incomes just above the poverty line. Although health insurance can reduce the possibility of impoverishment in case of illness, it does not eliminate this risk. It does, nevertheless, allow for easier access to health care. 97 The problem is larger inthe RS where the rates of non-coverage among agricultural workers, unskilled workers and service workers are significantly higher. 74 6.22 The percentages of the population using different health care services are shown in Table 6.3. These percentages, not surprisingly, vary with insurance and health status9*. Utilization rates are in virtually all cases systematically lower-sometimes dramatically so- for the uninsured (who are also more likely to be poor). There are also location-based differences in health services utilization, with higher percentages of people in urban areas utilizing services than people in rural areas99. However, these differences between urban and rural localities are only statistically different for insured people visiting a general practitioner and for people without insurance visiting a pediatrician. Table 6.3 Utilization of health care services across insurance coverage, health status a and regions Individuals Percentages. - Source: BiH LSMS 2001. Notes: Poor health status is define as having any limitation of intermediateactivities of a daily living; Questions regardinglimitations onhealthstatus were only askedto people older than 17 years; Four weeks is the recall period for all consultations with exceptionof gynecologist, dentist and hospitalizationwhose recall periodis equal to one year. 6.23 Although there are differences between the Entities in health care utilization, they only reflect differences in insurance coverage. Insured children visit a pediatrician more than twice as many times as non-insured children. The "Basic Package of Health Care" and legislation in both Entities guarantee health care for all children regardless of their parents' insurance status; more information i s needed to explain why there i s a difference in pediatrician visits between insured and non-insured children. 6.24 While public facilities provide most health care services in BiH, the number of private providers-mainly medical specialists-is growing in both Entities. Data from the LSMS show that 13% of people reported at least an outpatient visit in the last month '*Thedifference inservice utilizationbetween the insured and the non-insured might not be an indicator of a major threat to the health status of the non-insured if the higher percentage among the insured simply represented over-utilization by this group. However, once we divide the sample between those reporting any limitation to their normal activities and those who do not, the difference between the insured and the non- insured persists. 99 Although the data do not permit a deeper analysis, locational differences might first reflect differences in physical access to these services and also reflect barriers to access care in a different cantodentity. Insurance coverage i s not portable across cantons or Entities and health services and facilities (hospitals) are not evenly distributed across administrative boundaries. 75 (excluding visits to dentists) to a private provider.loO More than half of the people who visited a dentist went to a private provider, as did a quarter of the women who went to a gynecologist and a quarter of the people who visited other specialists. People of all income levels use these services, regardless of their insurance status"'. One of every ten persons visiting a private provider, not including general practitioners and hospitals, comes from the bottom 20% of the income distribution. It i s important to note that services received in private facilities are not covered by publicly funded health insurance. The significant level of utilization of private providers, including by those in low income households, points to a private sector that i s offering either more services, shorter waits, or better quality than those offered in the public sector. But since these services are paid out-of-pocket, poor people are at an evident disadvantage in accessing them. 6.25 Patients received health care mainly from doctors (physicians), and the system uses very few other providers. Data from the LSMS show that less than 1% of the total population of BiH visited a private nurse, midwife or paramedic in the 4 weeks before the interview; while about 16% went to a general practitioner. 6.26 Multivariate analysis of utilization of General Practitioner's (GP) services confirms the expected finding that, controlling for other factors (age, gender, education, reported health status, income level), the insured are significantly more likely to use medical (GP) services.'02 There are a few differences in the determinants o f utilization of these services across Entities: in the FbiH, for example, men are less likely to use these services than women. In both Entities the probability of service utilization increases, also not surprisingly, with deterioration of health status. The finding o f greater utilization by the insured holds for all groups, however, suggesting that lack o f insurance could be a barrier to access (see next para.); these results should be treated with some caution, however.'03 E. Barriers to access. 6.27 In both Entities there are major economic barriers in access to health service^"^. In the RS the major reason why people did not seek care when needed was the high cost, followed by minor illnesses that the person could self-treat. Not surprisingly, and given the loa However, as people visiting public facilities report more visits than those visiting private providers, the total number of visits to private facilities represent only 7% o f the total outpatient visits reported. Dentists' services even inthe pre-war system were mostly provided by private practitioners. lo' There are significant differences, however, in the percentages of people that use these services by income group. For instance, 33% of all those that went to a private dentist belonged to the richest 20% of the income distribution, while only 12%belongedto the poorest 20%. I O 2Results of the logic regression are not reported here, available in a background paper by M.E.Bonilla- Chacin. IO3 InbothEntities social insurance coverage is extended to all people working inthe formal sector. But contributions to social security are large, and there i s an incentive to work in the informal sector to avoid these payments. If this i s the case, the coefficient on the insurance variable does not represent the true effect that insurance coverage has on health care utilization. The effect might be exaggerated if people who select themselves out of the formal sector do so becausethey are healthier and do not need to use medical services, thus expecting to be better off by avoiding paying the health insurance fee. '04 People interviewed for the LSMS 2001 were asked why they did not obtain care when needed. This paragraph is basedon those answers. 76 relatively lower insurance coverage in the RS, not having insurance was mentioned as the third reason why care was not sought when needed. Inthe FBiHthe cost of health care was mentioned as the second biggest reason for not seeking care,lo5 and not having health insurance as the fourth major reason. Physical access i s also a barrier to care in rural and mixed localities. O f the respondents who cited distance as a reason for not receiving care, 93% lived in rural and mixed areas. People did not report poor service as a major problem, but it is not possible to determine whether this perception was true of the services offered in public facilities, because the relevant LSMS question did not differentiate between public and private providers. A study of corruption in BiH (World Bank, 2001a), based on a survey that also asked questions about quality in the health sector, suggests that the population i s more satisfied with the quality o f services offered by private providers than those o f the public sector. 6.28 Health service costs represent an major barrier to health care in both Entities. Even among the insured, one in every three who did not obtain health care when needed gave cost as a reason. In general, the poor are more likely to face economic barriers to entry, as they are less likely to be insured and less likely to be able to afford health care (see Figure 6.3). The high costs o f the system are in part related to its fragmentation. For example, some cantons in the FBiH do not have a full range of secondary and tertiary services-and since cantonal health insurance funds do not cover services provided in other cantons, patients from under-served cantons seeking treatment elsewhere are forced to make out-of-pocket payments for these services. Figure 6.3 People that mentioned an economic reason for not seeking care when needed by expenditure quintile. S! 14 A +Too expensive +No insurance E 4 F 6 aJ A Both n 1 2 3 4 5 Consumption Quintiles Source: BiHLSMS 2001. Note: The line marked with triangles includes people that reported that either it was too expensive, or that they did not have insurance. 6.29 The cost of transportation to health care facilities can also represent a possible barrier to accessing health services. This problem partly reflects fragmentation and non-portability (across Entitiesor municipalities) of health insurance. For instance, in some areas the closest health facility might be in a different health jurisdiction, but as insurance i s not portable, some patients are forced to pay unnecessarily high transportation costs to reach a facility in their ownjurisdiction that will accept their insurance. I O 5Self-treatable minor illnesses was the major reason mentioned for not receiving care when needed. 77 GP Pediatrician Gynecologist Dentist Other Doctor Hospital Average Expenditure R S 14.9 15.2 15.5 13.1 22.4 42.6 FBiH 11.1 12.0 10.4 13.6 16.5 39.0 Average Expenditure Urban 7.3 6.0 9.2 7.4 10.6 22.4 Mixed 10.7 13.1 11.1 11.3 18.4 44.5 Rural 16.0 14.8 13.5 16.2 22.5 44.0 Percent of usersincurring transportation cost Urban 31.5 32.7 41.2 31.2 39.4 45.0 Mixed 44.2 35.6 58.5 45.9 52.8 71.2 Rural 67.1 66.3 70.4 59.9 73.2 78.3 6.30 Table 6.4 shows how important transportation costs can be. For instance, the average transportation expenditure incurred in visits to a pediatrician in the FBiH i s about 12KM, more than twice the co-payment an insured person i s expected to pay to get a first examination by a medical specialist (5KM). Transportation costs vary sharply by locality. Rural dwellers pay on average about twice as much as urban dwellers to reach a health facility, and are also twice as likely to incur transportation costs. F. Health Care Finances: Public Expenditures 6.31 In 1999, total expenditure on healthcare inBiHwas close to 12% of GDP, one of the highest levels in the world. Of this total, public spending represented close to 7.7% of GDP (World Bank, 2002a). In 1999, the FBiH, which accounted for almost 80% of total BiH public health expenditure, had a deficit in its health accounts equal to 5.5% of its health sector income (DfID, 1999). The increasing gap between health revenues and expenditures results from: (i)infrastructure and staff levels that did not adjust to the sharp decrease in GDP compared to the pre-war period; (ii) inefficiencies in resource allocations; (iii) model a of health care that does not sufficiently emphasize relatively low cost primary care; (iv) a system that does not exploit economies of scale across jurisdictions; (v) a shrinking contribution base for health funds and; (vi) problems with revenue collection.. 6.32 In 1999 the Health Resource Accounts (DflD, 1999) estimated a per capita public expenditure equal to 172 K M ' 0 6 for BiH. But this average conceals large differences in per capita spending across Entities-218 KMinFBiH,and99 KMinRS. 6.33 Total 1999 health care spending in the FBiH was more than 518 million KM. More than a third of this expenditure was in primary care, followed by 27% in secondary care and 14% in tertiary care. This distribution of expenditure may conceal substantial secondary and tertiary care that some of the population of FBiHreceive in Croatia. Total expenditure inthe RS was close to 147 million KM, an.increase of almost 30% over the previous year. Direct lo6About 88 US dollars (using the 1999 end of period exchange rate of 1.947 KMper dollar), or only 58% of the annual per capita expenditure in 1991. 78 expenditures represented close to 90% of total spending, made up of primary care (37%), secondary care (35%), and tertiary health care (18%). G. Health CareFinances:PrivateExpenditures 6.34 The Health Resource Account (DflD, 1999) exercise estimated aggregate private expenditure at about 4.7% of GDP, close to forty percent of total sector spending. This i s surprising given that both Entities run social insurance schemes that in principle offer comprehensive health care to almost all the p~pulation.''~There are four types of payments in the health sector: (i)formal payments to private providers and pharmacists; (ii)co- payments in public facilities by those who are insured; (iii)formal payments in public facilities by those who are uninsured; and (iv) informal payments to the medical staff in public facilities as compensation for the services rendered. Unfortunately, the LSMS does not differentiate among these types of expenditure; therefore in this section (unless otherwise stated), payment refers to any type of expenditure made when visiting a health provider. 6.35 Table 6.5 shows that more than 80% of the BiHpopulation that seeks care incurs out- of-pocket expenditures. There are marked variations between Entities in the proportion of people who pay for consultation. More people pay for outpatient and inpatient care in the RS than in the Federation, reflecting different proportions o f insured and different co-payment exemption rules across Entities. Although in many instances the payments for health care seem to be lower in the RS, none of these differences are statistically significant. The cost of health care in both Entities can be a major drain on household income. The average household's total expenditure in one hospitalization can easily exceed 200KM in both Entities, which represents close to half o f the average monthly consumption o f a poor household."' More generally, data from the LSMS show that poor households spend, on average, an estimated 10% of their total consumption on health care, while households in the richest 20% spend only 5%. 6.36 The need to make informal payments (over and above official co-payments and other charges) was reported by participants in the Voices of the Poor study as one o f the major barriers to getting medical services of decent quality (World Bank, 2000). In a diagnostic survey on corruption, one in every five people who visited a health facility in the last two years reported making informal payments when accessing services. The major reasons given for making these payments were: to provide privileges for the patients; to secure good care for them; and to ensure that physicians would not overlook them and/or would permit the patient to "jump the queue" (World Bank 2001). The LSMS questionnaire does not distinguishbetween formal and informal payments. However, if we only consider the actual payments made to public institutions by the insured and compare them to the actual co- lo'Inother countries of the former Yugoslavia private expenditure on health as percentage of GDP is far lower-2% inCroatia and 1% inSlovenia and Macedonia in 1998. logThis canbe illustrated with the following simple back-of-the envelope calculation. If we take the 2200KWyr poverty line consumption level given in Ch I,para 14, multiply it roughly by 3 for an average family (getting 6600KWyr) and divide by 12, average monthly household consumption i s 550KM. The average poor have a consumption 24 percent short o f the poverty line, thus averaging to 428 KM;two hundred KMonan averageepisode of hospitalization is thus very close to one half of this average monthly total consumption of the poor. 79 payments they were supposed to pay, we can get an idea of the magnitude of informal payments. This will also give us an idea of how effective current insurance schemes are in protecting the population against catastrophic health expenditure. Avg. RS 22.4 18.9 28.1 46.4 35.5 296.5 Visits expenditurea FBiH 19.9 26.3 29.8 55.4 38.8 430.4 % paying RS 72.0 35.9 89.1 89.5 84.0 60.0 FBiH 39.2 14.7 68.9 76.1 47.0 42.7 AVP. RS 28.3 19.4 26.1 28.2 40.4 118.3 edicines expenditure FBiH 27.0 17.4 25.5 36.5 36.1 72.7 % paying RS 79.6 52.2 56.2 13.3 69.5 55.0 FBiH 74.1 59.4 51.7 12.1 63.1 35.6 Avg. RS 23.1 13.0 24.8 23.8 Lab. Tests expenditure FBiH 22.1 13.9 22.4 21.8 % paying RS 36.3 11.3 30.6 30.0 FBiH 20.9 5.9 33.2 21.3 Avg. RS 57.8 31.3 59.1 54.4 83.0 308.5 Totalb expenditure FBiH 43.3 26.7 52.1 61.2 60.9 280.3 % paying RS 95.6 79.5 96.0 93.3 97.2 89.3 FBiH 86.4 79.0 91.0 86.9 88.0 83.4 6.37 Table 6.6 summarizes the payments made on public health service providers by the insured and the non-insured. For those who paid for health care provided by a GP, pediatrician, dentist or gynecologist there were no significant differences in expenditure between those covered by social insurance and those who were not. Not being covered by insurance represents a barrier to system access, but unfortunately, being covered does not necessarily shield the insured from large health expenditure^.'^' The insured, however, are on average better off than the non-insured. Table 6.6 also points to an interesting result with respect to hospital visits. The proportion of insured patients paying for hospitalization i s little more than half the proportion of the uninsured who pay. The average expenditure by the insured patients who paid i s higher than the average payment of those without insurance, and it might reflect differences between insured and non-insured in the type of interventions received and possible extra services the former are receiving."' It is important to note that lo9The results of this kind of analysis require careful interpretation. Inout-patient services, similar expenditures by insured and non-insured might reflect payments for extra services that only the better off can afford. Nonetheless, it is unlikely that the public sector, with its budgetary problems, could offer many of these extra services and in any case there were no statistical differences inthe payments that poor and non-poor people made when accessing the health system (these results are not shown). I t i s also possible that the insured, being in general better off, incur more informal payments than those not covered by social insurance, or that co- payments are more enforced than fees for the uninsured. 'loThe population covered by insurance in the FBiHmust pay 100% of the costs for aesthetic surgery 80 hospitals are compelled to attend any emergency, regardless of the insurance status or income capabilities o f the patient. ''' Table 6.6 Out-of-pocket health expenditure in public facilities by insurance status (KM). GP Pediatrician Gynecologist Dentist Other Doctor Hospital Insured 20.3 21.8 21.7 32.1 34.3 401.3 Average Non- isits Expend. Insured 23.8 18.0 26.1 25.0 40.5 281.3 nsured 45.8 14.9 66.0 59.4 44.1 43.1 % paying Non- insured 79.0 42.9 85.5 82.1 66.9 78.8 I Insured 27.7 17.0 25.6 28.3 30.7 95.9 [Average bvon- 29.0 15.4 22.0 22.7 52.0 92.4 75.2 57.2 53.5 14.7 62.4 40.2 80.6 58.6 49.8 11.7 76.4 57.1 Insured 45.9 25.5 46.6 35.6 56.8 283.4 Average Non- otal" Expend. Insured 62.9 23.0 53.6 33.2 94.9 327.1 Insured 88.2 77.2 90.1 76.3 87.5 84.2 98.1 89.4 93.7 89.2 95.7 92.4 the health care facility. 6.38 Table 6.6 also highlights a different problem. Significant proportions of both the insured and the uninsured appear not to incur out-of-pocket expenditures. With respect to the insured, some of the non-payment may reflect visits by those who are exempt. Nevertheless, the data suggest that there may be problems with enforcement of health care fees among those not insured, and possibly of co-payments among the insured. 6.39 Using the LSMS we can infer whether or not some extra payments were made to health providers. The data show that on average an insured person visited the GP around 2.37 times and paid 20KM in the four weeks prior to the survey. On this basis, the average payment per visit was around 8 KM. But the official co-payment in the FBiH for a first examination by a general practitioner i s only 3KM, and then 2 K M for any repeated examination; in the RS these co-payments are even lower. The extra amounts actually paid might constitute informal payments,'12 a problem the government is trying to address. Finally, Table 6.6 also shows that, even though insured children are exempted from any co- corrections, and 80% of treatment in a natural spa. B y paying a co-payment a mother is allow to stay in the hospital with a child older than one year. 'I1 Unfortunately, the data do not differentiate between emergencies and non-emergencies, or between stays at a hospital or at a spa. Inpatient care expenditure differences between insured and uninsured might also reflect differences in the use o f non-emergency procedures or the use o f spas. 'I2 The data differentiate between costs incurred while visiting a health practitioner and costs o f drugs and lab tests associated with these visits. However, if other fees are charged, they might explain the difference between the official co-payment for the visit and the average payment per visit derived from LSMS data. 81 payment, 15% of the insured visiting a pediatrician reported having out-of-pocket expenditure, again suggesting the presence of informal payments. H. Conclusionsand Policy Implications 6.40 The findings of this Chapter can be summarized as follows: a. There i s a large pool of uninsured people in both Entities. The poor are less likely to be insured. b. Substantiallylarger percentages of the insured use health services compared to the uninsured (though payments made generally do not differ substantially between the two groups). The current system does not meet the goal of ensuring universal coverage. C. The poor allocate a greater proportion of their consumption to health care than the better off. d. The current costs of care negatively affect the access of poor people to health services. 6.41 These findings indicate important priorities for reform in the BiH health care sector. For example, even though both Eniities spend a large share of public resources on health care and both run health insurance schemes, there i s a high incidence of out-of-pocket payments. These payments constitute a real barrier to access to health services for people in poor households, especially for those who are not insured. Illness can impoverish households that are clustered just above the poverty line and can deepen the poverty of those below it. Policymakers need to address the resulting financial burdens and risks for the poor and the near-poor. 6.42 Large out-of-pocket expenditure and the inequity in this expenditure are in part caused by the fragmentation of the system and the lack of portability of health insurance. Great progress has been made toward reducing fragmentation with the creation of the Federal Solidarity Fund in the FBiH, and the Inter-Entity agreement that allow persons who live in one entity but work in another to receive health services in their place of residence. But much remains to be done in order to allow complete portability of health insurance across health jurisdictions, and to lessen the problems created by small and currently unsustainable risk pools. The Federal Solidarity Fund receives 8% of the payroll tax and equivalent contributions from the government. Unfortunately, because this year the budget allocation to Federal Solidarity was adopted after the budget was approved, the Fund only received a part of its allocation. 6.43 Another major area for reform i s expanding the coverage of the health insurance systems in both Entities. The large proportion of the population currently engaged in informal activities makes it impossible to cover the whole population through a payroll based health insurance system of the kind currently in place. Policymakers should find ways to cover those who do not fall into current categories of the insured, to cover those groups who cannot contribute to the schemes, and enforce the contributions of those who can and should pay contributions. For instance, more efforts should be made to limit the non-payment of health contribution by firms and institutions in the formal sector. In this regard more efforts 82 should be made to control and enforce the payment of payroll taxes in both Entities. In the case of agricultural workers and the self-employed, a revision of the regime to define payroll liabilities and ways to enforce them should be a priority. To ensure that those not falling within currently insured categories obtain coverage, a more comprehensive approach should be devised that takes into account the capacity of the Entities' budgets to cover these groups. Currently the World Bank and the Health Insurance Funds are trying to deal with these issues with a Social Insurance Technical Assistance Credit. 6.44 The health authorities in BiH are already allocating a large share of their income to health, but these resources need to be deployed in ways that better support equity. Other priorities include further improvements in coordination among health jurisdictions and a more efficient mix of levels of care, facilities, and staff. (The World Bank, 2002a). Better coordination with the private sector i s also essential. Private facilities are providing an important share of health services, and the scale of out-of-pocket payments suggests that many households are willing to pay for health care in both the private and public sectors. The public sector then might withdraw from the provision of some o f these services and use the resources saved to ensure more equitable access to health care for poor households. 6.45 More research i s needed to better understand the relationship between insurance status and out-of-pocket payments for health services. This should include studies of the existence and relative importance of informal payments, and of the enforcement of co- payments and fees in the health sector. A better understandingo f the situation will help in the design of policies aimed at increasing the financial protection of the poor in case of illness. 6.46 Finally, the government should undertake efforts to maintain and repair water and sanitation infrastructure, given its influence on the health status of the population. 83 84 7. EDUCATIONAND POVERTY INBOSNIAAND HERZEGOVINA By the time of independence, the education system in BiH has achieved impressive results compared to countries with similar income levels. But many problems had already started to emerge, and were exacerbated by the conflict and thepolitical fragmentation thatfollowed it. This Chapter examines access and equity issues in the education sector in BiH. The Chapter is divided into five sub-sections. The first offers an overview of the organization of the sector, the second discusses education status, the third analyzes possible problems with school enrollments, thefourth reviews the incidence of public and private expenditure on the poor, and, the last section offers policy recommendations. A. Overview 7.1 By the late 1980s, the education system in BiH had achieved impressive results. Primary enrollment rates were nearly 100% and about 80% of the students who finished the 8 years of compulsory basic education enrolled in a secondary school (UNDP, 1998). The gender gap in school enrollment had disappeared and in general the country benefited from a highly educated work farce. But even before the beginning of the conflict inthe early 1990s, a number o f problems were evident. Schools in BiH, especially secondary schools, were designed to provide students with specific skills to serve in the industrial economy of the Yugoslav era; these skills became less relevant as the country began the transition to a market economy. For instance, the primary school curriculum did not concentrate on cognitive slulls but rather on the transmission of factual information. The same problem was also present in secondary education, which was dominated by technical and vocational schools (which still account for about 80% of secondary students) instead of general academic secondary institutions, which more than 50% of OECD secondary school students attend. In addition to the questionable relevance of most secondary vocational programs for Bosnia's labor market and society, they are also relatively expensive, owing to low class sizes and the need for specialized equipment and training. 7.2 The conflict had a sharp and negative impact on the education sector. By the end of the war, 30% of schools had disappeared and many others were seriously damaged (UNDP, 1998). Enrollments had also fallen and, with a large proportion of the population displaced, had become unpredictable at the school level-a problem that still exists today. The adverse effects of the war were not confined to its effects on the education sector's physical infrastructure; the political fragmentation that resulted from it also created inefficiencies in educational service delivery, governance and financing. 85 7.3 Today the State level i s not responsible for education; this responsibility was delegated instead to the two Entities, the FBiHand the RS. In the FBiH, education finance and delivery was further decentralized to the cantons, which vary greatly in population and financial capacity. Differences across cantons and their reliance on their own resources to finance education have resulted in inequalities and efficiency problems. These problems will be aggravated as donor resources, which are predominately channeled directly to the Entities, start to decrease. Although some local reform efforts were undertaken after the war, they were mainly intended to "ethnicise" curriculums and textbooks, hence aggravating fragmentation. 7.4 Organization ofthe System The education system inBiHrelies almost entirely on the public sector, with only very few private schools operating in both Entities. For instance, in the year 2000 there were only 11private educational institutions in the FBiH.l13 Data from the LSMS for BiH as a whole show that less than 0.5% of all students enrolled in all three levels of regular education attend private institutions. This pattern i s different in pre-schools, where about 15% of the very few children enrolled attended private fa~i1ities.l'~Religious institutions also manage schools at the secondary and tertiary level, but offer services to fewer than 2% of secondary school students and to fewer than 1% o f higher education students. 7.5 Pre-school education i s intended partly to prepare children for primary school, and partly to serve as child care. During the Yugoslav period, pre-schools were usually connected to public enterprises; as many of these enterprises closed, and as others were pressured to cut costs, many of these schools have disappeared. Pre-school education i s not compulsory in either Entity; its supervision and (in some instances) financing depends on municipal governments. 7.6 BiHoffers 8 years of compulsory basic education. The compulsory age for entering primary school varies. In the FBiHthe starting age depends on the cantons but i s usually 7 years of age, as i s the case in the RS. The starting age does not appear to be strictly enforced: there are children in all localities starting school at either age 6 or 7. 7.7 Secondary education i s not compulsory. There are three main types of secondary schools: four year general academic schools (gymnasiums), three year vocational schools, and four year technical schools. Other types of schools, with small enrollments, include art schools, secondary teacher training schools, religious schools, and special education schools. 7.8 There are seven universities in Bosnia and Herzegovina. In keeping with regional traditions, however, the higher education system actually functions as a loose association of faculties and research institutions, each with its own legal status and separate funding. To date, only the University of Tuzla has made the transition to a modem, consolidated '13Two pre-schools ,3 primary schools, and 6 secondary schools. The informationfor pre-schools is from 1998 (FBiH, Federal Office of Statistics [2001]). 'I4This percentage should be treated with caution because pre-school education is so rare inBiHthat the LSMS included only a few instances of children enrolled init (91 non-weighted observations in total, with only a small proportion inprivate schools). 86 university management arrangement. In addition, as a result of the tensions created by the conflict, the academic programs that existed before the war in Sarajevo and Mostar have now split along ethnic lines. This further fragmentation adversely affects an already inefficient system, as the benefits of economies of scales cannot be exploited. B. EducationalStatus of the Population in BiH 7.9 Both Entities in BiHhave highly educated populations. According to LSMS data, the average years of schooling i s about 8.5 years for those over 15, and about 10.6 years for 15- 35 year olds, suggesting that in general people in BiH at least complete the full 8 years of compulsory primary education. Average years of schooling vary, however, by Entity, locality, gender, income level, and age. For instance, the population of the RS has half a year less schooling on average than the population of the FBiH. Women have on average two years less schooling than men; however this mainly reflects differences among older age cohort^."^ Adults in the richest quintile of the income distribution have a year and a half more schooling than adults in the poorest quintile; similarly, adults in urban areas have two years more schooling that those living in rural areas. All these differences are statistically significant. 7.10 Differences in years of schooling across income groups are worrisome, especially if they lead to dissimilarities in secondary school enrollment or completion. As will be discussed later, whether or not a person finishes secondary education has a strong impact on labor market outcomes. People with secondary schooling are more likely to be working and to enjoy higher wages. Figure 7.1 shows the proportion of people in the 16-25 year old cohort completing each year of education. Over the early years o f schooling, there i s little difference between the educational attainment of poor and non-poor individuals,"6 with more than 90% of both groups completing their 8 years of compulsory education. However, the proportion of students going from primary to secondary school falls significantly among the poor, only 60% of whom finish a ninthyear of schooling. The comparable proportion for the non-poor i s close to 90%. Figure 7.1 Educational attainment by income groups, cohort 16-25 years of age 25 - 3 1 0.8 u) c0 0.6 .- 0.4 g 'i50.2 o 1 3 5 7 9 11 13 15 17 19 Years of Schooling Source: calculationsbasedon BiHLSMS 2001 '"There are no gender difference in education outcomes for under 30 year olds. For those between 30 and 40, the difference i s less than a year of additional schooling for men. For older cohorts, the difference i s larger. '16For definitions see Chapter 1. 87 7.11 Enrollment rates vary substantially for the poor and the non-poor at all levels of education except for primary schooling. Pre-school enrollment helps prepare children for primary education, and, even more importantly, it redresses learning shortcomings that children from disadvantaged households may have. Without preschooling, children from poor households, who tend to receive less instruction or educational motivation at home, may enter primary school with a learning disadvantage that might never disappear. Today only 4.3% of children under 7 years of age are enrolled in a pre-school, and the vast majority of them are non-poor (see Table 7.1). '17 Table 7.1 Net enrollment rates for the poor and non-poor Pre-school a Primary Secondary Tertiary 0-6 7-14 15-18 19-23 Non-poor 5.6 93.0 76.4 27.3 Poor 1.1 92.2 57.2 9.3 Total (4.3)" 92.8 72.6 24.2 Confidence Intervals (3.0-5.7) (91.1-94.5) (68.3-76.9) (20.2-28.3) Source LSMS 2001 Notes: a These numbers have very large standard errors and should consequently be treated with caution. Secondary enrollment includes enrollment in all tracks: academic, technical, vocational and others. Some of the vocational tracks are only 3 year programs.' 95% confidence intervals are provided. 7.12 Primary school enrollment rates, although above 90%, are still below those of the pre- war era. These rates do not vary significantly across broad income levels, gender, or in terms of whether or not the child lives in a rural area. 7.13 Enrollment in secondary education i s also lower than before the war, with only 73% of children of the relevant age attending secondary schools -about seven percentage points less than before the conflict. To put this percentage in context, of all the transition economies for which data are available,'18 BiHhas the second lowest secondary net enrollment rate after Kosovo. While overall secondary enrollment i s low in BiH, the poor are particularly disadvantaged with only a 57% secondary enrollment rate. Moreover, of the total number of students attending a secondary school, only 1 in every 5 attended general academic or arts schools, while the rest attended vocational or technical schools. The latter were important in the Yugoslav era, as they prepared the labor force to work in the socially-managed industrial apparatus. The relevance of these programs i s questionable in the context of the country's transition to a market economy. These numbers are presented with the following caveat. To calculate primary enrollment rates, the starting age for the whole country was assumed to be 7 years. Although the starting age in some of the cantons inthe FBiHis 6 years, most children inbothEntities start school at age 7. However, inthe LSMS, aquarter of 6 year old children in the entire country are in first grade, reflecting the fact the children can start at age six in most districts if they will be seven years old by the end of the calendar year. "* The World Bank's WDI, 2001 has data on secondary enrollment rates inthe years 1997 and 1998 for the following transition economies: Azerbaijan, Bulgaria, Croatia, Czech Republic, Estonia, Georgia, Hungary, Kazakhstan, Latvia, Lithuania, and Slovenia. Data from Kosovo, the only economy from the FYR for which data are available, come from World Bank staff estimates using the Kosovo 2000 LSMS. 88 7.14 The age at which people attend higher education varies greatly, making information on enrollment rates at the tertiary level somewhat difficult to interpret. Nevertheless, available data can give us an idea of the disadvantage faced by young adults from poor households at the tertiary level. Only 9% of this group attend a higher education institution, while 27% of the non-poor in the same age group do. 7.15 In the specific context of BiH as a post-conflict society, it is noteworthy that no statistically significant difference was found in pre-school, elementary or secondary enrollment rates o f children from internally displaced refugee households as compared to enrollment rates among the general population. 7.16 This discussion of enrollment rates is based on LSMS data. Owing to the large numbers of displaced people and the lack of a representative household survey before the LSMS, there i s no comparable official information. Nevertheless, official data exist on the total number of students enrolled. Table 7.2 compares the number of students in primary and secondary schools estimated from the LSMS with the numbers from the official data.'" The discrepancy between the two sources i s very small (less than 2%) in primary education. In the case of secondary schools, LSMS estimates show 10% more students than the official data. 7.17 The LSMS data can only provide information on educational outcomes such as enrollment rates and educational attainment; they cannot offer insights about the pedagogic quality of the education offered in different parts of the country. Quality may well vary substantially across regions, given that there are 13 formal educational authorities in BiH, one for each Entity, one for each of the 10 cantons of the FBiH, and one for the Brcko District. Moreover, in FBiH cantons that are ethnically mixed and have significant minorities, there are parallel education administrators representing the two major ethnic groups in the region (World Bank, 2002e), and minorities are increasingly being segregated into different classrooms or schools. As the LSMS did not ask for ethnic affiliation, it i s not possible to determine whether enrollment rates for ethnic minority populations in a given area differ significantly from those of the majority population, let alone whether minorities are receiving a lower quality education. LSMS OfficialDataa RS FBiH Total RS FBiH Total Primary 127 257 383 124 259 383 students enrolledinpublic schools. ' 1 9When using these numbers to estimate net enrollment rates, it is important to note: (i) the official that numbers do not include enrollments in private institutions, (ii)that they include some students who are outside the relevant school age for each level and who therefore should not be included in the data. 89 C. Reasonsfor Low Enrollment Rates 7.18 Reasons for Low Pre- School Enrollment Rates. Households can face barriers and constraints in accessing pre-school services. These barriers may reflect inadequate supply, e.g., lack of or low quality of school facilities in the relevant area, or limited demand caused by resource constraints andor lack of interest ineducation. Inthe LSMS survey, households that did not send their young children to pre-school were asked why. Sixteen percent of households mentioned two supply side barriers: (i) the pre-school was too far away; and (ii) there was no transportation to the facility. This i s not surprising, given that many of the pre- schools that existed during the Yugoslav era have closed. On the demand side, the major reasons given for not enrolling were: (i)irrelevance of pre-school (reported by 40% of respondents); and (ii) the highcost of the school (reported by 20% of respondents). 7.19 As noted earlier, pre-school education helps children living in poor households to eliminate the disadvantage they can face when starting elementary school. This disadvantage can be important in systems where children enter their first year of basic education relatively late (as i s still the case in most of BiHwhere children enter when they are 7 years old). 7.20 Unfortunately, preschooling i s not currently functioning as a tool for equalizing the learning opportunities of children. On the contrary, it may actually be reinforcing the disadvantage that poor children may face when entering elementary school, because only children from the highend of the income distribution can afford it. 7.21 Reasons for Low Secondary School Enrollment Rates. Table 7.3 summarizes the reasons for dropping out of school given by secondary age children who had obtained a primary school diploma. The main reasons mentioned were demand side issues-in particular that student respondents considered their education as finished once they had obtained their elementary school diploma. Cost i s also an issue, as expenditure on secondary schooling can represent more than 20% of the average per capita yearly consumption o f a household. A third reason given was "lack o f interest." This response may be partly driven by the irrelevance of secondary curricula to the current labor market. A large number of secondary schools are vocational or technical schools. As noted earlier, these schools are over-specialized and were designed to teach the specific skills needed in the heavy industrial plants typical of the Yugoslav era.12o Reasons Percentage Considers It Finished 42.6 cost 31.5 No Interest 13.9 Too Far 6.0 Other 6.0 Total 100.0 In 1998, of the total number of public secondary schools inthe Federation,only 23% were Gymnasiums. In the RS the percentage was even lower at 12% (World Bank, 1999) 90 7.22 Are the poor systematically excluded from secondary education, or channeled to a low quality path often requiring grade repetition? Grade repetition i s one of the indicators of how well a child progresses through school. LSMS data can be used to estimate the proportion o f children lagging in their grade corresponding to age. Unfortunately, the estimate i s not very accurate, as the starting age varies between localities and i s not strictly enforced. A rough estimate of the proportion of children who repeat grades i s relatively from poor and non-poor households. small (about 5 percent), but there i s121 a large difference between the proportion of repeaters 7.23 Furthermore, multivariate analysis to establish the correlates o f secondary school enrollment shows that the probability that a child i s enrolled in a secondary school increases with the income o f the family, controlling for a number o f other socio-demographic factors. Moreover, there i s some evidence of inter-generationalinfluence: poorly educated parents are much less likely, other things being equal, to enroll their children in secondary education.122 D. Expenditure on Education 7.24 Expenditure on education in BiH i s quite high-private spending alone was close to 3.7% of GDP in 2001. 123 Combining this figure with public expenditure in the sector shows that BiH i s spending a higher share of its GDP on education than most high income countries. In the year 2000, total public expenditure in education was 6.8% of GDP in the FBiHand 3.6% in the RS, while the average combined public and private expenditure in education in OECD countries was about 5.55 percent in 1998. These percentages were similar to the mean o f OECD c ~ u n t r i e s . ' ~ ~ 7.25 Public e~penditure.'~~ Inthe year 2000, public education spending as apercentage of total public expenditure was 13.1% inthe FBiHand 12.3% in the RS. As shown inTable 7.4, per student spending relative to GDP exceeds European norms. Table 7.4 PublicExpenditurePer StudentRelativeto Per Capita GDP, 1999 (Percent) Primary Secondary Higher Bosnia and Federation 31 47 78 Herzegovina Republica Sprska 35 30 71 Europe 17 25 36 91 7.26 Fragmentation and lack of coordination among the subsystems in BiH have contributed to significantly different per capita expenditure patterns by jurisdiction, especially in the FBiHwhere each canton i s responsible for the financing and management of its own education structure. For instance, education expenditure per capita in West Herzegovina inthe year 2000 was equal to DM226, while the comparable figure for Gorazde was only DM 99. Each canton in the FBiHi s responsible for financing all recurrent costs of education in its jurisdiction, while the Federation level only finances capital costs, mainly usingdonor funds (World Bank, 2002e). 7.27 But inequalities in expenditure are not limited to geographical locality; they also appear to reflect unequal distribution of public spending by household income group. Table 7.5 shows the share of such spending on education by household consumption quintile. 126 7.28 Table 7.5 shows that total expenditure in education i s slightly pro-poor, with a concentration ratio12' of -0.05. A different pattern emerges, however, when expenditure i s subdivided by different levels of education. On this disaggregated basis, only expenditure on primary schooling i s pro-poor (Le., with the lowest consumption quintile attracting the largest percentage of funding, 27 percent of total), while spending on other levels, especially higher education, are pro-rich (with the lowest quintile attracting the smallest percentage, only 8 percent of total). The concentration ratio for secondary education of 0.05 suggests a regressive distribution of public expenditure at this level, which i s not surprising given the differences in secondary school enrollment between the poor and the non-poor (Table 7.1 ). Public expenditure on higher education i s even more regressive with a concentration ratio of 0.3. These concentration ratios are similar to those found in Kosovo, although in general BiH's public expenditure is less regressive than KOSOVO'S.instance, the concentration For ratio of public expenditure on primary education in Kosovo i s only -0.024 while the concentration ratio for tertiary education i s 0.367 (World Bank, 2001b). Table 7.5 Incidence of public expenditure in education a by consumption quintile 1:E;ation and type of school (shares of total and spending in KM),2001. I I I Consumption Quintile Total Concentration I I1 I11 IV V ratio Primary 0.27 0.20 0.24 0.18 0.11 283,997,890 -0.141 Secondary 0.13 0.23 0.22 0.24 0.18 145,290,904 0.046 Tertiary 0.08 0.12 0.14 0.27 0.39 93,597,952 0.303 Total 0.22 0.20 0.23 0.20 0.15 522,886,746 -0.052 126These results should be viewed with caution, as the estimates are made using 1999 data (World Bank, 2002e), the last year where informationdisaggregated by school level was available for the RS and FBiH. '*'The concentration ratio i s a measure o f inequality in the incidence o f public expenditure. This ratio is analogous to the Gini coefficient but because it is defined from the concentration curves and not from the Lorenz curve, it can have values from -1 to 1. Concentration ratios lower than zero indicate a progressive distribution of expenditures. 92 7.29 The same information on benefit incidence o f public expenditure i s summarized in Figure 7.2. It shows that public expenditure on primary education i s quite progressive as the concentration curve i s consistently above the 45-degree line of perfect equality. This i s not surprising, since primary enrollment rates are high at all income levels and poorer families tend to have more children. Expenditure on secondary education seems to benefit mostly the middle classes, while tertiary education i s highly regressive. Figure 7.2 Concentration curves of public expenditure ineducation by schoollevels 1 0.8 --e- Rimary 0.6 -E -Secondary 0.4 --- -Tertiary 0.2 -m-45degree 0 0 0.2 0.4 0.6 0.8 1 Percent of Households Source: Calculationsbasedon BiHLSMS 2001 7.30 Private Expenditure. In contrast to the pattern found in public expenditure, Table 7.6 shows that private expenditure on education i s progressive at both the secondary and tertiary levels.'28 Troublingly, however private expenditure on primary education i s very high for families in the lowest consumption quintiles; the poorest 40 percent of households account for 42 percent (22 + 20 percent) of total private expenditure on primary schooling. Education Consumption Quintile Level I I1 I11 IV V Total Primary 0.22 0.20 0.24 0.21 0.13 121,462,843 Secondary 0.11 0.19 0.22 0.23 0.25 134,212,795 Tertiary 0.06 0.14 0.13 0.26 0.41 109,411,895 Total 0.13 0.17 0.20 0.23 0.26 365,087334 7.31 Average yearly private expenditure per student in BiHi s quite high at almost 600KM (Table 7.7), the equivalent of almost 20% of average annual per capita consumption in the country. This average however hides large differences in expenditure by school level. Table 23 presents a summary o f private expenditure on education by level. Basically, all students incur out-of-pocket payments. Expenditure i s high, even at the primary level, representing close to 10% of national average annual per capita consumption. Transportation i s the '** "Progressive"inthe private expenditure context (as opposed to that for public expenditure) meansthat the expenditures incurred by the lowest income quintile represent the lowest share o f total private expenditure. 93 largest single expenditure item at all levels, representing as much one-third of total private expenditure on secondary and higher education. Tuition only constitutes a significant cost in higher education, since primary and most secondary public schools do not officially charge tuition fees. Tuition payments at these levels likely mainly consist of informal "under-the- table" payments. In contrast to low expenditures on tuition, the cost of textbooks and materials i s the second largest expenditure for primary school students and i s also important for the other two levels. Table 7.7 Private annual expenditure in education by item line, school level. inKM Education Level Primary Secondary Tertiary Average Transportation a 72.1 296.4 545.7 190.8 Tuition 3.7 54.9 188.9 40.0 Uniforms/Clothing 41.8 49.4 34.2 43.2 Textbookdmaterials 71.7 86.4 165.9 86.8 FoodLodging 69.6 103.2 197.3 93.5 ketal Other Expenditure 59.8 124.2 374.3 114.8 319 712 1,478 567 Source: calculations based on BiH LSMS 2001. Notes: a: Transportation cost was annualized by multiplying the weekly expenditure by 35 weeks, the total number of weeks o f schooling in FBiH general education. b: Other expenditures include: membership fees for parents' associations, payments for school repairs and maintenance, and total costs of education that the interviewee did not disaggregate. 7.32 Both capital expenditure and recurrent spending on school maintenance have been neglected in both Entities and have mostly depended on donor funds, which are gradually decreasing. Many schools are forced to depend on other sources of finance for maintenance, including charging students to cover the cost of school repairs. Figure 7.3 below shows the percentage of students who made payments for school repair and maintenance in primary and secondary schools. One in every five students in primary or secondary school makes some payment for school repairs. There were no differences in the proportion of children incurring this expenditure across income levels, but there were differences between Entities. Students inthe FBiHare almost twice as likely to incur expenditure for school repairs than students in the RS. However, the average payment made by households for repairs and maintenance was small; among primary school students who paid, it was about 10 KMin the school year. Figure 7.3 Percentage of students making payments for school repairs and maintenance s : 20% E 15% 0 10% 0% Source calculations based on BiHLSMS 2001 94 E. Conclusions and Policy Implications 7.33 Seven years after the end of the civil conflict, education outcomes in BiH have still not recoveredto the impressive levels that existed in the early 1990s. The above analysis has shown that education authorities inBiHconfront many challenges in securing equal access to quality education for all. But this effort will be essential for helping to eradicate poverty and preparing the population for the new demands of a market economy. 7.34 In summary, policymakers should focus on the following issues: a. The extremely low proportion of children of the relevant age attending pre- school. b. Primary enrollment rates which are still lower than those prevalent before the War. c. The comparatively low proportion of children continuing on to secondary school and the marked differences in secondary school enrollment across income levels. d. The large proportion of secondary school children enrolled in vocational or technical schools instead of in general academic schools, which only about 20% of the students attend. e. The regressive distribution of public expenditure in secondary and especially tertiary education. f. The economic barriers to access for some education services, such as pre- schools and higher education. 7.35 Policy options to ensure a pro-poor orientation of the education system include: 0 Pre-school education: In view of the currently very low level of pre-school enrollments in BiH and the already relatively high percentage of public spending devoted to education, the disadvantages of poor children when entering compulsory education cannot be adequately addressed by dramatically increasing publicly- subsidized pre-schools. The reduction of the primary school entrance age to six years throughout BiH combined with targeted remedial programs, such as core subject tutoring in the first year or two of primary education, are more sustainable policy options-particularly at a time when quality issues in compulsory education have not been effectively addressed. 0 Primary and Secondary Education: Primary school enrollments are high at more than 90%, but efforts are needed to raise them further. Low enrollment remains a significant problem in secondary schooling. Access to quality education i s one of the most powerful tools for reducing poverty and promoting economic growth. To ensure that the poor have equal access to secondary education, policy-makers should focus on two main constraints that households face: economic barriers to access and the lack of relevance and student interest ineducation. 95 0 Unfortunately, BiH i s not in a fiscal position to increase subsidies to secondary students for, e.g., transportation, housing, and subsistence costs. The provision of subsidized textbooks and education materials for poor students could, however, be increased on the basis of a family means test that has recently been developed for each Entity in the social protection sector. A specific means test should not be developed for education. One way of addressing transportation costs, especially at the secondary level where they are proportionately highest as a share of private spending, could be to develop a secondary education reform that focuses on offering general, broad-based Gymnasium-type programs rather than highly specialized vocational training. This would permit students to attend secondary schools of the former type closer to their homes. 0 Such a reform i s needed on broad educational, economic, employment and hence poverty reduction grounds. The current structure of secondary schooling and the curricula it offers continue to be broadly irrelevant to the needs of an emerging market economy offering better employment opportunities to labor market entrants. Today four of every five secondary school students in BiH are enrolled in technical and occupational programs that offer students very little flexibility in the current labor market. More broad-based skills are needed that can be used in a wider range of jobs. Examples of such skills include computer literacy and foreign languages, together with training in team work, problem solving, communication, and learning slulls. These are the competencies most urgently needed in a modernizing and rapidly changing economic environment. The design and introduction of programs based on such competencies for all students in the early years of the secondary curriculum would offer students more time to mature before committing to an occupation. In the latter years of secondary schooling, specialization should be restricted to a limited number of cross-occupational families (services, manufacturing skills, management, etc.) as opposed to large numbers of specific occupations. Short term degree programs and modular courses should then be offered by private training providers, public post secondary institutions, or by employers themselves, thereby completing a rational vertical articulation of the education system. More general and broad-based programs will also allow higher student teacher ratios and reduce the need for expensive equipment, which would free funds that could be used for targeting and quality enhancements. Some advances have been made on these issues in the context of ongoing European Commission support for VTE reform. VTE reforms will be the centerpiece of the next World Bank-funded education project for BiH, and further fundingis expected from the EC. 0 Higher Education: The politically-driven fragmentation of higher education organization and financing i s inefficient and effectively prevents qualitative improvements in programs. Important steps towards reducing inefficiency and providing an equitable distribution of resources include: (a) establishing Entity level management of higher education funding on the basis of a per-student funding formula; (b) consolidating autonomous university faculties into modem universities; (c) converting some of the existing universities/faculties into community college-like 96 institutions that offer short-term programs more relevant to current labor market needs; and (d) given the regressive incidence of public spending on higher education, gradually increasing fees and eliminating non-means tested fee-exemptions to students. 0 Education Finance: Finally, as mentioned above, education authorities should ensure a more progressive distribution of public resources earmarked for education, especially those going to secondary and higher education. Increasing private contributions to higher education will be an important part of achieving this goal, as i s increasing the relevance of both secondary and higher education programs. In addition, it i s important that geographically disadvantaged jurisdictions do not depend solely on their own resources to finance education in their region. Insufficient spending on education, especially at the basic levels, will result in poor educational outcomes that will eventually lead to low incomes, creating a vicious circle, with low incomes further limiting the availability of funding over time. This issue applies to the municipalities in the RS, which fund secondary education from their own revenue, as well as to the cantons in the FBiH,which have widely different education spending patterns. The Dayton Agreements do not preclude politicians and education stakeholders in BiH from rationalizing education funding to ensure that the poor benefit from a minimum level o f recurrent funding per student, and that investment resources are focused on high poverty areas. While cantons and, where applicable, municipalities should continue to control the organization and management of education (personnel, curriculum, books, etc.), an equalization fund for general education should be established to compensate poorer FBiH cantons or poorer municipalities within FBiHcantons and RS. All stakeholders in BiH will be familiar with "solidarity funds" that effectively reallocated resources across municipalities and republics in the former Yugoslavia to compensate for differences in the tax base or revenue shortfalls. 0 In addition, given the current fragmentation and inefficiency of the higher education system, the management of higher education funding should be moved to the Entity level in the Federation, while legislation should be adopted to ensure that universities maintain their political autonomy. The need for some revenue redistribution in education and the need to manage higher education finance at an Entity level should be taken into account in the context of assigning tax revenue to various levels of governance in BiH, including decisions on the assignment of the VAT tax currently under consideration. 7.36 Qualitative data on both gender and poverty show that both men and women in the FBiHandthe RSconsider the lack of marketable skills as one of the major causes of poverty; similarly, they consider that adequate and equal access to education i s one o f the ways out of it (World Bank, 2002~). The analysis of labor markets in Chapter V has shown that lower levels of education are associated with non-participation in the labor market, especially in the case of women. Thus improving education i s key for the long term success o f poverty reduction efforts. 97 98 8. TARGETING, EFFICENCYAND EFFECTIVENESSOF SAFETYNETS This Chapter analyzes the impact of public transfers on poverty. Itfirst describes the overall policy framework that determines the welfare regime (defined here in a broad sense to include all types of transfers to households, such as pensions, children's benefits etc.), and then discusses whether the poor benefit from the system and how particular types of poverty related benefits can be targeted. A. Background 8.1 In the former Socialist Federative Republic of Yugoslavia, the welfare regime was different from those in the socialist societies in Central and Eastern Europe that had remained within the Soviet sphere of influence. From the late 1950s onward, it was increasingly recognized that central planning could not meet all needs and eliminate poverty and social problems. This led to the formation of Centres for Social Work in most urban Municipalities in the early 1960s, with university level social work training beginning even earlier in Belgrade, Ljubljana, and Zagreb. Through a series o f reforms in the late 1960s and early 1970s, often with international assistance, a system of social protection emerged which was seen as both `very highly developed', and combining `development with decentralization' (World Bank, 1975). 8.2 The war changed the essential social contract underlying the welfare regime in BiH. The large-scale relief programs of the European Union, United Nations agencies (primarily UNHCR), USAID and other bilaterals, tended not to implement projects directly but to work through `implementing partners', initially INGOs such as CARE, Catholic Relief Services (CRS), the International Rescue Committee (IRC) and many others so that, by the end o f the war, over 200 INGOs were working in BiH. These agencies worked on an `external' model of assistance, which implied emergency-dictated simple categorization of beneficiaries. This flow was completely de-politicized, and rarely challenged "ethnic cleansing" (the wife of extreme nationalist Radovan Karadzic became President of the Red Cross in the Serbian territory, retaining this position in Republika Srpska for a time even after the Dayton Agreement). As a result, mistrust quickly became an intrinsic element o f a new diversified welfare regime - fragmented and localized, category-based rather than needs-based, and de- politicized in a sense that it was never perceived as coordinated social policy. 8.3 The evolution of social welfare and child protection spending in the 1990s involved virtually total dependence on foreign sources for food and other aid until 1996, followed by 99 sustained and significant NGO/donor support in 1996-1998, evolving into greater dependence on local budgetary sources from 1999 onwards. Two examples illustrate this evolution. First, from 1992 to 1996 donor/NGO food aid was available to almost all Bosnians, declining to around 15 percent of the population in 1997, and to 8 and 6 percent in RS and FBiHrespectively in 1998. It was largely ended for all but some elderly collective center residents from mid-1999 onwards.129 Second, donor/NGO financing for the sector declined from around US$ 110 million in 1998 to US$ 30-35 million in 1999. While later consolidated figures are not available, data from major agencies suggest an annual figure in 2000 of around US$ 12-14 million for the sector. 8.4 This phase-out of humanitarian assistance revealed deep rooted problems. The constitution agreed at Dayton allowed for the development of distinct social policies and social welfare regimes in the two Entities o f BiH-the FBiHand the RS. Within the FBiH social policy powers are further divided between the Federation and the cantons. The Federation, under Article 1, has exclusive responsibility for the creation of monetary and fiscal policy. Health issues and social welfare policy are defined as joint responsibilities of the Federation and cantons under Article 2 (111). Cantons are given responsibility by Article 4 for the creation and regulation of education policy, the creation and regulation o f housing policy, the creation and regulation of policy regarding public services, and the implementation of social policy and maintenance of social welfare services. In essence then, the Dayton Accord creates a contradiction between Entity-based rights and the need to raise resources to realize these rights at lower levels of the system-cantons and municipalities in FBiH, and municipalities in RS. The system thus promotes the creation of substantial but unrealizable social rights, completely disconnected from the revenue base. Indeed, as cantons and municipalities vary enormously in size, resources, and revenue-raising abilities, fundamental inequities in the implementation of rights and entitlements are also built into the system and its structure. Thus far the result has been growing pressure on Government resources with little accountability or assessment of effectiveness of the welfare system. 8.5 Current transfers in BiHcomprise transfers to social security institutions, households, utility companies owned by local governments, and inter- and intra-governmental transfers. They are broadly in line with those in other SFRY economies, though the region i s characterized by high transfers to households, mainly in the form o f pensions and health outlays. 8.6 Most transfers take the form of pensions, key features of which are outlined in Table 8.1. Old age and survivors pensions aside, a significant share of the transfers to households in BiHare "categorical" transfers intendedto improve the welfare of those most affected by the war. These include transfers to war veterans and invalids (most recently, close to 4 percent of GDP in both Entities) and refugees and displaced persons (less than 0.5 percent o f GDP in both Entities). Total budget current transfers represent about 17 percent of GDP in BiH (about 18 percent in the FBiH and 16 percent in the RS). They account for a sizable one- third of total on-budget expenditures in both Entities. Most transfers take the form of transfers to households, which account for about 15 percent of GDP in both Entities, with the lZ9See IBHI (2001) ans Stubbs and Gregson, eds. (1998). 100 rest mainly allocated to non-profit institutions. A recent Public Expenditure review130 shows that transfers to veterans in the form of benefits and pensions constitute the single major social transfer directly involving the budget, and represent a major budget constraint on the development of other social assistance programs. Average Pensioners Payroll Pension Indexation Monthly I Tax Spending I Pension, US $ Contribut Contributi GDP ors ons FBiH 77 0.69 17.0% 8.0% As per available resources RS 32 1.08 15.8% 10.8% As per available resources B. Household Perspective 8.7 There are two ways to assess how well the welfare regime works. One i s to evaluate how closely benefits administration matches the internal standards set in the legislation governing the program. It i s beyond the scope of this study to conduct a detailed assessment of whether or not particular claimants formally qualify for benefits. Instead, this chapter takes a second approach, which assesses the program in terms o f the extent to which benefits go to the poor. The results should not be taken as conclusive evidence that the program fails or succeeds on its own terms. Rather, our perspective i s based on the view that an ultimate aim of any welfare regime i s to alleviate poverty. 8.8 Table 8.2 presents the results based on LSMS data. It shows that public transfers cover every fifth person in the population (se last column). Coverage of Average Total public Total public What would be the population transfer transfers in total transfers in poverty rate without (a) per month consumption consumption transfers (and % (KM) (%) of recipients (%) reduction with transfers) Non-Poor* 16 146 5 15 0 Poor* 28 220 30 41 100 Returnee 21 195 13 28 30 ( -52%) Refugees /IDPs 21 174 14 29 49 (-34%) Total 19 174 11 24 32 (-40%) Source :BiH LSMS 2001. Based on pre-transferconsumption level. * 8.9 What i s more revealing i s that public transfers on average are equivalent to about 11 percent of total household consumption in the country (third data column). This i s a 130BiH:From Aid Dependency to Fiscal Self-Reliance, World Bank 2002a 101 relatively high level by international standards (although the comparable figure for Croatia i s higher, around 15 percent). 8.10 Assistance i s also quite substantial for the target groups that receive it, and the reduction o f poverty achieved by transfers i s particularly large for displaced population (returnees). In terms of recipients' pre-transfer poverty status, a substantial share of the population (13 percent - the difference between the actual poverty rate, 19 percent, and the simulated poverty rate that would prevail were all State transfers excluded from the consumption of households, 32 percent) relies on transfers to move out of poverty. But many poor people either receive nothing from the system, or receive too little. 8.11 As already noted, many households in BiH receive some type of transfer: cash benefits in some form are received by 20 percent of the population. But Table 8.2 shows that barely 30 percent of the poor receive any kind of transfer. Less than half of all households with poor members are coveredby any type of public transfer program. 8.12 Nevertheless, social protection programs, particularly pensions, do help to keep many households from falling into poverty. Table 8.2 shows that in the absence of social transfer programs, poverty rates would be 40 percent higher (32 percent of the population instead of the actual 19 percent). In terms of international comparisons, the poverty rate without transfers in BiH is similar to poverty measured in a similar way in Bulgaria, and, surprisingly, Croatia.131In Croatia the poverty rate without transfers is 25 percent ,while in Bulgaria it i s 29 percent; with transfers the rate i s 8 percent in Croatia and 12 percent in Bulgaria. Thus the poverty rate with transfers is much higher in BiH than in both these countries. In Bulgaria, for example, transfers reduce the poverty rate by 60 percent, while Croatia the reduction i s even greater. 8.13 These aggregate comparisons suggest that the poverty impact of the transfer system i s lower in BiH than in comparator countries. To see why this i s so, it i s necessary to look in detail at specific types o f benefits. C. The Pension System and Poverty 8.14 Although the main objective of pension benefits i s not poverty relief, these benefits are very important for poverty reduction. The elderly and those who are unable to work face a very difficult situation in both Entities. The collapse of the socialist economy in Bosnia negated many o f their expected old age support programs, and they have become more dependent on their children, their own efforts, and new social safety net programs. Many saw their children flee the country or lost them in the war, removing an important source of old age support. 8.15 There i s a concern that without savings, support from extended families, or social protection programs, nearly all elderly people not receiving pensions and living alone would fall into abject poverty (unless they continued to participate in the labor force). The pension system pays therefore a key role inprotecting this population group from poverty, and its role 13'For Croatia, see Croatia: Economic Vulnerability and Welfare Study. For Bulgaria, see Poverty Assessment. 102 can be assessed by comparing the level of minimum pension with the poverty line and coverage of the pension system. The poverty profile analysis showed, however, that the concern about adequacy underestimates the extent of various formal and informal safety net arrangements. The elderly, especially those not receiving pensions, are perceived as "deserving" poor and the combined efforts of communities, NGOs and private individuals have helped to move a substantial part of this group out of poverty. 8.16 In the post-war period key problems of the pension system were not problems of adequacy o f benefits in comparison to poverty line or pension system coverage o f elderly. The main issue was shrinking financial based on the pay-as-you-go system with falling contributions from the currently employed and growing number of pensioners with relatively well protected values of real pensions. This widening gap resulted in build-up o f arrears from pension fund to the beneficiaries and imminent threat of bankruptcy for the pension system as a whole. Progress towards strict financial discipline within the pension system (whereby pensions are determined on the basis of total revenues collected by both Entities' PensionFunds)achieved in 2001-2002 produced positive effects on the welfare o f pensioners by fostering more regular payment of pensions. 8.17 To achieve financial balance, both systems have compressed the distribution o f pensions, by introducing a maximum level o f benefit (2 times the average wage in the FBiH, 3.8 times the minimumwage in the RS, i.e., o f about 350 KMin the RS and 650 KM in the FBiH);by changing the indexing rules for the calculation of retirees' earning base (indexed to past benefit growth in the FBiH); or by using new coefficient related to the retiree's level of education when records of past working history were not available. 8.18 Both Entities have introduced minimumpensions. In the RS, this occurred in April 2001; the minimum has subsequently been fixed at S O W . The minimum in the FBiH i s higher (140 KM)for all municipalities. At the time of the LSMS survey, however, there were still two agencies in the FBiH: the Croat-populated cantors Pension Institute (with headquarters in Mostar) and the rest of the Federation PensionInstitute (with headquarters in Sarajevo). The minimumpension in the former area was 140 KMwhile the minimum level inthe latter area was 117 KM. 8.19 The financial viability of the system in the medium term is ensured by the rule adopted in the reform whereby pensioners are entitled to pensions above the minimum only within the limits of the current agencies' revenues. This mechanism eliminates the need for State budget transfers in the event o f a shortfall in current income, and the consequential accumulation of arrears ifthe State does not transfer enough money to cover the deficit, as occurred until the new mechanism was put inplace. 8.20 The actual distribution of pension benefits received by beneficiaries as measured by the pension funds i s very compressed in both Entities. The ratio of the average benefits o f the 9th decile to the average benefit of the first decile i s 2.3 in the RS and 2.8 in the FBiH. The average pension in both Entities i s about 40 percent of the average wage (41 and 47 percent inthe RS andthe FBiH,respectively). 103 8.21 Veterans' benefits have recently come to the forefront of debate over reform of the social protection system. The source of the controversy lies in the political sensitivity of reforming veterans' benefits, together with opposing views about whether they are targeted to those most in need. The LSMS data make it possible to estimate and compare the incidence of pension and veterans' benefits by population groups. While all sides view veterans' benefits as substantially distinct from the usual social insurance arrangements, analysis based on the LSMS shows that these benefits do not differ in their key distribution parameters from those provided by the pension system. 8.22 B y comparison with pensions, veterans' benefits vary widely in amounts paid. Invalid benefits can be as low as 19 KMper month inthe RS and 50 KMper month in the FBiH.At the opposite end of the scale, the highest level of benefit i s around 900 KMper month. The perception, however, that veterans are among the best-off i s not confirmed by the survey. Once classified by per capita consumption quintiles, veterans, whether family or invalid recipients, are present across the whole spectrum of distribution. Figure 8.1 (A) Incidence of benefits by ex-post consumption quintiles Concentration curve for Concentration curve for veteran's benefits pensions 1 i 2 0.8 G .$ku 2 2 0.6 .o- 0.4 w 2 0.2 U 0 U 0 1 2 3 4 5 0 1 2 3 4 5 Quintiles (frompoor to rich) Quintiles (frompoor to rich) 104 (B) Incidenceof benefitsby ex ante consumptionquintiles Concentration curve for Concentration curve for veteran's benefits pensions 1 L 0.8 B .P S 0.6 -.$5 O M 0.4 m 5 0.2 u 0 U 0 1 2 3 4 5 0 1 2 3 4 5 Quintiles(frompoor to rich) Quintiles (frompoor to rich) Sources: B M LSMS, 2001. Note: (B) shows the incidence of the benefit, estimatedinthe absence of the benefit. 8.23 Old age pensions and veteran's benefits seem to be fairly equally distributed by population groups ranged with respect to consumption levels, as shown in Figure 8.1. The most strikingresult of this analysis i s the similarity o f distributional patterns across quintiles. This suggests that both systems are achieving the same goals and using a common set of criteria to establish eligibility. 8.24 One important feature stands out from the analysis. The fact that veterans' benefits are so concentrated in the lowest pre-transfer quintile means that households of veterans rely very heavily on these benefits as their main source of income. While this i s appropriate for pensioners or the severely disabled, it i s not an optimal outcome for those who are capable of self-sustaining activity. As the recent analysis (see World Bank, 2003 c) has demonstrated, most veterans' benefit recipients are able to work. The best strategy for helping this group would be helping them to integrate into society through employment or entrepreneurship. The World Bank Pilot Emergency Labor Redeployment Project offers valuable experience with such policies, which are reviewed in detail in the next chapter. 8.25 On the surface, data on the amount of pensions received may imply that many pensioners or veterans' benefit recipients would be poor, as their pension level i s below the value of the poverty line. Increasing all pension benefits to the level of the poverty line would eliminate poverty among the pensioners, but the cost would be quite large, however, amounting to 4 percent o f GDP. 8.26 The great majority of pensioners and veterans' benefit recipients are less poor than the rest of the population. When poverty rates are compared by age and gender, and by age and type o f municipality, rates among pensioners are equal to or lower than those for the relevant population groups. Veterans are about as much better off as pensioners in the RS and slightly worse off in the FBiH, but still have lower poverty rates than almost all other population categories. 105 8.27 LSMS data suggest also that many recipients of the minimumpension are not among the poor. More than three-quarters of this group in the RS, and about 86 percent o f them are not poor. Those who do not contribute or contribute the minimum to social security are not necessarily the poorest. Thus increases in the minimum pension would mostly benefit non- poor pensioners, and hence would not be effective from the point of view of poverty reduction. Simulations show that the low risk of poverty for pensioners i s only partly due to the effect of pensions; private safety net and intra-family transfers are much more important. 8.28 On the other hand, many of the elderly are not covered by the pension system at all. Overall, pension and veterans' benefits combined reach only 59 percent o f the total population aged 65 and over. Pensions cover 46 percent of this population. In the RS, only one third of the elderly in rural areas are covered. Coverage among elderly women i s lower inboth Entities, butis especially low inthe RS, barely reaching 22 percent. The poverty rate among the elderly not covered by a pension, however, i s only slightly higher than for those covered. Nevertheless, the poverty rate for some groups of the uncovered elderly (those excluded from private support networks) i s high, reaching almost 50 percent among uncovered elderly IDPs and refugees.'32 This extremely high rate shows indirectly how important informal networks and private safety arrangements are for the elderly; IDPs and refugees are by definition excluded from local reciprocity networks of these kinds. 8.29 On balance, there i s a relatively weak link between poverty reduction and pension reform issues. Changing the pensions regime by increasing minimum pension or expanding coverage o f pension benefits i s not an efficient way to reduce poverty; it will either be unaffordable or require specific targeted interventions. Meanwhile, as already indicated, veterans' entitlements are liable to distort labor market incentives for recipients and to crowd out spending on other forms of social assistance. And finally, those not covered by the system have adjusted to current arrangements, and changing pension policies will not substantially affect their poverty levels. Important differences in poverty incidence by social groups do exist, but overall the elderly seem to be relatively well protected through pensions or private safety net arrangements. D. Targetingof SocialAssistance 8.30 Pensions and veterans' benefits represent the bulk of all social transfers in BiH. However, pensions are "special" social transfers in the sense that they have an income- smoothing function and may be regarded as deferred wages. This section focuses on how the social transfer system performs with respect to other, mainly smaller kinds of transfers whose primary function, in principle, should be to help poor households. 8.31 Social welfare assistance and child protection arrangements in BiH are increasingly facing a financial crisis involving an as-yet unfinished transition from humanitarian relief to a more coherent and sustainable system . Both Entities currently allocate extremely low shares of GDP to social welfare and child protection compared to several neighboring transition countries. The problem i s particularly acute in the FBiH. Moreover, the 13*This is another case where the limits of the sample size call for caution inthe interpretation of this result. 106 decentralized financing mechanisms in both Entities allocate resources in inverse relation to the regional incidence of poverty, with dramatically low and only ad hoc protection for the most vulnerable in poorer areas. At present, many areas in both Entities are effectively without a basic cash safety net for the poorest and most vulnerable. 8.32 Regulations in the FBiHprovide that social cash benefits are payable to persons who are unable to work, or lack sufficient income for living because they do not have family members who can provide support (as the Law on Family requires them to do). Cash benefits are also paid to civilian victims o f war who are invalids, and to their families in case of death (to widows if over 55, to widowers if over 65, to children if under 15, or 27 if they are students). Finally, special benefits are paid to children who live in low income families, and to unemployedmothers duringpregnancy and post-natal periods. 8.33 In the RS, several types of social protection benefits are payable to persons who live alone, persons who do not have sufficient income for living (the level i s not defined), persons who are disabled and need assistance and care, and all those who are not able to work and lack either property or relatives legally obliged to support them. Training-for-work allowances are paid to disabledpersonswho are able to work 8.34 Beneficiaries. Overall, there are about 260,000 beneficiaries o f all types of cash benefit and/or in-kind social service programs in BiH-equivalent to around7 percent of the population in each Entity. There are also about 125,000 registered beneficiaries of child protection programs (about 2 percent of the population in FBiH and 6 percent in RS). However, beneficiary numbers need to be treated with great caution, since double counting i s considered to be high andreporting rules are not consistently defined. 8.35 Roughly 2 percent of the BiHpopulation are registered as recipients of cash benefits from the social welfare and child protection systems. (This compares with around 1.8 percent in Croatia and around 5 percent in Serbia.) But a much smaller percentage actually gets somethingfrom the system. For instance, in RS only 13 percent of those registered actually receive benefits, with considerable variation across municipalities, ranging from as low as 2- 3 percent in some municipalities to over 20 percent in others. Resource constraints affect child protection programs particularly severely in the FBiH. Between 1998 and 2000, the number of child protection beneficiaries was halved. As other cantons terminated these programs, Sarajevo was the only canton still active in child protection . 8.36 The survey carried out for the Local Level Institutions Study (The World Bank 2002 d) considered social assistance to be nonexistent, an opinion held by more than one third (34.8 percent) of all respondents, and half (51.7 percent) of those belonging to local ethnic minorities. Inthe event o f a serious financial crisis, very few respondents would turn for help to any formal institution, and even Centers for Social Work were rarely seen as a source of help. Most people would rather turn for help to their personal acquaintances (relatives, friends or, to a lesser extent, colleagues and neighbors)-a finding that confirms the importance of informal types of material assistance and the predominance of private connections and informal networks as opposed to formal institutions. 107 8.37 The LSMS data throw light on how the system works. The survey indicates an individual's entitlements to permanent and temporary assistance, and reports the total amount of cash benefit received in the last year. There i s a concern that the questions in the survey were not precise enough to ensure that only government social assistance was reported and as a result, that individuals might report other type of assistance. The PEIR, however, reports that while ". ..there are also many donors and non-governmental organizations (NGOs) active in the social welfare sector.. .. many of the larger programs have now been terminated or have been significantly scaled back [leading to] ... greater dependence on SWCP [social welfare and child protection] spending on local budgetary sources, especially since 1999". (The World Bank 2002a). 8.38 LSMS data show that the targeting i s poor; only 4 percent o f the poor receive some targeted financial aid. Effectiveness of the system i s therefore extremely limited. This could reflect the small amount spent-but also the inadequate efficiency, as reflected in substantial leakage of the benefits to the non-poor: 75 percent of beneficiaries are not poor (measured by excluding social assistance from total consumption). Thus the system of cash transfers as a whole performs just a little bit better than a simple "random" distribution (the share of the poor in the population i s about 20 percent). 8.39 Not only are the amounts spent and number of recipients small, but the distribution of spending seems to be fairly irrational, with some receiving a lot more assistance than they need, and others getting hardly anything. The payment of social allowances "in excess of need" seems particularly prevalent among persons displaced by the war. This group represents 24 percent o f the poor who benefit from social assistance but 74 percent of those who receive the highest benefit inrelation to their need. 8.40 The results in the international context are presented in Figure 8.2. To make data comparable across countries poverty i s defined everywhere as bottom 10 percent of the population . Based on these comparable figures, the success of the social assistance system i s measured by how much of the (pre-assistance) poverty gap o f the bottom decile i s eliminated (effectiveness), and how much of disbursed social assistance i s received by them (efficiency). The scores are drawn as bars with corresponding figures on the left axis as percentages. Thus, the graph shows that in BiH the efficiency score i s among the lowest among the countries in the comparison (only Russia has less efficient system): slightly above 10 percent of all poverty benefit goes to the poorest 10 percent of the population. The effectiveness of the system i s even lower: less than 5 percent of all poverty gap i s eliminated by the transfer in BiH, as opposed to almost 30 percent in Hungary. 108 Figure 8.2 Internationalcomparisons of social assistance efficiency and effectiveness. 40 , ..... .... .. .... ....._ . __... . . . . ....-.I.._.. I_....____ _ ___ .-r1.2 ~ Bulgaria Estonia h g a r y Iahia Poland Russia Croatia BiH I(leftscale)-Effectiveness Efficiency (kft scale) -Role of Social Assistance (light scale) Source. Braithwaite, Grootaert and Milanovic (1999), BiHstaff estimates based on LSMS Note: Effectiveness in percentage score i s how much o f the (pre-assistance) poverty gap o f the bottom decile i s eliminated, efficiency inpercentage score i s how much o f disbursed social assistance i s received by the lowest decile. 8.41 The results presented on Figure 8.2 show that the very little money spent on social assistance seems to be not well spent. This low effectiveness o f social assistance benefit in BiH should not be explained by a meager finding of the system in BiH. The Figure 8.2 shows the overall size of the poverty benefit, as a share of private consumption (line market with triangles with right axis scale). Although the amount spent on social assistance i s small (around 0.6 percent of total private consumption), it i s actually higher than in many other countries in the comparison (Croatia, Latvia or Estonia). Simulating in BiHthe performance o f the social assistance in Croatia shows that an efficient and effective system that would increase the number of beneficiaries to include all the poor should cover on average about 20 percent of the poverty gap. Instead the current systems cover about 5 percent of the poverty gap of a very small portion of the current poor. 8.42 Child allowances in particular should be retargeted and enhanced. Current resource constraints reflect the fact that the systems are decentralized and financed at the local level. The limited available data suggest that only 25 percent of the very poor have been receiving cash benefits and that the number o f children covered has halved between 199.8and 2000, as many cantons in BiHhave terminated these programs. 8.43 The underfunding o f SWCP in BiH (with the notable exception of child protection in RS) is in stark contrast to other parts of the cash benefits system, in particular veterans' benefits, which are provided in nearly all cases regardless of income level, and consume resources many times those available for child benefits. This suggests that there i s major scope for reorienting the system towards better targeting o f cash transfers, involving both a more equitable allocation of existing social welfare spending, and reallocation o f resources from other parts of the transfer system (such as veterans' programs) that are weakly targeted and consume disproportionate shares of public spending. This will be challenging, given the political voice of veterans' groups relative to more marginalized and dispersed beneficiaries, 109 but i s a key area for reform if the PRSP commitment to addressing the needs of the most vulnerable i s to be credible. 8.44 Within the current social protection budget (KM10million), improvingtargeting and effectiveness would at most cover 8 to 20 percent of children in need instead of the 4 to 6 percent currently covered (depending the age of the children and the measure of poverty used). Expanding coverage to all poor children while covering the poverty gap they faced, assuming the best use o f the funds, would require KM70.8 to 83.9 million insteadof the 10.5 million currently spent. 8.45 This analysis confirms the conclusions of the PER that severe resource constraints sharply limit the number of beneficiaries of social assistance, but it also indicates that funds are not well spent. Thus a simple reallocation of money from other programs to social assistance i s not sufficient to reduce poverty. 8.46 Finally, according to the PIER, the lack of resources combined with the absence of regulations and clear responsibilities have led to local offices using a high degree of discretion in rationing beneficiaries. E. Unemployment Benefits 8.47 Both Entities adopted new unemployment insurance systems in the Fall of 2000.'33 Both systems are quite similar and provide a modest, affordable system of income support to the unemployed. Eligibility is limitedto workers with paid contributions who havenot quit a job nor been dismissed for cause. The duration of potential eligibility depends on the number of years of work experience and ranges from 6 to 12 months in the FBiH, and from 3 to 12 months in RS. The replacement rate'34ranges from 30 to 40 percent in the FBiH, and from 35 to 40 percent in RS. There are no ceilings or floors imposed on benefits. In comparison to other transition economies, the generosity of the benefit system i s on the low side, consistent with relatively limited financial resources. 8.48 Very large numbers of registered unemployed in Bosnia and Herzegovina are, in principle, eligible to participate in both active and passive labor market programs. Yet in the postwar period, there have been very few participants in such programs. In the FBiH, the number of recipients of unemployment benefits (workers who received benefits for at least a month) varied from 27 to 36 thousand in the 1998-2001 period; there were even fewer beneficiaries in RS (less than 2000 per year). The number o f unemployment benefit recipients i s so small that no statistically significant results were obtained in the LSMS (their share in the population i s below one percent, making general population survey with a margin of error of several percentage points totally inappropriate as a tool for monitoring of this group). ~ 133The Decision on Job Placement and Social Security of the Unemployed, Official Gazette of the FBiH,No. 55/2000; Job Placement Law, Official Gazette of RS, No. 38/2000. 134The replacement rate i s it the percentage of previous salary covered by unemployment benefit. 110 8.49 According to employment office data, 424 thousand workers were registered as unemployed in 2001, 275.8 thousand in the FBiHand 147.7 thousand in RS. Among these workers, 44.6 percent were women, which i s higher that the women's share in formal employment. In the full conformity with the LSMS, the registered unemployedincluded a relatively highproportion of young workers - 27.5 percent of unemployed were younger than 26, 41 percent were 27 to 39 years old, and 31 percent were older than 40. The educational structure o f the registered unemployed i s similar to the one obtained from the LSMS, with very few registered unemployed possessing a university degree (2.6 percent), and about one fifth possessing a highschool diploma. 8.50 The LSMS results shed light on the inefficiencies of the system. Only slightly less than a quarter of the registered unemployed qualified as unemployed under the standardILO- OECD definition, another quarter of them were employed, and 51 percent of them were out of the labor force. The main reason for registration thus appears to be free access to health services, an entitlement which i s obtained by registration. Not all the unemployed had registered, though 69 thousand (36.3 percent of the unemployed) did not approach employment offices and register. Thus registration statistics offers a distorted picture of the real situation on the labor market (World Bank 2002f). It i s even less appropriate to target transfers to those in need. 8.51 It is clear that the unemploymentbenefits system in its current state can hardly meet the challenge of absorbing a large employment shock with mass lay-offs and benefits actually being claimed by potential beneficiaries. The system is largely clogged by the "paper" unemployed, who use the system as a source of benefits (but apparently not unemployment benefit). Improving how the system functions will be an important step towards better protection against risks. F. Conclusions and Policy Implications 8.52 BiH's social protection system as a whole plays a substantial role in income support and keeping many households out of poverty. But different social assistance programs have had varied success inreaching the poor. 8.53 The pension system and the veteran's benefits keep many pensioners and beneficiaries above the poverty line. Inthe absence of these benefits, the poverty rate would be much higher. 8.54 A recently completed Local Level Institutions study (World Bank 2002d) found that pensions are absent from a "black list" of bad public services. This i s significant: pensions used to be cited in former studies as among the worst elements of the social protection system. The pension reforms of 1998-2000 have had a significant effect in stabilizing the financial position o f the PIOs in both Entities, and in regularizing pension entitlements and payments. As elsewhere the impact o f these improvements on poverty has been substantial. Nevertheless, fiscal pressures remain strong, making a more fundamental overhaul o f the pay-as-you-go (PAYG) pillar desirable in the medium term. Options for further adjustments of the rules governing pension entitlements, including further adjustment of accrual rates and action on the early retirement regime will need to be carefully reviewed, including from the 111 point of view of poverty implications. Apart from the PAYG pillar, it would be useful to assess the main constraints on the possible future introduction of funded pillars in the Bosnian systems. 8.55 Incomparison with many other countries inthe region, BiH's system of targeted cash transfers i s not effective and not well targeted. The system can and should become more pro- poor. Safety net priorities should include ensuring that the targeted benefits system i s sufficiently flexible to address future potential shocks. A particular concern i s the exposure of many households in BiH to high risks of catastrophic illnesses. Health policy and social protection should be more closely coordinated to ensure adequate management o f such risks. 8.56 As targeted programs command an inadequate share of overall Government social protection expenditures, the system has a limited capacity to protect the most vulnerable. Existing resources are misallocated to non-poor beneficiaries in politically driven social programs (such as veterans' benefits). What i s needed i s a serious reallocation of resources to address more adequately the needs of the poor. 8.57 The structure inherited from the past, involving multiple types of benefits managed by different agencies (with programs that often duplicate or even contradict each other) needs to be re-evaluated. Reform should focus on improved governance of the system, better coordination and specific targeting. 8.58 Institutional fragmentation and financial exhaustion also affect the policies of international donors. Many donor-financed projects encompass a limited number of beneficiaries, creating tensions at the local level, not only between NGOs implementing these projects and potential beneficiaries, but also within the local population itself. Negative feelings about donor policies are exacerbated by the fact that the criteria used for allocating donor assistance are not clear. Attempts to draw up lists of beneficiaries also give rise to complaints about "clientism" and corruption. The most frequent demands expressed by the population with regard to social benefits and donor support is the adoption of clear, comprehensive and transparent criteria for al10cation.l~~ 8.59 Support for war veterans needs to be rationalized and better targeted, since current arrangements place an enormous burden on State resources. Under the World Bank's Labor Redeployment Project, approaches would be tested which could demonstrate better ways o f targeting benefits, for example by using self-selection methods and by supporting veterans' reintegration into economic life. 8.60 More specifically the reform agenda would include: 8.61 Social Welfare Policies. Develop and legislate a new Social Protection Strategy in both Entities with an emphasis on the following key aspects: . Determining a realistic core set of entitlements, with clearly identified financing sources. 135See Local Level Institutions study for details. 112 ...Ensuring that Entity budgets finance one or more core "poverty benefits" equitably across each Entity. Creating incentives for adequate allocations for social protection by local governments. Developing sound targeting criteria (e.g., for means-testing andor beneficiary identification according to monitorable eligibility rules) in line with resource constraints and administrative capacity. 8.62 Balance between Resourcesfor Benefits and Resourcesfor Services. Inaddition to the above-noted options for improving financing arrangements, the authorities need to reassessthe balance between cash benefits and services to ensure that current resource allocations actually reflect community needs and priorities. 8.63 Institutional Framework. Redefine institutional responsibilities in social protection on the basis of the functional review of social protection to be undertaken in collaboration with the World Bank under the Social Technical Assistance Project (SOTAC, see World Bank 2003 c). 113 114 9. POVERTYREDUCTION: CHALLENGESAHEAD Thisfinal chapter discusses poverty reduction objectives and recommendations on the basis of the key findings about causes of poverty in BiH. The analysis presented in previous chapters indicates that, given current economic, political and social circumstances, poverty cannot be eradicated completely in the BiH PRSP time horizon (2003-06). Even gradual progress in reducing poverty will be difJicult, requiring substantial efsorts to redirect policies infavor of the poor. Continued growth is necessaryfor poverty reduction, but is not sufJicient to guarantee it: indeed, unequal, or anti-poor growth can reduce or wipe out the benefits of economic expansionfor the poor. Critical for achieving pro-poor growth will be enhancing the impact of growth on job creation, which in turn requires more decisive progress in implementing structural reform. The chapter concludes that, despite a qualitative leap achieved through the PRSP in understanding the nature of poverty in BiH and devising policy actions to address it, the country still faces a challenge with respect to creating a viable and sustainable poverty monitoring system that can trace the evolution of poverty over time and assess thepoverty impact of variouspolicies. A. Understandingthe Primary Causes of Poverty in Order to Fight it 9.1 The key recommendation of this report i s to change the focus o f policy making with respect to poverty and to review the assumptions underlying it. Past policies reflected the view that the majority of the population was poor, and that therefore any action that produces benefits to the population at large would be good for poverty reduction. This assumption i s exemplified by the UNDP Human Development Report of 1998 which claimed that two thirds of the population was poor. This approach needs to be carefully reassessed. 9.2 It is important to emphasize, however, that the majority of the poor still share many characteristics with a typical household, which in turn still calls for "universalist" policies, albeit different from the current ones. As demonstrated in Chapter 2, even a "typical" household in BiH faces a certain risk of falling into poverty in the future. In such an uncertain environment it i s important to make sure that every household i s covered by some social protection arrangements. Given that a majority of the population considers itself poor, universalism would also be more acceptable than targeting from a political perspective. 9.3 In the other hand, the report also finds that locality of residence is one of the distinctive factors associated with poverty. Directing programs specifically to poor localities would be one important way to improve targeting of policies, and should be incorporated as a delivery mechanism for some poverty reduction programs. Unlike universal arrangements, 115 targeting will provide for a more efficient use of public finds. Analysis of poverty causes helps to better assess the balance between the targeting and universalism. 9.4 The key conclusions from the analysis of poverty in BiH are closely aligned with the framework proposed by the World Development Report 2001 on Poverty, which views its primary causes as limited opportunities, lack of security, and powerlessness. In the BiH context, these core dimensions o f poverty need to be viewed against the backdrop of the transition to a market economy, combined with the constraints imposed by the aftermath of conflict and enduring ethnic divisions. 9.5 The conceptual framework that views poverty as an outcome of disadvantage across all three dimensions - opportunities, security and powerlessness - helps both to identify the core sources o f economic and social vulnerability, and to provide the basis for a strategy for poverty reduction. 9.6 Opportunities are limited for the Bosnian poor because, despite fairly robust growth since 1995, BiH has experienced only a modest net increase in aggregate employment (if any) compared to the pre-conflict period, and labor market participation i s still falling below the levels needed to make a decisive impact on poverty. Despite a strong base and success in rebuilding the infrastructure needed to invest in the human capital of a nation, BiH still faces a large gap between the needs of a modern market economy and the delivery o f health and education services. 9.7 Security, above all physical security, remains a problem (though not as acute as immediately after the conflict). Physical insecurity affects all groups in society and creates a permeating environment of uncertainty that drives out long term considerations. 9.8 Economic security i s also a problem. It has a number of dimensions-such as uncertainty about basic property rights (particularly evident in unsettled property rights in housing, affecting those displaced by the war); corruption; failure to adequately protect the population against economic and associated risks, such as disability, - all o f which are widely viewed as major factors contributing to the risk of falling into poverty in the future.'36 These factors help to explain why the this risk widely spread across population groups. 9.9 Both private and public social safety nets are extensive in BiH, but they seem over- stretched and do not adequately cover those in need. Targeted public safety nets are difficult to implement owing to the precarious state of public finances, the limited ability o f the government to provide services, and problems associated with a divided society marked by ethnic tensions. Private safety net arrangements are inherently limited by the many divisions runningthrough society and their focus on those perceived as "deserving poor". 9.10 The combination of insecurity and limited economic prospects leads many young people to leave the country. These factors also inhibit the return from abroad of almost a 136See World Bank 1999c. 116 million citizens, many of whom have the skills or capital that could help to revive the ec0n0my.l~~ 9.11 Finally, the powerzessness and voicelessness experienced by the poor exacerbates the problems they face in trying to escape from impoverishment. Powerlessness and voicelessness reflect the nature of institutional relations and connections between people (poor and non-poor), between poor people and governments, and between poor people and civil society. Poor people are generally excluded from the institutions that make the decisions and allocate the resources that affect their lives. Further, in an effectively decentralized system of resource allocation, powerlessness in Bosnia and Herzegovina i s primarily linked to the lack of social capital and weak community mobilization inpoor areas, together with disruption o f social capital incommunities affectedby the war. 9.12 Successfully attaclung poverty in BiH will require substantial and simultaneous action to promote opportunity, security and empowerment. And it will also require a commitment to positive change, to monitoring the effectiveness of policy actions, and to an across-the-board participatory approach-all of which are central principles of the PRSP process which BiHi s in the process o f completing. B. How Far Can Economic Growth Reduce Poverty in the Future? 9.13 Economic projections for BiHshow that the growth rate o f private consumption could range from 1percent per year (representing the "status quo") to 3 percent per year (under a "full reform" scenario). Future poverty trends are directly linked to consumption growth. Table 9.1 explores what would happen to the poverty rate and poverty gap in 5 and 10 years under various growth scenarios. Annual Poverty Rate I Poverty Gap consumption growth 5 years 10years 5 years 10years -3% 35% 74% -9% -27% -2% 26% 35% -6% -9% -1% 22% 26% -5% -6% 0% 20% 20Yo -5% -5%' 1% 17% 15% -4% -3% 2% 15% 12% -3% -3% 3% 12% 6% -3% -1% Source: Staff estimates based on BiH LSMS data. Estimates assumed a constant poverty line, and that the equivalent consumption distribution remains constant-in other words, everybody experiences the same consumption growth. 9.14 The table shows that in the coming 5 years BiH could reduce poverty b y 20 percent (from 19.5 to 15 percent) if it were to achieve a sustained real per capita consumption growth of 2 percent per year. Note that the poverty gap, which indicates how far the poor are below the poverty line on average, responds even more strongly to faster growth. Lower growth 13'See UNDPHDR (2002) 117 rates would reduce poverty more slowly; annual consumption growth of one percent would reduce the poverty rate to 17 percent over five years. 9.15 The critical assumption underlying these projections i s that consumption growth i s evenly distributed. In other words, everybody, both rich and poor, experiences the same real consumption growth rate. Even under this assumption, however, one cannot expect an eradication o f poverty in BiH. 9.16 But the assumption about stable inequality may be even a too optimistic view. Although BiHcurrently has a modest level of inequality, this could increase in the future, as the experience of other countries has demonstrated. To explore the linkages between growth and inequality, Table 9.2 shows the combined effect of different rates of growth of real consumption and change in its distribution leading to a given increase or decrease in ineq~a1ity.l~~ 9.17 Table 9.2 i s organized crosswise in four panels. The upper two panels show changes inpoverty rates, while the lower two show changes inthe poverty gap. The left handpanels display simulated changes in poverty incidence and gap over a 5 year time horizon, while the right hand panels take a longer (10 year) perspective. Each cell in the panel shows a simulated poverty statistic for a given combination of change in mean real consumption (arranged from top to bottom) and inequality (crosswise from decrease to increase). 9.18 Simulated figures under assumption o f constant growth or inequality (corresponding to 0 percent change) are shown inbold. The numbers runningdiagonally in the shaded boxes show the impact of simultaneous changes in consumption growth and inequality. In each panel the North East section shows the worst case scenario: falling consumption and increased inequality, while the South West section shows the best case: rapid growth with reduction of inequality. 9.19 Table 9.2 shows that if inequality rises, growth might not in fact lead to poverty reduction at all. Best possible outcome in terms of growth combined with increase in inequality (South East in each section in each panel) would see the positive impact of growth on poverty being completely offset by increase in inequality. 9.20 The message from this simulations i s a very important one: consumption growth by itself may do nothing to reduce poverty in BiHunless it i s accompanied by a series of public policies that ensure its pro-poor character. In the context of BiH, the share of resources that 13'We use BiHLSMS data 2001 as the basic tool. To project the impact o f economic growth with constant inequality, we simply multiply everyone's consumption by a given growth rate and compare the simulated level of per capita consumption with the constant poverty line. To simulate the effect o f a given percentage change in inequality (measured by the Gini index), we use the following simple representation for the proportionate shift inthe Lorenz curve: we find a new levelof per capita consumption y for household i so that to obtain l+a change inthe Gini, corresponding to a new Lorenz curve for the distribution of per capita consumption Y*~,, which can be simplified to finding a value o f !3, such as y*i =(1+ P)[(l-a) yi+ p(yi )], where p i s the mean. The new Lorenz curve is then given by L*(p*)=(l-a)L(p)+ a, and the change in Gini is Gini(y*)=( 1-a)Gini(y). See RavallionM.and HuppiM.(1994) for derivation. This simulation therefore i s a simple illustrationand ignores any BiH idiosyncratic dynamic effects (such as aging o f the population or change in household compositions). 118 can be re targeted to achieve any impact on overall inequality i s small. The pattern of growth by sectors and the ability of the population to respond to new opportunities is therefore of paramount importance for reducing inequality and poverty. 9.21 To show how the pattern of growth matters for poverty reduction, we attempt a series of basic simulations of sectoral growth rates.'39We use the classification by sectors that takes into account the main source of income for a household (work or transfers) , industry and type of ownership. We assume that consumption of all households in a given sector i s directly and proportionately related to the sectoral growth rates through productivity (and wage) increases or through the creation of new employment for secondary work or additional income earners. Table 9.2 Full Poverty Projections: Effect of Changes in the Mean Consumption and its Distribution on Poverty in5 and 10 years. Poverty Rate In5 Years with growth and inequality: 1Inequality (Annualchange) 1 Poverty Rate In10 Years with growth and inequality: J Inequality (Annual change) J -3% -2% -1% 0% +1% +2% +3% Growth -3% _ _ -2% -1% n9i - - 1 % +2% +?% - " ~ - - - - - -3% 27% * ' 35% 42% -3% 56% ' 74% 88% 26% 29% -2% 27%. 42% 20% 22% 23% -1% IIy'I 0% 12% 16% 17% 20% 21% 22% 26% 10% 2% 12% 16% 20% 22% 26% 33% I 15% 18% ' ' . Poverty Gap In5 Years with growth and inequality: Poverty Gap In10 Years with growth and inequality: 1Inequality (Annual change) 1 J Inequality (Annual change) 1 -3% -2% -1% 0% +I% +3%+2% Growth -3% -2% -1% 0% +1% +2% +3% -3% 6% 9 % 13% -3% 22% '27% 38% S% I- 6% 8% -2% 6% - 9 % 13% 1121 5% 570 6% -1% 5% 6% 8% 0% 2% 3% 4% 5% 6% 6% 9 % 10% 0% 2% 3% - 5 % 6% 9 % 16% I Source: Simulations based on the BiHLSMS data. 9.22 We explicitly look at the cross- sectoral variation of growth rates, but at this stage abstract from mobility across sectors. Further, these simulations assume no feedback effects, such as the possibility that growth in export sectors would lead to technological spillovers 139For discussion of methodology and example of application see: Poverty and Social Developments inPeru, 1994-97. The World Bank 1999a. 119 and higher productivity in other sectors, as generally occurs. Despite these limitations, the simulations provide crucial insights how different patterns of growth matter for the poor. 9.23 The simulation attempts to shed light on which growth pattern will be most beneficial for the poor and lead to eradication of poverty in BiH.The simulations show that poverty can be eradicated in BiH if the growth in each sector i s proportional to sectoral poverty gap (see explanation above); it would be sufficient under this extremely beneficial scenario for the national average real consumption per capita to grow by 60 percent (e.g. 4.8 percent p.a. for 10consecutive years). 9.24 We then can look at various alternative scenarios for the same overall growth rate: we can contrast this effect with equally shared growth, and with the growth that will mostly benefit the rich, i.e. workers in the sectors which have the lowest poverty rate according to LSMS, and finally with regionally concentrated growth (with all growth concentrated among households residing in Sarajevo and Banja Luka). These four difference scenarios are presented as columns in Table 9.3. For each type of growth the table shows the corresponding annual growth rate of a sector and its poverty reduction outcome in 10 years. 9.25 Table 9.3 shows that to achieve "pro-poor growth ",the sectors that will need to grow fastest are agriculture and manufacturing. It also underlines the importance of public transfers for achieving these outcomes. Pro-poor growth Equitable growth Pro-rich growth Pro-Metropolitan Sector: main income Source Growth Percent Growth Percent Growth Percent Growth Percent of the household rate poverty rate poverty rate poverty rate poverty reduction reduction reduction reduction Transfer recipients 6.2 -100% 4.8 -69% 0 0% 5.3 -6% Agriculture forestry etc. 5.3 -100% 4.8 -72% 0 0% 1.o -3% Manufacturing and utilities 4.8 -100% 4.8 -75% 1.4 -5% 2.9 -4% Construction 4.3 -100% 4.8 -82% 1.4 -14% 4.3 -10% Services, trade, communic. 3.8 -100% 4.8 -85% 9.6 -97% 6.7 -13% Total 4.8 -100% 4.8 -76% 4.8 -23% 4.8 -7% Source: LSMS 2002. Note: Simulations of 60% increase in real private consumption per capita are for 19 mutually exclusive groups. Equal growth rates are assumed for all households in a given group. Pro-poor growth is derived from Lorenz curve in each group and guarantee that the mean consumption in the lowest ventile i s above the poverty line. Pro-rich growth assumes that all growth i s equally distributed among groups with above the average per capita consumption. Pro-metropolitan growth assumes that all growth is concentrated in Sarajevo and Banja Luka. 9.26 It is hardto imagine, however, that the economy can sustain the level of redistribution suggested by the most pro-poor growth pattern: the simulation implies that real consumption of those relying on transfers has to increase by 6.2 percent per annum. This i s clearly beyond the fiscal capacity o f the Government. Thus this result can interpreted to indicate that the only solution for poverty eradication i s to provide job opportunities for those relying on transfers (without employment) who can work. On the other hand, those who are unable to 120 work and are in need have to benefit from a radically improved targeting o f social assistance to see their consumption levels improved. 9.27 A simple equitable growth scenario (under which everybody's consumption grows at the same rate, second scenario presented in Table 9.3) can produce substantial benefits to BiH's poor. On the other hand, if growth were to be concentrated in "rich" sectors, such as services and private formal sector firms, the benefits in terms of poverty reduction would be substantially reduced. While growth cannot be artificially tilted towards the pro-poor sectors, investment in these sectors will depend on putting in place a policy framework which at least does not discriminate against them. 9.28 Closely linked to this strong impact o f the sectoral distribution o f growth i s the impact of different regional patterns of development. Poverty reduction will be strongest if the rural sector and manufacturing in small and medium sized cities were to be the main sources o f economic progress in the coming years. If Sarajevo and Banja Luka were to continue their role as engines of growth, with new growth concentrated solely there, poverty reduction with the same aggregate economic growth rates will be the worse in all presented scenarios - only around 7 percent. 9.29 This analysis confirms that inthe longruninequality is a threat to social sustainability in BiH.14' To ensure the equitable or pro-poor distribution of growth benefits, it will be important to choose the right policies. For example, poor people need to have greater mobility - both economically and geographically - if they are to be able to use the opportunities presented by growth and move to more productivejobs. 9.30 The growth of employment in agriculture will be strongly pro-poor, that is, it will attract new workers from the ranks of those currently not employed who are now poor. Growth in manufacturing will also be a strong factor, but will have an equal impact on the rich and the poor. The growth of formal services (the bulk of which consists o f public sector employment) will provide employment opportunities for the "rich". 9.31 In the context of prospective declines in reconstruction aid inflows and high unemployment, strengthening the private sector will be critical. In addition to the challenge of accommodating new labor market entrants, there i s also an immediate need for job reallocation stemming from the on-going restructuring. According to projections preparedby one of Bosnia's bilateral donors, the transformation of moribund public companies will produce 50,000 additional dismissed workers by 2005 in the best-case scenario, or 76,000 in the low-case scenario, a burden far beyond the present capacity of Bosnia's weak social welfare systems. 140BiHPRSP. 121 BOX XV. Recentupswing inreturnee numbers The number o f registered minority returnees to Republika Srpska has leapt from less than the 10,000 per year before 2001 to more than 91,000 in January 2002, while the number of Serb returnees to FBiHhasjumped from 19,000 to 87,000. Given the inadequacies o f the registration process, the real figures are probably much higher. This process of reversing ethnic separatism has radically transformed Bosnia's reality. There are now mosques inPrijedor, Bosniac businessmen inDoboj, a strong Serb community inDrvar and hundreds of Bosniac children attending schools in Stolac. In Republika Srpska, the shift in attitude came later and still lags behind, but the changing reality inthe field over the past two years suggests that it will follow a similar path. The recently completed World Bank study on Local Level Institutions provides additional evidence on deep changes in attitudes towards returnees. In the Social Assessment (World Bank 1999), 74 percent of Bosniac IDPs, 29 percent of Croat IDPs, and 19 percent of Serb IDPs state that they wish to return to their pre-war place of residence. In 2001, about 80 percent of IDPs living in the Bosniac majority area, 50 percent of those living in the Croat majority area, and 30 percent of those living in the Serb majority area said they were willing to return. These decisions are influenced by factors such as employment, age, housing status, and level of education. I t would be an oversimplification to say that reintegration faces no major obstacles. As detailed by UNHCR, psychological trauma means that many will never attempt to return. Such trauma i s considered to be a major factor hindering the sustainable return of single female IDPs and refugees. In the 2001 LSMS survey, 20 percent of all respondent (and a quarter of women) stated that they still often suffer from the consequences of traumatic war experiences. A staggering 40 percent of women who remain refugees or IDPs admitted this as well. Inthe focus groups, many IDPs mention loss ofjob and fear of unemployment as the main reasonsthey do not want to return. The preoccupation o f IDPs and minority returnees with economic issues shows a change in attitude since 1999, when participants in the Social Assessment considered personal insecurity as a key obstacle to return. In the Social Assessment (World Bank 1999), 68 percent o f IDPs mention personal and property security as the main preconditions for return. Sources: ESI and UNHCR, Returns summary to Bosnia and Herzegovina from 01/01/96 to 31/01/02: www.unhcr.ba, DauntingProspects Minority Women: Obstacles To Their Retum And Integration, 2000 and Local Level Institutions and Social Capital Study, The World Bank 2002. 9.32 Finally, a significant challenge i s related to a very positive development: the dramatic recent upswing in the numbers of returning refugees and IDPs (see Box XV). Employment i s a critical condition for sustainable rebuilding o f BiH as a milti-ethnic State. C. Setting Priorities for Poverty Reduction 9.33 BiH is in the process of completing its first PRSP. The strategy lists the following policy priorities for poverty reduction: 1. Establishing a Single Economic Space in BiH; 2. Improving the business environment for domestic and foreign investment, and completing the process o f privatization; 3. Promoting the rule of law and establishing effective mechanisms for combating corruption; 122 4. Establishing sound social, health care and education systems; 5. Converging with EUintegration, through the Agreement on Stabilization and Association: 6. Fostering integration into the global economy by attracting foreign investment andjoining the World Trade Organization (WTO). 9.34 How well do these priorities accord with the causes of poverty identified in this report? Will their implementation lead to a significant reduction of poverty ? 9.35 Strictly speaking, only the fourth item listed in para. 33 above i s directly linked to human welfare. Evidently, this does not imply that it i s the only item that i s likely to have a significant impact on poverty. Indirect instruments may substantially benefit the poor as the present report has emphasized. 9.36 The first priority for BiH i s to achieve economic sustainability. Only then will progress with poverty reduction also be sustainable. To achieve rapid and sustainable growth, BiH needs to invest more and more effectively, from its own resources or by attracting foreign private investors. 9.37 Without continuing strong investment BiH cannot count on growing fast enough, or on competing in and integrating with European markets. To support investment needs while external aid flows decline, BiHwill need dramatically to increase domestic savings. There i s only limited scope for using public savings to expand investment, because the fiscal position remains fragile. So far, foreign assistance, with a grant component o f 73 percent, has provided the bulk of the financing needed to cover public sector deficits. As a result, despite large imbalances, deficits have not been translated into higher debt-to-GDP ratios and debt service has remained at comfortable levels. Nevertheless, in the period ahead, BiH's debt service burden will be on an increasing path; principal repayments started falling due in late 2001.The volume of concessional lending available to BiHhas also started diminishing. 9.38 While fiscal stability has been maintained to date, thanks to donor support and conservative fiscal policies, a significant fiscal adjustment i s required in order to consolidate macroeconomic stability under the Currency Board arrangement. The actual fiscal (primary) deficit i s substantially out of line with the requirements o f "financeable" debt-output ratios in the medium term. BiH's already high tax burden makes it clear that the bulk of the fiscal adjustment will need to be made in public expenditure. 9.39 In the first instance, this will require a comprehensive rationalization of public spending so as to lower domestic consumption. This will create room for increased public investment to refurbish public infrastructure over the medium term while maintaining public external indebtedness below 65 percent o f GDP. In addition, private savings rates will need to rise, notably through retained earnings in the corporate sector. This will require significant moderation in labor costs and ongoing efforts to improve the efficiency o f all aspects of corporate operations. 123 9.40 The provision of savings for investment can only be achieved through improved incentives to invest, i.e. an through improved business environment, which, in turn entails the following: P Corporate governance needs to be transformed through comprehensive privatization designed to provide for effective owner control of corporate activities and to bringknow- how and capital into firms; > Domestic relative price structures will need to be brought into line with international relative prices through further trade liberalization and the removal of domestic distortions; P A considerably more business-friendly environment needs to be established through reform of legal, regulatory, tax, and competition regimes that affect business activities; 9.41 In support of private sector development, the Government needs to continue to negotiate with its trading partners to increase market access, notably in the region and in the EU. It also needsto press forward with preparations for WTO accession. 9.42 Improving the business environment i s one of key priorities from the point of view of poverty reduction, because doing so i s the only way to create more and betterjobs. 9.43 The analysis of this report strongly supports a multi-pronged approach to poverty reduction, because it i s impossible to make a significant impact on poverty by addressing individual sector in isolation. Working towards the PRSP objectives clearly follows from the analysis of poverty and vulnerability. 9.44 In addition, the analysis of challenges in this section argues for the inclusion of macroeconomic stability among the priorities. This i s because macroeconomic policies, especially prudent monetary policies and responsible fiscal management, plays a role in maintaining economic stability and hedging the population against economy-wide risks. One of the key features in the BiH economy i s high vulnerability to poverty. A stable macroeconomic environment helps reduce risk and vulnerability. Moreover, other factors likely to affect the economy-including limited prospects for consumption growth, only limited early gains from trade integration, limited prospects for FDI, and declining international assistance all point to the need for strong and consistent macro stability. 9.45 Two other features of the analysis in this report deserve special mention as priorities for the necessary multifaceted strategy for poverty reduction in FBiH. The first i s improving labor market outcomes, and especially progress in raising female labor force participation. The poverty analysis made it very clear that whether or not women participate in the labor force makes a significant difference to family income. A lower dependency ratio means a much lower probability of being poor. The experience of other countries shows that women's labor force participation also has beneficial effects on the intra-household allocation of consumption (in favor o f children), reduces women's vulnerability to violence, and improves investment in human capital. High female labor force participation in other 124 countries is to a large extent built on the availability of child care programs and targeted transfers. The linkages between women's labor force participation, family incomes and child care could justify a renewed role for the government and the public sector, in partnership with the private sector, in ensuring that quality care i s affordable for all children. 9.46 The second area for special emphasis i s workmg to empower poor people. Ina world where political power (and economic power) i s unequally distributed, state institutions and rules will often work against the poor. From the point of view of a poor person, this often translates into a sense of voicelessness and powerlessness. Empowerment in this context comprises all the actions aimed at giving the poor voice; at increasing their ability to affect the decisions that affect their lives. 9.47 What are the main decisions/institutions that the poor are interestedin affecting? The first i s the legal framework (that i s the laws, rules and regulations that govern economic, social and political interactions). Of special importance for the poor are those aspects of the legal framework that regulate economic transactions - employment, rights to private property, ability to sell one's product, freedom of mobility (e.g., what might be called "pro- poor market regulations and policies"); as well as those that govern political rights (for example, the right to organize, to form associations, to participate in elections). A second key area i s that of institutional arrangements governing access to basic services (education, health, water, heating etc.). A third area i s access to recourse mechanisms andjustice. And a fourth i s that of the workings of state institutions and the public administration (local and central), particularly in the realm of economic policies (e.g. taxation, public expenditure) that directly affect the poor. 9.48 What meandinstruments do individuals (and especially the poor) have at their disposal to affect these decisions/institutions? These instruments are of four basic kinds: (i) instrumentsthatprovideaccesstoinformation (freedomofinformationacts,public information campaigns, disclosure requirements, free media, user surveys, user/ consumer/ small business associations, legal aid, representative labor unions); (ii) instruments of inclusion/ participation (democratic elections, political parties, user surveys and report cards, participatory budgeting mechanisms, social partnership agreements, social funds, representative labor unions); (iii) instruments of accountability (audit bodies, ombudsman, administrative tribunals, small claims courts, report cards) (iv) instruments to build local capacity (democratically elected local governments and representatives, social and investment funds, user/ consumer/ small business associations) 9.49 Empowering the poor requires protection of their rights. Therefore, continued support for judicial reform i s an important element of a poverty reduction strategy. The government's current reform program risks being undercut by a range of factors, including, e.g., judges not being paid on time, political interference in court cases, lack of enforcement 125 of court decisions, and, most critically, a corrupt procuracy. Collecting adequate information and formulating specific priorities for action are needed to keep judicial reform on track. 9.50 Another aspect of empowering the poor i s giving them an opportunity to manage programs that directly affect their lives. Most o f these mechanisms need to focus on the local level, because this i s the area where the positive impact i s greatest, in BiHas elsewhere. The Community Development Project financed by the World Bank offers a useful model for replication. Its objectives are to: improve basic services and facilities for low-income and poor communities in under- served municipalities (through investments in non-revenue generating socially oriented projects and programs). Improve the governance of local authorities and their capacity to deliver services to the poor through better partnerships between poor communities and municipalities in investment identification and decisions. The project will build the capacity of municipalities to lead community development processes, thus increasing the voice of local communities ininvestment decisions. 9.51 How one can help empower communities to examine their common needs, set priorities, and cooperate in achieving improvements in citizens'--especially poor citizens7- quality of life and prospects for the future? In many parts o f BiH, various donors are striving to use demonstration effects to help strengthen community cohesion, especially in localities where the ravages of conflict and associated material losses remain the most evident. Cohesion i s achieved through the identification of a shared priority, co-financing investments, participating in contracting and works supervision and, finally, undertaking full or joint responsibility for operations and maintenance. D. Monitoring Poverty Reduction: Statusand Recommendations 9.52 The final challenge for successfully attacking poverty, in BiH as elsewhere, i s the establishment of an effective system for monitoring poverty reduction actions and outcomes, based on the systematic collection and analysis of information on the implementation of interventions and on changes in outcome indicators. Monitoring can help to achieve: A. Improved interventions through detailed formulation of objectives, activities and expected results and through fine-tuning o f future activities B. Greater accountability through explicit commitment andmeans of control andmonitoring 9.53 The BiH PRSP makes a strong statement that improvements in the quality o f economic data will be essential for effective diagnosis and implementation of economic policies and program monitoring. The poor quality o f the statistical data base bedevils the Government's efforts to formulate policies. With the support of the European Union, the IMF,the World Bank, andbilateraldonors, the Government needs to undertake the necessary steps to improve the quality and coverage of economic data in general and poverty and inequality data in particular. 126 9.54 Monitoring i s essential to assess the quality of the growth process. To that end, the Government needs to improve coordination among the three statistical institutes and other data collectors with a view to creating countrywide statistical data, including consolidated government finance statistics and social sector and enterprise-level data. It i s important to produce the first set of national accounts on an expenditure basis, to prepare real GDP estimates for 1998-2002, to continue efforts to estimate the size o f the parallel economy, to produce a revised CPI and a revised industrial production index based on updated weights, and to improve the quality of data on foreign direct investment. At some point it will be necessary to initiate the first population census since the end o f the war, and to launch regular household budget, labor force, consumer and business surveys. 9.55 Monitoring consists o f compiling and following a set of indicators. In the PRSP context, indicators are being selected to reflect the main objectives of the national poverty reduction program, and measure progress toward those objectives. The choice of indicators needs to consider the availability o f relevant data sources at the points in time and the levels selected. The PRSP includes many activities and dimensions: each has to be followed with a specific set of indicators. 9.56 The analysis presented in this report shows that huge challenges remain in the area of monitoring the core set of indicators for PRSP - that i s the percentage o f people below the poverty line, enrollment rates and completion rates. 9.57 The key concern i s continuity in the efforts to collect and analyze nationally representative data on poverty. The LSMS survey was completed in 2001; since that time, no comparable data have been generated. The LSMS i s continuing on a panel basis (IBHI Panel survey "Living in BiH', conducted with DfID support), with a second round completed in 2002. The survey does provide very important data on access to health and education services and on labor market. But the consumption module essential to the measurement of poverty has been removed from the panel questionnaire. Thus, no follow-up data on poverty rate are available and there are no plans to collect such data in the Panel survey project. 9.58 The alternative source for representative data on poverty can be a special survey focused on household consumption behavior - Household Budget Survey (HBS). The project to carry out HBS in BiH with a support of the Statistical Agency o f Italy has been discussed since 1999. Household Budget Survey started to be implemented in 2003. But there i s an uncertainty about what its role as a monitoring tool might be. There i s also currently no plan for conducting the research to use the H B S as a poverty monitoring instrument. Thus, a political decision on how poverty i s going to be monitored over the PRSP time horizon i s urgently required. 127 128 REFERENCES Atkinson A B. 1998. "Social Exclusion, Poverty and Unemployment," inExclusion, Employment and Opportunity. A B Atkinson and John Hills eds. London: CASEpaper 4; Centre for Analysis of Social Exclusion, London School of Economics, pp. 1-20. Barberia, Lorena, Simon Johnson, and Daniel Kaufmann. 1997. "Social Networks in Transition." WilliamDavidson Institute: Ann Arbor, Michigan. working paper, 102. Bema@ Sabine, Informal Employment in Countries in Transition: A conceptual framework CASE paper 56 Centre for Analysis of Social Exclusion, April 2002 London School of Economics Bisogno M. and Chong S. (2000) "Bosnia and Herzegovina: Labor Market Study". Mimeo. World Bank, 2000 Bosnia and Herzegovina Council of Ministers, Ministry of Foreign Trade and Economic Relations, Office of BiH Coordinator for PRSP (BiH PRSP 2003) Development Strategy BiH (2003-2007) - Poverty Reduction Strategy Paper. Second Draft for Public Discussion, Sarajevo. Department for International Development, Know How Fund and Health and Life Sciences Partnership (DfID 1999) Health Resource Accounts in Bosnia and Herzegovina: Report of methods and Results. UK Department of InternationalDevelopment (DfID). London. Eurostat Task Force (1998), `Recommendations on Social Exclusion and Poverty Statistics', Document CPS 98131/2, Eurostat, Luxembourg. Feige, Edgar L. 1997. "Underground Activity and Institutional Change: Productive, Protective and Predatory Behaviour in Transition Economies," in Transforming Post-Communist Political Economies. FRYOM: Ministry of Finance. National Poverty Reduction Strategy. Skopje 2002. Foreign Investment Advisory Service (FIAS 2001): Bosnia and Herzegovina: Commercial Legal Framework and Administrative Barriers to Investment. March 2001. IBHI(2002). International Bureau of HumanitarianIssues (2002). Qualitative Survey of Gender and Poverty. Final Report. Sarajevo, IBHI,2002. ILO. 1993a. "International Classification of Status in Employment." International Labour Office: Geneva. ILO. 1993b. "Resolution Concerning Statistics of Employment in the Informal Sector." International Labour Office: Geneva. Resolution I1adopted by the Fifteenth International Conference of Labour Statisticians. International Crisis Group's (ICG 1999) report Why N o One Will Invest in Bosnia and Herzegovina, 1999; International Crisis Group's ICG's (2001) draft report Still Not Open for Business: Bosnia's Precarious Economy, July 2001. 129 Jakubowsky E. and Jaganjac N. (2001) "Restructuring Hospitals: Bosnia and Herzegovina." Eurohealth, v.7 no. 3 Autumn. Johnson, Simon, Daniel Kaufmann, and Oleg Ustenko. 1997. "Formal Employment and Survival Strategies After Communism," in Transforming Post-Communist Political Economies. National Research Council ed, pp. 177-202. Kaufmann, Daniel and Aleksander Kaliberda. 1996. "An `Unofficial' Analysis of Economies in Transition: An Empirical Framework and Lessons for Policy." Harvard Institute for International Development: Cambridge, MA. Development DiscussionPaper, 558. Kolev, A. 1998. "Labour Supply in the Informal Economy in Russia during Transition." European University Institute: Fiesole. Lokshin, Michael M.and Mroz Thomas. 2003. The Gender And Poverty: A Life Cycle Approach To The Analysis Of The DifferencesIn Gender Outcomes. World Bank: Washington. Lokshin, Michael M.and Ruslan Yemtsov. 2001. "Household Strategies for Coping with Poverty and Social Exclusion in Post-Crisis Russia." Transition Economies Working Paper. World Bank: Washington. Medica (1999) To Live Without Violence. Second Look 2. Final Report, Zenica, Medica, Infoteca 1999. Medica (2000) Violence Against Women. Women to Women - Udruzene - Medica. Banja Luka, Zenica, Sarajevo, 2000 Mesinovic, Haris, Zlatko Hurtic and Mersida Suceska (2002) "Foreign Trade Institutions and Policies. Country Study: Bosnia and Herzegoniva" In:Trade Policy and Institutions in South Eastern Europe.The World Bank, Washington DC. Murray, C.J.L., Lopez, A. (1996) (Eds) The Global Burden of Disease: a comprehensive assessment of mortality and disability from diseases, injuries, and risk factors in 1990 and projected to 2020. Harvard University Press. OECD (2001) Education at a Glance: OECD Indicators, Education and Skills. Prepared by the Statistics and Indicators Division of the OECD Directorate for Education, Employment, Labor and Social Affairs. Paci, Pierella (2002) Gender in Transition. The World Bank, Washington DC. Ravallion, Martin and Monika Huppi (1991) "Measuring Changes inPoverty: A Methodological Case Study of Indonesia Duringand Adjustment Period," World Bank Economic Review, Vol. 5, pp. 57-82, January 1991. Ravallion, Martin (2001) Setting Poverty Lines: Economic Foundations of Current Practices. The World Bank. Washington DC. Stubbs, P. and Gregson, K.(eds) (1998) Social Policy, Protection and Practice: the care of vulnerable groups inBosnia-Herzegovina. Sarajevo: Svijetlost. 130 UNDP (1998) Human Development Report: Bosnia and Herzegovina. Prepared by Independent Bureaufor HumanitarianIssues. UNDP, Office of Bosnia andHerzegovina,Sarajevo. UNDP (2002) HumanDevelopmentReport: Bosnia and Herzegovina. UNDPO Office of BiH and EconomicsInstitute,Sarajevo 2002. UNDP Early Warning system (2000-2003). UNHCR (2001) Health Care inBosnia and Herzegovinainthe context of the returnof refugees and displacedpersons. UNHCR, Sarajevo, 2001. UNICEF (2002) Statistical Data. Wall Street Journal's CentralEuropeanEconomic Review,November 1999. Winkelmann,Liliana andRainerWinkelmann(1998) "Why Are the UnemployedSo Unhappy" Evidencefrom PanelData", Economica,65: 1-15. World Bank (1975) Yugoslavia: development with decentralization. Baltimore: Johns Hopkins University Press. World Bank and European Commission's BosniaandHerzegovina(The World Bank. 1999): Review of the Priority Reconstruction Program and Looking Ahead Toward Sustainable Economic Development,The World Bank, WashingtonDC. May 1999. World Bank (1999a) Povertyand SocialDevelopments inPeru, 1994-97. The World Bank 1999. World Bank (1999b) PovertyandInequalityinBiH: The Legacy of War. Mimeo. World Bank (1999~)Bosnia and Herzegovina. Consultations with the Poor. A Study to inform the World Development Report 2000/1 on Poverty and Development. Poverty group, PREM. World Bank, WashingtonDC. World Bank (2000a) The World Bank Country Assistance Strategy (CAS) of the World Bank Group for Bosnia andHerzegovina,Report No. IDA/R2000-0060dated, dated May 18, 2000, World Bank (2000b) Implementation Completion Report on an IDA credit to Bosnia and Herzegovina for a War Victim Rehabilitation Project. Prepared by Nedim Jaganjac. The World Bank, WashingtonDC. June. World Bank (2000~)Group's World Business Environment Survey, The World Bank, Washington DC, Spring2000 World Bank (2000d) Making Transition Workfor Everyone: Poverty and Inequality in Europe and Central Asia. The World Bank, WashingtonDC. World Bank (2001a) Bosnia and Herzegovina: Diagnostic Surveys of Corruption. The World Bank, WashingtonDC. 131 World Bank (2001b) Croatia: Economic Vulnerability and Social Welfare Study. The World Bank, WashingtonDC. World Bank(2001~)Kosovo Poverty Assessment The World Bank,WashingtonDC. World Bank (2002a) Bosnia and Herzegovina: From Aid Dependency to Fiscal Self-Reliance. A Public Expenditure and Institutional Review The World Bank,WashingtonDC. World Bank (2002b) Bulgaria Poverty Assessment. The World Bank, WashingtonDC. World Bank and IBHI (2002~)Qualitative Survey "Gender and Poverty". Sarajevo, June. World Bank (2002d) Local Level Institutions and Social Capital Study: Finding and Recommendations. The World Bank, WashingtonDC, June. World Bank (2002e) BackgroundPaper on Education for the Public Expenditure and Institutional Review for Bosnia and Herzegovina. Preparedby NickBumnet and Peter Darvas. The World Bank, WashingtonDC. January. World Bank (20020 LaborMarketinthe Postwar Bosnia andHerzegovina: How to increase worker mobility andjob creation? Preparedby MilanVodopivec. The World Bank, WashingtonDC, May. World Bank (2002g)CAS ProgressReport. The World Bank No. 24978, October 15,2002. World Bank (2002j) Project Concept Document: Social Insurance Technical Assistance Credit. Preparedby Betty Hanan. The World Bank, Washington,DC. World Bank (2003a) Albania: Poverty Assessment.The World Bank, WashingtonDC. World Bank (2003b) WorldDevelopment Indicators. The World Bank,WashingtonDC. World Bank (2003~) Program Document of the IDA Second Social Sector Adjustment Credit to BosniaandHerzegovina.Report No. P7536-BIH.The World Bank.WashingtonDC. 132