Report No. 40723-UZ Republic of Uzbekistan Living Standards Assessment Update August 27, 2007 Human Development Sector Unit Central Asia Country Unit Europe and Central Asia Region FOR OFFICIAL USE ONLY Document of the World Bank This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. FOR OFFICIAL USEONLY ACKNOWLEDGEMENTS This report was prepared by a team led by Cem Mete (Senior Economist, ECSHD) and comprising Kinnon Scott (main author of chapter 1, Senior Economist, DEC), Jan J. Rutkowski (main author of chapters 2 and 3, Lead Economist, ECSHD), Dina N. Abu-Ghaida (main author o f chapter 4, Senior Economist, ECSHD), Stefania Rodica Cnobloch (co-author of chapter 5, Consultant, ECSHD), Lucian Pop (main author of chapter 6, Senior Social Development Specialist, ECSSD), Dilnara Isamiddinova (Country Officer, ECCUZ), Eskender Trushin (Economist, ECSPE) and Flora Salikhova (Operations Officer, ECSHD). Mrs. Olga Nemirovskaya provided valuable inputs into the social protection chapter by providing a summary of recent changes in the social protection system and also by carrying out qualitative fieldwork to document the delivery arrangements for social assistance and child protection cash benefits. Peer reviewers for the report were Ruslan Yemtsov (Senior Economist, ECSPE), Martin Raiser (Adviser, ECSPE) and Akiko Maeda (Sector Manager, MNSHD). The team is grateful to Martin Raiser, Arup Banerji, Maureen A. Mclaughlin, Annette Dixon, Dennis de Tray, Lilia Burunciuc, Andriy Storozhuk, Michael Mills, Ernest0 Cuadra and John Langenbrunner for their support and advise at various stages of the project. Invaluable assistance for the analytic work was provided in Uzbekistan especially by the Ministry of Economy, Ministry of Labor and Social Protection, Ministry of Public Education, Ministry of Health, State Statistics Committee; Center for Social and Economic Research and the Center for Economic Research. As part of this task, additional work was also carried out by local researchers on a variety of development topics including "Qualitative aspects of access to tertiary education and impact of education quality on welfare improvement" by E.Z. Imamov and N. Khusavova; "Development of Social Welfare Index and Identification of Regional Differences in Socio-Economic Development in Uzbekistan" by Ms. Karimova; "Increasing Labor Force Participation in Uzbekistan" by Lapygin V.E.; "Macroeconomic, Sectoral and Gender Aspects of Personal Income Growth" by Chepel S.; "Labor Migration as an Alternative to Poverty and Unemployment" by L. Maksakova; "Liberalization of Land Relations in Agriculture" by U.M.Muminov; "Tax aspects of increased access to higher education in Uzbekistan" by Rasulev Sh. B.; "Study of the quality of health services in the Republic of Uzbekistan in the context of public welfare improvement" byN.D.Sharipova. The final views expressedinthis report arethose ofthe World Bank team. This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not be otherwise disclosed without World Bank authorization. 1 Table of Contents EXECUTI?TSUMWRY 4 CHAPTER1: Profile of thePoor 13 Introduction 13 Welfare Levels inUzbekistan2000-2003 14 Data Sources and Methods 14 Living Standards 15 Sensitivity Analysis 17 Economic Growth 19 Characteristicsof the Populationby Welfare Level 21 Household demographics 21 Correlates of Household Consumption 23 Areas of Change 25 Annex 1: MeasuringPoverty 29 Annex 2: Issuesof Comparability 32 Annex 3: Probabilityof SchoolEnrollmentamong 15-18 year olds 34 Annex 4: Data sources 38 Chapter 2: Labor Markets andpoverty 39 Main Findings 39 Part 1: Labor Market Outcomesand Poverty 40 Macroeconomic Conditions and Labor Market Outcomes 40 Labor Market in Uzbekistan-Key Features 42 Labor Market Status and Consumption 53 PART 2: The investmentclimate and job creation 60 Box I Thecosts of doing business in Uzbekistanhave beensignificantly lowered in recentyears but there is roomfor improvement 61 CHAPTER3: Education and Poverty 71 Main Findings 71 EducationSystemand Demographic Trends 72 Pre-schooling 73 Basic and Upper SecondaryEducation 75 Enrolment and Attendance 75 Quality ofEducation 79 Uzbekistanandthe MillenniumDevelopmentGoals 84 Higher Education 85 Annex 1. EducationSystem 91 Annex 2. Enrolmentand AttendanceRates 92 Annex 3. Quality of Education 95 2 CHAPTER 4: Health and Poverty 96 Main findings 96 The healthsystem 97 A comparativelookat healthindicatorsand healthcare spending inUzbekistan 98 Health,povertyand inequality 100 Children'sHealth outcomes 100 Disabilities 102 Health information, behavior andprevention 103 Access to and utilization of health care services 104 Seekinghealth care 105 Out-of-pocket servicepayments -formal and informal 108 Annex 1.Child healthindicatorsinselected countries 111 Annex 2. Females' decisionson healthcare, results from a Probit model 112 CHAPTER 5: Socialprotection and thepoor 114 Main Findings 114 Overview of the Social ProtectionSystem 115 Effectiveness of Social ProtectionPrograms 118 Review of key socialprotection issues 127 Annex 1.Statisticson the coverage, targetingand adequacy of social protectionprograms-129 Annex 2. Overviewof the SocialProtectionSystem 147 3 EXECUTIVESUMMARY The purpose of this Living Standards Assessment is to update and expand the first Living Standards Assessment report that documented the poverty situation in 2000/2001, This report is a product of the broader Programmatic Poverty Assessment work, which aims to support the development of the Government's Welfare Improvement Strategy (WIS), to inform the operations of the Bank and other international organizations, to provide inputs into the design of the Household Budget Survey (in collaborationwith the Asian DevelopmentBank) and to build capacity inthe area of poverty analysis (in collaborationwith the UNDP and UNICEF). These objectivesare consistent with the UzbekistanInterim Strategy Note. They also support the existing Bank operations in Uzbekistan, through in-depth analysis of the linkagesbetweenpoverty, employment, education and healthoutcomes and by identifyingthe most vulnerablegroups that needto be targetedby the socialprotectionprograms. Key FindingsandPolicyImplications Poverty trends over time 1. Poverty fell between 2000-01 and 2003 in Uzbekistan, but extreme poverty remained unchanged (Table 1). In fact, there was an alarmingincrease in extreme poverty in several regions. This is a major concern because the extreme poverty line used here is that amount of money needed for a very sparse 1500 calories. This level of consumption is not sustainable over time and can lead to, at worst, serious health problems and, at best, decreased labor productivity. Also, by having slipped further from the poverty line, it may well be harder for the householdsto helpthemselves move out ofpovertyand also for the governmentto assist them. Table 1: Head Count (Poverty Rates), 2000-2003 Extreme Poverty (0.01) (0.70) (0.53) (0.66) Source HouseholdBudget Surveys, 2000-01,2002,2003 Calculationsby authors Note: percent of the population, standard errors inparentheses 2. Of all the characteristics associated with low levels of welfare, the regional location of the household and the education of the head of the household are key. Why and how regional disparities affect welfare levels is worth further investigation. Specifically, additional work is neededto understand the change in circumstances of the populationinthe Mirzachul region of the country: and understanding how the Southern regionmanagedto decrease povertywould also merit further study. Economic growth has had a modest impact onpoverty reduction 3. The growthelasticity in Uzbekistanis quite low relativeto other countries: aone percent increase (decrease) in meanconsumptionleadsto only a one percent decline,(increase) in poverty. 4 Pro-poor growth can be achievedthrough a combination of cross-sectoral reforms 4. Inorder to makepoverty reductionmoreresponsiveto economic growth inthe short run, two sets of reforms can be considered. First, there is substantial roomto improvethe efficiency and flexibility of the labor markets. Cross-country comparisons reveal that Uzbekistan can especially benefit from (further) reducing administrative barriers, moving towards a competitive financial system, tackling governance challenges and improving infrastructure. Second, the coverage and (especially) targeting performance of the social protection programs can be improved. In the long term continued public investments in education and health would be needed to ensure that the poor contribute to and benefit from economic growth. Selectedfindings andpolicy recommendationsonthese topics follow. Poverty,labor market outcomesandjob creation Economic growth has not been suflcient to absorbpopulation growth, leading to labor market slack 5. The population of working age has grown considerably faster than formal sector employment since the mid 1990s. This excess labor supply was partially absorbed by the informal sector, alongside labor force withdrawal. 6. The Labor force participationrate is especially low among older men: only 47 percent of men aged 55-64 are employed in Uzbekistan, compared with 62 percent in the OECD. Possible reasons for this includelow retirementage, lowerskills ofolder workers and poor healthstatus of older individuals. 7. The Labor intensity of growth is relatively low (Figure 1). A 1 percent increase in GDP is associated with only 0.37 percent increase in employment. Unemployment and underemployment is particularlyhighamongless educatedworkers. Figure 1: PopulationGrowthand SelectedEconomicIndicators Dynamics of key macroeconomic indicaton 1995.100 1885 1998 1887 1898 1898 2000 2001 2002 2003 2004 /-Population ofworking age- - . Employment- -- -GDP Labor PraduclivltY--Wage81 Source:Ministry of Macroeconomicsand Statistics; Ministry of Labor andSocialProtection; and IMF; Bank staff calculations. 5 8. Low open unemployment masks significant labor market problems. Many existingjobs are of low productivityand do not protectworkers from poverty. Laborforce participationis low, largelydue to the lack of productivejob opportunities. Underemployment is pervasive, especially in rural areas, as evidencedby short workinghours. 9. The informalsector looms large, drivenby hightaxes on labor. Although it does providejobs, it often seems to be an employer of last resort offering only subsistence income. All these features contributeto poverty amongthe working age population, includingthose withjobs. Creation orjobs would benefitporn inpastructure investments, improvedgovernance, less administrative barriers and a competitivefinancial system 10. Large gains in laborproductivityare a positivephenomenonand are characteristics ofvirtually all transitioneconomies. However, the low paceof new firm creation is worrying giventhe widespread firm downsizing. The costs of doing business in Uzbekistan have been significantly lowered in recent years but there is roomfor improvement. 11. Poverty reductionin Uzbekistan requires improvingthe labor market outcomes: a faster pace of job creation and stronger sustainable productivity growth. There needs to be more jobs and the jobs should be better. They need to be more productive and thus offer higher wages. This necessitates improvements in the investment climate so as to encourage new firms to enter the market and existing firms to invest and expand. In order to increase formal sector employment, the most bindingconstraints need to be removed. Analysis of Business Environmentaland Enterprise Performance Survey (BEEPS) data indicate significant payoffs to investing in infrastructure, improving governance, removing administrativebarriersand developingan efficientand competitivefinancial system inUzbekistan. Educationand poverty 12. Public investments in education have been traditionally high in Uzbekistan. As a result, compared to other countries with similar GDP per capita, Uzbekistan ranks favorably with nearly universal school enrollment rates for the 7 to 15 age group. An explicit recognition of the linkages between poverty and education is requiredfor identifyingand addressingthe remainingchallenges in the education sector. Expansion of pre-schooling education would benefit rural residents and thepoor 13. At less than 20 percent of 3 to 6 year-olds, pre-schoolenrolment rates, although on the rise again since the late nineties, are nowhere near the highs of close to 40 percent observed in the late eighties and early nineties. Enrolmentrates are markedly higher in urban than in rural areas and for the better-off, with rates for therichest quintileat 3-4 times fhose for the poorestquintile. Basic and upper secondary education enrollments are high but school attendance is vulnerable to agriculture work and cold weather 14. Near universal enrolment rates are observed for the 7-15 year-old age group, with comparable overall male and female as well as rural and urban enrolment rates, and no clear variation across welfare quintiles. There is a five percentagepointdifferencebetweenenrolment and attendancerates. 6 15. Enrolmentrates for 16-18 year-olds exhibit an upwardtrend (at 67 percent in 2005, close to the pre-transitionlevel) with a two percentage point gender gap in favor of males and increasing rates by welfare quintile. 16. Survey findings confirm anecdotal evidence on students, particularly those aged 14-18 years, having to miss two months of study in the fall for purposes of agricultural work (presumably picking cotton). Children's school attendance rates can be improved by reducing the prevalence of seasonal agriculturework andthroughimprovingschool infrastructure(especiallyheatingin winter) The reduction of the resourcegap between urban andrural schools requires targeted interventions 17. Schools with students from more disadvantagedbackgrounds and schools in rural areas are more likely to enroll above capacity and have teachers with lower qualificationswho are further less likely to receive training. Geographic remoteness is particularlydetrimental, so that schools in rural communities that are remotefrom boththe rayonand oblast centers fare worst. Only 38 percent of schools in these communities have sufficient numbers of rooms or desks, as compared to roughly 80 percent of urban schools. The resource gap between urban and rural schools can be reduced by adopting capitation(per pupil) financingmodels as describedby the 2006 Uzbekistan ProgrammaticPublicExpenditureReview. Access to tertiary education is currently very limited: supply and demandside interventions are needed to provide higher education opportunities to childrenfrom middle and lower income households. 18. At 5 to 7 percent, enrolment rates in higher education for 19-24 year-olds are low and demonstrate wide variation by gender and welfare quintile, so that higher education today is the prerogative o fthe well-to-doin Uzbekistan (contrary to historicalachievements). 19. Higher education is associated with a roughly 60 percent increase in earnings relative to the attainment of basic education. As the transitionto market economy progresses, there is reasonto expect even larger returns to higher education, as seen in other transition countries. As a result if higher education institutionscontinue to serve exclusivelyto the elite, social mobility across generations will be very limited. 20. Given the substantial returns to higher education, the key reasons for low higher education enrolments emerge as supply-side rationing of available "seats" at this level of education and also the highout-of-pocketcosts of universityattendance. The out-of-pocketcosts for universitylevelaveragesix times the out-of-pocket expenditures for general secondary education, with university tuition alone accountingfor one-thirdoftotal costs. 21. Roughlyforty percent of universitystudents benefit from scholarships and subsidies, but boththe beneficiaryrate for as well as the average amount of scholarship received by the two richest quintiles are on the order of twice the rates and amounts for the three bottom quintiles. Thus, scholarships and subsidies for university students were found to be starkly regressive in allocation. This represents a missedopportunityto alleviatethe financialburdenofhigher education for the disadvantagedand reverse current negativetrends in enrolment. 7 Figure2: Benefitincidenceof scholarshipsandsubsidies, 2004 g 100,-. .. . . . . . ...... -.,.. expenaiture --A--Grades 59 schoaarshlps -.x- . Uniwnitv 0 20 40 60 80 100 scholarships 1 22. Increasing higher education enrollments with demand side interventions (such as tax brakes for higher education expenditures and scalingup programs that offer grants/scholarships to poor students to finance their tertiary education) alongside supply side reforms (staffing, infrastructure and curriculum reforms to increase availability of higher education beyond a small share of population) taking into account the equityand quality aspects ofthe full cycle of education. Health and poverty 23. Healthinessis not only an indicator of living standards in itself, but it is also a key determinant of education and productivity of individuals. In order to achieve the most favorable population health outcomes under severe resource constraints, there is a need to pay attention to the health status and characteristics of the poor. The timing and targeting of policy interventions should be adjusted for different healthconditions, however. Subjective measurements of health conditions should not be used to evaluateprogress towards the MDGs and oficial statistics should be usedwith caution. 24. The poor and the uneducated are less healthy. However, this relationship is visible only if one uses objective measurements of health conditions such as anemia and stunting. The self-reported health indicators typically included in household surveys produce misleading insights if they are used to establishlinkages with poverty. Thus, they shouldnot be usedto evaluateprogresstowards the MDGs. 25. Analysis of regional survey data revealthat almost 12 percent of all individuals ages 7 and older have at least one serious difficulty or a full limitation (in seeing, hearing, movement, learning, communicationor self-care). But official disability status is grantedonly to 3.8 percent of the population in the regions under investigation(national average figure was 3.2 percent in 2003). The proxy used to capture disability status matters for the analysis of living standards. While official disability status of childrenis strongly associatedwith lower school enrollmentrates among childrenages 7 to 14 as well as with lower labor force participation rates among adults (smaller effects observed for reporting a full- limitation), the indicator that is most closely associatedwith poverty is reportingo f a full limitation(and not the official disability status, presumably in part because of the differences in the probabilityof being eligibleto receive disabilitypensions). 8 Pro-poor targeting at the householdlevel is needed to reduce theprevailing stunting rates 26. The health-socioeconomicstatus gradient emerges early amongchildren in Uzbekistanifone focuses on stunting, a long-termindicatorof malnutrition. Although, the wealth effecton anemia is larger for the older children (ages 3 to 6). The differences in health outcomes of children within households by their gender or birth order are negligible in most cases. In contrast, regional targeting holds promise for reducingthe prevalenceof anemia, both becauseof high anemia rates and because of the regionalnature of some interventionssuch as flour fortification (althoughmore elaborate targeting could be considered for the distributionof ironpills). Deteriorating health care infiastructure in rural areas and costs of medicines emerge as main barriers against the health care utilization of thepoor 27. Access to health facilities is not the main obstacle for the poor, but health infrastructure has deteriorated significantly inruralareas (wherethe poor are more likely to reside) and the poor are unable to afford the costs of medicines even if they do consult with a health service provider in the case of sickness. A first step in tacklingthese issues -which are essentially byproducts of existingpublic and privatehealthfinancing arrangements-would be re-examiningpublic resource allocationarrangements taking into account size and characteristics of population,their health care needsetc. (as describedby the 2006 Programmatic Public ExpenditureReview); and also by,better documenting the prevalence of out- of-pocketpaymentsfor healthcare. Thepoor and the uneducatedwould benefit the mostfiom health information campaigns 28. The poor andthe uneducated are also significantly disadvantagedwhen it comes to knowledge about diseases, for example TB and AIDS. Furthermore, poor females are much less likely to have a say in their own health care. These differences in access to and possession of health related information by schoolingand wealth of individuals suggest the needto target information campaigns to the uneducated andto the poorestsegments ofthe populationifthey are to be effective. 9 SocialProtectionand the Poor 29. Previous analyses of the Uzbekistanmeans-testedbenefits (benefitsfor low income families and benefits for families with childrenunder 16 years old) praisedthe socialassistance schemes implemented by Mahallasas being "a clear break with the past when benefits were often administered by state-owned enterprises, with entitlement either being universalor linked to households' formal cash income alone". Indeed, the introductionof means-testschemes and delegating their implementationto local communities represented a significant improvement of the social assistance system in Uzbekistan. However, this Poverty Assessment, as well as the previous one (2003), signal that the performance of the social protectionsystem is affected by errors of inclusionandexclusion, as well as by horizontalinequity,which leadto lower targetingperformance ofthe Uzbekistan'ssocialassistance benefitsas compared with other countries in the region. 30. The slow pace of transitionto market economy in Uzbekistanmay have spared the country from an even larger decline in GDP per capita during the early years of transition (as indicatedby the Bank's first Living Standards Assessment), but the absence of a functioning market economy remains a major obstacle for economic development inthe longterm. Learningfrom the early reformexperiences of other countries in the region, it is importantto ensure that some of the negative side-effects of reforms in the short term are absorbed by a well functioning social protection system. In this context, the weak coverage, adequacy and targeting performance of unemployment benefits should be addressed by redesigning the relevant programs so that they offer adequate protection against temporary loss of income. The coverage of social protectionprograms are modest, although thepension benefits offer protection to thepoor 3 1. Social protectionspending as percent of GDP decreased in recent years. This drop in spending is the result of slower growth of social protection expenditures relative to GDP growth, which led to a relativelysignificantreductionof benefits generosity relativeto the averagewage. 32. In2003 (the most recent year for which HBS data was available) about 57 percent of the total population,respectively62 percent of the poorest quintile, was covered by at least one social protection program. Despite a decrease in the generosity relativeto the average wage, the pension benefits offer a significant protection to the poor, by covering almost half of the populationin the poorest quintile and morethan 40 percentofthe poor beneficiaries'consumption. As thepopulation ages and economic reforms are implemented, social protectionprograms will need to adjust to satisfi the emerging needs 33. Some population groups did not benefit much from economic growth and they started to rely more on social protectionprograms instead: especially the elderly, those who reside in rural areas and in the poorer regions of the country. As time progresses, the employment and social protection programs will needto focus more on the needsofthe elderly. The overall improvementsinthe functioningof labor markets - such as the reduction of taxes on labor which would serve to discourage informal sector employment that is outside the Government's tax base -should aid the elderly, since at this time they are the onesthat are more likelyto be left out of employment. 34. The unemployment benefit reaches only about 0.4 percent of the unemployed looking for a job. Its incidence on registered unemployed is low as well, with only about 5 percent of this group receiving 10 unemployment benefits. Due to the low level of benefits, the adequacy of the unemployment benefit is also very low, at about 5 percent of the beneficiary's consumption, meaning that it does not succeed to offer sufficient protectionagainst temporary loss of income. Targetingperformance of social protectionprograms can be improved by allocatingpublic funds across regions by using a reliableproxy poverty criteria: and also by establishing a moreprecise methodology to evaluate householdwellbeing. 35. Both coverage and targeting of the Uzbek social assistance programs are rather modest when compared to other countries in the region. If only the means-tested programs are compared (all non- contributory programs based on income-, means-, or proxy means-test mechanisms), ignoring other programsthat formthe mix of social assistance, Uzbekistanperforms relativelywell on coveragebut poor on targeting. For example the low income benefittransfers more than halfof the funds to the middle and upper quintiles. The establishment of a more precise methodology to evaluate household wellbeing, taking into account the differences in oblasdregion level prices, would be useful to reduce the arbitrariness at the local level. Furthermore, distribution of social protection funds across regions and communities can be improvedthrough the usageof a reliableproxypovertycriteria. Figure4: Coverageandtargeting of social protectionprograms Performance of non-contributorybenefits,includingprivileges 50 -.-T z x?1 40 P 0 g 30 '0 g cr .-m I 20 4.4 P 4.4 10- I 1 I 1 0 20 40 60 EO 100 coverage of the poorest qulntlle (%) ECA Targeting Peflormance. WorldBank 36. The targeting performance of means-tested transfers is low. The means-tested programs are affected by significant inclusionand exclusionerrors. By design, too much space is given for discretion and arbitrariness at local level through the lack of more precise guidelines/methodology regardingthe evaluation of household wellbeing and imputation of incomes. Combined with the absence of (proxy) poverty criteria for the distribution of funds between regions and communities, this leads further to 11 horizontal inequity: wealthy regions/communities can receive as much, or even more, funds as poor regions/communities, and, as a consequencethe poor will not receivesimilar treatment in all regions. 37. The introduction of the cash compensation mechanism for utilities using the eligibility criteria inheritedfrom privilegesdoes not offer protectionto low incomegroups. Timely collection, distribution, analysis and discussionof data are requiredfor informed policy making 38. The successful implementation of the public policies that aim to improve living standards requires timely collection, distribution, analysis and discussion of data. At this time, Household Budget Survey data that are collected by the State Statistics Committee are made available to very few institutionswithin or outside the Government -often with substantial delay, if access is granted at all. Other household survey data, such as the HealthExaminationSurveys, contain valuable informationbut they are not implemented regularly because of one-time funding arrangements by international organizations. 12 CHAPTER1: PROFILEOFTHE POOR INTRODUCTION In 2004 the Government of Uzbekistandeveloped an (interim) Welfare ImprovementStrategy Paper (I- WISP) for the 2005-2008 time period. The I-WISP lays out the Government's objectives for poverty reductionand embeds these in a broader macro-economic framework. It summarizes for the first time the various reform initiatives already underway and offers a comprehensive assessment of the development challenges facing Uzbekistan.It also contains a matrix of target development outcomes, with a first effort to define measurablemonitoring indicators. In2005 the Government will embark on the preparationof a full WISP. This effort will require further prioritization of objectives, integrationwith the budget and additional refinement and development of monitoring indictors. The Living Standards Assessment (LSA), carried out in 2003 provided a good starting point for the design of policies and much of the analysis from that study has been incorporatedinto the I-WISP. Additional analytic work that can feed into the WISP has been undertakenby UNDP', the Asian DevelopmentBank' and USAID.3 To ensure that the upcomingWISP is as relevant and accurate as possible, analysis of new household level data is warranted. The purpose of the present document is to updatethe basic profile of welfare that was done usingthe first round of the HouseholdBudget Survey in 2000-01 with data from the subsequent two rounds. Careful attention has been paid to ensure the comparability of the analysis of the three years of data. The data themselves are largely comparable: the sampling methods have remained the same, as have the questionnaires and the basic implementation of the survey. However, some changes have occurred, primarily due to improved field work: these have been taken into account in the present analysis. The chapter is divided into three parts. An overview of the welfare levels in Uzbekistan from 2000-01 through 2003 is presented in section one, along with a discussion about the data sources and their strengths and weaknesses for multi-year analysis, and an analysis of the relationshipbetween economic growth and welfare. In section two the basic indicators of living standards over the time period are presented. Section three contains informationon two topics of interest for short-termand medium-term welfare: labor andeducation. IUNDP. M~IIcmum DevelopmentGoalsEspsnencc.2004 1 Paolo Vcm. "UzbelustanCountryProfileAnalysis". 2005 3 The Demographicand HealthS w e y fundedby USAID in 2004 prowdes iubrtanltal new dam for snammngt68ucs of healthand fenilit). 13 WELFARE LEVELSUZBEKISTAN2000-2003 IN Data Sources and Methods In 2000, the State Committee on Statistics (SCS) of Uzbekistan began a new round of what had previously been called the Family Budget Survey. In its new incarnation as the Household Budget Survey (HBS), the sample design moved from quota to probability sampling with a 10,000 household sample distributedthroughout the country and throughout a 12-monthperiod. The data collectionfor the 2000-01 HBS with the new sample and an improvedquestionnaire was carried out from April of 2000 until Marchof 2001. Since the 2000-01 H B S survey, five other rounds of the HBS have beencarriedout, annually from 2002 to 2006 when data were collected from January through December. Only the first three roundsofthe datawere available for the presentanalysis: 2000-01,2002 and 2003. The methodology employedto measure welfare levels is the same as that used inthe previous analysis of the H B S 2000-01 for the LSA. A full review of the analytical technique used and detailed findings from the analysis can be found in the Living Standards Assessment of 2003 (LSA)4, The basic descriptionof the methods is reproduced from that document in Annex 1. Even though this approach, the Lanjouw- Lanjouw method, is not that standard it is chosen here both to be consistent with previous work and because of the favorable results from the validity analysis reported by the 2003 LSA. Essentially, total householdconsumptionof food (be it purchased, received as gifts or produced by the household itself)is the welfare measure. The per capita value of total household consumption of food is compared to the value ofa food poverty line. The food povertyline is the amount of soums neededto obtain2100 calories per adult per day duringthe year. The constructionof the food poverty line takes into account boththe consumptionpatternsofthe population(specificallythose whose food consumptionranks inthe secondto third deciles ofthe population)as well as the pricesfaced by this populationgroup. The first food poverty line was constructed in October 2000 prices. To carry out the analysis here, the value ofthe food poverty line needs to be adjusted for inflationfrom 2000 to 2002 and 2003. To do this one couldeither usethe consumer price index (CPI) or construct a new CPIusingthe prices of food items directly from the Household Budget Surveys. Ifthe CPIwas sufficiently detailed(Le. providingfigures for each month, regionand consumer-goodtype) and reliable,one would expect to obtain similar poverty estimates regardlessofthe procedureusedto take into account variations inthe prices of goods over time. However, the available CPI informationis not detailed enough andthere are questions about reliability of the figure^.^ Not surprisingly, following this route leads to substantially different (and implausible) povertyfigures. To avoidthe problems of usingthe CPI, the household survey data were used to construct the new value of the poverty line. The potential drawback of usingthe H B S prices for 2002 and 2003 to update the value ofthe povertyline is that, from the data, it appearsthat consumptionhas changedinthe three years. Thus, there are not prices for all of the goods includedin the 2000-01 poverty line. Typically, to adjust the poverty line for inflation, the quantities of every food item includedin the poverty line from 2000-01 are revalued usingthe 2002 (2003) prices found in the HBS of 2002 (2003). The problemthat arises in this case is that several goods consumed in 2000-01 are not being consumed in 2003 (among the second and third deciles of the populationwhich is the base group). Thus the poverty line of 2000-01 cannot be revalued exactly. To handle this problem, a new poverty line was created to estimate inflationusingthe quantities of 2000-01 for food items consumedjointly in 2002,2003 and2000-01. 4 World Bank (2W3) UzbekistanLimg Standards Assessment Policies 10 lmprove Living Standards RepanNo.25923, Waslungton D.C. 5 There are wa CPI indices ulat one can use the official CPI and ulalcalculated by the IMFbasedan quanerly regional SNdlCE The wo indicesme nota1all alike r f Octabsr 2WO is the base (hdex=IM)) h e officlal CPI csUm1cs the mdex 81 169 93 m Oetabsr 2003 while he IMFindex is 21 148. 14 Some concerns about usingthis method exist. If the observed changes in consumption are a result of increased poverty levels, this method of adjusting for inflation would not capture this welfare decline. However, in the processof carryingout the analysis in this document, a detailed reviewof the changes in consumption was undertaken. It was found that many of the `changes' in consumption are actually the result of improved survey techniques: fewer aggregated categories of food goods are reported by householdsandthe more detailedcategoriesof foodconsumption prevail.6 Inshort, there does not seem to have been a shift in consumption: the 2002 and 2003, data appear simply to better reflect the consumption of households in the country. Thus, using the HBS data to construct a CPI for 2002 and 2003 shouldprovide accurateresults. An additionalfactor which helpsto reinforcereliability of usingthe H B S for this purpose is that all of the items that were purchased or consumed in 2000-01 that do not appear in 2002 or 2003 represent a very small fractionoftotal items purchasedor consumed, a very small share of total householdexpenditures (in the case of purchased items) and a very small physicalamount. Forthese reasonswe are reasonablyconfidentthat the changes in consumption do not adversely affectthe use of the HBS 2002 and 2003 prices to update the poverty line. For the interested reader, a brief descriptionofthe changes that were found inthe consumptiondata can be found inAnnex 2. In order to carry out the analysis of the three years of data, the original data from 2000-01 was re- processed. There were several reasonsfor this re-processing. First, the sample of households used inthe original analysis included those households with some incomplete data. While the consumption data appeared to be complete, the SCS recommendedhighly that these households be dropped to ensure that no data problems existed. This led to the use of 9647 households in the revised version, compared to 9,709 households in the original estimates. The sample weights were adjusted accordingly. Second, becausethe first year was largelyan experiment intestingthe new methodology and questionnaires, there were some (fairly small giventhe scope of the survey) problems at the data collectionphase. There were more obvious errors in the data in 2000-01 than in the two subsequent roundsof the survey that hadto be resolved. For all years some data cleaning (resolving problems of outliers, missingdata and incorrect data) was required and a standardized approach across the three surveys was implemented, that varied somewhatfromthat used inthe originalanalysis. The result of the requiredrevisionto the analysis of the 2000-01 HI3S led to changes in the level of the welfare indicatorscomparedto those published inthe LSA (3 1.5percent povertycomparedwith 27.5). It did not, however, change the relativewelfare among various groups andor areas ofthe country.' While there can be some discussion on the absolute levels of welfare in 2000-01, the benefit of having re- processedthe 2000-01data is that we canbe certain ofthe comparabilityofthe data inthe threeyears and thus are able to focus on assessing changes inwelfare over time. Living Standards Usingthe comparable poverty figures (2000-01 revised, 2002 and 2003), it can be seen that the overall headcount rate (percentage of the populationwith per capita food consumption below the food poverty line) fell between 2000-01and 2002 but did not change, in a statistically significantway, between 2002 and 2003. In contrast, extreme poverty, based on a food line allowing for only 1500 calories-- an unsustainable levelof consumption-- has not changed inthe three and a halfyear period. The results are shown in Table 1. 6 The SCS r ~ c a e dB delailed document analyzingthe changesin consumptionand conoboraledthe findingthal the bulkof the changes were due Io improvedfield work of the HBS. 7 For example.the onginal showed the head counlrate 10 be hghcr in mllhanurbanareas and highenm the Southernreglo& this IS me of the rwisedversionBIwell 15 Table 1: HeadCount (Poverty Rates), 2000-2003 ExtremePoverty (0.01) (0.70) (0.53) (0.66) Source HouseholdBudget Surveys, 2000-01,2002,2003 Calculations by authors Note: percentof the population, standard errors in parentheses * This column contains the original estimatesfor Food and ExtremePoverty in Uzbekistan from the LSA 2003. Notethat these numbersare not strictly comparable to the 2002 and 2003 figures. For comparisonsover time, the 2000-01 revised figures should be used. The gap in welfare levels betweenthe urbanand rural areas ofthe countryworsened slightly in the period under study (see Table 2). Regionally,the gaps have closed, the differenceinthe headcount rate between the poorest andrichestregionhas gone from 27 percentagepointsin 2000-01 to 20.4 percentagepoints in 2003. What is most striking, however, is the reversal in rankingof the regions. In 2000-01 the Southern Economic Region had the highest head count rate, 47 percent. This fell dramatically by 2003 to 27.6 percent. At the same time, the Mirzachul region saw its headcount rise to 37.9 percent, making it the poorest region in the country. In the Mirzachulregion the extreme poverty rate also rose substantially while that inthe Southern regionfell by half. Inthe Northernregion, although overall poverty stayedthe same, extreme poverty rose every year. These shifts can be seen in Table 2 and Figure 1.' Future research can contribute to our understanding of the determinants of regional variations in poverty in Uzbekistan, by usingthe morerecentwaves ofthe Hl3S data. Table 2: HeadCount (Poverty Rate)by GeographicAreas 2000-2003 General Poverty Extreme Poverty Area Rural Urban Economic Region Tashkent Mirzachul Ferghana Northern Central Southern National Source HouseholdBudget Surveys,2000-01,2002,2003 Calculations by the authors *The shadedcolumns show figures from the original analysis and are not strictly comparableto 2002 and 2003 figures. For comparisonpurposes the revised2000-01 figures should be used 8 Noit ha1the vends over time are the same regardlessof whether the ongmal or mwwd figurer from 20uO.01 arc usedfor four oflhe SIX regions In Tarkeenl. Leong31ul tigurcs. campared10 2W2 and 2003, would show povsny mcrwmgandthen falling Usingthe revised2M)Odl figurcr showspveny slayngLhe m e andthen fallmg Inthe Cenwl region. the ongmal figures would show and increarmgtrendwhile the rnised figures show a dropand thenan m c mmpoveny 16 Figure 1: Changes inWelfare,2000-01 to 2003, GeneralPoverty andExtremePovertyHeadCountRates 50 45 40 35 30 25 EaPov Gem Pov. 20 15 10 5 0 01 02 03 01 02 03 01 02 03 01 02 03 01 02 03 01 02 03 Tashkent Mirzachul Ferghana Northern Central Southern Source HouseholdBudget Surveys, 2000-01,2002,2003 Calculations by authors SensitivityAnalysis The dramatic changes in the overall and the extreme head count rates may be the result of clusteringof the populationaroundthe poverty lines. In other words, ifmany people have per capita consumptionthat is close to the poverty line, small changes in their consumption can have large effects on the head count rate. One way to test this hypothesis is to look at how the headcount would change ifthe value of the poverty line itselfwere increasedor decreased. As can be seen inTable 3, ifthe value of the poverty line is increasedby 10 percent, in most cases the head count rate would change by substantiallymorethan 10 percent. This indicatesthat many householdsdo have consumptionlevelsright below andright abovethe poverty line, thus explaininghow notable shifts in the headcount ratescan occur inthe short run. Table 3: Sensitivityof HeadCountRatesto a 10% Changeinthe Value ofthe Poverty Line,2000-01to 2003 2000-01 2002 2003 Low High Low High Low High Tashkent -31.1 35.0 -21.3 23.4 -25.8 9.6 Mirzachul -25.5 30.3 -7.1 7.4 -7.4 5.4 Ferghana -31.3 26.4 -23.3 31.2 -30.7 26.5 Northern -30.2 26.9 -23.2 18.8 -10.4 12.1 Central -33.8 32.7 -25.3 27.7 -19.7 19.4 Southern -14.8 15.7 -36.4 33.4 -27.1 45.1 National -27.3 26.1 -25.3 26.7 -23.0 22.6 Source: HouseholdBudget Surveys,2000-01,2002,2003. Calculations by the authors. What is particularlyinterestinginthe table is what has happenedinthe Mirzachulregioninthe pastyears. In 2000-01, a 10 percent change in the poverty line, in either direction, would have changed the 17 headcount rate by 25 to 30 percent. In 2002 the poverty rate rose in the region (as shown in Table 2), specifically the extreme poverty rate. This decline in consumption shows up in 2002 and 2003 where a 10 percent change in the value o f the poverty line only affects the head count rate by 5 to 7 percent. In other words, the poor are poorer than previously, their consumption levels are further from the value o f the poverty line, and thus, moving the poverty line up or down makes less o f a difference to poverty rates. The same decline in the sensitivity o f the poverty rate occurs in the Northern region where extreme poverty has also risen. In contrast, in the Southern region the decline in extreme poverty means that consumption levels among the total pool o f the poor in that re ion have increased. So, the sensitivity to the poverty rate in the Southern region has increased over time. F The sensitivity o f the head count rate to the value o f the poverty line can also be seen in Figure 2. In 2000-01, the cumulative density functions for all o f the regions, with the exception o f the Southern one, are quite steep around the poverty line (in log terms the poverty line is 8.2 in 2000-01). Hence the high volatility o f head count rates if the value o f the line were to shift. Because the.cumulative density functions for the other regions cross at different levels o f consumption, the relative rankings o f these regions would also change if the poverty line were set higher or lower than that use here. This is even more apparent in the 2002 and 2003 periods. In short, the head count rates are fairly sensitive to the poverty line. Policies aimed at raising living standards will need to take this into account. Figure 2: CumulativeDensityFunctionsof LogPer Capita Consumption by Regionand Year Panel a: 2000-01 Source: HBS 2000-01,2002,2003. Author's calculations. Note: x-axis is the natural log of per capitaconsumption;the y-axis is the percent of the population 9 The inability lo cowtmcla fillconsumptionaggregate (nOlJuSlfaad) wing the HBS meanslhalalher m e w sensitive lo the distance of the poor from the pverty linec m o l k calculatedrushas lhe pverry gap and xvenly indica Nor can inequalitymeasuresbe canmaed The hmre rsvimnof the HBS will permit such m&cators 10 be c o n ~ t ~ ~ l e d 18 Panel b: 2002 Note: x-axis is the natural log ofper capita consumption; the y-axis is the percent ofthe population Panel c: 2003 Ferghana -- Note: x-axis is the naturallog of per capita consumption; the y-axis is the percent o fthe population. Economic Growth Economic growth is one of the key components of raising living standards in the country. In most countries there is a strong relationship between economic growth (as captured by overall consumption 19 growth) and poverty reduction. If one measures the change in the head count ratio (poverty) over the change in meanconsumption(growth), a one percentchange in consumption is typically associatedwith a two percent change inthe headcount ratio. To determine the impact of economic growth on the poverty rates Uzbekistan, the growth elasticitiesare constructed here. In order to allow for international comparisons, this exercise was carried out using a total consumption aggregate (that includes food and non-food) converted into PurchasingPower Parity (PPP) adjustments. While the problems with the total consumption aggregate remain (see Annex 1 for a discussion of this), the use of this variable in PPP dollars is the best approximationto the work done in other countries. Caution, of course, should be used when doing the comparison. The PPP conversion factors are from 2000 and the poverty line used to calculate the head count rate is the equivalent of two dollars a day, which producesa 47 percentpovertyrate. The growthelasticity inUzbekistanis quite low relativeto other countries. As can be seen in Table 4, for much of the country, a one percent increase(decrease) in mean consumption leads to only a one percent decline (increase) in poverty. The findings from the labor markets analysis, presented in chapter 2, suggestthat key reasons for this relativelylow growthelasticityare administrativebarriers, absenceof an efficient and competitive financial system, governance challenges and underdeveloped infrastructure, which in turn lead to small pace of new firm creation and defensive restructuringwith few new hires amongexistingfirms. Table 4: Elasticity of Poverty to Changes in Consumption Change in Change in Elasticity Consumption Head Count Tashkent 32.18 -46.83 -1.46 Mirzachul -15.77 20.68 -1.31 Ferghana 1.11 -1.21 -1.09 Northern -12.56 -0.35 0.03 Central 16.17 -15.93 -0.99 Southern 29.23 -23.47 -0.80 National 12.50 -12.89 -1.03 Source: ECAPOV, (forthcoming). Note: Uses atotal consumptionmeasure (food and non-food)that is converted into PurchasingPowerParity dollars for 2000. The national figure does hide some within country variations. In the Mirzachul region, the 16 percent decline in consumption led to a 21 percent increase in the head count. In contrast, in Tashkent a 32 percent increasein meanconsumption ledto a 47 percentdecrease inpoverty. In order to explainwhy the povertyelasticity of growth is relativelysmall, we also present data as decile ratios. The consumption of households that are in the very bottom of the consumption distribution (the first three deciles) and the consumption of those who are at the top of the distribution improvedmuch more than others. Since at around the poverty line there is not as much movement, the poverty elasticity of growth is relativelymodest. 20 Ratio in comparison with national average ratio 0.04 0.03 - i 0.02 - .- /--- 0.01 - 0.00 -0.01 1 2 3 4 5 6 7 8 Q 10 IL,\ I -0.02 '. -0.03 - r' -0.04 1--- - --A Decile CHARACTERISTICSOF THE POPULATION BY WELFARE LEVEL A comprehensive look at the characteristics of those living below the food poverty line compared to the rest of the population was carried out in the LSA using the 2000-01 H B S data. One would expect that many of these characteristics of households would not change in the short run (from 2001 to 2003). In this section we first recalculatethe basic tables from the 2000-01 survey using the 2002 and 2003 data sets to provide an update. The results confirm that there has been little change in most of the basic household characteristics. However, there are some aspects of households that could be expected to change inthe short run and that the basic tables here indicatedo change. For this reasonthe next part of the sectionlooks more closely at two areas where change has occurred inthe period2000 to 2003: labor market status and educational enrollment levels. The latter topic is especially important given the concerns with the results from the HBS 2000-01 and efforts to improve more recent data. The section ends with a discussion of the key correlates of poverty and how these have changed in the three-year period. Household demographics The compositionof households has not changed much in the 2000-2003 period, nor have the differences between those living below the poverty line and those above it. As is shown in Table 2, the poor have larger householdswith more childrenbut these represent a smaller share of total household membersthan among the non-poor in 2003. The dependency ratio, the number of householdmembers below and above working age divided by the number o f working age adults in the household, has fallen slightly but is marginallyhigher amongthe poor. Table 5: HouseholdComposition by Welfare Status andYear 2000-01 2002 2003 Poor Non-poor Poor Non-poor Poor Non-poor Household Size 7.6 6.0 7.2 5.9 7.2 5.7 No. of children 0-6 1.5 1.O 1.2 0.9 1.2 0.8 No. of children 7-17 2.1 1.7 2.1 1.6 2.0 1.6 No. of adults of working age 3.9 3.2 3.8 3.2 3.9 3.1 No. of pensionersaged 0.5 0.5 0.5 0.5 0.5 0.4 Dependencyratio 1.2 1.1 1.2 1.O 1.1 1.O Source: HouseholdBudgetSurveys, 2000-01,2002,2003. Calculationsby authors. 21 The characteristics of the head of household are fairly stable over time and do not vary much between groups with some exceptions. Interms of education, twice as many of the non-pooras the poor live in a household with the head having higher education: rates of completion of technikum are also higher among the non-poor although the gap appears to be diminishing. Persons living in poor households are more likely to have a headof householdwho is not inthe labor force or who is unemployed. Inthe three year periodunder study, the levels ofunemploymenthave fallen for bothgroups, although the poor suffer disproportionatelyfrom unemployment. Interestingly, the percent ofthe poor living in householdswhere the head of household is disabled" has increased substantially in the three years. The absolute share is low yet it is a disquietingtrend. Table 6: Characteristicsofthe Heado fHousehold Poor Non-Poor National 2000-01 2002 2003 2000-01 2002 2003 2000-01 2002 2003 Gender of Head Male 82.5 84.1 84.6 80.9 81.2 80.3 81.4 82.0 81.5 Female 17.5 15.9 15.4 19.1 18.8 19.7 18.6 18.0 18.5 Age of Household Head Under 30 yrs. of age 6.5 5.1 4.6 6.9 6.5 6.4 6.8 6.1 5.9 30-45 yrs. of age 38.9 41.2 41.9 43.3 41.1 42.9 41.9 41.1 42.6 46-60 yrs. of age 28.8 29.5 29.1 26.6 28.4 27.1 27.3 28.7 27.7 Over 60 years of age 25.8 24.3 24.4 23.2 24.1 23.6 24.0 24.2 23.9 Education of Head Zero through 4 years 10.3 8.4 5.8 7.6 4.8 4.7 8.5 5.7 5.0 Five to nine years 16.3 14.4 13.4 12.1 13.6 11.7 13.4 13.8 12.2 Secondary completed 53.4 52.3 56.4 45.0 47.6 47.8 47.6 48.9 50.2 Technikum Incomplete 2.6 1.7 2.3 3.0 1.6 1.8 2.9 1.6 1.9 Tecnikum Complete 9.3 13.7 13.3 15.4 16.2 16.7 13.5 18.5 15.8 Higher Education 8.1 9.6 8.7 16.9 16.3 17.4 14.2 14.5 15.0 Civil Status of Head Married 84.1 83.7 83.6 82.9 81.7 81.6 83.3 82.3 82.1 Single 0.5 0.5 0.4 0.8 0.7 0.7 0.7 0.7 0.6 Divorced or Widowed 15.5 15.7 16.0 16.3 17.5 17.7 16.0 17.1 17.3 Labor Force Status of Head Active 53.2 55.7 55.4 59.6 59.9 60.0 57.6 58.8 58.7 Unemployed 3.6 2.0 2.1 3.0 1.4 1.2 3.2 1.6 1.8 Inactive 46.8 44.3 44.6 40.4 40.1 40.0 42.4 41.2 41.2 Pensioner 68.0 65.2 67.3 69.2 69.I 68.2 68.8 68.0 68.0 Disabled 4.7 5.4 8.8 6.7 7.3 8.1 6.0 6.8 8.3 Other 27.3 29.4 23.9 24.1 23.6 23.7 25.2 25.2 23.7 Source: HouseholdBudget Surveys, 2000-01,2002,2003. Calculations by authors. Note: The table refersto the percentofthe populationliving in householdswhose headhas the indicated characteristic.The active populationare all those of working age (16-54 for women and 16-59 for men) who are employed or unemployed..Unemployedrefers to persons not presently working but looking for employment. 10 The definition ofdisabled !hat individualsusedhereIS not precise People were asked10 le11their mtusmterms of laboractivitieswth 'disabled' kingoneof Ihe Options I1IS notclear Ifa person who IS dmbled butwho also works would define himor herselfas Lsabled InIhe survey 22 Correlatesof Household Consumption To identify the key facets of households and their membersthat affect poverty, it is importantto look at the marginal effect of each characteristic. For example, what impact does additional schoolinghave on welfare levels holding all other characteristics (variables) constant? A linear regression model was run for eachyear to identifythe variables that are stronglyassociatedwith welfare andthe relativeimportance of each of these in explainingwelfare (consumption) levels. The dependent variable is the log of per capita annual consumption andthe explanatory variables consist of household demographic information (number of children of different age, number of pension age adults in the household), characteristics of the head of household (age, gender, marital status, education level, ethnicity)and geographicvariables. As can be seen in Table 7, greater household size is associatedwith decreased consumption levels as is the presenceof childrenaged 14-17 although the effect is not large. Only in 2003 does the presenceof a female headof household affect welfare and does so by increasing it, a slightly counter-intuitivefinding. It may indicatethat poorer women and their households have a greater chance of beingincorporated into other householdsthan male-headedhouseholds, in other words only those female-headedhouseholds that are able to support themselves remain as separate households, or female-headedhouseholdsmay receive more privateand publictransfers, but further analysis would be neededto determinethe accuracy ofthese hypotheses. In 2000-01 the number of pensioners in the household was associated with increased consumption: this is notthe case inthe lattertwo years. The education of the household head has a strong and statistically significant impact on welfare levels. Highereducation and finishingtechnikumare consistently found to have a positiveeffect on consumption (relative to having less than 5 years of schooling). The magnitude of the impact is quite large. Other characteristics of the head of householdthat affect welfare are ethnicity: ethnic Russians and those from outside CentralAsia havehigherwelfarethanhouseholdswhose headis Uzbek. The regionalvariables appear to increase their importancein explainingwelfare levelsover time. Living in any other regionthan Tashkent in 2003 loweredper capita consumption: this was not the case in 2000- 01 where only living in Ferghana or the Southern region negatively affected consumption. Holding all other factors constant, the negative impact of living outside of the Tashkent is striking. The effect of living in a rural area, compared to an urban one, is insignificantin two out of the three years and even in the year inwhich the differenceis significant, the effect is small. 23 Table 7: Correlates ofHouseholdConsumption 2000-01 2002 2003 Childrenages 0-6 - -0.022 0.008 0.003 [O. 0081** [0.008] [O. 0081 Childrenages 7-13 -0.021 -0.013 0.007 [0.009]* [0.009] [0.008] Childrenages 14-17 -0.030 -0.020 -0.030 [O.O 101** [0.009]* [0.009]** Number of pensioners 0.045 0.001 0.008 [O.0121 ** [0.013] [O. 0141 Householdsize -0.030 -0.03 1 -0.045 [0.006]** [0.006] ** [0.005]** Female headof household -0.012 -0.003 0.051 [0.019] [0.021] [O.O 191 ** Head aged 30-45 0.008 -0.014 -0.009 [0.021] [0.021] [0.022] Headaged 46-60 -0.003 -0.015 0.002 [0.029] [0.0251 [0.023] Head ages 60 or older -0.033 0.037 0.041 [O. 0391 [0.03 I] [0.0301 Head Married -0.040 -0.017 0.034 [0.020]* [0.022] [0.020] Head Education: 5-9 years -0.009 0.127 0.029 [0.023] [0.028]** [0.029] Head Educ:Secondary complete 0.020 0.138 0.023 [O. 0281 [0.026]** [0.028] HeadEduc:Technikumincomplete 0.091 0.137 0.056 [0.031]** [O. 0471** [0.046] Head Educ:Technikumcomplete 0.138 0.201 0.114 [O. 0281 ** [0.029]** [0.032]** Head Educ: Higher 0.222 0.290 0.203 [0.026]** [0.027]** [0.031]** Head Ethnicity: Russian 0.064 0.200 0.119 [0.026]* [O. 0341** [0.033]** Head Ethnicity: Other Cent. Asia 0.022 -0.064 -0.038 [O. 0281 [0.028]* [0.0281 Head Ethnicity: Other 0.122 0.130 0.108 [0.026]** [0.031]** [O. 0301 ** Mirzachulregion 0.056 -0.181 -0.333 [O. 0361 [0.0381** [0.043]** Ferghana region -0.094 -0.098 -0.264 [0.026]** [0.0261** [0.034]** Northernregion -0.037 -0.109 -0.264 [0.033] [0.03 1]** [0.038]** Central region 0.009 -0.065 -0.250 [0.033] [0.030]* [0.046]** Southernregion -0.303 -0.253 -0.300 [0.054]** [0.032]** [0.040]** Rural 0.013 -0.038 -0.043 [0.020] [O.O 191 * [0.026] Constant 8.754 9.387 9.610 [0.048]** [O. 0481 ** [0.047]** Observations 9636 9568 9487 R-squared 0.22 0.18 0.24 Sources:HBS2000-01, 2002,2003. Calculationsby authors. *significant at 5%; ** significant at 1%. Standarderrors inbrackets. 24 Areas of Change As the above has shown, welfare levels are affected by a range of factors. The regional differences, particularlyinterms of extreme poverty,are clearly a subject that requiresfurther study to determine why the headcount rates have changed so dramatically andto understandthe variation in the linkagebetween growthandwelfare levels. Two other areas which are addressed in this section are also closelyconnected to household welfare: labor market status and education. A person's participation in the labor market affects his or her welfare levelinthe short-run, as well as those of other householdmembers. Incontrast, education, specificallyenrollment levels, will have a mediumto long-termimpact on welfare levels as the returns to today's investmentin human capital are realizedwhen students move intothejob market. This sectionprovidesa brief overview ofthese two facets ofwelfare. Education The estimates of the correlates of consumption shown above demonstrate the importance of education in terms of welfare levels. As was shown in the LSA [World Bank, 20021, overall levels of educational attainment are quite high in Uzbekistan. There was, however, a problem with enrollment rates that appeared in the 2000-01 data. Subsequent work by the SCS showedthat the low rates might have been due to field work problems related to the first round of the HBS. As can be seen in Table 8, the differences in enrollmentrates captured in 2000-01 and the later two years are quite striking and indicate that the low enrollmentrates seen in 2000-01were probablydue moreto dataproblemsthanto actual low enrollments. Table 8: Gross EnrollmentRatesby Area, 2000-01 to 2003 (Children6 through 18 years o f age) 6- 14 years of age 15-18 years of age 2000-01 2002 2003 2000-01 2002 2003 Urban 72.2 88.2 91.4 51.9 59.0 69.0 Rural 69.2 89.4 91.5 43.4 58.3 65.4 Tashkent 82.0 89.0 89.3 59.0 68.8 68.7 Mirzachul 79.9 88.8 92.2 49.7 52.8 60.6 Ferghana 76.3 86.8 91.4 48.9 58.8 68.5 Northern 67.8 90.6 91.8 42.8 50.3 62.3 Central 82.9 89.8 92.3 55.4 59.2 67.2 Southern 36.3 90.5 92.1 21.7 58.1 66.5 National 70.2 89.0 91.5 46.0 58.5 66.5 Source: HBS2000-01,2002,2003, Calculations by authors. Gross enrollmentrates are fairly consistent across urban and rural areas as well as across regions, (Table 8) and welfare status (Figure 3). This is particularlypositive given the sharp rise in extreme poverty in two regionsthat couldbe expectedto have ledto drop out rates increasing. The drop out rate after age 14 is much higher than in younger ages: by age 18 only two-thirds of children are still in school. There seems to have been a rather substantial improvement in this indicator between 2002 and 2003 which may reflect recent investments in the secondary school system. Poor children are more likely to drop out at earlier ages than the non-poor. As the regressions above showed, the real impact on welfare comeswith completedtechnikum andhighereducation: levelsthat are not attainedby those leavingschool at fourteen or fifteenyears ofage. Itwill be importantto continue to monitorthis variableover time. 25 Figure3: Gross EnrollmentRatesof6-18 Year Oldsby Welfare Status 2002-2003 120 100 g 80 4.3 e 2002: Non-pool = 60 E : 2002: Poor 2003: Non-pool 0 u) 2003: Poor u) 2 40 (3 20 0 6 7 8 9 10 11 12 13 14 15 16 17 18 Age ~~~~ ~~~ ~ ~ Source HBS2000-01,2002,2003, Calculations by authors To understandfurtherthe factors affectingenrollmentrates amongthose 15 through 18 years of age inthe country, a probitmodelwas estimated. The dependent variable takes on the value of one ifthe person is enrolled in school duringthe relevant school year and zero otherwise. The explanatory variables are the educationofthe headofhouseholdand hisor her ethnicity, the age and sex ofthe individual, andwhether or not the individual lives in a household with consumption below the poverty line." Three different specificationsofthe modelwere runperyear. The resultsare presentedin Table 9 and inAnnex 3. The first column (and columns 3 and 5) contains the results of the model run without regional dummies, while the second model includes these economic regions (columns 2, 4, and 6). In all models and all years, children who live in a householdwhose head has achieved higher education are more likely to be enrolled in school. Additionally, as would be expected, the older the child the lower the probabilityof enrollmentis. Povertyhas a significant negative effect on schoolingin all years in the first model. But, in 2000-01 this effect is eliminatedwhen regional variables are taken into account. For 2002 and 2003 poverty, even taking into account regional variables, poverty had a negative impact on enrollment. In 2003, children living in povertywere three percent less likely to attend school thantheir peers, all else beingequal. This is a worrying trendthat may warrant further study to determine specific programs to increasethe school enrollmentrates ofthe poor. 11The means and standard deviations of the variables used here can be found In Annex 3, Table A3.1. 26 Table 9: Probabilityof SchoolEnrollment,Ages 15-18,2000-01 to 2003 (Marginal Effects) 2000-01 2002 2003 Headsecondaryor technikum 0.008 0.023 0.021 0.026 0.099 0.102 [0.023] [0.0241 [0.025] [0.025] [0.025] ** [0.025]** Hhldheadhighereducation 0.107 0.134 0.115 0.112 0.144 0.147 [0.029]** [0.030] ** [0.028] ** [0.029]** [0.023]* * [0.023]* * Male 0.016 0.016 0.023 0.022 0.026 0.027 [0.016] [0.016] [0.016] [0.016] [0.014] [0.014] Years ofage -0.230 -0.249 -0.252 -0.255 -0.222 -0.222 [0.008]** [0.008]** [0.007]** [0.008]** [0.007]** [0.007]** RuralArea -0.088 -0.050 -0.002 0.009 -0.027 -0.021 [0.018]** [O.O 19]* * [0.017] [O.O 181 [0.O161 [0.016] Headofhhld:Russian 0.145 0.072 0.112 0.028 0.059 0.041 [0.053]** [0.057] [0.048]* [0.0571 [0.052] [0.0561 Headother CentralAsian -0.067 -0.048 -0.053 -0.022 -0.007 0.008 [0.024] ** [0.025] [0.026]* [0.0271 [0.022] [0.022] HeadOther Ethnicity, not CA 0.184 0.116 0.042 -0.017 0.114 0.108 [O.0461 ** [0.0501 * [0.049] [0.053] [0.036] ** [0.037]** Not soddaughter of HhldHead 0.022 0.022 0.045 0.035 0.044 0.043 [0,0251 [0.0261 [0.0261 [0.026] [0.023] [0.023] PC Consumptionbelow pov. line -0.074 -0.021 -0.052 -0.043 -0.036 -0.032 [O.O17]* * [O.O 181 [O.O 181 ** [0.018]* [0.O161* [O.O 161* MirzachulRegion -0.122 -0.242 -0.105 [0.032]** [0.0361 ** [0.035]** FerghanaRegion -0.134 -0.142 -0.013 [0.027] ** [0.0301 ** [0.0271 NorthernRegion -0.190 -0.233 -0.091 [0.028]** [0.034] ** [0.031]** CentralRegion -0.047 -0.118 -0.031 [0.028] [0.032]** [0.028] SouthernRegion -0.432 -0.166 -0.037 [0.020]** [0.033]** [0.0281 Observations 4849 4849 4808 4808 4789 4789 Source: HBS 2000-01, 2002, 2003, Calculations by authors. Note: Omitted categoriesare headwith educationbelow secondary, Tashkent region(in the equationswith regional variables), and headof householdwith Uzbek nationality. *Significant at 5%;** significant at 1%. Standarderrors inbrackets. There is no gender gap inenrollment, which representsa substantialachievement inthe region. However, in 2003, males were 3.5 percent more likely to attend school than females. Separate probitswere run for males and females to determine how the variables included here affect enrollment differently for males and females. As can be seen in Tables A3.2, 3, and 4, the equation does a much betterjob of explaining male enrollmentthan female enrollment: very few variables in the model are statistically significant for females. Thus, male and female enrollment decisions appear to be determined by different factors: an issue that needs to be investigated further in order to ensure that the impressive investment in female education inthe country innot put at risk. Labor Force The labor force situation merits an additional review in the present document as it is one area where changes have occurred in the three-year period. As can be seen in Table 10, the labor force participation rate (those working or seeking work divided by the number of persons of working age) has stayed the 27 same at the national level, but inTashkent the labor force participation rate declined between 2000-01 and 2002 and only partially recovered by 2003. And in Mirzachul, some of the poor individuals who faced a decline in welfare levels appear to have responded by dramatically increasing their supply o f labor: the labor force participation rate for the poor rose from 45.4 percent in 2001 to 63.3 percent in 2003. Table 10: Labor ForceParticipationRates by Economic Regionand Level, 2000-01 to-2003 Poor Non-Poor National 2001 2002 2003 2001 ' 2002 2003 2001 2002 2003 Tashkent 56.8 54.9 56.1 64.4 58.1 60.0 62.8 57.4 59.3 Mirzachul 45.4 61.5 63.3 63.0 65.6 64.5 58.3 64.2 64.0 Ferghana 52.4 55.6 52.9 54.1 56.4 58.4 53.5 56.1 56.6 Northern 45.5 52.6 51.1 56.4 61.8 57.1 52.8 58.8 55.4 Central 47.0 47.4 48.0 59.2 60.7 57.8 56.1 57.6 55.3 Southern 62.3 61.5 66.1 65.2 35.4 65.6 63.8 63.8 65.8 National I 53.2 55.7 55.4 59.6 59.9 60.0 57.6 58.8 58.7 Source: HBS 2000-01,2002,2003, Calculations by authors. Given the rise in extreme poverty in Mirzachul, it i s not clear that the increased supply of labor has been successful in increasing incomes. As can be seen in Table 11, the unemployment rate also rose in the region among the poor. The supply o f labor increased but it is not clear that the demand also rose or that the income from availablejobs was enough to keep households out o fpoverty. Nationally, the drop inthe unemployment rates has been quite substantial in the three year period. The lack of a corresponding decline in poverty suggests that there may be more issues related to the quality of employment than the demand for labor (with, perhapsthe exception of the Mirzachul region). Poor Non-poor National 2000-01 2002 2003 2000-01 2002 2003 2000-01 2002 2003 Tashkent 6.9 6.0 3.4 3.5 3.5 5.5 4.2 4.0 5.1 Mirzachul 6.0 6.8 8.2 7.0 3.8 4.1 6.8 4.8 5.6 Ferghana 64. 2.4 4.1 4.9 2.0 2.6 5.4 2.1 3.1 Northern 5.5 3.5 2.6 4.8 2.4 2.3 5.0 2.6 2.3 Central 10.1 3.9 1.9 6.1 2.4 2.1 6.9 2.7 2.1 Southern 5.9 2.4 3.2 4.9 0.9 0.6 5.3 1.5 1.3 National 6.7 3.7 3.8 5.0 2.4 2.8 5.5 2.7 3.1 I I I Source: HBS 2000-01,2002,2003, Calculations by authors. 28 ANNEX1:MEASURING POVERTY (From the LSA 2003) 1.1 Common Approach The first step in standardpovertyanalysis requires the constructionof a total consumption aggregate from householdleveldata. This consumption aggregate is a money-metricmeasureof welfare and can be used to rank individuals in the society from lowest to highest level of welfare. With this informationone can easily compare how individualsdiffer on key characteristic by their welfare status. Usingsuch a measure, a common analysis is to divide the population,based on welfare, into populationquintiles (deciles) and comparethose inthe lowestquintilewith those in higher ones. Such an analysis, however, says nothingabout absolute levels of poverty. In the context where poverty may be a serious concern and where monitoring of changes in it are desired, some absolute measure of poverty is needed. In other words, one needs to determine below what level of consumption a person should be classified as poor. Thus, the next step in the poverty analysis is the constructionof a poverty line. The basic, or food poverty line is set at the value in monetaryterms of a minimum level of calories requiredfor basic health. Three pieces of information are used to determine the value of the food poverty line. First, the actual consumption patterns of the populationare identifiedto determine, on average, the share of calories that are obtained from different foodstuffs. This informationis obtainedfrom the diaries offood consumption kept by the householdsinthe survey. Second, the minimumrequiredlevelof calories is determined. The third piece of informationrequiredis the cost of obtainingthis minimumlevel of calories. To do this the actual consumptionpatterns of the population(quantities of each foodstuffconsumed) are converted into monetary values. The cost of consuming the minimum level of food based on actual consumption patterns is the value of the food poverty line. The value of an individual's consumption is then compared to this food poverty line: all those whose total consumption i s below this line would be classified as extremely poor, while all those whose consumptionis abovethe linewouldnot be. Recognizingthat individuals have other needs beyond food, the next step is to construct a more general poverty line that includes non-foodconsumption. Giventhe impossibilityof consensus on exactly what bundle of goods should be in the non-foodportionof consumption, no effort is actually made to do this. Instead, an allowance for non-foodspending is added to the value of food consumption. This allowance is calculated as the share of total spending that is dedicated to non-food consumption among all individuals whose food consumption is at the level of the extreme poverty line. The logic being that if one could, one would increase one's food consumption. If a person is foregoing additional food, however, in order to obtain other items, the assumption is that this other consumption represents a fundamental need. Once this general line is constructed, the value of an individual's total consumption is compared to the value ofthis line. Iftotal consumption is belowthe linethe person is consideredto be poor. Note that, by definition, anyone who is extremely poor is also poor. 1.2 Approach UsedHere The approach used in this chapter for measuring poverty in Uzbekistan differs in several ways from the common approach outlined above. The divergence is due to several limitations in the data that were available for the analysis. The data used are from the Household Budget Survey of April 2000-March 29 2001 (HBS). While a strong Incomeand Expenditure Survey, the H B S was not designed specifically to provide poverty data and, thus presents some serious drawbacks for such analysis. To overcome these constraints a different methodology for measuring poverty has to be used. While allowing an overall assessment of poverty to be carried out, the method does limit both what can be analyzed and how comparable the analysis that is done will be to the 'standard' approach. What follows outlines the approach used and some of its advantages and disadvantages. 1.2.1 Welfare Measure-Consumption Aggregate: The data from the HI3S do not allow the construction of a total consumption aggregate. Only a food consumptionaggregate can be constructed. For the other components of total consumption (flow of services from housingand durable goods and all other non-foodconsumption) the data are either missingor inadequate for constructinga reliablemeasure. The flow of services cannot be calculated due to no information having been collected on the value of housingor durablegoods. And for other non-foodconsumption, a two-week reference periodwas usedto collect information: a reference periodthat is appropriate for the consumer price index goals of an HBS, but so short as to create enormous noise in the data when trying to measure household welfare, not national consumption. Non-foodpurchasestend to be very lumpy, thus, the day in which the household i s interviewedwill affect the overallwelfare levelthat is recorded for that household, regardless of its true welfare level. l2 1.2.2 Food Poverty Line: The food consumption data from the HBS do allow the construction of a food poverty line. The linecalculated here is basedon a level ofrequiredcalories set at 2100calories per person per day. The FoodandAgricultural Organization(FAO) has developedsix basic caloric levels depending on the primary staples consumed in a country. The 2100 caloric level is from the FA0 wheat- baseddiet and should be appropriate inthe Uzbekistancontext. The line was constructed inthe following way. The actual consumptionpatterns (from the HBS) of the population in the 11' to the 29'h percentile were identified as was the share of the required 2100 calories coming from each food item. As this reference group of individuals did not, on average, consume a minimum of 2100 calories per day, it was necessary to adjust the total calories consumed from each food so that the total food consumption per person per day would be the minimumrequired2100. This consumptionwas converted into sourns using the average prices paid by the reference group, in other words using the prices faced by the poorer population. The value of the food poverty line in October 2000 prices is S/ 3601 per month. Note that this is 50 percent higher than the value of the minimumwage in 2000. The value of the line is, however, less than one half of average wages in2000, ItTable A.l: Value of FoodPovertyLine, MinimumandAverage Wages 1-1 1 Food Poverty Line, nominal2000 Source: FBS, 2000-01: Value of food poverty line calculated by author; value of minimum wage in 2000 from SDOS, average wage in 2000 as quoted in World Bank 2001a. Up to this point, it has beenpossibleto follow a standardapproachto povertymeasurement. But, because it was not possible to construct a total consumptionaggregate, it is not possible to construct a general I 2 lmag.me the m e of two householdsLhal purchse the same amount of clothing for theircluldrenm Ihe fall pnor to Ihe kgimng ofthc achwl year Ifone householdIS inletviewed during the purchaseweek and Ihe other IS mtcrvwwsd hro weeks Iaicr,the data would lead you to concludeLhalthe firsl householdwas bener oflthw Ihe latter. while m fact there 1s no d~~Terence between Ihem. 30 poverty line: the allowance for non-food consumption cannot be estimated. This is the problem that requiresthe applicationofa differentmethodologyto resolve. 1.2.3 Method Used A methodproposedby Olsen Lanjouwand Lanjouw (OL-L, 2001) allows oneto determine ageneral headcount rateusingonly food consumption, and comparingthis consumption to the food poverty line.13 Essentially,this is a methodfor calculatinga headcount ratiothat is robustto variations in the comprehensiveness of the consumption measure used. The three assumptions underpinningthis method are that: (i)all components of consumption that are used follow an Engel's relati~nship;'~ (ii)consumptionpatternsarestableacrossgroups;and(iii)thereisnomis-measurementin the data or at least no differenceinmeasurementerror acrossthe indicatorsusedor the populationgroups. Ifthesethree assumptionsholdtrue, a headcount ratiocalculatedusingatotal consumptionmeasureand an upper bound approachto setting the general poverty line will be the same as the ratio calculated using a less complete consumption measurecomparedto the foodpovertyline. It is importantto notethe two maindisadvantagesofthis method. First, it allows one to rank individuals on the basis of food consumption but provides no guidance on the distance between individuals in terms of total consumption. Thus, inequality measures cannot be calculated from the consumption data. Second, comparisons with poverty rates in other countries are limitedto those where an 'upper-bound' poverty line has been used: poverty rates calculated on a 'one-dollar a day line' are not comparable. Giventhe data available, however, this methodallowsthe best possibleuse ofthe data and is usedhereto providea head count index for Uzbekistan as well as to provide some insights into the characteristics of the poor. 13 The FonwGrecr-Thorkke (1984) cims of povertyindicators is: The headwunt rauo i s the FGT indicatoris the case wherethe parametera = 0 The poverty ime inquestionmuat use an 'upper bund' approach This entiis ConsUvclinga tolal consumpionagpgale, using!his (anda mimmumsalonc rcquiremel)to determinethe ememe or food, poverty lineand then addingan allowance for "on-food Fonsumption 'This allowance IS caimisted as the share ofconrumption dedicdtedto non-fooditem by householdswhose foodconsumptionio equivalentto the food poverty line Ravaiiion(1994) alx, dirclunes B second method for calmlaungthe 'allowance'for "on-food which IS to scale up the food povmy line by L e amount spent on "on-food consumptionby households whosetotal consumpianis a1the level of the food pavcny line The OL-L methodIS not robust with respct to this austere approach I4 Engel'slaw nates that total psr capitaconsumpuon mer ala nondecrearmngrate wth food wnrumpuon. 31 ANNEX2: ISSUES OF COMPARABILITY Carryingout the analysis for 2002 and 2003 to ensure strict comparabilityamong the three years of data requiredthe resolutionof a series of issues. 1. Consumer price index: The value of the original poverty line is in October 2000 prices. To determine the value of the poverty line for 2002 and 2003, the original line needs to be adjusted for the inflationbetween October 2000 and October 2002 and October 2003. The simplest methodwould have been to use the national CPI for the period to make the inflation adjustment. However, the original poverty line was priced using the prices faced by the poorest (2nd and 3'd deciles) population. The nationalCPI does not make this distinction. To the extentthat the prices faced by the poor have changed at a different rate than those faced by the non-poor, using the national CPI could distort the poverty comparisons. 2. Constructing a CPI: To avoidthe problems of usingthe nationalCPI, a CPIcan be constructed using the household survey data based on the inflation faced by the second and third deciles of the populationwhich is our basegroup. To do this, the quantities of every food item includedin the poverty , line from 2000-01 are multiplied by the prices found in 2002 (2003) HBS. A problem arose in this process, however, as many goods that were consumed in 2000-01 were not consumed in 2002 (2003). Thus the poverty line of 2000-01 cannot be revalued exactly. What we have done is to create a new poverty line usingthe 2000-01 quantities of all food items consumed in all three years. To obtain the value ofthe line ineachyear, these quantities were multipliedby the October prices in eachyear. A potential problem with this method is that, if the changes in observed consumption are a result o f increasedpoverty, the method would obscure this welfare loss. A review of the changes in consumption over the time periodin questionwas carriedout andthe findings indicatethat this is probablynot an issue inthe presentcase. The review and findings are discussed here. 1. There are two files ineachyear that containall ofthe informationconcerningthe number offood products purchased(expenditures) and consumed (consumption). The analysis of the changes in consumptionwas carriedout on the raw data files, withoutany cleaningor imputation,as the goal was to identify changes in the data itself that could be problematic. (The poverty analysis presented in the main body of this note is based on cleaned data.) The review was done using both weighted and unweighted data: as the results do not vary much we do not present both versions. 2. There is a change inthe number offood items purchasedand consumedby households amongthe three years. The mainchange is from 2000-01 to 2002. (See Table A2.1.) However, the average number of items purchasedor consumedper household is essentially the same over the three-year period, as is the rangeof items. 32 Source: HBS 2000-01,2002, 2003, calculations by author. Note: # of items refers to all of the food codes used in each year Meanitems per householdrefers to the average numberof food items listed by householdsas havingbeen purchasedor consumed in each year. a. A review of the categories of foodproductsconsumedor purchasedshow that many were only consumed in 2000-01. The majority of these categories that do not appear in the subsequent data sets were `aggregated' categories (i.e. not the most detailed levelof food code) and an additional group was `other' categories. In other words, the bulk of the codes which do not reappear were `imprecise' codes. One would expect to see these changes if the fieldwork of the survey itself were changed: if interviewers and coding were improved. This is what the SDOS claims they have done and the data supports this claim. Thus the changes in food categories that appear among 2000-01 and the two subsequent years do not appear to reflect a change in consumption, but an improvement in the survey itself. b. We also lookedat the importance ofthe itemsthat appear only in2000-01 andthat do not reappear in 2002 or 2003. The concern was that the decline in the number of items purchased (consumed) could represent a decline in expenditures or consumption. The results show that the share of expenditures or consumption represented by these omitted categories was very small: the difference in the value of daily consumption in 2000-01 includingand excludingthe food itemsonly found inthat year is two soums. 33 ANNEX3: PROBABILITY OF SCHOOL ENROLLMENT AMONG 15-18 YEAR OLDS Enrolled 0.460 [0.009] [0.009] [0.009] Head's education0-4 yrs. 0.074 0.051 0.042 [0.007] [0.005] [0.005] Head's education5-9 yrs. 0.113 0.113 0.099 [0.007] [0.008] [0.008] Head's educationsecondary or technikum 0.671 0.681 0.702 [0.012] [ O . O l l ] [O.O lo] Head's educationhigher 0.141 0.151 0.152 [0.007] [0.008] [0.008] Male 0.491 0.495 0.491 [0.007] [0.008] [0.008] Years of age 16.441 16.455 16.484 [O.O 131 [0.015] [0.014] RuralArea 0.694 0.687 0.687 [0.012] [0.011] [0.011] TashkentRegion 0.147 0.144 0.156 [0.011] [0.009] [0.o101 MirzachulRegion 0.072 0.077 0.077 [0.005] [0.005] [0.0041 FerghanaRegion 0.256 0.263 0.265 [0.011] [0.011] [0.008] NorthernRegion 0.139 0.135 0.129 [0.009] [0.007] [0.007] Central Region 0.205 0.182 0.191 [0.013] [0.010] [0.009] SouthernRegion 0.181 0.199 0.182 [0.012] [0.011] [0.008] Headethnicity: Uzbek 0.828 0.850 0.842 [0.018] [0.014] [0.014] Headethnicity:Russian 0.019 0.016 0.012 [0.0021 [0.0021 [0.002] Headethnicity: other Central Asian 0.127 0.109 0.120 [0.017] [0.014] [0.014] Headethnicity:other, not CA 0.024 0.021 0.021 [0.0041 [0.003] [0.003] Not soddaughter of Head 0.163 0.149 0.142 [0.009] [0.008] [0.009) PC Consumptionbelowpov. Line 0.356 0.310 0.320 [0.O 161 [0.014] [0.014] Observations 4849 4808 4789 Standarderrors inbrackets Source: HBS2000-01,20002, 2003. Authors' calculations. Note: Standarderrors inbrackets. 34 Table A3.2: Probabilityof Enrollmentamong 15-18 Year OldMales andFemales, 2000-01 Males Females Headsecondarv or technikum 0.059 -0.004 [0.034] [0.034] Hhldheadhigher education 0.137 0.146 [0.042]** [0.044] ** Years of age -0.239 -0.258 [0.011]** [0.011]** RuralArea -0.027 -0.072 [0.027] [0.026] ** MirzachulRegion -0.116 -0.125 [0.047]* [0.044] ** FerghanaRegion -0.150 -0.113 [0.038]* * [0.039]** NorthernRegion -0.205 -0.172 [0.038]** [0.0411 ** Central Region -0.076 -0.014 [0.0401 [0.041] SouthernRegion -0.449 -0.413 [0.028]** [0.0301 ** Headofhhld: Russian 0.067 0.091 [0.075] [0.083] Headother Central Asian -0.024 -0.074 [0.035] [0.0351 * HeadOther Ethnicity,not CA 0.003 0.220 [0.0721 [0.067]** Not soddaughterof HHldHead 0.095 -0.036 [0.0381* [0.0361 PC Consumptionbelow pov. Line -0.036 -0,005 [0.026] [0.025] Observations 2408 2441 Robuststandarderrors inbrackets * significant at 5%; ** significant at 1% Source: HBS 2000-01. Author's calculations. Note: Omitted categoriesare: headwith less than secondary education, Tashkenteconomic region, heado f householdwith Uzbek ethnicity *significant at 5%;** significant at1%. Standarderrors in brackets. 35 Table A3.3: Probabilitv of Enrollmentamong 15-18Year Old Males andFemales. 2002 Males Females Headsecondary or technikum 0.019 0.033 [0.035] [0.037] Hhldheadhigher education 0.129 0.091 [0.0381 ** [0.043]* Years ofage -0.246 -0.266 [O.O lo]** [ O . O l l ] * * RuralArea 0.018 -0.000 [0.0251 [0.025] MirzachulRegion -0.240 -0.245 [0.050]** [0.051]** FerghanaRegion -0.156 -0.126 [0.043]** [0.043]** NorthernRegion -0.228 -0.241 [0.047] ** [0.047] ** CentralRegion -0.135 -0.099 [0.045]** [0.046] * SouthernRegion -0.174 -0.158 [0.047] ** [0.046] ** Headofhhld: Russian 0.004 0.059 [0.077] [0.083] Headother CentralAsian -0.032 -0.013 [0.0381 [0.0391 HeadOther Ethnicity,not CA 0.022 -0.057 [0.076] [0.074] Not soddaughterof HHldHead 0.029 0.044 [0.037] [0.0361 PC Consumptionbelow pov. Line -0.037 -0.050 [0.0261 [0.026] Observations 2416 2392 Source: HBS 2002. Author's calculations. Note: Omittedcategoriesare: headwith less than secondary education, Tashkenteconomic region, head of householdwith Uzbek ethnicity. *significant at 5%; ** significantat 1% Standarderrors in brackets. 36 Table A3.4: Probabilityof Enrollmentamong 15-18 Year Old Males andFemales, 2003 Males Females Headsecondaryor technikum 0.094 0.108 [0.0351 ** [0 0361 ** I Hhldheadhigher education 0.144 0.149 [0.030] ** [0.034]** Years of age -0.217 -0.227 [O.O 1O]* * [O.O lo]** Rural Area -0.022 -0.020 [0,0221 [0.024] MirzachulRegion -0.115 -0.097 [0.051]* [0.047] * FerghanaRegion -0.041 0.013 [0.038] [0.0371 NorthernRegion -0.092 -0.090 [0.044] * [0.044] * Central Region -0.019 -0.046 [0.039] [0.0401 SouthernRegion -0.063 -0.014 [0.041] [0.040] Headofhhld: Russian -0.054 0.118 [0.088] [0.069] Headother Central Asian 0.011 0.003 [0.0311 [O.0321 HeadOther Ethnicity, not CA 0.098 0.115 [0.052] [0.053]* Not soddaughter of HHldHead 0.073 0.012 [0.030] * [0.034] PC Consumptionbelow pov. line -0.034 -0.032 [0.023] [0.023] Observations 2385 2404 Source: HBS 2003. Author's calculations. Note: Omittedcategoriesare: headwith less than secondary education, Tashkenteconomic region, head of householdwith Uzbek ethnicity. *significant at 5%;**significant at 1%. Standarderrors inbrackets. 37 ANNEX4: DATASOURCES This poverty assessment utilized several surveys, each with its strengths and weaknesses. A summary of key features is as follows. HouseholdBudget Survey (HBS) The HBS data are used and described by the first chapter. The advantages of the HBS are that they are nationally representativeand data are available for three years (2000-01,2002, and2003). It is, however, limited inits coverage of education, health and labor topics. In addition, response categories in education, for example, are not consistent (for attendance, 5 categories, while for attainment, 12 categories) and with specification problems (e.g. for attainment, one category combinesthe technical PTU/SPTUtrack ofthe old system with the academic lyceumtrack ofthe new system). HouseholdEnergy Survey (HES) This 2005 survey is nationally representative and comprehensive in its education treatment, including questions on all levels from pre-school up, attendance, enrolment, private expenditures on education, etc. However, the approach adopted for the questionson household consumption is different from that inall other surveys inthe country. Thus, the available welfare indicator is based on total household income rather than consumption expenditure, so that welfare indicators are not comparable across HES and other surveys. UzbekistanRegional Panel Survey (URF'S) This first wave of this survey (the one analyzed for this assessment) was conducted in February and March 2005 in Kashkadarya oblast; in Andijan oblast; and in Tashkent City. It is, therefore, not nationally representative. The advantages of the URPS include the implementation of a comprehensiveLSMStype householdquestionnaire alongside a community survey. The coverage of education, health, agriculture and labor i s rich, including the full range of household-levelinformation, and inaddition information on school facilities, although unfortunately the education level responsecategory "lyceum" i s omitted. UzbekistanHealth Examination Survey This survey collected information on 5,806 women between the ages of 15 to 49, 2,640 men between ages 15 and 59 (all eligible men in Tashkent were surveyed and one in three for the rest of the country) and 3,488 children. Hemoglobin was measuredfor 3,243 children; of these, 1496 had no anemia, 1153 had mild anemia, 558 had moderate anemia and 36 had severe anemia. Measurementson height and weight of 3,276 children are also available, indicating stuntingand wasting rates of 20 percent and 7 percentrespectively, for childrenyounger than 6 years old. 38 CHAPTER2: LABORMARKETSAND POVERTY This chapter looks at the relationship between labor market performance and poverty in Uzbekistan. The first part analyses the relationship between labor market outcomes and macroeconomic developments; discusses the key features of the labor market in Uzbekistan; and examines the link between labor market status and consumption. In order to explain the observed relationship between poverty and labor market outcomes, the second part turns attention to the main obstacles of firm growth in Uzbekistanand examines howthe investment climateaffects firm restructuringand employment. MAINFINDINGS 1. The populationof working age has grown considerably faster than formal sector employment since the mid 1990s. This excess labor supply was partiallyabsorbed by the informal sector, alongside labor force withdrawal. 2. Labor intensityof growth is relativelylow. A 1 percent increasein GDP is associatedwith only 0.37 percent increase in employment. Unemployment and underemployment is particularly high among less educatedworkers. 3. Low open unemployment masks significant labor market problems. Many existingjobs are of low productivity and do not protect workers from poverty. Labor force participation is low, largely due to the lack of productivejob opportunities. Underemployment is pervasive, especially in rural areas, as evidenced by short workinghours. 4. The informal sector looms large, driven by high taxes on labor. Although it does providejobs, it often seems to be an employer of last resort offeringonly subsistence income. All these features contribute to povertyamongthe working age population, includingthose withjobs. 5. Labor force participationrate is especially low among older men: only 47 percent of men aged 55- 64 are employed in Uzbekistan, compared with 62 percent in the OECD. Possible reasons for this include low retirementage, lower skills ofolder workers and poor healthstatus of older individuals. 6. Poverty reduction in Uzbekistan requires improving labor market outcomes: a faster pace of job creation and stronger sustainable productivitygrowth. There needs to be morejobs and the jobs should be better. They needto be moreproductiveandthus offer higherwages. This necessitatesimprovementsinthe investment climate so as to encourage new firms to enter the market and existingfirms to invest and expand, as discussedby the previous chapter. 7. The costs of doing business in Uzbekistanhave been significantlyloweredin recent years but there i s room for improvement. 8. Large gains in labor productivityare a positive phenomenon and are characteristics of virtually all transition economies. However, the low pace of new firm creation is worrying given the widespread firm downsizing. 9. In order to increase formal sector employment, the Government should try to remove the most binding constraints. Analysis of BEEPS data indicate significant payoffs to investing in infrastructure, improving governance, removing administrative barriers and developing an efficient and competitive financial system. 39 PART1: LABOR MARKET OUTCOMES AND POVERTY Macroeconomic Conditionsand Labor Market Outcomes This section examines the impact of economic growth on labor market outcomes and poverty in Uzbekistan since the mid 1990s. It finds that economic growth in Uzbekistan, as other transition economies, has been driven largely by productivity improvements and brought about relativelyfew jobs." At the same time the population of working age has been growing rapidly, causing an increasing gap between the number of available (formal sector) jobs and the number of available workers. Productivejob opportunities are thus scarce despite low unemployment, which is reflectedin a relativelylow employment-to-populationratio, and contributes to poverty. But improvements in labor productivity have translated into fast real wage growth, which is likely to reduce povertyamongthose withjobs. Economic growth has not brought about enoughjobs to keep pace with a rapid increase in working age population butfuelled wagegrowth Labor market conditionsclosely depend on macroeconomic developments. In particular, they are relatedto the GDP growth, which is a measure of the dynamic of product demand. GDP growth is a necessary although not a sufficient conditionfor employment growth. The latter depends also on the labor intensity of growth. If labor intensity of growth is high, then GDP growth is translated into employment growth. Conversely, low labor intensity of growth means that the growth in output is largely achieved through improvements in labor productivity with employment growth lagging behind. But improvements in labor productivity are translated into higher wages. So, economic growth benefits workers either by bringing about newjobs, or higherwages, or both. However, unemployment declines as a result of economic growth only ifthe rateofoutput growthis higher thanthe rate at whichthe populationofworkingage is growing. In other words, thejobs growth needsto be at least as highas the rate of working age populationgrowth. Ifthis condition is not met, then unemployment may increase despite economic growth. What was the pattern of growth prevailing in Uzbekistan in the recent period?Figure 1 shows dynamics of key macroeconomic indicatorssincethe 1995 while Table 1 presentsdata for the last five years. Figure 1: Economicgrowthhasnotbeensufficient to absorbpopulation growth, leadingto labor market slack Dynamics of key maoroeoonomloIndioators 18S6.100 - - - -- -- 1995 19% 1997 1998 1999 ZWO 2001 2002 2003 2004 j-PopuI.tion offorbnpapc -Employment -0DP Labor producuvlty Wages] Source: Ministry of Macroeconomics and Statistics; Ministryof Labor and Social Protection; and IMF;Bank staff calculations. I S A discussion of UzbShstan~ o n t ecan k found m Linkng Macroeconomic Policy to Poveny Reduction,2WS. Centerfor EconomicResearch,Tarhkenl Uzkbnan For cross-counuyevidencesee Enhancing Job Opponuniuer: ~ t EarternEuropeand the FormerSaviel Union.the World Bank. Warbnpton D C 40 Table 1Keymacroeconomicindicators, 1999-2004 1999=100 2000 2001 2002 2003 2004 Annualaveragerateof growth Populationofworkingage 103.1 106.4 109.8 113.2 116.8 3.1 Employment 101.1 102.8 105.0 107.9 111.5 2.2 GDP (real) 103.8 108.2 112.5 117.2 126.2 4.8 Labor productivity 102.7 105.3 107.2 108.8 113.5 2.6 Wages 99.1 104.2 104.7 109.6 137.4 6.6 Source: Ministry of Macroeconomics and Statistics; Ministryof Labor and Social Protection; and IMF; Bank staff calculations. Three main observations emerge from the inspectionof the data. First, the disparitybetweenthe population growth and employment growth. The populationof workingage (16-64) has grown considerably faster than (formal sector) employment since the mid 1990s. This is an obvious source of labor markettension as there are more people entering the market than newjobs beingcreated. Inother words, potentiallabor supply has grown faster than labor demand. To illustrate, in the last five years the number of persons of working age increasedby 2.1 million, while the number of formal sectorjobs increased by only 1 million. Such a large gap usually gives rise to an increase in unemployment: however, open unemployment at around 4 percent is still low in Uzbekistan (see below). This means that the excess labor supply was absorbed by the informal sector as well as by labor force withdrawal. The latterphenomenon-when people of working age withdraw from the labor force because they cannot find work - is known as the "discouraged worker" effect. To absorbthe fast populationgrowth in the recentperiodthe formaleconomywould needto createnewjobs at a rate about 1 percentage point higher than the one prevailing so far. This poses a considerable policy challenge, and calls for the improvements in the investment climate so as to support firm entry and growth (see the sectiononthe investmentclimate). Second, the disparity between employment growth and GDP growth. While the GDP was growing at an annual rate of 4.8 percent in the last five years, formal sector employment was lagging behind and was growing at a rate of only 2.2 percent. This means that labor intensity of growth was relatively low. A 1 percent increase in GDP was associatedwith only 0.37 percent increase in employment. Output growth has been thus driven mainly by the growth in labor productivity,rather than by the growth in labor input. This pattern is quite typicalof all transition economies, where the increase ineconomic efficiency, includinglabor productivity, is an essential feature of the transition.16 Firms exposed to a soft budget constraint and both internaland external competitioneliminate overstaffinginheritedfrom the socialist past and shed redundant labor, which leads to substantial gains in labor productivity. Thus in Uzbekistan, as in other transition economies the productivity-employment trade-off is resolved in favor of productivity growth which manifests itself in the lowjob content of growth. It is importantto note, however, that this trade-off is of short-term character. In the longer run, productivity gains are translated into higher real incomes and consequently higherdemand, which inturnengendersemployment growth. Finally, the disparitybetweenreal wage growthandthe growth inlabor productivity. As a rule, inthe longer term wage growth closely follows productivity growth. But this is not the pattern that was prevailing in Uzbekistanin the last years. Instead wages grew much faster than labor productivity. For example, in the last five years wages grew at an annualratethat was 2.5 times higherthan that of labor productivity(6.6 and 2.6 percent, respectively). Put differently, a 1 percent increase in GDP per worker was associatedwith 2.8 percent increase in real wages. This impliesa substantialincrease in unit labor costs, i.e. the cost of laborper unit of output. In the short-run this should not be a problemas wages are still low by regional standards." 16 it may seem paradoxrcal,that rigniflcanl prcducuvityImprovementshave occurreddespitethe fan ULlt markstonented reformshave been slowUzkkiim(see Chaptsr2 on the mvestmenlcbmate) However, once (slate owed) firms have staned la face a harderbudgetwnswnt andhave beenexposedto some campuuve pressure (from thc emergingpnvateseaor) lhey munhavecut cam,including laborcasts, m order IO survive They thus $Wedto shed redundant laborwhichresultedm a hgheroutputper worker, lhal~6m laborprodunwily grouth. l b r is a symptomof "defensive resuuctmg where prcduniviryimprovementsresult largelyfrom ehrmnaungowstaffng ralherthanfrom technological or orpumtional i ~ w a t i ~ nFuter reformswould encouragefirms IO engagem "avatep nmctunng that IS m seekmgprcducuvitygams lhmughinvement and ~ O V B U Oand mlngthem into market expanlion (see r R Pan2 on mvertmentclimate). 17 Inpwciple, lo determine the Mmpetibvenersof Uzbekfirms one would needto relatewage levelsto labor prcdustmty. However,informatian on labor prcductiviryis not available 41 However, ifthis trend persists in the longer run it may negativelyaffectthe competitivenessof Uzbek firms. This is because an increase in unit labor costs translates either into higher prices (in an imperfectly competitive market) or into lower profits, which in turn may limit investment and job creation. But one needsto bear in mind that rapidwage growth since the mid 1990s is characteristic of virtually all transition economies, where real wages are rebounding after the dramatic drop at the beginning of the transition.18 Thus to some extent wage growth in excess of productivitygrowth simply reflects laborrestoringits share in total output. But this processhas its limits and cannot be sustained for a longer periodwithout undermining the competitivenessof Uzbek firms. To conclude, the relatively strong economic growth observed in Uzbekistan since the mid 1990s translated itself more into higher wages and less into higheremployment. In other words, economic growth benefited the insiders - workers with jobs - rather than outsiders, i.e. the jobless, new labor market entrants, and workers with casual, precariousjobs in the informalsector. This has clear poverty implications. As we will show later, the incidence of poverty among outsiders (especially among the discouraged workers) is significantly higher than among the insiders. Such a pattern of economic growth, which reduces poverty among some privilegedgroups but increases poverty among disadvantagedgroups, brings about an increase in socialinequalities. Labor Market in Uzbekistan-Kev Features This section examines labor market outcomes in Uzbekistan in 2005 and provides an assessment of labor market performance. It argues that despite low open unemployment labor is underutilizedin Uzbekistan. Hiddenunemploymentor underemploymentis pervasive, especially in rural areas. Labor force participation i s relativelylow, largelybecausemany workers become discouragedby the futility oftheirjob search efforts and withdraw from the labor force. Many of the employed work short hours, which is a symptom of underemployment. Jobs are frequently of low productivity, especially those in the large informalsector and in agriculture. Box The Labor Force Survey inUzbekistan The labor force survey, known as the Balance of Labor Resources (BLR) was introducedin Uzbekistan in 2002 and since then has beencarriedout on a quarterlybasis. The survey is administered by the Ministryof Labor and Social Protection(MoLSP) and implemented by the district (rayon) level Employment Offices. The sample size is approximately 18 thousand households (about 110 thousand individuals). The survey uses definitions which are broadly consistent with the L O definitions of unemployment and employment. The introductionof the survey significantly improvedthe available labor market informationand its results informMoLSP's policy. However, survey results are not inthe public domain which limitstheir usefulness. Accordingly, it would be desirable to makethe results available to facilitate labor marketresearchand public discussion of labor issues. It is noteworthythat BLR results are broadly consistent with the results of the URPS which are used in this report. For example, accordingto the BLR2005 the mainlabormarket indicatorsare as follows (Table A): Table A Main labor market indicators according to the BLR 2005 (percentages) Unemploymentrate 3.1 Laborforce participationrate 70.0 EmploymentPopulationratio 67.8 Informalsector employment (as % oftotal 34.5 employment) Source:Ministry of Labor and Social Protection; Bank staff calculations. 18 Yemlaev, Ruslan. RcdicaCnoblochand Ccm Mete (2006), "Evolutmnof [he Rdinorr of E m g s h g Transation"manuscript. the World Bmk 42 Labor is underutilizedin Uzbekistandespite the low unemploymentrateTheunemploymentrate-the most commonmeasureof labormarket slack-accountedfor 4 percent in2005 and is low by standards of transitionecon~mies.'~However, this simply meansthat fewjoblesspeopleare lookingfor ajob. Butthe crux ofthe labor marketproblem inUzbekistanis that many personsofworkingage (16-64) do not search for ajob becausethey do not believethat they canfind one. Suchpersons,called"discouraged workers", account for over 7 percentof the workingage populationin Uzbekistan. Accordingto a narrow definitionof unemployment, discouragedworkers are out ofthe laborforce. Butthe so called"augmented unemploymentrate" takes into accountthe discouragedworkers (who are also referredto asjob-wanters)and is a broadermeasure of labor market conditions. The augmented unemployment rate accounts for 11 percent in Uzbekistanand is almost 3 times as high as the traditional unemployment rate. This means that there is a large pool of available workers who are without work. Thus the low unemployment rate masks difficult labor market conditionsand the scarcity of job opportunities. Underutilization of labor in Uzbekistan is also reflected in the low labor force participation rate and, consequently, in the low employment-to-populationratio (Table 2). Only six out of ten workers of working age are employed in Uzbekistan. That is, the employment-to-populationratio is 5 percentagepoints lower in Uzbekistanthan the OECD average. It is still lower than the target of 70 percent adopted by the EU (the so calledLisbontarget), which is met or even exceededin many low-unemploymentEU countries (e.g. Nordic countries, Ireland,the UK). Thus some additional 10 percent of the working age populationof Uzbekistan could work if the labor market were more efficient. This underutilizationof human capital translates into lower output, incomesandhigherpoverty. Table 2 Indicatorsof labor forceutilizationby gender andage. Uzbekistancomparedwith OECD Persons aged 16-64 years (percentages) Uzbekistan(2005) OECD (2004) Unemployment Employment/ Labor force Unemploy Employment/ Labor force rate population participation -ment rate population participation ratio rate ratio rate All workers 15 to 64 4.0 60.7 63.2 6.9 65.3 70.1 15 to 24 6.4 44.9 48.0 13.4 43.2 49.9 25 to 54 3.3 72.5 74.9 6.0 75.7 80.6 55 to 64 2.5 30.2 31.0 4.7 50.7 53.1 Men 15 to 64 3.9 72.7 75.7 6.7 75.0 80.3 15 to 24 6.2 54.5 58.1 13.6 47.4 54.8 25 to.54 3.1 85.3 88.0 5.7 86.9 92.1 55 to 64 3.5 47.0 48.6 5.0 61.5 64.7 Women 15 to 64 4.2 49.3 51.5 7.2 60.1 55.8 15 to 24 6.8 35.6 38.2 13.1 39.0 44.9 25 to 54 3.5 60.4 62.5 6.4 64.8 69.2 55 to 64 0.0 16.1 16.1 4.3 40.4 42.2 Source:URPS 2005 and OECD (2005), Bank staffcalculations. 19 The unemployment rate reponed m llus study WY calculatedusingthe rtandardILOdefinitionafunemplo)ment and datacomingfrom the firs wave of the W S Usingthe ILO definrtionand householdbmeddam allows one 10 obtain internationallycomparabledauan unemplo)ment. We do not use the unemploymen!rate calculaled usingdaucomingfrom the unemploymentregister. sine regismuonis lnnuencedby ~oscnuvcsto reprter (cnutlementr to ~anoussocial benefits)and Ihus Ihe numberof rspslercdunemployed is nota reliableindicator of actualuncmploymcnl. 43 The relativelylow labor force participationin Uzbekistan reflectsnot only labor market conditions but also cultural factors. This is particularlyevident inthe relativelylow labor force participationrate amongwomen (especiallyolder ones). It should be noted, however, that in this respect Uzbekistancompares favorably to other Muslim countries in the region, such as Turkey, where the labor force participationrate is much lower than inUzbekistan(27 percentand 52 percent, respectively). A strikingfeature of the labor market in Uzbekistanis low laborforce participationamong older men. Only 47 percent of men aged 55-64 are employed in Uzbekistan, compared with 62 percent in the OECD. However in Turkey the employment-to-populationratio for men in this age group is the same as in Uzbekistan. The most likely explanation for this pattern is the low official retirementage, which is 60 years for men. Although this is typical of the FSUcountries, it is importantto stress that in most OECD countries and CentralEuropeantransitioneconomies (such as Poland), the retirement age for men is 65. So, the low retirementage effectively reduces labor supply with a likely negative effect on output and not necessarily a positive effect on the labor market. An increase in the retirementage and the associated increase in labor supply generally do not lead to higher unemployment as additional labor supply tends to create additional demand2' A contributingfactor can be the lower skills of older workers which makesthem less competitive in the market. Finally, it may also be the poor health status of older workers, which prevents them from engaging in productiveactivities. Further research is necessaryto determine the relativeimportance of these variousfactors in explainingthe low labor force participationof older workers inUzbekistan. A positive phenomenon is relatively low unemployment and high employment among Uzbek youth. High unemployment among young workers enteringthe market is a serious social problem which bedevils many countries, includingmost transitioneconomies of ECA. But fortunately, Uzbekistanis spared this problem. As many as 45 percent of workers aged 16-24 are employed in Uzbekistan, which is even slightly above the OECD averagee2'This is an encouraging sign, as research shows that youth unemployment tends to be associatedwith poverty. Open unemploymentin urban areas and hidden unemploymentin rural areas Open unemployment in Uzbekistan is largely an urban problem but hidden unemployment is pervasive in ruralareas. Again, this pattern is typical of mosttransitioneconomies, especially those with a large share of agriculture.22The unemployment rate in urban areas, close to 8 percent, is substantiallyhigherthen in rural areas, where it is less than 2 percent (Table 3) However, when one looks at the augmented unemployment rate, the picture changes, and the discrepancy becomes much less. The augmented unemployment rate in urban areas is only some 3 percentage points higher than in rural areas (13.1 percent and 9.7 percent, respectively). This reflectsthe fact that the fraction of discouragedworkers -those who could work but are not lookingfor ajob becausethey do not believejobs are available - is much larger in the rural areas (close to 6 percent) than inurbanones (less than 4 percent). 20 The "hard 10 employ"worker goups maybe mexceplionlo ths rule. And some older workersmay indeedbe "hard10 employ",especially givenlhe changmgrldllprofile ofpbr which IS a salient sharaclcnsucofthe economic Vansition Newly crcalcdjar differ in t e r n oflhe sbllconten1from the oldjobs lhatare kingdemoyed, and older workersmayfind if pvtlcularlydifficult 10 moveacrossjobs 21 OECD 2WS. Emplopen1(hrtlwk Pans,France 22 World Bmk (2005), EnhancingJob Oppanmiuer.EmemEuropeand the Former Sowel Union. Warhin5on DC. 44 Table 3 Openunemployment inUzbekistan is largely an urbanproblembut hidden unemployment is pervasive inrural areas Persons aged 16-64years Urban areas Rural areas Unemployment rate 7.7 1.7 Augmentedunemployment rate a) 13.1 9.7 Employment/population ratio 55.1 64.7 Labor force participation rate 63.3 65.8 Discouragedworkers (% of working age population) 3.7 5.9 Hours of work (per week) 43.2 33.7 Percentage of workers working less than 40 hours per week 26.6 57.0 a) The augmented unemploymentrate takes into account the "discouraged" workers, i.e. jobless workers who do not search forjob becausedo not believejobs are available. Source:URPS2005. Bankstaffcalculations Lack o f labor demand inrural areas manifests itself also inthe high incidence o f part-time workers an inthe short working hours. The proportion o f workers working less than 40 hours per week (part-time workers) approaches 60 percent in rural areas, compared with less than 30 percent in urban areas. In a similar vein, the rural workers work on average 10 hours less per week than urban workers. Underemployment is also reflected in the nature o f jobs in Uzbekistan. Many jobs are in low productivity agriculture, in the informal sector, and o f temporary and precarious nature. This explains the striking fact that many o f the poor in Uzbekistan are the "working poor" (see Table 7 and corresponding analysis below). Over a quarter o f the workforce is employed in agriculture (Table 4), but as we will show later, farmers and agricultural workers are disproportionately represented among the poor. Table 4 Many workers inUzbekistan are employed in low-productivity agriculture or have irregularjobs Workers Workers Industry Industry without with Self- employment employment casual/tempo employed contract rary jobs Percent Nationaleconomy 100 42.3 27.0 16.2 Agriculture 26.6 22.8 35.2 7.3 Industry 8.3 26.0 17.1 11.6 Construction& utilities 3.2 8.3 11.4 4.6 Trade & catering 5.6 78.9 41.1 62.9 Transport 2.2 12.3 4.5 7.0 Business& financial services 29.6 92.2 40.8 24.2 Public and personalservices 21.4 3.2 1.9 5.6 Other activities 3.1 39.6 31.3 15.9 Source:URPS2005, Bank staff calculations. 45 Large informal sector A substantial fraction of the workforce is employed in the informal sector, lackingjob protectionand the benefits of formal sector employment. In addition, in low-incomeCIS countries - and Uzbekistanis not likely to be an exception - informalsectorjobs tend to be of lower productivityand provideonly subsistence income.23 Using the lack of employment contract as the criterion of informality, the informal sector is estimated to account for over 40 percent of total employment in Uzbekistan (Table 5). The informal sector not surprisinglydominates in trade and catering (nearly 80 percent of total employment) but also in business and financial services (over 90 percent of industry's employment). In a way "business services'' is just another name for informal employment and is a "catch-all" category comprising various poorly defined business activities. And this is a dominant industry in Uzbekistan, which accounts for 30 percent of total employment. The informal sector provides quite different types of employment opportunities. Some workers and firms preferto operate in the informalsector in order to avoid and evade burdensomeregulations and taxationand to earn higher incomes. Others are forced to operate in the informal-fee entry - sector because of the lack of productivejob opportunities in the formal sector. For them the informal sector is an employer of last resort and a source of subsistence income. In other words, some workers are pulled into the informal sector by better earnings opportunities, while others are pushed into it by lack of formal sector work. Unfortunately,becauseofthe lack of relevant datawe are not ableto determinethe relativeimportanceofthe "pull" and "push" factors in Uzbekistan. To discriminate between the two factors one would need to compare earnings in bothsectors, or to examine the movementso fworkers betweenthe sectors. Neithertype of data is available for Uzbekistan. But evidence from other low-income CIS countries suggests that informal employment is largely an escape from unemployment and poverty, and not a source of higher incomes(Verme, 2006). It seems likely that the same patternprevailsin Uzbekistan. Informal sector employment has been driven by high taxes on labor and by low expected benefits of formality. Until 2003 the tax wedge on labor was 44 percent, which was close to ECA average, but highfor a low-incomecountry.24Since then the tax wedge has been gradually reduced and now (2006) accounts for 38 percent of labor costs. The 6 percentagepointsreductionin the tax wedge representssignificantprogress in improvingincentivesfor formal employment. However to entice more firms and workers to moveto the formal sector the benefits of formality need to be strengthened, too. On the firm side this includes, for example, improvingaccess to credit and foreignexchange. Onthe worker side this includes buildingtrust in the pension system. The actual ension replacement ratio is currentlyonly about 30 percent (MOLSPdata), down from 60 percent in 2000!5 This has lessened the opportunity costs of informal employment. The recent introductionof the second pension pillar may help to reverse this trend and strengthenthe incentives for workers to seek formal sector jobs. In short, reducinginformality requires lowering the costs of doing businessinthe formal sector (see Part2 onthe investmentclimate). Many workers have only temporary or casualjobs, which is associatedwith employment insecurity. This is in a strong contrast to the pre-transitionperiod, when regularjobs were the norm, andworkers were enjoying a high degree of employment protection. However the changing nature ofjobs and the shift from regular jobs in the formal sector to irregularjobs in the informal sector is characteristic of virtually all transition economies (World Bank 2005). Obviouslythe intensityof this processvaries from countryto country. This shift tends to be more pronounced in low-incomecountries than in more developed transitioneconomies of CEE. InUzbekistanthe incidenceof irregularemployment is relativelyhigh: one worker in four has a casual or temporaryjob (Table 4). Naturally, the high incidenceof irregularemploymentto a large extent overlaps with a high incidence of informalsector employment. Thus casual employment is most commonintrade and cateringas well as in business services, accountingfor as muchas 40 percentoftotal employment. 23 Vcrme, Paulo (2w6) "Consumntsto Oroulh and Job Crsauon in Low-incomeCIS Countnes."Policy ResearchWorkingPapsr NO.3893. The World Bank Washington DC, and alw World Bank (2W5) 24 The tax wedge On labor was calculatedas lolal labor cost 10 employer mnusnet employeewags, expressed as B percentageof laborcosu In 2w6 lalal labor cos1was 125 percenlof the posr employee wagc The net emplo)ee wage was 77 5 prsenl of the gross wags (20% income tax and 2 5% social xcunrycontribution) The tax wedge was calculated for a worker carmng the average wags whxh 8% taxes ma(nomind) rate of 20% 25 World Bank (2W3),UzklurlanLiwng Standards Arserrmenl Policies10 Improve Living Standards, WaslungtonDC 46 Self-employment has becomea frequent alternativeto dependent employment intransitioneconomies, and is often regarded as a sign of a developing entrepreneurial spirit. In Uzbekistan the incidence of self- employment is less than the average for low-incomeCIS countries. Accordingto URPS, the self-employed account for 16 percent of all workers, although according to official statistics, they reach 25 percent. But in any case this is substantially less than the low-income CIS average of 50 percent. Whether this low figure indicates barriers to entrepreneurshipis an openquestion. Informal employment is prevalent not only in rural but also in urban areas. Expectedly, the incidence of informal employment is higher in rural areas, where roughly every secondjob is in the informal economy (Table 5). But even in urbanareas one worker inthree has an informaljob. The profile of informal sector employment differs from that of formal sector employment. Informalsector workers tend to be younger, less educated and less skilled than formal sector workers (Table 5). They predominantlyhold manual or farmingjobs and tend to work part-time. Interestingly,there are no gender differences inthe incidenceof informalemployment. For example, young workers (16-24 years of age) hold close to one-thirdof all informaljobs and less than one-fifthof formaljobs. Workers with secondary general education hold 60 percent of informal jobs and only 40 percent of formal jobs.26 And the higher the educational attainment, the lower the incidence of informal sector employment: a worker with less than secondary education is four times as likely to work in the informalsector as a worker with tertiary education (Figure 2). Farmers and unskilled manual workers are less likely than other occupations to have a formal sectorjob andjointly account for three-quartersof informalemployment. 26 The hghmcidenceof informalemployment(and underemployment)amongworkersullhsecondarygeneraleducationshould not be lnfcrprnedBI anupmt against the usefulnew ofthistypc of educauon,especiallyansompansan with rssondarytechnical education Workers who completedgensml rewndaryeducationfall intotwo groups:those who proceedto teniary education.andlhose for whomthnIS the endofformaleducauon W e latter poup wmpnrsr workers ulth no specific vocational skills. Were workers ,deedfind 11 difiisultlo find formalempiaymsnt. But they are not representativeof all those, who completedgeneral education,an issuedb o w as lhe seiec11oobias See Chapter 4 on education for more informationon demandfor differenttws of educationin Uzbebslurm See also Yemlbov,Mete and Cnobloch (2006)on labor market outcomes for differenteducationalgroups m ECA One mainfinding of this study is that m general the pnvate sector valuer generalscconm graduatcr more ulanvocational secondary graduatcr 47 Table 5 Formaland informal sector employment: compositionand incidence, 2005 Formal sector Informal sector a) Composition Composition Incidence Percent All workers 100.0 100.0 42.3 Location Urban 40.9 31.8 36.6 Rural 59.1 68.2 46.I Gender Men 58.3 58.4 42.4 Women 41.7 41.6 41.6 Age16- 24 18.5 31.1 55.2 25 -54 77.2 64.3 37.9 55 + 4.3 4.7 44.4 Educarion Basic (grades 1-9) 5.6 12.8 62.7 Secondary general 39.9 59.3 52.2 Secondary technical 11.9 10.3 38.9 College 18.2 11.7 32.1 University 24.5 5.9 15.0 Occupation Managers, officials and professionals 9.5 I.o 7.0 Techniciansand assistants 22.6 1.6 4.9 Clerks and service workers 11.8 8.0 33.5 Skilled manual workers 13.1 13.4 43.1 L'nskilled workers 16.1 22.2 50.5 Farmers 27.0 53.8 59.6 Working rime (hours per week) less than 20 10.1 28.1 68.7 20 - 39 19.9 36.1 58.6 40 r 70.0 35.5 28.4 a) Workers without employment contract Source,URPS2005. Bank staff calculations. Figure 2: Less educatedworkers are more likely to work inthe informal sector Incldenc. of Informal S . 0 1 ~ Employment by Edwatlon 2006 Source: URPS 2005, Bankstaff calculations. Interestingly, most - two out o f three - of informaljobs in Uzbekistan are part time (less than 40 hours per week) and informal sector workers on average work less than their formal sector counterparts (33 hours against 41 hours per week). This is in contrast to the pattern prevailing in other countries, where informal 48 employment tends to be associatedwith longer working hours (Avirgan and others, 2005).*' This is largely due to the fact that a substantialpercentageof informalsector workers are farmers, ofwhom majority-close to three-quarters - work less than 40 hours per week. Thus short working hours in the informal sector are likely to be a symptom of hidden unemployment, rather than reflect voluntary welfare maximizingchoices ofworkers. To summarize, given the scarcity of job opportunities in the formal sectors workers in Uzbekistan have resorted to various coping strategies. They include employment in subsistence agriculture, in the informal sector, casual jobs and finally self-employment, However the latter is much less prevalent in Uzbekistan than in other low-incomeCIS countries, which may suggest that the entrepreneurial spirit has not developed yet in Uzbekistan. In additionto agriculture,the two most important industries which provideemployment opportunitiesfor workers lackingregularjobs are trade and catering, andbusiness services, which comprise a wide range of notwell defined moneyearningactivities. Educated workersface better labor marketprospects Workface in Uzbekistanis well educated. Less than 10 percent of working age populationdoes not have secondary education, and about 17 percent has university education, which is impressive by standards of developing economies. Employer based surveys indicate that there is no shortage of skilled workers and there does not seem to be a ``skills gap" in Uzbekistan (see Part 2 on the investment climate). Unemployment in Uzbekistan is thus mainly an issue of insufficient labor demand, not of inadequate labor supply - While labor market prospectsof highly educatedworkers are good, those of less educatedones are poor. As in most other transitioneconomies, there is substantial demand for high skills in Uzbekistanbut apparently low demandfor simple manualskills. Accordingly, low skilledworkers are more likely to be unemployedor out of the labor force, and are more likely to be underemployed(work short hours). The employment-to- populationratio for workers with basic education is only slightly above 40 percent (Figure 3). This is very low: after all the majorityof workingage personswith basic educationare without work. Incomparison, the employment-to-populationratio among workers with secondary vocational/technicaleducation approaches 70 percent. But those who do work tend to be underemployed: over half of workers with basic education work less than 40 hours per week; and the average hours worked per week is 20. Again, for workers with secondary vocationaVtechnica1education the percentageof part-timeworkers is well below 40 percent, and the average hours worked per week is over 50. This data clearly show that job opportunities for workers with less than secondary education are few.28This is important, as less educatedworkers face higher than the averagepovertyrisk and are preponderantamongthe poor. Underemployment - although not open unemployment - is also pervasive among workers with secondary general education. In this category, one worker in two works less than full-time, and average hours worked per week is only slightly more than 20. But given the relatively high employment rate for this educational category, it is not clear to what extent short workinghours reflect voluntary welfare maximizingchoices, and to what extent lack of labor demand. Another noteworthy fact is relatively poor labor market prospects of graduates of post-secondary (non- university) schools. Only half of persons of working age in this educational category are employed, and the incidence of part-timework is relativelyhigh(40 percent). This either suggests that this type of education is not particularlyvalued by employers, or that labormarket attachmentofthis worker group is ratherweak. 17 Avirga Tony,L.Josh Biveni and Sarah Gammge. eds (2005). GoadJobs. BadJobs. NOJobs Labar Marketsand Informal Work 8" Eglpt.El Salvador. India, Russiaand South Afnca. Economic Pollcy Insuae. WashngianDC 28 Gbmously.low educationalattainmentIS likelyto coincide with wal locationand agricultural emploment. 49 Figure 3: Unemploymentandunderemploymentis particularlyhighamong less educatedworkers PanelA Unemployment Rate and EmploymentlPopulation Ratio by Education 2006 80 c 7.0 70 0 6.0 60 0 5.0 50 0 4 0 i( 400 # 3.0 30 0 2.0 20 0 100 1.0 0 0 0 0 Basic (grades 1-9) Secondary general Secondary technical College University I I Employmentlpopulation ratio +Unemployment rate1 PanelB Workers working less than 40 hours per week and average hoursworked 2006 60 50 40 37 X 30 36 c 20 10 0 Basic (grades 1-9) Secondary general Secondary technical College University iID/, worK ng less inan 40 n o m +avg no,rs womea Note: population ofworking age (16-64) Source: URPS 2005, Bank staff calculations 50 Unemploymentis of short duration and affects mainly secondary earners Unemployment in Uzbekistanhas two important features which limit its social cost. First it is of short duration. Second, it is concentrated among secondary earners. These two features explain the seeming paradox that in Uzbekistan - unlike other countries - unemployment does not translate into poverty (see below). The median duration of (uncompleted) unemployment spells is 6 months in Uzbekistan, much less ,thanin most other transitioneconomies ofECA, especially in CEE. Accordinglythe share of long-termunemployed (those unemployedfor morethan a year) is very low (Figure4). Only around 5 percentofthe unemployed in Uzbekistan are searching for work for more than a year, compared with 40 to 60 percent in other ECA countries. Even inTurkey, where the labor market is quite dynamic, the share of longterm unemploymentis over 20 percent. Short duration of unemployment is a highly positivephenomenon because (a) workers are deprived of earnings only temporarilywhich helps them stay out of poverty, and (b) they do not lose skills, which facilitates their finding a new job. In other words, unemployment does not seem to lead to social marginalizationwhich mitigates its poverty impact. However, it is an open question what proportion of those who exit unemploymentfound ajob, and what become discouragedand withdrew from the labor force. Ifunemploymentendswith findinganewjob this indicatesa dynamic, buoyant labormarket. But ifinstead it ends with labor force withdrawal, this indicatesa depressed labormarket. This is an importantissue which requiresfurther research. Figure 4: Unemploymentspells are relativelyshort inUzbekistan Unemploymentby duration (months) 1 D1.6 0 6 - 12 0 1 2 tI Source: URPS 2005, Bank staff calculations. Unemployment seldom strikes household heads. Instead, the unemployed tend to be secondary earners. Some 60 percent of the unemployed are sons and daughters of the household head, i.e. family memberswho provide supplementaryrather than main income. Some additional 15 percent of the unemployedare spouses of the household head. Thus only one unemployed in five i s a household head. This limits the poverty impact of unemployment: it is the loss of ajob by the householdheadwhich raises the risk of poverty the most. Inflows into unemployment seem to be pretty high. Usingthe fraction of workers who are unemployed for less than one month as a proxy, we find that the monthly inflow into unemployment is about 10 percent. 5 1 This is significantly more than in transition economies of CEE, where the inflow rate into unemployment varies from 3 to 6 percent. Such a high rate may indicate either intensiveenterprise restructuring,or high worker turnover associated with the temporary and seasonal nature ofjobs. Given the high incidence of informal and casual jobs in Uzbekistan (see above) as well as rather slow progress of the transition, the second explanation seems more likely. In either case, unemployment in Uzbekistan seems to be a high turnover pool, which is in a sharp contrast to transition economies of CEE, where unemployment is a stagnant pool. This high worker turnover may raise the overall unemployment rate, but as already mentioned, is a positivefeature ofthe labor market inUzbekistan. Employment Ofjlces serve only afraction of the unemployed Few of the unemployedregister with the EmploymentOffice. Accordingto URPS, only 5 percent of those who do not have ajob and are lookingfor one are registeredas unemployed. There are two possiblereasons for this. First, the short durationof unemployment spells may meanthat it does not pay to register. Second, the benefits of registration - services provided by the Employment Office and benefits, such as unemploymentbenefit or entitlement to social assistance-may be limited. In fact, the unemployed seldom use job search assistance provided by the Employment Offices. The majority of the unemployed- close to 60 percent - look for a job through friends and relatives. Only 17 percent uses as their mainjob search effort assistance providedby the Employment Office. The third most important job search method is directly contacting employers, which is used by 13 percent of the unemployed. Persons who are registered with the Employment Office are genuinely unemployed. That is those who register meet the two maincriteria for unemployment:they do not have ajob and are lookingfor one. This is in contrast to the situationinmany countries, where the numbers ofregisteredunemployedare inflatedand many persons who are not interested in obtaining a job register only to get entitlement to certain social benefits. EmploymentOfficesinUzbekistan serveonly the truly needy. But it appearsthat not all those who needjob search assistance andcounseling receive it. Ifthis is indeedthe case, then there is scope for improvingthe functioning of the EmploymentOffices and makingthem more effective at matching the unemployed with jobs. The key to the efficacy of Employment Offices is their ability to acquire job vacancies from the employers and offer them to the unemployed. If Employment Offices can offer few job vacancies then their services are of little use to the unemployed. In other words, job searchassistance is effectively only whenjob openings are available. Ifthe effectivenessofEmployment Offices in Uzbekistanis limited by the fact that they do not have information on available vacancies then measures needto be taken to increasethe vacancy penetrationratio, i.e. the proportionof available vacancies that is reported to the Employment Office. There are different methods of doing so, includingdeveloping better contacts betweenthe Employment Officeandthe businesscommunity. Many workersmove in search of jobs, mainly to other CIS countries Many Uzbeks are believed to work abroad -mainly in Russia and Kazakhstan -in search o fjobs and better earningopportunities. However, partlyillegalnature of Uzbekmigrationmakesit difficult to precisely estimate its magnitude. According to some estimates external and internal migration accounts for 8 to 10 percentofthe laborforce, which meansthat 800 thousandto 1million people migrated in2005 driven by the need to improve their standard of living (Maksakova, 2006). This indicates considerable migration flows, with a significant impact on local and the national labor markets. External migration helps to reduce the excess supply of laborwhile internalmigrationhelpsto equalize locallabor marketconditions. According to URPS data, the magnitude of migration in search of work in is smallerh Uzbekistan. The overal migration is estimated at around 10 percent of the working age population, which is consistent with the estimates quoted above. However, less than three percent of persons of working age left their household to look for ajob in a different location(Table 6). This figure may be biaseddownward, as respondentsmay be reluctantto revealthe true motiveofmigration,given that most migrants work illegally. 52 Consistent with other sources, the majority (56 percent) of labor migrants search for jobs abroad, almost exclusively in other CIS countries (Table 6).29 Internal labor migration is somewhat less frequent. Labor migration, especially internal, is mostlytemporary or seasonal. Less than 10 percent of all internalmigrants leave their place of residence for more than six months. In contrast, 30 percent of internationalmigrants leave their place of residence for more than six months. But still international labor migrationis largely of short-termnature. To conclude, there is evidence that migration is an importantphenomenon in Uzbekistan, influencinglabor market conditionsandthe standards of living. Migrantsmoveto where thejobs are, earn higher incomes and support their families by sending remittances. But precise estimates of the size o f migration flows in Uzbekistanare difficult to obtain due to largely illegal, or at least informalnature of migration. Thus labor market effects of migrationinUzbekistanmeritfurther research. Table 6 LabormierationinUzbekistan.2005 Migration Total migration 10.4 Percentageofworkingage population Labor migration Percentage ofworkingage population 2.9 Percentage ofthe laborforce 3.7 Internal labor migration Percentageofall migrantsofworkingage 6.6 Percentage ofall labormigrants 44.2 International labor migration Percentage of all migrantsofworkingage 7.7 Percentageofall labormigrants 55.8 Labor migration for more than 6 months Internalmigrants(%) 8.6 Internationalmigrants e!) 31.6 Notes: Migrantsare persons who were absent from the household in the last 12 months for one month or longer. Labor migrants are migrants working either in another part of the country or abroad. Datarefers to populationof working age (16-64). Source:URPS2005, Bank staff calculations. Labor Market Status and Consumution Inthis section we examine the relationshipbetweenboththe individual's and the household's labor market status and the consumption level. The individual's labor market status comprises the labor force status (employment, unemployment, inactivity) and job characteristics. The household's labor market status comprisesthe number of earners inthe householdandtheir earnings. As to the consumption status, we focus on the bottom quintile (bottom 20 percent) of consumption distribution. For short we call those in the bottomquintileas "relatively poor". Hence, we look at the estimatedprobabilitythat a workers or household with a given labor market status falls intothe low consumptioncategory, or inother words, is relativelypoor. Incidence of low consumption isparticularly high among the discouraged workers, rather than among the unemployed Surprisingly, unemployment does not elevate the risk of low consumption in Uzbekistan. To the contrary, the unemployedinUzbekistan are less frequently inthe low consumptiongroup thanworkers withjobs. The labor force group which faces the highest risk of low consumption is discouraged workers, that is workers who would like to work, but are not looking for a job because they don't believe that jobs are available. 29Russiais most probably the maindestination,but we lack statisticaldata to confirmthis conjecture. 53 Table 7 illustratesthis pattern. Only 12 percent of the unemployed are in the bottom quintile of per capita expenditure distribution. For the employed workers the risk of low consumption is 5 percentage points higher. And it is as much as 18 percentage points higher for the discouraged workers. So it is the discouragedworkers, ratherthanthe unemployed, who are the most disadvantagedworker group. The high incidence of low consumption among the discouraged workers is a bad news from a policy perspective. This is becausethe labor market attachment of the discouraged workers is weak, weaker than that of the unemployed. The unemployedhope to find work and look for a job. The discouraged workers don't. Thus it takes more efforts to bring them back to the labor market and to encourage them to start searchingfor work. The bigpolicychallenge is to restore hopeamongthis group, which can be done only be creatingmorejob opportunities. Why is it the unemployed who face the least risk of low consumption? This apparent paradox can be resolved by recallingthat the unemployedtend to be the secondary earners and that unemployment is rather rare amonghouseholdheads (see above). Put differently, the unemployed in Uzbekistanare often members of well-off familieswho in a way canaffordto be without ajob. Incontrast, inpoor familiesjoblessness is a non-affordable option. Poor workers are compelledto take anyjob to earn subsistence income. Table 7 Labor force status and the risk of low consumption Incidence Share in Laborforce status o f low low consumpti consumpti on ongroup Percent Employed 17.7 55.5 Unemployed 12.2 1.6 Discouraged 29.9 7.7 Out of the labor force 17.0 35.3 Note incidence of low consumption=percentageof personswith a given characteristicswho are in the bottomquintile of householdper capitaexpenditure distribution Source URPS 2005, Bank staff calculations. Agricultural workersface thegreatest risk of low consumption Doesthe type ofjob affect ones income and consumption? Are somejobs associatedwith higher risk of low consumption than others? What are the "bad" jobs, i.e. jobs that do not protect from falling into the low consumptioncategory? And what are the "good" jobs? The single most important job characteristic that is closely associated with low consumption is its rural , location. Accordingly farmers and other agriculturalworkers face a much higher risk of low consumption than other workers. Table 8 shows that the incidence of low consumption(bottom quintile) is only 5 percent among workers with urban jobs and as much as 26 percent among workers with rural jobs. And most tellingly, the bulk ofthe workingpoor-90 percent-have ruraljobs. The comparison between farmers and unskilled manual workers is also illustrative. The incidence of low consumptionis closeto 30 percent amongthe former and less than 20 percent amongthe latter group (Figure 5, Panel A). Finally, 30 percent of agricultural workers are in the low-consumptiongroup compared with less than 10 percent of industrialor trade workers (Figure 5, PanelB). Rural employment does little to protect workers from poverty which is closely linkedto low productivityof ruraljobs. The way to lift rural workers out of poverty is thus through raisingagriculturalproductivityand throughgenerating off-farmemployment opportunitiesinrural areas. Migrationto urbanareas is yet another option. It should be noted however, that the growth in agricultural productivitytends to be associatedwith employment reductions, which may contribute to unemployment unless the expansion of other industries absorbsredundantagriculturallabor. 54 Table 8 The risk of low consumption byjob characteristics, 2005 Job characteristics Incidence of low Share in low consumption consumptiongroup Percent Education Basic (grades 1-9) 22.3 10.9 Secondary general 24.8 68.3 Secondary vocational/technical 15.0 9.5 College 9.4 8.3 University 3.4 3.1 Occupation White collar 7.8 12.3 Skilled blue-collar 12.0 8.9 Unskilled blue-collar 16.9 17.8 Farmer 28.2 61.0 Location Urbanjob 4.6 9.7 Ruraljob 25.5 90.3 Job type Dependent 11.8 33.3 Self-employment 10.6 10.2 Agricultural 25.8 56.4 Formal/informal job a) Formal 16.8 54.2 Informal 18.9 45.8 Permanentkemporary Permanentjob 16.3 67.2 Temporaryicasualjob 21.4 32.8 Firm ownership State administration 6.7 1.2 State owned enterprise 9.8 13.2 Collective 30.8 37.7 Private 11.8 12.8 Individual 13.1 11.1 Other 24.8 24.1 Industry Agriculture 30.1 44.8 Industry 9.2 4.3 Construction and utilities 15.8 2.8 Trade and catering 6.1 1.9 Transport 4.1 0.5 Businessand financial services 19.3 32.3 Public and personal services 9.8 12.1 Other activities 7.6 1.3 Hours worked (per week) less than 20 18.7 21.7 20 - 39 18.8 33.0 40 and more 12.6 45.3 a) Employment contract Note: Incidenceof low consumption= percentage of persons with a givencharacteristicswho are in the bottom quintile of householdper capitaexpendituredistribution. Source: URPS 2005, Bank staff calculations. 55 Interestingly,the incidence of low consumption is quite high amongworkers employed in the "business and financial services" industry (Figure #, Panel B). In most advanced transition economies these are high paying activities being part of a modern economy, but in Uzbekistan this industry apparently comprises mostly traditionalsmall scale business services, which often provide only subsistence income. The outlook for this industry is important from the viewpoint of poverty reduction as it employs about one-third of all workers. Whether the modern business services sector develops soon in Uzbekistan, and whether workers currently employed in this sector will be able to make a transitionto the modern sector remains an open question. Figure 5: Agricultural workers face by far the highest risk of low consumption, but workers inthe "business services" industryare also likely to be relatively poor PanelA Fraction of Worken in the BottomQuintile of Par Capita Expenditure Distrlbutlon by Occupation 2006 30 25 20 # 15 i o 5 0 While collar Skilled blue-collar Unskilledblue-collar PanelB Fraction of Worken in the Bottom Puintlie of Per Capita Expendltun Distribution by Industry of Employment 2005 35 30 25 20 # 15 10 5 0 Agriculture 8~~10888 and Construotlon and Public and Industp/ Other sctlvltles Trade and Transwn financial sewices Utilities personalsewice8 catemg Source: URPS 2005, Bank staffcalculations. 56 Workers with privatesectorjobs are somewhat more likelyto be relativelypoor thanthose with public sector jobs. But the differential is not substantial: the proportion of private enterprises workers who are in the bottom consumption quintile is some 12 percent, compared with 10 percent of SOEs workers. This may reflect the nascent character of the private sector in Uzbekistan, which so far has developed largely in less productive branches of the economy. An important issue, which needs to be further explored, is to what extent the apparently higher earnings in the SOEs reflect higher labor productivity, and to what extent rents appropriatedby workers. Informal sector jobs are more often associated with relative poverty than formal sector jobs. This is an expectedresult, as the informal sector tends to be perceived as offering "bad", low-payingjobs. However, the difference in the incidence of low consumption between those two types of jobs is small, only 2 percentagepoints. This suggests that contrary to the commonview, informal sectorjobs are not necessarily inferiorto formal sectorjobs, andthe "job quality" differentialmight infact be less than oftenassumed. Interestingly, on average there is little difference between self-employment and dependent employment in terms of the associated consumption level. In other words, self-employment neither provides an income advantage nor a disadvantage. Accordingly, based on the data at hand one cannot determine ifworkers are pushedinto self-employment by lack ofjob opportunities, or pulledby highexpected earnings. Expectedly, the consumption status of workers with casualjobs is markedlyworse than that of workers with regularjobs. The incidence of povertyamongthe former is 5 percentagepointshigherthen amongthe latter. This supports the widespreadperceptionofcasualjobs as "bad jobs". Earlier analysis suggests that part-time jobs in Uzbekistan are more likely to reflect lack of full-time employment opportunities than voluntary choices by workers. The relatively high incidence of low consumption among part-timeworkers supports this supposition. Close to 20 percent of workers working less than 20 hours per week are in the bottom consumption quintile, compared with 13 percent of full-time workers. Finally,jobs that require secondary general education more often are associated with relative poverty than other jobs. Secondary general education leads to a relatively weak labor market position. As many as 25 percent of workers with secondary general education are in the bottom consumption quintile, comparedwith 15 percent of workers with secondary vocationaVtechnica1 education. However, one needs to take into account that secondary general education is associated with other job characteristics, such as occupation, industryofemployment, etc., which may be equallyor more importantindeterminingthe income status than educationalattainmentper se. To sum up, job characteristics have a significant impact on the worker's income status. Jobs in agriculture and, more broadly, in rural areas are often associatedwith relative poverty. So are the casualjobs in the informal sector as welljobs in the "business services" industry, On the other hand, "good jobs", i.e. those associated with relatively high consumption, are located predominantly in the public sector (state administration,SOEs) in such industries as manufacturing, trade and transportation. This demonstratesthat poverty reduction in Uzbekistan will require a shift from low productivity agriculture towards more productive industries, especially towards modern services sector. Moreover, it will require the informal sector firms and jobs to move to the formal sector, which hinges on the improvements in the business environment. Finally, it will requirethe development of a modernproductiveprivate sector, which pays for productivityand adequately rewards its employees. Insome cases this may also require better enforcement ofkeyworker rights. Poverty is more related to earnings of householdmembers, less so to labor supply decisions Household's labor income is determined by two factors: (a) the number of earners, and (b) their average earnings. The surprisingfinding of our analysis is that the number of earners has a rather weak influenceon the households' income status in Uzbekistan. By implication it is average earnings of household members which are a decisive factor. These results are at odds with those for European economies, including 57 transition econ~mies.~'In most European countries families with two earners are virtually free of poverty. This is notthe case inUzbekistan. The relationshipbetweenthe number of earners in the householdand the risk of low consumption is rather weak in Uzbekistan. The increase in the number of earners in a household lowers the risk of low consumption but only weakly and not monotonically (Figure 6). Expectedly, the risk of poverty is the highest in families with no earners and the lowest in families where most or all family members havejobs. But among working families consumption not necessarily increases with the number of earners (relative to family size). For example, families where all family members of working age are employed on average are not richer than families where only 3 out of 4 family members of working age are employed. And importantly, among rural households the incidence of poverty is high even if all household members are employed. In rural areas one person if four is in the bottom quintile of expenditure distributiondespite the fact that all household members are working. Thus, employment not necessarily protects families from poverty,especiallyinruralareas. Poverty in Uzbekistan is related not so much to the lack of employment, but to low-productivity employment. Accordingly the way out of poverty for Uzbek families is finding more productive employment, rather than through increasinghousehold's labor supply (Le. through increasing the number of earners). This entails increasing the productivity of rural jobs, but also movements from rural to urban employment. It is better - more productiveand higher paying--jobs that are neededto lift families out of poverty. Why does increasing the number of earners within the household have little impact on its consumption status? There are a number of possible explanations, which are complementary rather than mutually exclusive. First,there is likely to be a negativecorrelationbetweenthe household's earningpotentialandthe number of earners. Households whose earning potential is low (e.g. due to low skills) compensate this by increasing the number of earners. Incontrast, households whose earningpotentialis highcan affordto have fewer earners. In other words, labor supply decisions of poorer households are driven by the income effect, while those of richer ones are driven by the substitution effect. Second, there is likely to be a negative correlation between the number of earners in a household and the effective labor input measured by the number of hours worked. This pattern is particularly pronounced in rural areas, where most household members of working age are employed (see below), but tend to work short hours (see above). Third, wages account for only about 30 percent of income in Uzbekistan, which lessens the impact of employment on income. One reasonfor this is a substantialpart playedby remittances sent by migrants. Finally, inthe case of ruralhouseholds income i s oftenmore determined by the size ofthe available plot and less by the number of people working on the plot. This especially refers to small subsistence farms (such as dekhan and LPKh farms). The point that it is job productivity rather than employmentper se that matters most for the household's income status is best illustratedby the urban-ruraldivide. On averagerural householdshave more employed members than urban households. Nonetheless, poverty incidence is much higher among rural than urban households. To illustrate, the proportionof households where more than a half of household members of working age (16+) are employed is 57 percent in ruralareas and only 36 in urban areas. At the same time, a rural householdwhere all family members are employed is about four times more likely to be poor than an urban householdwhere only one person out of four persons of working age is employed. Obviously, this dramatic difference in the poverty incidence betweenrural and urbanhouseholds is explained by differences injob productivity,not by differences in householdlabor supply. 30 Mani k e and Gene Verbin (1991),"Law-paid work the household msomepackageand pveny. A nprs-counyanalysii,''Paperpresentedat the Ewopcan Low.Wage EmploymentR e w c h Network Conferenceon the Analysis of Low-wage EmplaymencLondon, Desemkr 12.13: Rutkowr~, (19991,'"LabarMarketsand Poverly in Bulgaria". SP DiscussionPaper NO9918. World Bank W&ngton Jan DC 58 Figure 6: The number of earners in a householdis not a goodpredictorofthe household'sconsumptionstatusinUzbekistan Proportion of PersonsInthe BottomQuintiie of Per Capita ExpenditureDi8t1IbutlOn by the Number Of Earnersin a Hou8ehold 2006 45 40 35 30 25 8 20 15 10 5 0 None 1 in 4 2ln4 3 In4 All Number of earners relative lo the number of household members of worklng age Source: URPS 2005, Bank staffcalculations. Another paradoxicalfinding is that households with an unemployed household member are less, not more likely to be relatively poor than households where there are no unemployed. The incidence of relative povertyis 6 percentagepointshigher in householdswhere there is no unemployedpersonthan in households where a household member is unemployed. This is again at odds with the pattern observed in other countries, where unemployment of a householdmember greatly increases the risk of poverty. But this result should come as no surprise, since we have already seen that unemployment in Uzbekistan, concentrated among secondaryearners, not necessarily translates intopoverty. The bottom line is that labor supply - the number of earners in a household - has little impact on the household's consumption status. By implication it must be productivity, reflected in earnings, that has a decisive effect. Ifso, then the key policychallenge for Uzbekistan is not only to foster the creation of more jobs, but also the creation of "good", that is more productivejobs. This entails both sectoral shifts ofjobs and workers as well as improvements in productivity within industries and firms through enterprise restructuring, investment and technologicalprogress. But the necessary foundation for productivitygrowth i s a favorable investmentclimate, which encouragesenterprise restructuringand investment. 59 PART2: THEINVESTMENT CLIMATE ANDJOB CREATION A key to poverty reduction is the creation of more and better - more productive -jobs. This in turn depends on the creationof and growthof firms. For employment to increase and unemploymentto fall more firms needto enter the market and existing firms need to have incentivesto invest, expand and hire more workers. A set of factors which influences firms' decisions regardingenteringthe market, capitalformation, output and employment is referred to as the "investment climate". A favorable investment climate encourages firm growth andjob creation. In contrast, an inhospitableinvestment climate deters new firms from enteringthe market and discourageshiringnewworkers. There is thus a direct link betweenthe quality of the investment climate, employment and poverty. In the longer run, improvements in the investment climatetranslate into morejobs andless poverty. A favorable investment climate is particularly important during the periods of intensive enterprise restructuringand massivereallocationofjobs and labor, such as the economic transitionthat is under way in Uzbekistanand other economies of Centraland Eastern Europe. First, the transition is associatedwith high rates ofjob destruction, which reflectsthe eliminationof old low-productivityjobs. Accordinglythe rates of job creation need to be commensurately high, so that the "new" sector can absorb workers displaced in the "old" sector. Second, the transitionis associatedwith downsizing, that is firms sheddingoff redundant labor and eliminatingoverstaffinginheritedfrom the communist past. However, ifmost firms downsize, then the number of firms needs to increase to offsetjob losses resultingfrom firms cutting on employment. In this section we examine the main obstacles to firm growth in Uzbekistan and examine how the investment climateaffects firm restructuringand, consequently, empl~yment.~~ The economy of Uzbekistan grows at a high pace (see Chapter 1). Yet despite the strong economic growth the rate of job creation is low and productivejob opportunities are scarce. The employment intensity of economic growth is low inUzbekistan, the phenomenonalso knownas "jobless growth". Joblessgrowthis a result of two factors: (a) widespread downsizingby firms, which shed redundant labor, and increase output by improvingproductivity, and (b) limited creation of new firms and thus the limited capacity of the new sector to absorb labor displacedinthe old sector. The low employment content of growth, such as that in Uzbekistan, is characteristic of an early stage of economic transitionand of slow reforms. At the early stage of the transition the "old" sector is downsizing while the "new" sector is still weak and creates relativelyfew jobs. Most firms are engaged in "defensive restructuring" that is improvingproductivitythrough shedding off redundant labor (WorldBank, 2005). The economy of Uzbekistan falls into this category, as attested by the EBRD index of the progress of transition (Figure 1). Only once Uzbek firms beginto engage in "strategic restructuring", that is turning productivity gains into investments so as to gain market share and expand, the employment content of growth will increase, alleviatinglabor markettensions. 3 IInOUT analylis of firm renruclunngand Ihs mnvcrtmentcIunale we use the resultsa specialfirm b dn w e y ofthe business environmentm ECA,referredIO a BEEPS 111Since BEEPS nl w samsd out ~n2005 some major reforms 10 investmentclimate have been put m placs These are rummanredm Box I. 60 Figure 1: Uzbekistanhas beena slow reformer EBRD index of Progressof Transition, 2003 2 -15 1 -0 5 0 0 5 1 Daviatlonsfrom riplon average Source: Bank staff calculations. Majority of firms inUzbekistan are doingwell and increasedsales over the last three years (Figure 2A). But at the same time, many firms reduced employment and few firms hired additional workers. Interestingly, employment reductions occurred even in firms which expanded their output (Figure 2B). This implies that firms in Uzbekistanrestructure by improvinglabor productivity and eliminating overstaffinginheritedfrom the era of central planning, ratherthan expanding and increasingtheir market share. Box 1 Thecosts of doinn business in Uzbekistanhave been simificantlv lowered in recentyears but there is room for imurovement The investment climate assessment presented in this study is primarily based on the results of the BEEPS 2005 survey. These results help explain the current labor market conditions in Uzbekistan. However, in recent years the government made considerable efforts to lower the costs of doing business (not fully reflected in BEEPS 2005), which may positively influence labor market outcomes in the future. The most importantreforms include: Consolidationof taxes paid by firms. In 2005 a uniJied turnover tax of 13 percent replaced numerous taxes which firms hadto pay earlier. Lowering of payroll taxes. Payroll taxes paid by employers were gradually lowered from 39 percent until 2003 to 25 percent since 2006. This is a substantial reductionwhich lessens the adverse effect of payrolltaxes on labor demand inthe formal sector. Reductionin the number permits. As many as 12 permits were abolished in 2005, which lowered the barriers to entrepreneurshipand limitedthe scope for corruption. Reductionin the frequency of inspections. Accordingto new regulationsan enterprise can be inspected only once every two years. Inspections by various controlling bodies were consolidated into one comprehensive inspection. Furthermore, enterprises must now receive advance notice of planned inspections, which eliminates unpredictability. These measures significantly limited the scope for bureaucratic harassment. Reduction in the frequency of reporting. Enterprises switched from time-consuming monthly to quarterlyreporting(tax and statistical reporting). 61 While these measures represent substantial progress, business environment can and should be further improved to foster firm growth and job creation. For instance, high import taxes and custom duties, and limitations in access to foreign exchange hamper imports and as such limit production growth and thus job creationn3* Figure 2: Most firms inUzbekistan increasedsales over the last three years but decreased employment Panel A Productivity Improvements: while sales increase, employment declines Sates Employment Panel B Employmmntchang. by firm .alms p~rformanco1002-2006 0 5 -20 =2s No change Source: EBRD - World Bank Business Environment and Enterprise Performance Survey (BEEPS) 111, 2005; Bank staff calculations Large gains in labor productivity are a positive phenomenon and are characteristics o f virtually all transition economies. After all, more efficient use of resources, including labor, is a salient feature of economic transition, Moreover, an increase in labor productivity translates into wage growth and thus reduces poverty 32 IFC (2005). Business environment In Uzbekistan as seen by small and medium enterprises,Tashkent, Uzbekistan 62 among working households. The worrisome feature of the transition in Uzbekistan is that the employment reductions in existing firms, which occur largely in the old sector, are not compensated by job creation in new firms. Inall transitioneconomiesbusiness start-ups create a large fraction-from one-fourthto one-half -ofallnewjobs(WorldBank2005). ThereissomeevidencethatinUzbekistanentryofnewfirmstothe market is limited. For example, the proportionofyoung firms (up to five years old) is relatively smallat less than 20 percent (less than the average for the CIS countries) and points to a low pace of new firm creation (Figure3).33 Figure 3: Young firms are relatively few indicating limitedfirm creation Dlstrlbutlon of firms by ape [ p a n ) . 2005 [El-5 m e - 10 0 1 1 . 1 5 O l e + ] Source: EBRD - World Bank Business Environment and Enterprise Performance Survey (BEEPS) 111, 2005; Bank staff calculations The low pace of new firm creation is particularlyworryinggiventhe widespread firm downsizing. There are not enoughnew firms to offsetjob lossesinthe old firms which implies a fall inthe number of formal sector jobs. This fall translates either into informal sector employment, or unemployment, or labor force withdrawal. Ineither case, low firm dynamics inUzbekistan contributes to the underutilizationof labor. Why do so many firms in Uzbekistan reduce employment? The reason is that firms cut costs in order to stay profitable and survive in an increasingly competitive environment. They do this through reducing overstaffingwhich allows them to achieve given output growth with less labor. In other words, gains in labor productivity are turned into lower employment. This mode of adjustment is often referred to as "defensive restructuring". Relatively few firms in Uzbekistan are engaged in "strategic restructuring" whereby firms strive for increasing their market share by investingand usingproductivitygains to increase employment and expand. For example, less than 30 percent of firms in Uzbekistan increased their fixed capital over the last three years. Growthin labor productivityis thus largelydrivenby staffreductions rather than investments in new technology, which is indicativeof defensive restructuring. Job losses take place mainly in the old sector of the economy. It is old, large and privatized enterprises which shed most labor (Figure 4). Young, small and de novo private firms also reduce employment, although on a much smaller scale. Defensive restructuringin the old sector is characteristic of all transition economies and thus it is not surprising. What is surprising is the lack of dynamism of the new sector in Uzbekistan. The new sector instead of creating new jobs and increasingemployment, as it does in most other transitioneconomies, actually contributes to employment decline. Too few firms in the new sector are engagedinthe strategic restructuringfor employment to grow. 33 Far cxmplc. the proponionof young f i n s ascounlifor 23 percenlI"Armmiaand Km!hstan and for 29 per~~nlRussia m 63 Figure4 : Old, large privatized f m s reducedemployment the most Panel A Employmentchange by firm age 2002-2005 0 -5 -10 2 ;-15 -20 -25 -30 -35 4 0 - 5 6-10 11 15 - 16 + years 1 PanelB Change in employment by firm size 2002-2005 -I__- - 10 11-50 51 -250 251+ firm size Panel C Employmentchange by firm ownership 2002-2005 0 e -5 -10 d -15 -20 Privatized De now private ownership Source: EBRD- World Bank Business Environmentand EnterprisePerformanceSurvey (BEEPS) 111,2005; Bankstaff calculations 64 Why do firms in the new privatesector reduce rather than increaseemployment? What is the reason behind the dominanceof defensive restructuring? Oneway to look at this issue is to see what would be the level of employment if there were no constraints to hiring or firing workers. It turns out that if there were no costs associatedwith hiring additionalworkers many Uzbek firms (some 30 percent) would create morejobs and employment would be some 5 percent above its current level (Figure 5).34 This means that there are investment climate obstacles to employment growth in Uzbekistan, and that once the constraints are removed, then job creation is likely to pick up pace. However, there is a significant fraction of firms (20 percent) which would reduce employment if there were no firing costs. This implies that labor hoardingis still present in many firms in Uzbekistan, mainly in the old sector. There is still room for more efficient use of labor resources, although in the short-term this is likely to be accompanied by lay-offs and a likely increase in unemployment. Figure 5: Many firms wouldincreaseemploymentiftherewere nohiringcosts ""' force 40 Source: EBRD - World Bank Business Environmentand EnterprisePerformanceSurvey (BEEPS) 111, 2005; Bank staff calculations Skills do not pose a constraint for firms hiringmoreworkers. In fact, employers in Uzbekistanfind it easy to find a worker with adequate skills, and fill available vacancies for both skilled and less skilled workers. There is a balance betweenthe skill profile of labor demand and labor supply and in this respect Uzbekistan compares favorablyto many other transitioneconomies, especially in Centraland Eastern Europe (Table 1). For example, it takes only 1.6 weeks to find a professional with appropriate skills in Uzbekistan, compared with an ECA average of almost 4 weeks. Similarly, to find a skilledworker requires less than half the time necessary in other countries in the region. However, this positive pictureneedsto be seen in the context of firms predominantlyfiring rather than hiring workers. Once firms engage in strategic restructuringand start to increaseemployment, skills ofthe workforcemay becomea constraint. ~~~ ~ 34 This enlmalei s denved from the answers Io the following qucslioninthe BEEPSIIs w e y : "If couldchange the numberof regular full-bmeworkersyour firm surrenlly employs without any remaionr, what would be your opllmd you level of cmplo)menl BI a pcrcenlof your exining aorkfororce?" Giventhc hypatheticalmlwsof lhe quenion, the flgurc needsIO be rcgardsdas purclyIIIUNBIIVO. 65 Table 1:Findinga worker with appropriate skills is relatively easy inUzbekistan Time taken to fill vacancy for: Non- Manager Professional Skilled Unskilled worker worker production workera) Weeks Uzbekistan 1.6 1.6 1.4 1.1 1.1 Central and EasternEurope 5.7 4.9 3.7 2.1 2.8 Southeastern Europe 4.5 3.9 2.6 1.5 2.0 Middle-income CIS 4.3 4.3 3.8 1.8 3.O Low-income CIS 2.4 2.7 , 2.2 1.3 1.4 Europe and Central Asia 4.0 3.9 3.2 1.7 2.3 a) Administration, sales, etc. Source:EBRD- World Bank BusinessEnvironmentand EnterprisePerformanceSurvey (BEEPS) 111,2005;Bank staffcalculations. The scarcity o f productive job opportunities in Uzbekistan reflects anemic growth o f the new sector and defensive restructuring in the old sector. Both o f these phenomena can be traced back to high costs o f doing business in U z b e k i ~ t a n . ~What are the main obstacles to firm operation and growth, and thus to job ~ creation? One way o f answering this question is to examine what employers themselves perceive as top constraints for their businesses. In employers' view, there are three main obstacles to firm operation and growth in Uzbekistan (Figure 6). The first obstacle is high tax rates and onerous tax administration that raise the cost o f doing business. The latter means bureaucratic harassment: burdensome tax inspections, arbitrary interpretation o f tax regulations and often extortion. Second, high cost and poor access to financing which reflect an underdeveloped financial system and limit firm entry and growth. The final obstacle consists o f numerous business licenses and permits, which hamper business activity. 35 IMF (200% Republicof Uzbekistan: Interim Poverty Reduction Strategy Paper loint Staff Advisorf Note lists liberalizing businessenvironment as one key Structural reform. In particular, the document encouragesthe authorities to advance the private sector agenda: reducestate planning, input and output distribution, and interventions in enterprise and bank operations; dissolve parastatal industrial associations; and substantially accelerate privatization and foreign trade liberalization. Furthermore the document recommends a comprehensive review of a11 regulations, restriction and barriers to entry for private business. 66 Figure6: Taxation, access to financing andpermitsare amongtop business concerns inUzbekistan Proportlon of Finnsthat Report the Followingaa Major Obstacle Tax rater Tax administation Cost of financing Business licensingand permits Access to financing Uncertaintyabout regulalo~ypolicies Macroecanomic instability Contractviolations Customi and bade regulation8 hreet come.then and disorder Curuption Anti-competitivepractices Eleceicity Functioningof the judicary Organized crimelmafia Skills and educationof wailable workerr Access to land Traniportation Title of land Labor regulations Telecommunications 0 2 4 6 8 10 12 14 16 18 20 A Source: EBRD - World Bank Business Environment and Enterprise Performance Survey (BEEPS) 111, 2005; Bank staff calculations Some of these concerns (such as taxation) are common for most modern economies, others (such as an underdeveloped financial system) are characteristic of the transition economies. But there are also some obstacles specific to Uzbekistan, i.e. ones that are seen as more severe than in other transition economies of ECA (Figure 7). The three that are most importantare (a) poor or unreliable infrastructure(e.g. electricity, transport), (b) administrative barriers (e.g. custom and trade regulations and numerous permits), and (c) burdensometax admini~tration.~~ It is noteworthy,that labor regulationsare not seen as a significant obstacle to firm operationinUzbekistan. In fact, they rank as one of the least important factors with a negligible proportion of firms complaining about them. This perception is consistent with objective indicators of the strictness of employment protection legislation. For example, the difficulty of firing, and the rigidity of employment indices are both below the ECA average, indicatinga relativelyflexible labor market.37Thus, the underutilizationof labor in Uzbekistan(see Part 1) can hardly be explainedby labor market rigiditie~.~~ 36 r3~ectiveindicatorsalso indlcate the inefficiency of tax sdmimstrationIn Uzbekistan For example, the numberof Laxes psd b) firms uno1rcssnllywas 118. well above the ECA average in Turkey firms pay only 18 &xes(World Bank 2W6) The n~ationhasimprovedonly recentlywlh the lnuoductionof the unfied mover tax (see Box i). 17 Both indiceswme from the World B dCost of DoingBusinerr2006database. On a reds from 0 to i W (wth 1Mmeaninge m e m ngrdity)the drffmlty ofhnng indexmounUlo 10and the ngrdityofempiopent index IS 11in Uzbekistan By cornpansoninTurkeythese indices are 40 and 55, respectively 38 Simriar wnclusion uas reachedby the prcvlour Uzkkirlan Poveny Assessment study. "The Labor Ccdc heavilyrelieson heold Soviet Code but surpnrmgl) IS quitsmodest m terms of social guaryltepr.For example.severance pay for redundantworkerr m Uzkkistan equalsonly one manlhlywage. andcan be exiendedto mmmum three monfhlywages only ifthe redundant worker registeredat the laborbureaucannotfind alob with" lhalpenodof time '' (Murmgarra and Tracy, 2003) But the reponalso emphastzcslhat labor market nexibliity m UrbslurtanIS achevedby w& enforcement of laborrcgulauonr. World Bank (2045) arguer lhalnextbllity throughnon-enforcementIS a SOsIaiIyinfenor outcome. A prcferabisapproachIS that labor regulationsare limledm scope andfocus on protectingwre worker nataandare effectivelyenforced 67 Figure 7: Concerns about infrastructure,administrativebarriersandtax administrationare morepronouncedinUzbekistanthaninothertransitioneconomies of ECA ECA average Source:EBRD-World Bank BusinessEnvironmentandEnterprisePerformanceSurvey (BEEPS) 111,2005; Bank staff calculations Pervasive administrative barriers tend to be closely associated with widespread corruption in Uzbekistan, which amplifies their negative impact on business growth and job creation. For example, firms in Uzbekistanmuch more frequently than in other ECA countries report that unofficial payments (bribes) to obtain business permits or to deal with tax collectionare common (Figure 8). Also lengthy inspectionstax the management'stime and imposean additionalburdenon firms (Table 2).39 Figure 8: Firms inUzbekistanfiequently haveto makeunofficialpaymentsto obtainbusinesspermitsor to deal with tax collection PanelA Unoficial payments to obtain business licenses and p e n l t s K V Uzb Tal RYS AIb K U Ukr SAM SUI Bel &e Rom MW Mol Arm Lll rur Slk GeO C X POI BlH Cro La1 E81 Hun SI" 0 2 4 6 8 10 12 14 16 18 20 'h offirms reportingMusualor common 39 The frequency of inspectionswas recently significantly reduced, see Box 1. Ais0 the processof seiectinp businessfor inspection has become more transparent. However, inspections continue to have punitive nature: most of them result in penalties, both official and unofficial (IFC 2005). 68 PanelB Unofflclal payment8 to deal with taxes and tax collaotlon 0 5 10 15 20 25 30 35 40 x dnrmimportlng a. W Y ~ or common I Source: EBRD - World Bank Business Environment and EnterprisePerformance Survey (BEEPS) 111,2005; Bank staff calculations Table 2: Businessinspectionsare lengthyinUzbekistan andimpose an additionalburdenonfirms ~~ Average duration of an Regionlcountry inspection Municipal Custom police agency ~~~ ~ hourshisit Uzbekistan 4.0 4.7 Central and Eastern 2.3 3.7 Europe Southeastern Europe 3.0 2.8 Middle-income CIS 3.2 3.2 Low-income CIS 2.9 3.6 Turkey 1.6 1.5 Europe and Central 2.7 3.4 Asia Source:EBRD- World Bank BusinessEnvironmentand Enterprise PerformanceSurvey(BEEPS) 111, 2005; Bank staff calculations Firms in Uzbekistan are also at a disadvantage when it comes to access to modern information and communicationtechnology (ICT), which weakens their competitive position and inhibitstheir growth. For example, only 16 percent of firms in Uzbekistan use the internet in interactions with their clients and suppliers, which is a negligible proportioncompared with about to 40 percent in other low-incomeCIS and over 60 percentinmiddle-incomeCIS. To conclude, firms in Uzbekistan face substantial constraints to their growth, which translates into a slow pace of job creation and the scarcity of formal sector job opportunities. Thus the way to increase formal sector employment and reduce unemployment is through removing the most binding constraints. Specifically,the government should focus its efforts to improve the investment climate with the following four priorities: 69 Investingin infrastructure. This ranges from improvingthe transport system, to makingthe electricity network more reliable, to improving access to modern communications and informationtechnology. Underdeveloped infrastructureundermines the competitiveness of Uzbek firms and limits their growth prospects andconsequently inhibitsjob creation. Improving governance. This entails improving the quality of regulations so as to promote economic efficiency, limiting the discretionarypower of bureaucrats, improving the administrative capacity (e.g. tax administration),reducing policy uncertainty, and last but not least, reducingwidespread corruption. Poor governancehas beenprovento reduce long-termeconomic growthandtherebyjob creation. Removing administrative barriers. Administrative barriers are a significant obstacle to firm operation and growth in Uzbekistan and as such discourage formal sector employment. Thus removingthem is key to fosteringjob creation. This includesreducingto a necessaryminimumthe number of licensesand permits, simplifyingregulations (e.g. custom andtrade regulations, tax regulations), simplifyingbusiness registrationprocedures, etc. Developingan efficientand competitive financial system. Currently the high cost of and poor access to credit severely limit the creation of new firms and the growth of existingfirms. Therefore overcoming these constraints by inter alia developing a sound banking system is critical for firm growth and job creation. The above are broad areas where reforms aimed at creating more and better jobs need to be concentrated. However, the governmentalso needsto identify specific priorities. The way to proceed is to instituteregular consultations with an authentic representation of the business community, including small firms from the new private sector. Also investment climate surveys can be a useful tool to determine obstacles to employment growthand informgovernment policy. A good example of such an approach is IFC(2005). 70 CHAPTER3: EDUCATIONAND POVERTY The 2003 World Bank poverty assessment for Uzbekistan documented the signs of stress in the education system during the early years of the transition. Although coverage rates in basic education were found to have pickedup and reached levelshigherthan those at independence, available householdsurvey data were not conclusive on the extent of actual participation in basic education. Falls in incomes combined with emergenceof formal and informalcosts associatedwith sendingchildrento school made it more difficult for the poor to keep their children in school. In addition, despite increases, all other levels of education still demonstratedcoverage levels below pre-transitionlevels. In particular,the sharp and prolonged declines in higher education enrolments were particularlyunusual. Insufficientjob opportunitiesfor graduates from the education system were considered an important contributor to declining enrolments by lowering the perceivedbenefits fromthe educationsystem. The present assessment, by contrast, finds that educated workers face better labor market prospects, especially those with a university education (see Chapter 3). In addition, the following analysis of the education sector finds continued trends of recovery at the pre-schooling, basic, and upper secondary levels, with varying success. Enrolments in higher education, on the other hand, remain alarmingly low. This chapter's mainfindingsare summarizedbelow. MAINFINDINGS Pre-schooling 1. At less than 20 percent of 3-6 year-olds, pre-schoolenrolment rates, although on the rise again since the late nineties, are nowhere near the highs of close to 40 percent observed in the late eighties and early nineties. 2. Enrolmentrates are markedlyhigher in urbanthan in rural areas and for the better-off,with rates for the richest quintileat 3-4 times those for the poorestquintile. Basic and Upper SecondaryEducation 3. Near universal enrolment rates are observed for the 7-15 year-old age group, with comparable overall male and female as well as rural and urban enrolment rates, and no clear variation across welfare quintiles. There is a five percentagepoint differencebetweenenrolment and attendancerates. 4. Enrolmentrates for 16-18 year-olds exhibit an upwardtrend (at 67 percent in 2005, close to the pre- transition level) with a two percentage point gender gap in favor of males and increasing rates by welfare quintile. 5. Survey findings confirmanecdotalevidence on students, particularlythose aged 14-18 years, having to misstwo monthsof study inthe fall for purposesofagriculturalwork (presumably pickingcotton). 71 6. Schools with students from more disadvantaged backgrounds and schools in rural areas are more likely to enroll above capacity and have teachers with lower qualifications who are further less likely to receivetraining. Geographic remotenessis particularlydetrimental, so that schools in ruralcommunitiesthat are remote from both the rayon and oblast centers fare worst, e.g. only 38 percent of schools in these communities have sufficient numbers of rooms or desks, as compared to roughly 80 percent of urban schools. Quality of education is recognized by the Government as a primary concern and a nationalMDG focuses on improvingit inbasic education in particular. This analysis therefore shows that there are payoffs to targetinginterventionsto the poorestcommunities. Higher Education 7. At 5-7 percent, enrolment rates in higher education for 19-24 year-olds are low and demonstrate wide variation by gender and welfare quintile, so that higher educationtoday is arguably the prerogative of the well-to-doinUzbekistan (contraryto historicalachievements). 8. Giventhe substantialreturns to highereducation, the reason for low enrolments probably lies inthe high out-of-pocket costs of university attendance, which average six times the private expenditures associated with general secondary education, with university tuition alone accounting for one-thirdof total costs. Roughlyforty percent of universitystudents benefitfrom scholarships and subsidies, but boththe beneficiary rate for as well as the averageamount of scholarshipreceived by the two richestquintilesare on the order of twice the rates and amounts for the three bottom quintiles. Thus, scholarships and subsidies for university students were found to be starkly regressive in allocation. This represents a missed opportunityto alleviate the financial burden of higher education for the disadvantaged and reverse current negative trends in enrolment. EDUCATION SYSTEMANDDEMOGRAPHIC TRENDS The structure of the education system in Uzbekistanis in transition (see Annex 1 Figure 1). Preschools continue to enroll childrenages 3-6 years. On the other hand, under the old system, after the nine years of compulsory basic education, students went on to one of three options: 1) two years of upper secondary, thereby completing the eleven years of the academic general secondary education cycle; 2) four years of vocational education at a technikum, generally considered to produce mid-level managerial staff, and theoreticallyallowing graduates to continue on to higher education; or 3) one or two years of terminallower- level professional training at professional technical schools (PTUs) usually affiliated with specific line ministries and catering to their needs. Under the Government reform program known as the National Program for Personnel Training (NPPT), launched in 1997, students are now requiredto complete twelve years of schooling, startingwith nine years of basic education, followed by three years of either academic lyceum or professional college. Therefore, three-year academic lycea replace the two grades of academic upper secondary and are often attached to universities and staffed by university faculty. In parallel, professional colleges with close links to enterprises and flexible curricula responsive to local employment needsreplacethe country-widenetworkof technikums andPTUs. General secondary schools continue to be managedby the Ministry of Public Education; the new academic lycea andprofessional colleges, however, are managedby the Ministry of SpecializedSecondary and Higher Education. In addition, education in Uzbekistan is overwhelmingly publicly provided, at all levels. Accordingto data from the 2003 HouseholdBudget Survey, the overwhelmingshare of pre-schoolstudents attend state-run institutions (87 percent), while 12 percent attend enterprise-run pre-schools, and 1 percent privatepre-schools. Student data from the first wave of the 2005 Uzbekistan Regional Panel Survey (URPS) 72 show that all general secondary andPTUstudentsare enrolledinpublic institutions,while less than 1 percent of students intechnicalschools,professionalcolleges, or at universityare enrolledin privateinstitutions4'. Lookingaheadto the future in terms ofthe size ofthe student age population, Uzbekistanexhibitssomewhat varied trends depending on the specific age group. As Figure 1 shows, relative to the year 2005, demographic projections for the numbers of 3-6 year-olds exhibit declining numbers until 2010 that then begin to rise again, reflecting projected declining fertility and mortality rates. The size of the cohorts correspondingto basic and upper secondary education, i.e. 7-15 and 16-18 year-oldsrespectively,is roughly decliningthrough2020 andrebounds slightlythereafter. Alone the number of 19-24year-olds is on the rise, but only until 2010 when it too begins to decline. Thus, alone the pre-schoolage group is projected to be largerin size in2025 than it was in2005. Figure 1:Trends instudent body size, 2005-25 120 110 100 90 80 70 60 Year 3-6 year-olds ----7-15 year-olds 16-18year-olds -.-._19-24year-olds Source: World Bank HNPStats PRE-SCHOOLING The ongoing decline in the size of the 3-6 year-olds age group before it begins to increase again in 2010 providesthe country with a much neededdemographic window of opportunity. Pre-school enrolment rates, although on the increaseagain since the latenineties, are nowherenear the highs ofclose to 40 percent of3-6 year-olds observedinthe late eighties and early nineties (see Annex 2 Table 14'). Thus, administrativedata4* for the school year 2004/05 report an average enrolment rate of 21 percent nationally,with Tashkent city a clear outlier at 53 percent (see Annex 2 Table 2). As described earlier, accordingto data from the 2003 HBS, the overwhelmingmajority of pre-schools are state-run. In addition, the share of those attendingstate-run pre-schools is higher in urban than in rural areas, while the shares of children in enterprise-run and private pre-schoolsare higherinruralareas. According to household survey data presented in Table 1, overall pre-school enrolment was at 15 percent in 2003 (HBS 2003) and 19percent in 2005 (HES 2005). Inaddition,there is clear variationby urban and rural location, with overall urban rates at over 30 percent inbothyears, but rural rates at 7 and 13 percent in 2003 40 The 2005 URPS does not includelyceumsas a category for student enrolment. Sce Chapter IAmex 4 for a descnptionof the different surveysanalyzed for purposesof this chapterand their vavng sucnghsand rhoncommgs. For the URPS tn pmcular, data analysis revealedproblemswith the quertionron rshwl enrolment and anendance A nzcable numterof observations(171). t.e the ovcnuhelming malonty of those reponingnan-anendance,also reponnon- enrolmmt The swey firm suggestedlhatthe cxplanauonformsanomalyIS that m1-S iomeumerdo nothave required re~ruauanof permanentor temporaryresidencefor childrenlo be formally enroIIed, yct their chldrcnattend educational inltiNtionr However, wing the W S dataOn mtgrauoh 11was not possible to confirm ULlllhore reponing non-attendanceas wcll as nondnrolmentare indeeddisproponionately the childrenof migratinghousehold heads. Thus, inlho end 11was decidednotto rely oncomparing the enrolment andanendancedatafromlhe UF3S far purpaws of lhir analysis. 41 This table IS emacled from data from the TlansMONEE2005 Dalabsse(UNICEF hmenu ResearchCenue,Florence)and is therefore data collated directly from the Uztek Government usinga standardizedternpiate 42 State Comminsefor StauNcr. 2004. Main Indicatorsof Developmentm Educaiionand Perso~eTnuNng AnnualStatlmd Bulletin l 73 and 2005 respectively. Thus, the overall increase in pre-school enrolment rates according to these surveys is almost entirely attributable to increases in enrolment in rural areas. Pre-school enrolment rates also rise markedly with welfare quintile, with rates for the richest quintile at 3-4 times those for the poorest quintile, and the richest quintile in urban areas exhibiting rates above 50 percent. This variation in pre-school enrolment by welfare and geographic location is particularly oignant since early childhood development programs have been found to contribute to greater social equity', so that the enrolment differentials reported here are contributing to inter-generational transmission o f poverty. Table 9: Pre-school enrolment rates by welfare quintile and geographic location, 2003 and 2005 Quintile and Household Budget Survey 2003 Household Energy Survey 2005 geographic location Male I Female I Overall Male I Female I Overall Poorest quintile Rural 4.2 3.6 3.9 9.9 13.8 11.7 Urban 14.8 12.8 13.8 32.1 24.3 27.6 Total 7.2 6.1 6.7 12.4 15.7 13.9 Quintile 2 Rural 5.4 5.6 5.5 7.2 7.5 7.3 Urban 16.2 22.1 19.0 16.5 18.2 17.4 Total 7.9 9.3 8.6 9.1 10.9 9.9 Quintile 3 Rural 7.6 6.1 6.9 16.4 20.9 18.3 Urban 28.6 17.9 22.2 18.0 33.3 26.2 Total 11.8 9.5 10.6 16.8 25.0 20.5 Quintile 4 Rural 8.8 7.5 8.2 14.5 7.6 11.5 Urban 33.6 33.4 33.5 39.7 26.0 32.6 Total 17.6 18.4 18.0 22.4 14.6 18.8 Richestquintile Rural 8.9 17.7 13.1 23.9 24.9 24.4 Urban 55.2 46.6 51.0 53.2 56.3 54.7 Total 37.6 35.6 36.6 40.5 43.9 42.1 Overall Rural 6.6 6.6 6.6 12.4 13.7 13.0 Urban 33.2 29.0 31.1 34.2 33.0 33.6 Total 15.2 14.5 14.9 17.9 20.1 18.9 43 "Why Inveri m E C D (World B a t ) 81hnp//web worldbat 0 r g A Y B S I T E I E ~ R N A P l c s ~ ~ D u c A n o N ~ c D l n , , ~ ~ l ~ ~ ~ K : 2 n ~ o 7 7 4 7 - ~1489S6piPK216618-theSiicPK34493~,~~.hVnl e ~ ~ K ~ s 2 7 ~ 8 ~ g e P K 74 BASICAND UPPERSECONDARYEDUCATION Enrolmentand Attendance44 This variation in enrolment rates across geographic location45and welfare quintile is not observed for the basic education age group (7-15 year-olds). Instead, data from the nationally representative 2003 HBS and 2005 HES show near universal enrolment for the 7-15 year-old age group at roughly 99 and 98 percent respectivelywith comparable overall male and female as well as ruraland urban enrolment rates (see Annex 2 Table 4). In addition, it is not possible to speak of any clear variation in enrolment rates across welfare quintiles. However, a closer look reveals slightly higherenrolment rates for males than for females in urban areas, and the reverse, i.e. higher rates for females, in rural areas. The 2005 HES collected data on attendance inadditionto enrolment, andthe resultshere indicatean overallgap, for bothmales and females, of slightly more than 5 percentage points between enrolment and attendance rates (see Annex 2 Table 4). This gap between enrolment and attendance rates is slightly higher than 5 percentage points for both males and females in urban areas, and slightly less for males and females in rural areas. Even with this gap, however, participationof 7-15 year-olds in educationremains at rates well above 90 percent. For the age group of 16-18 year-olds, on the other hand, Table 2 shows enrolment rates on the order of 59 percent based on the 2003 HBS and 67 percent usingthe 2005 HES data. This arguably wide gap between the two surveys, which are two years apart, may point to enrolment expansion in upper secondary under the Government's NPPTreformdiscussedabove, which extends the durationof compulsory educationfrom nine years to twelve years, amongst other policies. Indeed, earlier analysis of enrolment rates for 15-18 year- old^^^ (Le. extendingto one year younger in age than the present analysis) of the 2000-01, 2002, and 2003 H B S revealed an upward trend from 46 to 59 to 67 percent enrolment for this age group, showing again substantial increases. These rates are therefore comparable to the OECD 2003 country average enrolment rate for 15-19year-olds(again, a slightly larger age group) of 79 percent47.Overall, boththe 2003 HBS and 2005 H E S show a roughly2 percentagepoint gap betweenmale and female enrolment rates. The evidence is mixed, however, with respect to geographic location: in the 2003 HBS, rural enrolment rates are close to 6 percentagepoints lowerthan urbanrates, while they are roughly 2 percentagepoints higherinthe 2005 HES. The 2003 HBS data also show increasingrates of enrolment by welfare quintile4*.Finally, the gap between enrolment and attendance rates is on the order of 3 percentage points, and is again higher for males and females in urban as opposedto ruralareas. The 2005 RegionalPanel Survey (UWS) also collecteddataon enrolment, attendance, as well as the number of weeks missedduringthe academicyear (other thanholidays), andthe reasonsfor failure to enrollor attend or for missingstudies. The reasons span a range of circumstances, includingbadweather, financial reasons, health reasons, quality of schooling, as well as the need to work. Officially, Uzbek law discourages child labor. Yet, the requirements ofthe nationaleconomy, where cotton is one of Uzbekistan's maincommodity exports, continueto outweigh the country's obligationsto internationalstandards. Thus, Uzbekistan, despite its membership o f the International Labour Organization (LO), has ratified only one of twelve o f its conventions banningchild labor4'. In addition, anecdotal evidence abounds on students missingas much as two monthsof studies inthe fall to harvest cotton, andthen againtwo months inthe springto plantcotton. 44 The mrolrnent rater reponedzn thissectionare age-spCElfic enrolment mc8, as opposedto not or gross enrolmnt rates 85 a result of the data lim181msencountered m the s w e p and outlmed in Chapter I h e x 4, m parucularthe lack of consistencyin response categonor and the omission of certainupper secondarycatcgansa 45 According to ole Man lndicatori of Development m Educationand PersamclTmmng h u a l Slatisucal B~lletln(State C O ~ L SforSS18tatlsticsZW4). total general~scondarj(grades 1.1 I)enrolmentmountedto 6.2 millionnudens m I003iW. ofwhich roughlytwo-thirdr were atudenlsm mland one-Uurdstudentsm urban a r m (set AMCX 2 Table 3). 46 See World Bank 2W5. "An Updateon Living Standardsin Uzbskirtan 2000-1W3." Discussion DraR 41 OECD 2005. EducaUonala Glance OECD Indicators2W5, Table C I 2, p 240 48 The m e cam01bs sud for the 2005 HES data, but we arc less camfortable wth ulalrurvey'r income-baxdwelfarememm 49 Accordingto the UrutedNaUon~Integrated Regional Informat~onNetworks. LM Offlce for the Cwrdnauon of HumarumanAffam, at httpi i w innncus.orgircpona r p ? R c p o ~ l D = 3 9 3 ~ 5 & S s l ~ ~ ~ ~ ~ = C e n u a l _ A s ~ & S ~ l ~ ~ t C o ~ ~ U ~ ~ E K l S T A N 75 The 2005 URPS data indicatethat the need to work in agricultureor undertake other types of work was not an important reason for failing to enroll in school in the first place5'. Since the URPS was conducted in February and Marchof 2005, it does not capturethe full academic year, and inparticularwould only include days missed for harvesting, as opposed to planting, cotton. With this caveat in mind, Table 3 reports by student's age on the share missing studies in the first place, the average number of weeks these students missed, the average number of weeks missed by all students, and the seven most prevalent main reasons providedfor missingschool. It shows several interestingpatterns. First, the share of those reportingmissed weeks of studies is highestfor ages 14-18years, with at least 30 percentof each age havingmissedat least 1 week of studies. The average number of missedweeks increases with age, with those aged 14-18who miss at least 1 week of studies reportingover 5 weeks on average. In terms of reasons for missingschool, the most prevalent reason for 7-9 year-olds is overwhelmingly own illness, with the second most prevalent reason bad weather or death or illness in the family. For 10-13 year-olds (roughly speaking), while own illness remains the most prevalent reason, the second most prevalent reason becomes agricultural work. However, for 14-18 year-olds, agriculturalwork becomesthe most prevalent mainreasonfor missingstudies, followedby own illness. Focusingon those who reported agricultural work as the main reason for missing studies amongst 14-18 year-olds who missedat least 1 week of study, Table 4 looks in more detail at the number of weeks missed by welfare quintile and geographic locationof students. The overall number of weeks missed, 8.3 weeks, corresponds to the anecdotal evidence on students havingto miss two months of study to harvest cotton in the fall. Furthermore, rural students haveto miss moreweeks than urban students and constitute the bulk of students missing school for agricultural work. Females in rural areas miss roughly the same number of weeks as males, but urbanfemales missroughlytwo-thirdsthe number ofweeks missedby urban males. 76 Table 10: Enrolment and attendance rates of 16-18 year-olds, by welfare quintile and geographic location (2003 and 2005) Quintile and Household Budget Survey 2003 Household Energy Survey 2005 geographic location Male I Female I Overall Male I Female I Overall Poorest quintile Rural Enrolment 59.2 54.8 56.9 66.2 62.1 64.2 Attendance 66.0 59.7 62.9 Urban Enrolment 56.3 49.1 52.9 64.9 75.7 71.1 Attendance 60.0 64.4 62.5 Total Enrolment 58.4 53.4 55.8 66.0 65.1 65.6 Attendance 65.0 60.7 62.8 Quintile 2 Rural Enrolment 55.5 52.1 53.6 69.5 66.1 67.8 Attendance , 68.4 60.9 64.6 Urban Enrolment 63.8 57.4 60.4 60.9 74.6 67.5 Attendance 58.9 71.3 64.9 Total Enrolment 57.3 53.2 55.1 67.1 68.4 67.7 Attendance 65.8 63.6 64.7 Quintile 3 Rural Enrolment 61.5 56.2 58.9 76.1 73.1 74.6 Attendance 75.1 71.8 73.4 Urban Enrolment 66.9 59.5 63.2 61.6 71.8 66.1 Attendance 59.0 68.9 63.4 Total Enrolment 63.0 57.1 60.0 70.8 72.7 71.7 Attendance 69.1 71.0 70.0 Quintile 4 Rural Enrolment 61.8 59.9 60.9 78.0 64.1 71.6 Attendance 68.9 61.3 65.4 Urban Enrolment 56.4 69.6 63.4 68.7 64.4 66.9 Attendance 62.7 54.7 59.3 Total Enrolment 60.0 63.7 61.8 73.8 64.2 69.5 Attendance 66.1 58.5 62.8 Richest quintile Rural Enrolment 60.0 55.0 57.7 60.5 60.8 60.6 Attendance 57.0 63.2 59.6 Urban Enrolment 68.5 67.5 68.0 62.8 60.3 61.7 Attendance 58.2 * 60.3 59.2 Total Enrolment 65.1 62.9 64.0 62.0 60.5 61.3 Attendance 57.8 61.3 59.3 Overall Rural Enrolment 59.6 55.3 57.4 70.3 66.0 68.2 Attendance 68.0 63.1 65.6 Urban Enrolment 63.3 62.7 63.0 64.0 68.3 66.0 Attendance 59.9 63.4 61.5 Total Enrolment 60.8 57.9 59.3 68.0 66.8 67.4 Attendance 65.0 63.2 64.1 77 << I ! < c .-8m w w 0fi yl b c JI c L Table 4: Missedschoolweeks due to agriculturalwork, by quintile andgeographiclocation (ages 14-18years who missedat least 1week of study) Quintile and IA geographic Male Female Total location Poorestquintile Rural 8.2 I Urban 4.0 4.0 Total 8.2 8.7 II 8.5 Quintile 2 Rural 8.2 Urban 8.0 Total 8.2 Quintile 3 I Rural 8.8 9.1 9.0 Urban 4.0 2.0 2.5 Total 8.7 8.6 8.7 Quintile 4 Rural 8.9 6.6 7.6 Urban 7.6 8.0 7.8 Total 8.8 6.7 7.6 Richest quintile Rural 9.8 8.4 9.1 Urban 7.3 7.3 Total 9.3 8.4 8.9 Overall Rural 8.6 8.2 8.4 Urban 7.1 4.4 6.0 Total 8.5 8.1 8.3 Source:URPS, wave 1,2005 Note: Timing of survey in Februaryand March2005 means that only the cottonharvest season is capturedin this table, and not the cottonplanting season inthe spring, which couldbe an additionaltwo months of missed school. Oualitvof Education As already mentioned (see Chapter 1 Annex 4 on the available surveys for analysis), the 2005 URPS includes a community questionnaire in which school officials were asked about various issues, including whether schools have sufficient numbers of teachers, rooms, desks, boards, manuals, and heating equipment, and whether those physical materials available are in good condition. In addition, the Ministry of Public Education undertook a thorough inventory of general secondary schools in 2004 (known as "passportization") in an effort to assess gaps inphysicaland humanresources as a baseline for the NationalProgram for School EducationDevelopment adopted by presidentialdecree in 20045'. This "passportization" dataset was merged with the household data from the 2005 URPS to provide further insightson student backgroundand corresponding school of enrolment. Thus, these two sources of data providecomplementary findingsto those availablefrom administrativesources. 51 This Rogram spans the pen& 2004-2009 and mge&mengUlemngand development of Ihc m f r m c l w e ofbaric educationschwls. fumirbng rchwlr with upladale leacbng and laboratoryequipment,computers.axlbocka. and tcachmg mlmsl~, improvement of tcachmgmelhodolagyand C ~ C Uand~ , I improvcmentoflhe qualificationroftcaching staff (Governmen1of Uzbelustanand Uniled Nalions Counlry Team 2006. Mille~ium Development Goals Repon). 79 According to the latter, the size of general secondary schools averaged nationally 647 and 637 students in 2002-03 and 2003-04 respectively (see Annex 3 Table 1). Schools are larger in urban than in rural areas with average size largest in Tashkent city (1,101 and 1,064 students in 2002-03 and 2003-04 respectively). Finally, school size has decreased across the board between the academic years 2002-03 and 2003-04, both as a result of an increase in the number of schools as well as a decrease in the number of students. As part of the "passportization" data collection effort, data on both the total number of students enrolled as well as the school capacity were collected, and Table 5 below shows the enrolled-to- capacity ratio by geographic location and student welfare quintile. It shows that, overall, school enrolment is on the order of a third more than school capacity, and although rural schools are smaller than urban schools, their enrolled-to-capacity ratio is higher. Furthermore, enrolment is 60 percent over capacity for the poorest quintile of the student population, and schools become less crowded the higher the welfare quintile, so that the richest students are in schools with only 10 percent enrolment above capacity. There is, however, no clear trend by student welfare when data are further disaggregated by geographic location. Both welfare quintile and geographic location demonstrate correlations with the percentage share of teachers who received any training over the period of 1999-2003. In particular, the data indicate that overall 62 percent of primary (grades 1-4) teachers received training, with the share in urban areas higher than in rural areas. Furthermore, the share of urban primary teachers who received training is stable at roughly 63-65 percent for all welfare levels, while it varies for rural teachers from a low of 53 percent to a high of 70 percent for the richest students. Secondary (grades 5-11) teachers are less likely to receive training than primary teachers (only 52 percent did in 1999-2003) but here too urban teachers fare better than rural teachers52.Giventhat urban teachers are already more likely to have higher qualifications than rural teachers - the share of teachers with higher education is higher in urban than in rural areas (see Annex 3 Table 2) - this differential in training further disadvantages the learning outcomes of rural students. On the other hand, student-teacher ratios are lower in rural than in urban areas (see Table 6 below). At 13.7 students per teacher in general secondary education, student-teacher ratios in Uzbekistan are in fact comparable to the 2003 OECD averages of 16.5 in primary and 13.6 in secondary education53. Table 5 further shows that schools that serve poorer children are less likely to have a phone connection or access to piped water in satisfactory condition, with rural areas faring worse than urban areas. In addition, the schools that serve the wealthiest quintile of the population are less likely to have at least one building in unsatisfactory condition (45.8 percent compared to roughly 60 percent for all other q~intiles).'~Finally, the last column in Table 5 shows a clear trend of schools serving better-off student populations being closer to the district center. 52 It 16, however, difficult to speak of any clearpatternwhen conridenngVaimngof secondaryleachersby nudentwelfare IsvcI. 51 OECD 2005 Educauona1a Glance: OECDIndicatorsZO(1S. Table D2.2. p. IS]. 54 There m y be some vanauon m lhe wal~auonand reponingof the son&uon of schwl inframnwe.but 11IS not pcsrible Io speculateon lhc dvection ofbiar 80 Table 5: School attributes, by welfare quintile and geographic location (2005) Primary Secondary Phone Piped water in At least one Distance teachers teachers connection satisfactory building in from district Quintile and trained in trained in (percent) condition unsatisfactory center to geographic Enrolled-to- 1999-2003 1999-2003 (percent) condition school (h) location capacity ratio (percent) (percent) (percent) Poorestquintile Rural 1.6 52.9 49.3 40.6 41.5 62.8 16.4 Urban 1.2 63.4 62.5 53.2 46.5 43.6 9.5 Total 1.6 53.6 50.1 41.4 41.8 61.6 16.0 Quintile 2 Rural 1.5 58.7 48.9 57.6 45.5 63.8 11.0 Urban 1.6 63.1 44.8 48.1 45.9 76.5 14.7 Total 1.5 59.3 48.3 56.3 45.6 65.6 11.5 Quintile 3 Rural 1.5 59.8 44.7 48.8 39.4 55.8 15.6 Urban 1.5 64.3 51.8 65.8 61.2 67.3 6.0 Total 1.5 62.0 48.1 57.0 49.9 61.4 11.0 Quintile 4 Rural 1.4 56.1 38.2 34.5 23.6 60.6 21.9 Urban 1.2 64.2 51.4 79.6 59.0 58.0 4.4 Total 1.2 63.4 50.1 75.3 55.6 58.3 6.1 Richest quintile Rural 1.7 69.6 56.4 71.7 42.9 85.9 5.8 Urban 1.1 64.9 58.1 89.7 61.8 44.8 3.7 Total 1.1 65.0 58.0 89.3 61.4 45.8 3.8 Overall Rural 1.5 57.5 47.3 49.6 41.3 61.6 14.2 Urban 1.2 64.5 54.3 80.8 59.9 54.3 4.7 Total 1.3 61.9 51.7 69.3 53.1 57.1 8.2 Source:MergedURPSwave 1 (2005) and "passportization" (2005) dataset. ral secon ry educal n, by obla (2003/04) Region Overall Rural Urban Rep.of Karakalpakstan 9.9 9.2 11.2 Andijan 15.1 14.4 17.0 Bukhara 12.7 12.4 13.9 Djizzak 13.8 13.2 15.6 Kashkadarya 12.2 11.4 15.7 Navoi 12.0 10.9 14.4 Namangan 16.7 15.5 19.2 Samarkand 12.9 12.2 16.6 Surkhandarya 14.1 13.8 16.3 Syrdarya 14.3 12.8 18.5 Tashkent oblast 15.5 13.8 19.5 Fergana 14.5 13.5 18.1 Khorezm 12.3 11.5 15.9 Tashkent city 19.9 19.9 Total 13.7 12.7 16.8 Source: State Committee for Statistics. 200 Main Indic Having established this link between welfare level of the student population and geographic location of the school, data from the community survey of the URPS are used to further explore the importance of 81 geography for schooling inputs.Thus, Table 7 documents.various school attributes not merely by urban and rural location but further by the remoteness of the rural location from the rayon or oblast center. However, first it is important to note that the overall shares of schools with sufficient numbers of inputs, or inputs in good condition, leave much to be desired. Thus, for both urban and rural communities, only 83 percent of schools have sufficient numbers of teachers, and 79 percent sufficient numbers of boards. The next range is for rooms and desks, where 72 and 66 percent respectively o f schools overall have sufficient numbers. Finally, the greatest need lies in manuals and heating equipment, where only 57 and 56 percent respectively of schools overall have sufficient numbers. Interms of overall quality o f schools, only 40 percent of schools are deemed in good or excellent condition, while 18 percent have experienced deterioration intheir condition over the last five years. Beyond this sobering big picture, schools in urban communities fare better in almost every aspect, followed by schools in rural communities that are close to the rayon and oblast centers. Clear disparities are apparent between urban schools and rural remote schools, whether these are remote from the oblast or rayon center or both. Indeed, it is of interest to note that distance from the oblast center in many instances has more of an impact on school attributes than distance from the rayon center. Across the board, however, schools in rural communities that are remote from both the rayon and oblast centers fare worst, in many instances exhibiting as little as half the share that is observed in urban schools, e.g. only 38 percent of schools in these communities have sufficient numbers o f rooms or desks, as compared to roughly 80 percent of urban schools. 82 Table 7: School attributes, by geographic location and remoteness of school (2005) Geographic location of community Both Urban urban Ri and rural Close to rayon Remote from (percentage share o f schools, and oblast Remote from Remote from rayon and unless otherwise soecified) oblast centers I centers oblast center rayon center Number o f observations 190 96 33 43 42 24 Sufficient number o f teachers 83.2 84.4 84.8 79.1 73.8 66.7 Sufficient number o f rooms 72.1 84.4 69.7 44.2 54.8 37.5 Rooms in good condition 66.8 79.2 54.5 46.5 50.0 33.3 Sufficient number o f desks 66.3 77.1 60.6 46.5 50.0 37.5 Desks ingood condition 61.1 67.7 60.6 48.8 42.9 33.3 Sufficient number o f boards 78.9 83.3 87.9 62.8 66.7 58.3 Boards in good condition 75.8 81.3 84.8 55.8 61.9 50.0 Sufficient number o f manuals 57.4 64.6 54.5 37.2 52.4 37.5 Manuals in good condition 68.4 76.0 66.7 51.2 61.9 54.2 Sufficient number o f heating equipment 55.8 70.8 45.5 32.6 42.9 37.5 Heating equipment in good condition 56.8 74.0 42.4 32.6 42.9 37.5 Schools' quality excellent or good 40.4 47.9 29.0 27.9 35.7 25.0 Schools' condition deteriorated in last 5 years 17.6 15.6 6.5 34.9 28.6 45.8 The mediandistanceto the rayon center was 1 2 h andto the oblast center 37km. A school was consideredremotefrom either the rayonor oblast center ifthe distanceto it exceededthe respectivemediandistance. The sum ofthe rural categories exceeds the total number of rural observations(94) becausethe categories remotefrom oblast center, remotefrom rayon center, and remotefrom oblast and rayon center are not mutually exclusive. Overall quality of schoolswas specifiedas either excellent, good, fairly good, or poor. The conditionof schools was specifiedas either havingimproved, deteriorated, or remainedunchanged duringthe last 5 years. 83 Uzbekistanandthe MillenniumDevelopmentGoals Uzbekistan's commitment to educating its citizens is demonstratedbothby the fact that it becamea party to the Conventionon the Rights ofthe Childin 1992 andthe InternationalCovenant on Economic, Social, and Cultural Rights in 1995, thereby affirmingthe right to education, as well as the sheer size of public expenditures dedicatedto education, amountingto 28 percent oftotal public expenditures and9 percentof GDP in 2005, and projected to increaseto 32 percent of public expenditures and 11 percent of GDP by 2009. By contrast, the 2002 OECD averagefor education's share ofpublic expenditures and GDP is 12.9 and 5.4 percent respectivelys5.Public educationexpenditures in Uzbekistan are therefore on the order of twice the average of OECD countries. The country, however, exhibits a large variation in per student public expenditures across oblasts: as Table 8 shows, the overwhelming share of current expenditures (above 90 percent) are wage expenditures, and total per student expenditures at the oblast levelvary from a low ofninetenths to 1.3 times the nationalaverage. Table 8: Per studentgeneralsecondarypublic educationexpenditures,by oblast (2005) (rar ed from lowest to highestper student exper iture) Region Per student current exp. ('000s soums) - . Total per student exp. Wage exp. Non-wage exp. Total as percent of avg. Namangan 69 6 75 88 Andijan 72 I 79 93 Fergana 73 6 79 93 Kashkadarya 76 5 81 95 Samarkand 77 4 81 95 Bukhara 76 6 82 96 Khorezm 78 6 84 99 Tashkent city 76 11 87 102 Surkhandarya 83 5 88 104 Tashkent oblast 80 8 88 104 Djizzak 86 6 92 108 R.Karakalpakstan 94 8 102 120 Sirdarya 97 6 103 121 Navoi 106 8 114 134 Overall 79 6 85 Source: Avanesyan, Vahram. 2006. "Per PupilFundingFormulafor UzbekistanGeneral Education," Draft. Undoubtedly,this commitment to education has contributedto Uzbekistan's progress towards achieving the MillenniumDevelopment Goals (MDGs) related to education. Indeedthe Government of Uzbekistan has deemed the country to have achieved bothMDG2 on universalprimary education and MDG3 on the eliminationof gender disparity in primary and secondary education. Instead, Uzbekistanhas adopted a set of national MDGs and targets, which for the education sector are about improving the quality of education and promotinggender equality and empowering women (see Table 9). Specifically,the target for this first national MDG aims at maintaininguniversal access in basic educatiod6(which has been documentedabove for the 7-15 year-oldage group) and improvingeducation quality by 2015. This target seems plausible given the launching of the National Program for School Education Development described above that has quality improvement as its primary focus. With respect to the second national MDG, the first target on gender equality is arguably achievedfor enrolment in basic educationbut for the 16-18 year-old age group, there remains a two percentage point difference in enrolment between males and females, as discussed above. The next section on higher education addresses the second target on improvinggender balance in higher educationby 2015. 55 OECD. 2005. Education at a Glance: OECD Indicators 2005, Table 04.1, p. 205, In fact, the OECD figures include public subsidies to households which are not included in the figures for Uzbekistan, thus further widening the gap between the two. 56 "Basic education" would encompass "primary" and .basic secondary education", the terms used in Table 9 and quoted from the Millennium Development Goals Report. 84 Table 9: Uzbekistan'sNationalEducationMDGsand Targets NationalMDG Target Improvethe quality of education in Improveby 2015 the quality of primaryand basic primary and secondary schools secondaryeducation while maintaininguniversalaccess Promote gender equality and empower Achieve gender equality in primary, basic secondary, and women vocational educationby 2005 Improvegender balance inhigher educationby 2015 Source:Government of Uzbekistanand UnitedNationsCountry Team. 2006. Millennium Development Goals Report. HIGHER EDUCATION At 5 percent based on the 2003 HBS and 7 percent from 2005 HES data, enrolment in higher education for 19-24 year-olds is and demonstrates wide variation by gender, welfare quintile, as well as geographic location (see Table IO). Female enrolment in higher education is on the order of two-thirds that of male enrolment, as is rural enrolment when compared to urban enrolment (although this is in part due to the tendency of higher education institutions to be located in urban areas). In addition, enrolment rates in higher education increase with welfare level, whether for rural or urban areas, with considerably higher ratesfor the richest segments ofthe population(the top one or two quintiles), indicatingthat higher education in Uzbekistan is primarily the domain of the well-to-do. This was not always the case. As Figure 2 shows, Uzbekistan exhibits clear differences in educational attainment across generations. Using the 2003 HBS data on educational attainment of non-student adults58,it is clear that successive cohorts attain higher levels of education: where for 65-year-olds and older, attainment of grades 5-9 represents the largest category with 34 percent of the cohort, for those aged 55-64 the largest category is grades 9-11 (35 percent of the cohort), and the size of this category remains the largest for the two youngest cohorts, reaching 49 percent for 35-54 year-olds, and 56 percent for 25-34 year-olds. Unfortunately, HBS attainment data do not distinguishbetween the more technical PTU/SPTU and the more academic lyceum track - but for this combined category, the share is larger for successive cohorts. This is also true for attainment of technikum or college. 51 The average 2W3 OECD enrolment rate for lhe admrnedly largerage groupof 20-29 ysar-aldais 24 percent (Source OECD. 2WS. Educationat a Glancs OECD Indimton 200s. Table C I 2, p 240) 58 'The supulauon that adults not be students leadsto omissionof 42 abrewattons. out of a total of 7.342, of lhc 25-34 ysar4dr age group,and no obsewalionr from my of the other age groups 85 Table 10:Highereducationenrolmentratesby welfare quintile andgeographiclocation, 2003 and2005 Figure 2: Educationalattainment, by age cohort -Age 25-34 --8- -Age 35-54 ...*...Age55-64 -.Y.-Age 65 p l u ~ Education level attained jource: HouseholdBudget Survey 2003. However, attainment of higher education exhibits an anomalous pattern: while the share of those with higher education increases from 9 percent of 65-year-oldsand older to 17 percent of 55-64 year-olds, this share i s smaller at 16 percent for 35-54 year-olds and drops precipitouslyto 9 percent for 25-34 year-olds. Inother words, attainment of highereducationexhibitsaworrying reversetrend. 86 Yet there are clear returns to higher education in the country. Table 11 reports on average monthly earnings by education level, showing generally higherwages with increasededucational attainment, with the single largest spike in monthly earnings occurring for those continuingbeyond general secondary to higher education. It further shows similar wage levels for those with grades 5-9 or those who complete general secondary (grades 10-11). Females earn lower wages across all education levels except for females with postgraduatedegrees, who are rare entities. Table : Monthlyearningsby educatior evel, wage rners age 1f 55 years (200: soums) I Level of education Female M a l e Overall No schooling 14,497 21,281 18,462 (5) (6) (11) Grade 4 or less 8,456 15,723 13,363 (3) (6) (9) Grades 5-9 13,809 23,540 20,121 (122) (242) (364) Full secondary(grades 10-11) 15,052 21,405 19,170 (1,169) (2,189) (3,358) PTUiSPTUlacademic lyceum 20,373 29,035 26,788 (239) (738) (977) Technikudcollege 20,887 30,177 26,021 (913) (1,156) (2,069) . Higher 29,728 36,851 34,076 (676) (1,079) (1,755) Postgraduate 40,112 37,657 38,225 (4) (14) (18) w c e : HBS 2003 Note:numberof observations in parentheses. Multivariable regression analysis confirms the above broad findings (see Table 12), e.g. a full general secondary educationis not associatedwith significantgains above basic education (the omittedcategory). Higher education, on the other hand, is associatedwith a roughly 60 percent increase in earnings relative to attainment of basic education, while the gains from the vocational track are only on the order of 30 percent (regressions I-IV). Males earn over a third more than females with similar qualifications (regressions 11-IV). Urban residence is associated with a roughly 40 percent increase in wages (regressions 111-IV). Finally, returnsto education vary greatly by ethnicity:Russians and Ukrainiansearn a third more than Uzbekswith similar educational background, while Kyrgyz, Karakalpak, Turkmen, and Tatars earn20-40 percent less. 87 Table 12: Extended earnings function, wage earners age 16-65 years (dependent vaiial =log monthly ean 1gs) Variable I I1 111 Iv Level of education No schooling 0.0733 0.0936 0.0215 -0.0183 (0.1489) (0.1359) (0.1330) (0.1422) Grade 4 or less -0.4422 *** -0.4520 *** -0.4855 *** -0.5243 * (0.2617) (0.2551) (0.2517) (0.2015) Full secondary (grades 10-11) -0.0612 -0.0643 0.0050 0.0168 (0.0512) (0.0484) (0.0443) (0.0441) PTUiSPTUiacademiclyceum 0.2800 * 0.2450 * 0.2577 * 0.2505 * (0.0570) (0.0556) (0.0526) (0.0512) Technikudcollege 0.3154 * 0.3447 * 0.3488 * 0.3499 * (0.0525) (0.0494) (0.0462) (0.0455) Higher 0.5968 * 0.6069 * 0.5735 * 0.5799 * (0.0524) (0.0497) (0.0461) (0.0453) Postgraduate 0.8378 * 0.7992 * 0.6917 * 0.7177 * (0.1073) (0.1 135) (0.1116) (0.1137) Age 0.0190 * 0.0250 * 0.0302 * 0.0349 * (0.0059) (0.0059) (0.0058) (0.0056) Age squared -0.0002 * -0.0003 * -0.0004 * -0.0004 * (0.0001) (0.0001) (0.OOOl) (0.0001) Male 0.3396 * 0.3671 * 0.3826 * (0.0178) (0.0172) (0.0169) Urban 0.4196 * 0.3709 * (0.0327) (0.0335) Ethnicity Russian 0.3295 * (0.06 10) Tajik -0.1272 (0.0860) Karaka1pak -0.2802 * (0.0484) Kazakh 0.0780 (0.0904) KYrgYz -0.4125 * (0.1223) Turkmen -0.2778 * (0.0875) Ukrainian 0.3211 * (0.1227) Tatar 0.1836 ** (0,0865) CrimeanTatar 0.4339 * (0,1435) Korean 0.0867 (0.1288) Azeri 0.03 18 (0.2071) Constant 9.2527 * 8.9360 * 8.6493 * 8.5706 * (0.1 109) (0.1103) (0.1111) (0.1084) No. of observations 8,56 1 8,561 8,561 8,561 R-squared 0.1176 0.1619 0.2297 0.2469 Inificant at 1%;** sigiicant at 5%;***signi: mt at 10%. Omitted 88 Given the benefits to attaining higher education described above, the fact that enrolment rates are low is probably attributable to the high cost of enrolling in higher education. Table 13 provides data on private expenditures for education, includingon transportation, uniforms, textbooks, meals, maintenance of schooling institutions,tuition, as well as a valuationof in-kindcontributions. It shows that, overall, privateexpenditures are lowest for basic and general secondary education, they triple roughly for secondary special education and technical school or college, and university costs are on the order of six times the cost of general secondary education. Uniforms account for over half of private expenditures in basic and general secondary education, and textbooks for roughly one fifth of expenditures. For secondary special education and technical school or college, transportation and meal costs become important, with transportation and uniforms accounting for roughly one third of expenditures each, and meals for one fifth. Transportation costs remain high for I universityand account for one fifth of total costs, but the largest single category at university level are tuition payments, which are one third of total expenditures. 'I ble13:Privateeducationexpenditures,bylevelofeducation(2005soums) Level of education Ty !of expenditure Total Transport. Uniforms Textbooks Meals Maint'ce Tuition In-kind Other Absolute value Grades 1-4 1,643 19,346 5,830 2,115 936 480 202 926 3 1,476 Grades 5-9 1,861 22,780 8,290 2,473 1,229 456 315 1,458 38,862 Secondary general 6,160 28,793 9,313 3,751 1,100 828 389 2,812 53,146 Secondary special 42,897 40,583 10,496 30,72 1 10,588 16,486 2,240 849 154,860 Technical school, college 53,165 39,676 11,125 23,510 1,588 10,546 286 4361 144,459 University 69,578 42,348 16,892 45,510 2,715 106,121 142 25,436 308,742 Percent share Grades 1-4 5.2 61.5 18.5 6.7 3.0 1.5 0.6 2.9 100 Grades 5-9 4.8 58.6 21.3 6.4 3.2 1.2 0.8 3.8 100 Secondary general 11.6 54.2 17.5 7.1 2.1 1.6 0.7 5.3 100 Secondary special 27.7 26.2 6.8 19.8 6.8 10.6 1.4 0.5 100 Technical school, college 36.8 27.5 7.7 16.3 1.1 7.3 0.2 3.2 100 University 22.5 13.7 5.5 14.7 0.9 34.4 0.0 8.2 100 Source: URPS, wave 1,2005 The 2005 URPS includes information on scholarships and subsidies received in support of education (100 students in each of grades 1-4, grades 5-9, and at the universitylevel). As Table 14 shows, the beneficiaryrate (percent of those enrolledreceiving a subsidy or scholarship) is relatively low overall for both grades 1-4 and grades 5-9, and average amounts received are approximately one third and one fourth, respectively, of costs reported above in Table 13. In addition, although the beneficiary rate is highest for the poorest quintile for both grades 1-4 and grades 5-9, the average scholarship amount is the lowest. For university education, on the other hand, the overall beneficiary rate of above 40 percent is high and the average amount is comparable to average tuition payments, so that there is potential for university scholarships to relieve the financial burden for disadvantaged students. However, the beneficiaryrate for university students rises with welfare level of the student, as does the amount of scholarship received. Indeed, both the beneficiary rate for as well as the average amount of scholarship received by the two richest quintiles are on the order of twice the rates and amounts for the three bottomquintiles. 89 Quintile Grades 1-4 Grades 5-9 University Beneficiaryrate Avg. scholarship Beneficiaryrate Avg. scholarship Beneficiary rate Avg. scholarship (YOof enrolment) amount (% o fenrolment) amount (%of enrolment) amount Poorestquintile 13.1 7,072 6.2 6,860 21.3 54,000 2 6.3 8,438 3.8 8,041 18.6 37,333 3 7.6 9,647 3.1 12,028 23.1 52,429 4 10.5 14,817 3.5 17,33 1 43.2 92,606 Richestquintile 6.0 10,435 5.2 12,798 48.6 92,141 Overall 9.1 9,533 4.3 10,545 41.2 87,885 Consequently, Figure3 shows that educationsubsidies and scholarshipsfor grades 1-4 and grades 5-9 are pro- poor: the concentration curves for education subsidies and scholarships for these two levels of education lie above the Lorenz curve for aggregate consumption expenditure, implying progressivityo f the subsidies and scholarships. Inother words, the grant amounts are distributedmore equallythan consumption expenditure is in the country. In addition, the subsidies and scholarships for grades 1-4 succeed in targeting the poorer segments of the population: their concentration curve lies entirely above the diagonal. The concentration curve for university scholarships, on the other hand, is below the Lorenz curve for aggregate consumption expenditure and is therefore clearly regressive. The fact that university scholarships are more unequally distributed than consumption expenditure represents a stark missed opportunity to counteract the negative trend in enrolment inthis sub-sector by relievingthe financialburdenfor the most needy segmentsin society. Figure 3: Benefitincidenceof scholarshipsandsubsidies,2004 Consumption expenditure -c-Grades 1-4 scholarships -t-Grades5-9 scholarships -.x- - Uniwsity 0 20 40 60 80 100 scholarships Cumulative distribution of population (%) Source:UWS,wave 1,2005 90 ANNEX1. EDUCATIONSYSTEM Figure 1:Past and current education system University/ institute University/ institute Technikum Basic Basic Old system New system 91 4 ANNEX2. ENROLMENT ATTENDANCERATES AND Table 1: Trends inenrolment rates, by level ofeducation I_; Level of education Year 1989 - --- 1996 1997 1998 1999 2000 2001 2002 Pre-primary a 36.8 19.5 17.6 16.1 16.2 18.2 19.4 19.9 Basic b e 92.1 88.4 88.9 89.2 88.9 97.0 97.8 97.5 Total upper secondary e 33.4 ---- o/w general 69.4 47.7 50.6 53.5 55.9 55.2 53.1 68.4 36.3 26.2 28.0 29.6 30.9 22.7 21.2 34.9 I 1 I o/w voc./technical 33.1 30.0 27.8 26.8 25.3 21.5 22.6 23.9 25.0 32.5 31.9 33.5 Higher 15.0 15.2 14.8 13.4 11.3 - 6.5 6.6 -- 7.3 7.9 aNetrates, percent ppopul on aged3-6 'Gross Gross rates, percent of pop ationaged 7-15 rates, percentof populationaged 15-18 Grossrates, percentofpopulationaged 19-24; IRC estimatebasedon numberof students innon-degreeor degree-grantingtertiary education Break intime series in 2000 due to change in educationsystem Source:TransMONEE 2005 Database, UNICEF InnocentiResearchCentre, Florence Table 2: Pre-: hool enrolment. y oblast (2004/05) Region Pre-schools Children Coverage(YO) Rep. of Karakalpakstan 390 32,058 15.8 Andijan 580 50,972 18.0 Bukhara 518 40,714 22.9 Djizzak 193 3 1,370 21.8 Kashkadarya 397 41,930 10.2 Navoi 235 23,982 10.6 Namangan 651 52,462 22.0 Samarkand 696 50,077 12.3 Surkhandarya 426 29,013 11.0 Syrdarya 193 15,306 18.5 Tashkent oblast 55 1 45,247 20.4 Fergana 938 76,036 26.4 Khorezm 434 31,801 19.0 Tashkent city 500 84,052 53.4 Total 6,702 605,020 21.0 Source: StateCommittee for Stc ;tics. 2004. Main6 icators ofDevelopment in Education L ?d Personnel Training Annual Statistical Bulletin. 92 T )le 3: General se mdarv student enrolment. bv oblast (2003/04) Region Total number Oblast share Student distributionin oblast ( O h ) of students o ftotal (%) Rural Urban Rep.of Karakalpakstan 364,608 5.8 62.3 37.7 Andijan 566,801 9.1 71.2 28.8 Bukhara 344,007 5.5 74.8 25.2 Djizzak 266,2 10 4.3 72.4 27.6 Kashkadarya 628,768 10.1 76.9 23.1 Navoi 201,467 3-2 63.9 36.1 Namangan 516,814 8.3 64.5 35.5 Samarkand 741,915 11.9 79.2 20.8 Surkhandarya 521,263 8.4 86.8 13.2 Syrdarya 156,199 2.5 67.3 32.7 Tashkent oblast 521,442 8.4 62.5 37.5 Fergana 690,333 11.1 73.6 26.4 Khorezm 345,819 5.5 76.2 23.8 Tashkent city 375,517 6.0 100.0 Total 6,24 1,163 100 68.4 31.6 ource: State Committee for Sta ;tics. 2004. Main It cators of Development in Education and Personnel TrainingAnnual 'tatistical Bulletin 93 Table 4: Enrolmentand attendance of 7-15 year-olds, by welfare quintile andgeographic location (2003 and2 )5) Quintile and Household Budget Survey 2003 Energy Survey 2005 geographic location Male I Female 1 Overall Male I Female I Overall Poorest quintile Rural Enrolment 98.9 99.5 99.2 96.2 98.2 97.2 Attendance 92.9 92.0 92.5 Urban Enrolment 98.9 97.8 98.3 98.4 100.0 99.0 Attendance 80.0 92.4 85.0 Total Enrolment 98.9 99.1 99.0 96.6 98.5 97.5 Attendance 90.5 92.1 91.2 Quintile2 Rural Enrolment 99.0 99.2 99.1 97.0 98.4 97.6 Attendance 94.5 93.8 94.2 Urban Enrolment 99.7 99.0 99.4 100.0 100.0 100.0 Attendance 93.1 88.1 90.3 Total Enrolment 99.2 99.2 99.2 97.6 98.9 98.2 Attendance 94.2 92.1 93.2 Quintile3 Rural Enrolment 99.6 99.2 99.4 98.5 98.8 98.6 Attendance 96.4 94.7 95.6 Urban Enrolment 99.2 97.8 98.4 99.1 98.3 98.8 Attendance 94.7 95.0 94.8 Total Enrolment 99.5 98.8 99.2 98.7 98.6 98.7 Attendance 95.8 94.8 95.3 Quintile4 Rural Enrolment 99.1 99.2 99.2 98.6 99.3 98.9 Attendance 94.6 96.5 95.4 Urban Enrolment 99.4 98.8 99.1 100.0 97.0 98.5 Attendance 92.5 92.2 92.4 Total Enrolment 99.2 99.1 99.2 99.1 98.3 98.7 Attendance 93.7 94.6 94.1 Richest quintile Rural Enrolment 99.4 99.5 99.4 98.6 100.0 99.2 Attendance 90.6 91.1 90.8 Urban Enrolment 99.4 98.9 99.2 99.5 95.0 97.3 Attendance 94.2 92.6 93.5 Total Enrolment 99.4 99.1 99.3 99.1 97.4 98.3 Attendance 92.4 91.8 92.2 Overall Rural Enrolment 99.2 99.3 99.2 97.5 98.7 98.0 Attendance 93.9 93.5 93.7 Urban Enrolment 99.4 98.5 98.9 99.4 97.9 98.7 Attendance 91.5 92.0 91.7 Total Enrolment 99.2 99.0 99.1 98.0 98.4 98.2 Attendance 93.2 93.0 93.1 94 ANNEX3. QUALITY OFEDUCATION Table 1: Ger a1secondary school si Region ov )all Rural 2002103 2003104 2002103 2003104 2002103 2003104 Rep.of Karakalpakstan 502 477 388 373 976 887 Andijan 761 756 678 679 1,083 1,052 Bukhara 657 635 619 604 801 747 Djizzak 499 489 430 421 856 853 Kashkadarya 575 576 523 537 931 759 Navoi 547 534 430 423 1,042 998 Namangan 766 766 672 668 1,050 1,042 Samarkand 624 613 578 566 924 896 Surkhandarya 627 631 597 603 931 906 Syrdarya 518 507 45 1 438 755 750 Tashkent oblast 596 581 515 496 821 814 Fergana 762 755 706 701 970 961 Khorezm 653 643 611 607 833 792 Tashkent city 1,101 1,064 1,101 1,064 Total 647 637 564 559 957 I 915 jource: State Committee for S1 stics. 2004. Main Indicators ofDevelopment in Education and Personnel TrainingAnnual Statistical Bulletin. Table 2: Educational bat ground o fteachers, by oblast (2 03104) Region Overall (%) Rural (%) Urban (%) Higher Incomplete Secondary Higher Incomplete Secondary Higher Incomplete Secondary education Higher Special education higher Special education higher Special R. Karakalpakstan 60.9 3.7 35.4 57.8 3.8 38.4 67.1 3.4 29.5 Andijan 68.2 5.3 26.5 66.4 5.3 28.3 73.6 5.0 21.4 Bukhara 80.2 1.7 18.1 78.8 1.7 19.5 85.0 1.7 13.3 Djizzak 70.8 3.4 25.9 69.2 3.7 27.1 75.5 2.4 22.1 Kashkadarya 62.5 2.6 34.9 60.3 2.5 37.2 72.7 2.8 24.5 Navoi 72.6 3.9 23.6 71.1 3.7 25.2 75.9 4.4 19.8 Namangan 66.8 3.2 30.0 64.7 3.1 32.3 71.5 3.6 24.9 Samarkand 67.9 2.2 29.9 67.7 2.0 30.3 69.2 2.9 27.9 Surkhandarya 58.4 4.2 37.5 56.5 4.2 39.3 72.9 3.6 23.4 Syrdarya 72.4 4.6 23.0 71.6 4.6 23.9 74.9 4.8 20.3 Tashkent oblast 66.4 4.1 29.6 65.1 3.9 31.0 69.4 4.4 26.2 Fergana 69,l 4.1 26.8 67.0 4.3 28.7 76.9 3.3 19.7 Khorezm 72.1 3.0 24.8 72.4 2.8 24.8 71.1 3.9 25.0 Tashkent city 81.8 7.9 10.3 81.8 7.9 10.3 Total 67.9 3.6 I 28.5 65.7 3.4. -. -- 30.9 74.3 4.2 21.5 Source:StateCommitteefor Stat ics. 2004. Main Indicators of Development in Educai n and Personnel TrainingAnnual Stati. :a1 95 CHAPTER4: HEALTH AND POVERTY This chapter displays the linkages between poverty and health status to inform the design of public policies that are critical in achieving the Millennium Development Goals. In order to obtain basic insights into the factors that influence health outcomes, analyses of trends in health care behavior, access to care, status of rural healthfacility infrastructure, healthcare utilization,choice of provider and publicandprivatespendingon healthcare follow. MAINFINDINGS 1. The poor and the uneducatedare less healthy, However, this relationship is visible only if one uses objective measurements of health status of conditions such as anemia and stunting. The self-reported healthindicatorstypically includedin householdsurveys producemisleadinginsights ifthey are usedto establish linkages with poverty. Thus, they should not be used to evaluate progress towards the MillenniumDevelopmentGoals (MDGs). 2. The health-socioeconomic status gradient emerges early among children in Uzbekistan (especiallyif one focuses on stunting, a long-term indicator of malnutrition) indicating the importance of early interventions. 3. Pro-poortargeting at the household level is neededto reducethe prevailingstuntingrates. However, there are no statistically significant differences in health outcomes of children within households by their gender or birthorder. 4. Regional targeting holds promise for reducing the prevalence of anemia, both because of high anemia rates (close to 80 percent in some regions in 2002) and because of the regionalnature of some interventionssuch as flour fortification. UNICEF (2006), focusingon three regions, demonstratesthat interventions such as flour fortification and distribution of iron pills do indeed produce significant results. 5. Analysis of regional survey data designed specifically for measuring disability rates reveal that almost 12 percent of all individuals ages 7 and older have at least one serious difficulty or a full limitation(in seeing, hearing, movement, learning, communicationor self-care). But official disability status is granted only to 3.8 percent of the population in the regions under investigation (national average figure was 3.2 percent in 2003). Thus the formulation of policies to prevent disabilities and improve the living standards of the disabled emerge as a major development challenge for Uzbekistan, in part because of strong linkages between disability and school enrollments (official disability status is associatedwith 40 percentagepoints drop in school enrollments among 15 to 18 year olds), labor force participation(with a 52 percentagepointsreductioninthe probabilityof beingeconomicallyactive) and household welfare (having a full limitation is associated with a 7.8 percent decline in household consumption; with no statistically significant linkagewith official disability status presumably because of the relativelygenerous but low-coveragedisabilitybenefits). 96 6. Access to health facilities is not the main obstacle for the poor, but health infrastructure has deteriorated significantly in rural areas (where the poor are more likely to reside) and the poor are unable to affordthe costs of medicines even ifthey do consult with a health service providerin the case of sickness. 7. The poor and the uneducatedare also significantlydisadvantagedwhen it comesto knowledge about diseases, for example TB and AIDS. Furthermore, poor females are much less likely to have a say in their own health care. These findings indicatethe need to target informationcampaigns towards the poorest segments ofthe populationifthey are to beeffective. Inthe remainder of the chapter, section 2 describes the healthsystem. Section 3 offers a cross-country perspective on health indicators, while section 4 focuses on the linkages between health, poverty and inequality. THEHEALTHSYSTEM At the time of independence, Uzbekistan inheriteda health infrastructurethat enabled universal access to healthcare. Beforetransition, the provisionof health services was financed by public revenue and it was free-of-charge. Yet the Uzbek health system, which was centrally organized and managed, also suffered from major inefficienciesin part becauseof a strongfocus on secondary andtertiary levelcare. After independence,the Government of Uzbekistanestablishedthe priority areas within the health care systemas women's and children'shealth, infectiousdiseasesprophylaxis,environmentalprotection,and development of primaryhealthcenters(PHC). The health sector currentlyconsists of public and privateproviders, as well as traditional hea1e1-s.~'This last category is relativelysmall, makingup only 2.6 percent of the total health care visits according to the 2003 HBSe60The public sector is organized in a three-tier system. The first tier includes sanitary- epidemiologicalservices and health centers. The health centers are the rural feldsher-akusher points (FAPs), small rural clinics/polyclinics(SVAs), rural medical centers (SVPs), and polyclinics in urban areas. The second tier consists of city polyclinics that provide basic specialist care; rural district hospitals provide basic secondary care, first-aid, and maternity services; central district hospitals, regional and city hospitals; maternity homes, and specialized hospitals, The third tier consists of national facilities and the National Institute of Health. There is also a parallel system of health care servicesprovidingcare for employees and officials of certainorganizations, enterprisesand ministries. The first-tier institutionsare the most common option at the health district level, and all the first-tier institutions excludingthe FAPs provide a mix of first and secondary care services. The FAPs were typically serviced by 1-3 staff including one feldsher and one midwife, and offering preventative, curative, antenataland basic postnatalcare. Their number was significantlyreducedbeginningin 2002. The SVPs were introduced in 1997, to become primary providers for the rural population instead of FAPs. The SVPs have 1-5 physicians and 4-8 nurses and auxiliary health personnel. In addition to basic diagnosis and treatment, they provide reproductive health care, and health education and prevention. The SVAs have around 4 physicians, usually an internalmedicine specialist, a pediatrician, an obstetrician, and a dentist. The city polyclinics have usually 10-20 staff, offer diagnostic and 59 See Health Care Systems m Transiuon Uzbebslan (2001) publishedby the EuropeanObsewaloryon Health Care Systems for a detmled descnpuon of the UrbehrtanhealIhryslem 60 Having said that, the share visits to traditional providers can be quite high in some regions around 11 percent in Navoi and Kashkadarya, and 7 percent in Jiuak, as opposedto 3 percentin Karakalpakstanand below 2 percent in all other regions 97 therapeutic services, and their equipment level varies between rural and urban areas. There are polyclinicsfor adults, for children, andfor women's health. The modern private sector is poorly developed. The privatization of the health sector began in 1993 covering pharmacies, optician practices and ancillary services in some health care institutions. A second wave of privatizationwas supposed to increase the number of privatehospitals and polyclinics. However, more often private units in state-run hospitals and polyclinics are encountered. The state started issuing licenses for private practice in 1995; in 1999, most of the private practitioners were dentists, pediatriciansand internalmedicine physicians. There are also some NGOs involved in health- related activities. Privatizationrefers mostlyto giving permission to charge (unregulated) out-of-pocket payments; for a smallfee, privatedoctors are given a licenseto practice. A COMPARATIVE LOOKAT HEALTH INDICATORS AND HEALTHCARESPENDING IN UZBEKISTAN This section presents cross-country statistics on health spending and on selected health outcomes to provide a context for the analysis of the relationshipbetween poverty and health behavior and health outcomes inUzbekistan. In2002, the Uzbek total public expenditures on healthmadeup 2.5 percent ofthe GDP (Figure 1). This level of spending is higher than its Central Asian neighbors, but smaller by 29 percent compared to Russiaandby morethan 60 percent comparedto EMU and OECD. As importantly,spendingas a share of GDP declined since 1996. The private expenditures as 8 percent of the GDP decreased during the same period, but its share inthe total healthexpenditures increased. In2002, private spendingon health as a percent of the GDP was again higher than in any of the neighboringcountries and EMU, but lower than the OECD average. The private expenditures on health include official user fees at state facilities:' semi-officialcharges for drugs and medical equipment (sometimes requiredfor inpatients), informalpaymentsand fees in the privatesector. Figure 1. Healthexpendituresas percent of GDP. I 6 5 5.5 4.5 s 0 3.5 2.5 1 5 0 5 uzB.1996 LeB KAZ KGZ RUS TX EMU OEC Source WDI-GDF,2005 Unlessspecified, the figures are for 2002 Compared with the neighboring Central Asian countries, Uzbekistan has relatively good health indicators: among the highest life expectancy at birth, high immunizationrates, the lowest (estimated) maternal mortality rate, and the lowest TB incidence. However, Uzbekistanhas the highest HIV/AIDS 61 By 2005, only the PHCr. emergency xMCCS. and healthcm for infectiousdiseases. msnld dxordcrs, cmcer,endcccnncdirsascr. and matcrmtyhealth wcrcIObe prowdedIrcc+f-chargc 98 incidence (Table 1). Uzbekistanalso fares poorlywhen comparedwith either OECD or EMUcountries, except for the prevalenceof immunization. Table 1. Regionalcomparison o f healthindicators Immunizations Mortality rates (YO of children Life 12-23 m.0.) TB HIV TFR expectancy incidence incidence Maternal Infant (estimated) Under-5 DPT Measles 2003 2003 2000 2003 2003 2003 2003 2003 2004 Uzbekistan 2.3 66.7 24 57 69 98 99 115.0 76.1 Kazakhstan 1.8 61.3 210 63 73 99 99 145.4 45.4 KYrgYz Republic 2.4 65.0 110 59 68 98 99 124.4 30.1 Tajikistan 2.9 66.3 100 76 95 82 89 168.3 31.4 Russian Federation 1.3 65.7 67 16 21 98 96 112.2 238.6 OECD 1.7 78.5 13 5.4 6.6 95 92 16.2 EMU 1.5 78.9 9 4.3 5.5 94 89 13.4 Source: WDI-GDF (August 2005). For maternal mortality regressionestimates. For infant and under-5 mortality rates WE?estimates using - - data from UN and UNICEF, State of the World's Children. For HIV incidence: ECEM AIDS (newly diagnosedHIV infection rates per mill. people). Time trends in basic populationhealth indicatorsare mixed (Table 2). For example, the percentage of children 12-23 months immunized against DPT3 and measles increasedover the last decade, but so did the infant and under-fivemortality rates and the incidence of HIV and TB.62Simultaneously, the life expectancyat birthdecreasedin Uzbekistan, as did in all the neighboringcountries and in contrastto the increasing EU/OECD trends. In 2003, female life expectancy was 6.5 years higher than male life expectancy. This differenceis mostlystable over time (6.3 years in 1997). Table 2. Uzbekistan,selected indicators 1995 1996 1997 1998 1999 2000 2001 2002 2003 Population,total (mill) 22.8 23.2 23.7 24.1 24.4 24.7 25.0 25.3 25.6 Populationgrowth 1.8 1.9 1.9 1.6 1.5 1.0 1.3 1.2 1.3 Fertility rate, total (births per woman) 3.6 3.3 3.1 2.8 2.7 2.3 2.3 Life expectancy at birth, total (years) 69.2 67.0 66.7 Maternalmortality 39 Infant Mortality Rate 49.1 52.0 61.7 Under-5Mortality Rate 59.3 73.3 Proportionof 12-23 monthsold immunizedagainst DPT3 89 97 96 99 99 96 97 98 98 Proportionof 12-23 monthsold immunizedagainst measles 91 95 88 96 96 99 99 97 99 YOBirths attendedby healthstaff 97.5 95.6 Incidenceof HIV 0.0 0.3 0.1 1.1 6.2 21.7 38.2 70.4 Incidenceof TB 76.2 71.3 79.4 84.3 87.5 91.4 100.1 107.6 115.0 Source: WDI (2005) unless othenvise noted. Maternal mortality: TransMonee. IMR and U5MR - DHS (1996) and UHES (2002). HIV prevalence (infection rates per mill. people): ECEM AIDS. MMR is computed per 100,000 live births, U5MR and IMR - per1,000 live births, TB incidence per 100,000 population, and HrV incidence -per mill. population. - 62The survey data and official data on infant mortality rates differ, in part because of different live-birthdefinitions 99 The officially reportedtotal fertility rate decreasedby 35 percent between 1995 and 2003. The survey- computedTFRs are higher thanthe officially reportedones: 4.0 in 1996and 2.9 in 2002, but the trendis similar.63 A more detailed picture of TFR is captured by the 2002 UHES: rural TFR is 28 percent higherthan the urban TFR, and females with higher educationtend to have fewer childrenthan females with general secondary or less education. Despite the decrease in TFR, the overall population is increasing, but at a decliningpace over the last decade. This decline is due to a decliningbirth rate (per 1,000 population)from 34.5 in 1991to 20.2 in 2003, and a slightly increasingdeathrate from 6.2 to 6.4 (per 1,000 population). The cross-country comparisons listed above should be interpreted with some caution due to the differences in data collection across countries. For example, the statistics reported above suggest universal immunizationcoverage among 12-23 months olds in Uzbekistan. However, the 2003 Health Facility Survey (HFS) found that for the four oblasts included in the survey -Ferghana, Syrdarya, Navoi and Tashkent-only in 55 to 85 percent ofthe cases did the practitionerscheck the vaccination status of the 0 to 5 year old children coming for consultations. Of the 6 to 18 percent of children identified with missing vaccinations and who were not given the vaccinations immediately, only 30 percent inFerghanawere askedto come later for vac~ination.~~ HEALTH, POVERTY AND INEQUALITY Now that we have outlinedthe key features ofthe health system and health spendingpatterns, a more in depth analysis of healthoutcomes follows. Children's Health outcomes To discuss the linkages between poverty and poor health status, this section focuses on the children sample. Malnutritionat early stages of life has been linkedto poor health status later, as well as to poor schoolingoutcomes. To highlight trends in malnutrition, three indicatorsare considered here: stunting (an indicator of chronic malnutrition),wasting (to capture short-term malnutrition)and anemia. In 2002 one-fourthof the childrenaged 0-6 years were stunted, while less than one-tenth were wasted (Figure 2). Among the malnourished children, almost 40 percent of the stunted were severely stunted, while only over 15 percent of the wasted were severely affected. Stunting rates peek among 12-23 montholds, decreasingafterwards, while anemia is morecommon amongolder children. Table 3 shows that 57 percent of the Uzbek children aged 0-6 were affected by anemia, of which 35 percent were affected by moderate or severe anemia. The prevalence of stunting and anemia were higher in rural areas and lower in the wealthiest 20 percent of the population, while the opposite happened with the prevalence of wasting (see Table 3) - interestingly though, multivariate analysis does not reveal statistically significant differences in healthoutcomes of childrenwithin households by their gender or birth order.65Among households with four or more children aged 0 to 6, there was more than a 90 percent chancethat at least one of the childrenwas anemic; while there was almost a 45 percent chance that at least one of the children was stunted (multiple occurrences of anemia and stunting are shown in Figure3). 63 Source DHS m 1996 and 'LR(ES ~n2001 64 Still. the same survey found lhalvaccination cqtupmsnland supplier for ~ a ~ ~ i mare~fullyr availableand accessible u n 65 See Mete and Cnobloch ( 2 M ) for ulc resultsof rnuluvmal~analysis on the prcdinorsO f healthoutcomes. 100 Figure 2. Nutritional status of 0-6 year oldchildrenby age. I I I Mldanenia bb2eratdsevere W e d SRleretystlnted Alwasted Saveretywasted -a m (111 mrrths o 12-23months 24-35 ma;4hs H3647rrmths 04871 rrmths Source:UHES,2002, own computations. Table 3. Nutritionalstatus of 0-6 year oldchildrenby poverty status. Anemia Stunting Wasting No Mild Moderatehevere anemia anemia anemia All stunted Severely stunted All wasted Severely wasted Poorest 37.4 42.1 20.6 25.4 9.8 7.7 1.3 quintile 35.3 40.2 24.4 23.2 7.9 6.3 1.2 3rdquintile 39.8 37.8 22.3 21.0 7.5 5.1 1.2 4'h quintile 47.9 34.0 18.2 17.0 6.8 5.3 2.0 Wealthiest 53.9 30.8 15.3 12.7 4.4 11.1 2.3 Total 43.0 36.9 20.1 19.8 7.3 7.1 1.6 Source:UHES2002, own computations Figure 3. Prevalence of multipleoccurrencesof anemiaand stunting amonghouseholds,by number of children0- 6 year olds. I Ij Anemia -multiple occurences within households Stunting muttiple occurences within households - 1W% 90% 80% 70% 60% 50% 40% 30% 20% 10% , . . 0% .,.... ...... .,.. .. . . . 1 2 3 4* number of children 06 y.0. Number of children lessthan 6 y.0. 00 occurences 0 1 occurence 2 occurences 8 3 occurences 0 4 occurences Source:UHES, 2002, own computations. The nutritional status improved when compared with 1996: 31.7 percent of children were stunted in 1996 versus 22.8 percent in 2002, and 11.3 percent were wasted in 1996 compared with 6.1 in 2002 (Figure 4). The anemia levels only slightly diminished over years: of the children 6-35 months, 61.2 percent were anemic in 1996, and 56.8 percent in 2002. For anemia, a recent UNICEF study that collected additional data in three oblasts found that, as a consequence of the iron-folic acid 101 supplementation program anemia plummeted between 2002 and 2005: by 21.4 percent in Karakalpakstan, by 22.6 percent in Fergana, andby 44.9 percent inKhorezm.66 Figure 4. Evolution ofthe nutritional status of childrenunder 3 years. ~ 60 50 * 40 fa 30 20 10 0 Stunting Wasting Anemia (any form) 0 1996, lowest quintilem2002, lowest quintile 1996, all 2002, all Source: DHS, 1996 and UHES, 2002. For stunting and wasting children 0 to 35 monthsold are included, for anemia children 6to 35 months old are included. Disabilities Analysis of regional URPS data designed specifically for measuring disability rates reveal that almost 12 percent of all individuals ages 7 and older have at least one serious difficulty or a full limitation (in seeing, hearing, movement, learning, communication or ~elf-care).~' But official disability status is granted only to 3.8 percent of the population in the regions under investigation(national share of the officially disabledwas 3.2 percent in 2003). Official disability status reduces enrollments by about 24 percentagepoints among children of ages 7 to 14; this negative effect increases to 40 percentage points drop among 15 to 18 year olds. If one considers alternative indicators of disability the enrollment effect is smaller but still sizable. For example having at least one full limitation in seeing, hearing, movement, learning, communication or self-care is associated with about 10 percentage points drop in enrollmentsamong 7 to 14 year olds and 35 percentagepointsdrop in enrollmentsamong 15 to 18 year olds. Disability status is strongly correlated with being out of labor force and being poor, although these relationshipsshouldbe interpretedwith caution because it is difficult to speculate about the direction o f the effect. For example poor adult healthcan leadto poverty through its negative impact on labor force participation and productivity; but poverty can also be expected to contribute to poor health status of adults. With this proviso, the data revealthat official disability status is associatedwith a 52 percentage points reduction in the probability of being economically active (37 percentage points reduction if one considers having one full limitation). Havinga full limitation is associatedwith a 7.8 percent decline in household consumption; with no statistically significant linkage with official disability status presumably becauseofthe relativelygenerousbut low-coveragedisability pensions. 66 Tho w e y Largcted 3,000 mdividualr m Karakalkaplntan,Khorem and FerganaOblw " A n e m RcvenuonandControl P r o m Evaluation m Uzbclustan" UNICEF(2005) providsr informauonabut methddogy, samplingand findings 67 Ses Scott and Mete (2006) for detailson various alternative indicatorsof disability,and*SI relauonstdp ullh mmllmenti, labar force paRlcipationand welfare. 102 Health information, behavior andprevention Since the previous section established the health disadvantages of the poor, the next step is to explain the key factors that yield this association. 68 One such set of factors has to do with access to information. Available evidence suggests that this may indeed be an important issue in the Uzbekistan context. For example, the 2002 HES data show that only 28 percent of the men and 38 percent of the women measuring hypertensive (at either lSt or Zndmeasurement) at the time of the survey were ever told by a healthprofessionalthat they hadhypertension. In 2002, 81.3 percent of the men and 83.9 percent of the women of reproductiveage had heard about TB, of which 72.6 and 81.5 percent know at least one TB symptom respectively, and 68 and 71.3 percent know the correct way of TB transmission in the same order. However, the poor are much less likely to have this informationas shown by Figure 5. The rural population is disadvantaged, and the urban-ruralgaps6' for the three indicatorsdepict a different picture for men and women. It is larger for females interms of hearingabout the disease, while it is larger for males interms of knowingsymptoms or transmission. The gender-wealth interactiondeserves attention as well. Females are 14 percentage points more likely to know at least one symptom of TB, but this gap reduces to half among the wealthiest females and males. A similar trend is visible when comparing the percentages of those who knowthe correct way of TB transmission. Even though 90.2 percent of females and 94.5 of males (of reproductiveage) have heard about AIDS, only 58 and 62.7 percent know at least one main correct'wayof preventionre~pectively.~'The rural- urban gap is wider for women. When compared with urban areas, 6 percent less women versus 2 percent less men in rural areas had heard about AIDS, while 20 percent less women versus 15 percent less men in rural areas know at least one correct way of prevention. This knowledge of prevention varies from around 50 percentinthe poorestquintileto above 70 percent inthe wealthiest. Figure 5. Knowledge of TB and AIDS andpoverty. I ~ U l T B InfonnationUl#Ds 1 ice - m - 0 E e 70 . n 0. / / Source: UHES, 2002, own computations Smoking is another health concern in Uzbekistan among the male population, especially given that lungs and respiratorysystem diseases are listed as the second most common type of chronic illness in 2005.7124.1 percent of the males smoked at the time of the survey, and smoking is more common 68 The lnfomationtn this ~ m o comes from the 1002 W n S and refers 10 women OfreprOduNVoage (15-49-ycar-clds) and menofiqrcduai~~ (15-59-ycar-olds) age 69 Expressedas (urban-rural)rate * ICQlurbanr a t 70 "At lesst one w n way of prevention" for HIVIAIDS refers 10 recognuingat least oneof absunenscfrom ICX, condomwage, and havingoneparmer only. 71 According10 the KES dam, 14 percentofthe chrorucally 111list the lungsandrespiratorysystemas king the body pBn most affccfed. 103 among the wealthier and more educatedmen: 18.9versus 32 percent for rural/urban, 15.1 versus 37.6 percent for poorestlwealthiest, and 21.9 versus 27.9 for basic-secondary general/ higher education attainment. This might meanthat price can be a deterrent to smoking amongthe poorest. Females are much less likely to smoke. In2002, only 0.9 percentof reproductiveage females smoked at the time of the survey, while 1.3 percenthadever smoked 100cigarettes. Among the mothers of children 0-6 years, only 0.3 percent were smoking when pregnant or during pregnancy, but 8.5 percent of the children under 6-year-oldwere exposed to smoking during their 40 days of life for an average of 17 hours per week (see Table 4). This behavior towards children is slightly more visible among the wealthiest, and much more visible in urban areas (11.1 percent comparedwith 6.9 percent in rural areas). In2002, 97.8 percentof ahildrenless than 6 monthsand 95.7 percentof children6-11months oldwere still being breastfedat the time ofthe interview,ofwhich 83.7 and respectively99.4 percentwere not exclusivelybreastfed. The same table also shows the trends in usage of Vitamin D drops, which supports bone growth and calcium absorption. Vitamin D deficiency is relatedto weak bones, poor bone development and bowed legs (althoughthere is some disagreement on whether vitamin D drops are neededfor childrenwho are beingbreastfed and exposed sufficientlyto sun). About 37 percent of the childrenunder age of 6 (and 8.9 percent of the 1-2 months, and 10.4 percent of the 3-4 months) ever received these drops for an average of 2.8 weeks (respectively 1.6 and 3 weeks). The percentage of childrenreceivingthe drops in the wealthiest quintile is more than 3 times higher and their average time to receive the drops is 40 percenthigherthanthe percentageof childrenbelongingto the poorest quintile. Table 4. Children'sexposureto smokingandthe usage of vitaminD drops Poorest 2nd 3rd 4th Wealthiest Total quintile quintile quintile %ofkids0-71 m.0. exposedto smoking in the first 8.5 5.9 8.6 8.9 10.8 8.5 40 days of life average no. hourdweekkids 0-71 m.0.were 13.7 11.5 22.1 20.5 15.5 16.9 exposedto smoking in the first 40 days of life (conditionalto beingexposed) % of kids 0-71 m.0.ever receivingvitamin D drops 20.1 25.0 35.3 42.0 67.1 37.1 average no. weeks kids 0-71 m.0.receivedvitamin 2.5 2.5 2.4 2.7 3.5 2.8 Ddrops (conditionalto ever receiving) Source:UHES,2002, own computations. Information only for those kids whose mothers are 15-49 year olds, participatinginthe survey Access to and utilizationof health careservices In Uzbekistan, access to health facilities is relatively widespread, even in rural areas (Table 5). According to the SVA/SVP Passport database, Uzbekistan had 2105 SVNSVPs as of January 2004, each servicingan average of 4752 people and havingan operational radius of 5.6 km.72However, the characteristics of those rural health facilities vary significantlyacross oblasts. Only 10 percent of the facilities reported having transportation means, of which 30 percent were in bad or unsatisfactory condition. The utilities of the SVP/SVAs were also in poor condition: 21 percent of the facilities had central cold water, 7 percent of the facilities had central hot water and there was no hot water in 74.7 percent of the facilities. Central sewerage was present in 6 percent of the facilities, while the heating was providedcentrally in 8 percent, from boilers in 33 percent, from gas in 35 percentand from coal in 24 percent. Only half of the facilities had communication means (mainly telephone), of which 81 percentwere in good or satisfactorycondition. 12 The averagenumber of ppulation reNed vaned between below 3,Slx) in Bukhm. Karakalpakrlanand Syrdarya. and over 6.100 mFerghanaand Kharezm AI lhe m e Ume. the average mdusof service fluctuated between2 8 !amBuwlara.andover I2lunmKarakalpakrtanandNavoi 104 Table 5. Proximitv to a health center YOof population having the facility within the Average km to the closest facility, if facility not PSU within the PSU Rural Urban :;yt Wealthiest 60% Total Rural Urban ziyt :Elthiest Total Andijan 51.2 53.4 59.5 46.5 51.7 6.2 2.2 6.4 ' 4.1 4.8 Hospitals Kashkadarya 46.5 81.4 ' 54.5 52.6 53.4 19.4 2.4 21.0 15.1 17.8 Tashkent city 41.3 44.9 43.4 4.0 3.5 3.7 Andijan 38.6 77.1 50.5 45.3 47.4 7.6 2.9 7.4 6.2 6.7 PolYclinic Kashkadarya 61.5 81.6 61.0 68.4 65.4 19.7 2.2 21.9 14.1 17.8 Tashkent city 67.7 65.8 66.5 3.2 2.5 2.7 Andijan 54.7 7.2 52.4 38.2 43.9 11.4 2.5 6.0 6.2 6.2 PHC facility Kashkadarya 74.1 32.4 72.0 61.8 65.9 15.3 4.0 16.1 13.9 14.5 Tashkent city 16.2 21.4 19.3 3.2 3.3 3.3 Ambulance Andijan 32.7 48.2 44.8 30.5 36.2 7.5 3.2 7.6 5.8 6.4 station Kashkadarya 42.9 62.9 45.9 47.5 46.9 23.0 2.9 23.7 19.1 21.2 Tashkent city 34.9 39.4 37.6 3.7 3.7 3.7 Maternity AndiJan 70.5 58.4 74.0 63.6 67.8 9.3 2.6 5.9 6.7 6.4 consulting Kashkadarya 76.0 77.6 76.1 76.4 76.3 17.4 0.6 18.0 12.4 14.7 center Tashkent city 61.9 64.5 63.4 3.5 3.0 3.2 Andijan 58.7 92.6 67.6 65.5 66.4 5.7 0.0 5.9 4.7 5.2 Drugstore Kashkadarya 71.1 100.0 70.7 80.8 76.8 20.1 24.2 15.9 20.1 Tashkent city 83.8 82.4 82.9 1.7 1.7 1.7 Andijan 72.7 71.2 74.3 71.1 72.4 6.4 2.9 6.2 4.6 5.2 Pediatrician Kashkadarya 79.8 93.5 76.5 86.5 82.5 16.2 0.5 20.4 7.2 14.9 Tashkent city 64.8 61.5 62.8 3.3 2.8 3.0 The official statistics on health care utilization are useful to display unconditional utilization rates and broad time trends, while the survey data can shed light on linkages with poverty and inequality. Table 6 shows significant decreases in inpatient care admissions from the mid-1990s, with some recovery starting from 2000. In contrast, outpatient contacts per person per year have increased steadily throughout the period. Table 6. In-patientand outpatient data 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 In-patient care admissions per 100 19.2 18.6 16.4 15.9 12.9 12.9 13.3 13.8 14.0 14.2 Outpatient contacts per person per year 5.8 6.7 7.5 6.7 7.1 8.1 8.4 8.3 8.5 8.5 Source:WHO-HFA. 2005. Additional evidence on health care utilization conditional on reporting sickness comes from household surveys. Table 7 shows conditional utilization rates at around 70 percent between 2000 and 2003. Even though utilization in the poorest quintile is lower than in the wealthiest quintile in 2000/01, this pattern reverses in 2002 and 2003. Health care utilization rates in urban areas tend to be lower than in rural 105 areas. Bothtrends should be interpretedwith some caution, since poor individuals(or those who reside inrural areas) faced with a healthconditionmay be less likely to report it as sickness to start with. Table 7. Utilization rates (conditionalon beingsick), last 30 days Poorest 2ndquintile 3rdquintile 4Ihquintile Wealthiest Total quintile quintile 2000101 Rural 65.5 79.3 77.1 80.1 73.3 75.5 Urban 69.1 65.9 66.8 63.5 67.8 66.6 Total 67.2 73.2 72.0 71.4 69.5 70.7 2002 Rural 71.4 71.9 77.4 75.7 74.1 Urban 77.6 74.6 60.0 67.3 65.8 67.1 Total 73.8 74.2 66.5 72.7 67.9 70.3 2003 Rural 73.3 77.6 78.3 71.1 74.6 75.0 Urban 54.1 58.9 69.3 73.5 62.4 64.4 Total 65.8 71.8 74.7 72.5 64.7 68.9 Source: HBS, own computationusingfood quintiles. The conditionalutilizationrates for females slightly decreased over the periodof three years from 71.9 percent in 2000/01 to 68.5 percent in 2003, while the male rates stayed the same at approximately 69 percent duringthe same period. However, gender inequalitiesare probably more severe than what the conditionalutilization rates may imply. In 2002, less than half of the females expressed final or joint- final say on issues regardingtheir own healthor (to some extent) controllingthe money to buy drugs for their own use, while one-fourthof females are only allowedto go accompaniedor are not allowedto go at all to a health center (Table 8). The situation is worse for the poor and for those who reside in rural areas. Working females have more control over their healthcare: 44.9 percent report havingowdjoint final say in their own health care; 72.2 percent report (partially)controllingmoney to buy medicine for their own use, and 18.6 percent report being allowed to go alone to a health center. In a multivariate setting, education, wealth, beingemployedandbeingeither Russianor of othednon-Uzbek ethnicityand residing in Tashkent city increase the probability of making decisions, while living in rural areas, the share of male adults in the householdand a larger family decrease the probabilityof makingdecisions (see Annex 1 for a complete set ofthe multivariateresults). Table 8. Females'decision makingpower on healthissues, 15-49 year olds. Poorest 2nd 3rd 4th Wealthiest Total quintile quintile quintile quintile quintile % having own/joint final say on own health care 26.3 34.2 38.5 46.1 68.0 42.7 YO(partially) controlling money to buy medicine for herself 32.9 42.3 42.7 51.1 62.9 46.5 %allowed to go aloneto health center (vs. not alonehot at all) 69.0 77.8 74.1 76.9 80.6 75.7 Source: UHES 2002, own computations. Other reasonsfor not seeking health care The main reasons for not seeking health care in case of sickness are (i)having consulted a doctor or traditional healer previously and following instructions (33.9 percent in 2003 from 27.8 percent in 2000/01), and (ii)knowinghow to cure oneself (35.2 percent in 2003 from 25.7 percent). One possible concern about these two leadingreasons for not seeking care is the presence of poor communications betweendoctors and patients. The HFS found that the share of caretakers (for children0-5 year of age) who understood the treatment varied between 18 percent in Navoi and Syrdarya and 34 percent in Ferghana. Additionally, between65 and 80 percentof patients or caretakers were assessedto knowhow to administerthe prescribeddrugs. 106 Inability to afford health care consultations does not come up as a main reason for not seeking consultations (Although, affordability of medicines i s a major issue, as discussed later.): 12.9 percent reported having no money for either transportation or for treatmenvdrugs in 2000/01, versus only 7.8 percent in 2003. In 2005, findings from URPS show that the transportation costs for seeking care (excluding hospitalizations) for the average household amount to about 10 percent o f total household expenditures on health in Andijan and Kashkadarya, and 6 percent in Tashkent city (with hospitalizations, the percentages are 8 and 5 respectively). Choiceof providers Seekingcare at polyclinichospital in town was the most common health care option in Uzbekistan in 2003 (an increase of 27 percent occurred between 2000/01 and 2003, explained mainly by the large increase of this choice in urban areas). Inurban areas, the dominating choice of providers is going to a polyclinic/hospital in town, going to a polyclinic/hospital in the rayon center, and calling the doctor to the house, which account for a total of 78.3 percent of all choices in 2000/01 and 82.6 percent in 2003. The wealthiest 40 percent of the population are more likely to go to a polyclinichospital in town, while going to a polyclinic/hospital in the rayon center or calling the doctor to the house are more often selected by the poorest 60 percent of the population. In rural areas, people opt for more types of providers as their main choices -around 85 percent of people go to out-patient clinics (rural medical point or ambulatorium), polyclinicshospitals in the rayon center, feld'sher-akusher points, or small rural clinics or polyclinics, or call the doctor to the house (see Table 9). Table 9. Choice ofproviders Poorest 2nd 3rd 4th Wealthiest quintile quintile quintile quintile quintile Total Rural Out-patient clinic (rural medical point or ambulatorium) 42.0 29.1 23.3 25.4 27.6 29.4 Feld'sher-akusherpoint 11.2 15.8 5.8 17.8 21.1 14.1 Small rural clinic or polyclinic 9.7 11.5 14.0 7.9 10.3 10.8 Polyclinic/hospitalin the rayoncenter 20.4 20.9 22.1 14.4 17.2 19.1 Polyclinic/hospitalin town 1.7 6.4 12.5 9.8 8.3 7.8 Called doctor to the house 5.0 11.7 15.4 12.3 9.2 10.9 Traditionalhealer 1.8 1.5 3.7 7.4 2.2 3.3 Other 8.2 3.1 3.2 5.1 4.1 4.7 Urban Out-patient clinic (rural medicalpoint or ambulatorium) 2.1 0.0 3.4 4.7 1.3 2.3 Feld'sher-akusherpoint 8.0 9.4 8.7 7.0 3.2 5.5 Small rural clinic or polyclinic 2.7 0.0 0.0 1.5 0.4 0.8 Polyclinic/hospitalinthe rayoncenter 19.1 10.3 16.8 13.2 9.4 11.9 Polyclinidhospital in town 45.8 58.8 43.1 54.3 63.4 57.4 Calleddoctor to the house 11.5 19.2 16.5 9.1 14.2 13.3 Traditionalhealer 0.0 0.0 4.0 3.4 1.6 2.1 Total Out-patientclinic (ruralmedicalpointor ambulatorium) 29.1 21.7 15.8 13.5 7.1 14.7 Feld'sher-akusherpoint 10.2 14.2 6.9 11.5 7.1 9.4 Small rural clinic or polyclinic 7.4 8.6 8.8 4.2 2.6 5.4 Polyclinic/hospitalinthe rayoncenter 20.0 18.2 20.1 13.7 11.1 15.2 Polyclinic/hospitalintown 15.9 19.7 24.0 35.5 51.4 34.6 Called doctor to the house 7.1 13.6 15.8 10.4 13.1 12.2 Traditionalhealer 1.2 1.1 3.8 5.1 1.7 2.6 Other 9.0 2.9 4.8 6.1 6.0 5.8 Source:HBS, 2003, own computations. 107 Out-of-pocket servicepayments -formal and informal After 1991, the health services went from a free-of-charge to a fee-for-servicesystem. Nevertheless, some services and medicines are still provided for free for exempted groups (chronically ill,low- income, widows with many children, Chernobyl and Afghan war veterans) and certain services. The inpatients should be provided with free medication, while outpatients would have to fully pay for medicines. However, in practice, inpatients are often required to bring their own drugs and medical supplies. As discussed previously, amongthe reasons for not seeking healthcare are the travel costs and service fees, both formal and informal. There are four types of out-of-pocket payments: official user-fees for public-provided services, semi-official user-charges for consumables (such as drugs and medical supplies), informal payments, and private fees. The analysis of out-of-pocketpayments uses the 2005 URPS (first wave, March-April,three regions) data. The total cost for services includes laboratorytests, vaccinations, consultations, and informalpaymentsto medicalstaff providingthe services. Aggregating over the three available regions, almost half of those seeking care paid service fees, and their average payment was approximately 12,500 soums. A larger percentage of the wealthiest individualswho seek care pay for services, and their payments are larger. While the same percentage of rural and urban population at the extremes of the wealth distribution (poorest and wealthiest) pay for care, the expenditures are larger in rural areas, possiblybecause of the added transportation costs. For the mid- range o f the distribution, a larger percentage of the urban population is paying for care, and their payments are 11to 27 percentlargerthan ofthe rural population. Of the people paying for services, almost 40 percent reported making informal payments. The incidence of informal payments is lowest for state doctors, and largest for other qualified health personnel. The service fees for privatedoctors are morethan twice those madeto state doctors, butthe incidence of service fees for privatedoctors is relativelylow. This can be partiallyexplained by the fact that as of 2001, "all privatehealth care institutionsstill haveto provideat least20 percent of healthcare within the state system. Thus, the state sets and pays at least 20 percent of service fees."73 The average payments in rural areas are 1.6-2.3 times larger than in urban areas, and the average payments in Tashkent city are larger than in urban Andijan and Kashkadarya. The trends across socioeconomic quintilesvary by regionand location. Another dimension of the out-of-pocketpayments is the informal payment^,'^ which might be a serious impediment in seekinghealth care. The percentageof patients payinga positive(unofficial) amount for health care ranges from 11.4 in Andijan to 22.7 in Kashkadarya. The ruralhrban differences are noteworthy in Kashkadarya. In Andijan and Kashkadarya, other qualified staff (nurses, feldsher, dentist, pharmacist) category has a larger proportion of people making informal payments, while in Tashkent city, the same is true for the private doctors. In Andijan and Kashkadarya, there is no a clear trend by socioeconomic status, while in Tashkent city individuals from the wealthiest 20 percent are four times more likelyto make informalpaymentscomparedto the poorest40 percent. To illustratethe importanceof informalpayments, it is usefulto consider the ratioof informal-to-formal payments. The informal payments are 0.6 times the formal payments in urban Andijan and Tashkent city, while being almost equal with the formal payments in urban Kashkadarya. In rural areas, the informal payments account for thirty to forty percent of the formal ones. There is no robust trend by socioeconomic characteristics. 73 Health Care Systems10Traniition Uzbekistan(2001). publishedby ths EuropsanObservatoryon HealthCare Synemr 14 We differentme between formal and tnhrmal pqmentsfor ~wlcerwing the following two questions."How muchdid your householdpay officially for there medical sewices for [NAME],includingpspentr for laboratory tests, ~accinsfl~n~consultationsm the last 30 days (do not include Ihe con of medicme)?"and "What was the value (in soums) of unofficial expslwas of you householdm the lm 30 days. includingg i f i and iemcer 10 the and mdcal staff that providedmedicalcareto INAME]?'' 108 Table 10. Incidenceof paymentsfor services (incl.laboratory tests, vaccinationsandconsultations) Poorest 2nd 3rd 4th Wealthiest State Private Other quintile quintile quintile quintile quintile doctor doctor qualified Total Rural % paying, out of those seekingcare 42.1 34.4 41.8 48.6 47.1 46.9 54.8 43.6 42.9 averagepayment, for those paying 11013 15975 6954 7919 20294 12113 21720 5496 12806 Urban YOpaying, out ofthose seekingcare 41.2 47.1 51.2 58.5 48.4 47.8 59.9 63.4 50.8 average payment, for those paying 6806 17686 8820 9283 15305 9467 23717 7292 11886 Total %paying, out of those seekingcare 42.0 41.3 47.5 54.3 47.6 47.4 57.4 50.3 46.6 averagepayment, for those paying 10324 17041 8171 8762 18197 10740 22791 6265 12337 Source: URPS, March-April 2005, averages over Andijan, Kashkadarya, and Tashkent city. "Other qualified" category comprises nurses, feldshers, dentists, and pharmacists;the omitted categorycompriseshealers, midwifes, and others Table 11.Incidenceof informalpayments for services Poorest znd 3rd 4th Wealthiest State Private Other quintile quintile quintile quintile quintile doctor doctor qualified Total Rural % makinginformal payments 17.4 15.9 9.1 16.3 17.2 15.3 16.0 23.0 15.6 informal-to-formal payments 0.49 0.26 0.45 0.19 0.24 0.44 0.18 0.22 0.34 Urban %making informal payments 11.6 15.9 20.2 23.2 25.0 17.8 28.0 28.1 20.2 informal-to-formal payments 0.36 0.82 0.69 0.43 0.84 0.56 0.79 0.64 0.65 Total %making informal payments 16.5 15.9 15.8 20.2 20.5 16.6 22.2 24.7 17.8 informal-to-formal payments 0.47 0.61 0.60 0.34 0.48 0.50 0.50 0.41 0.50 Source:URPS, March-April2005, averages over Andijan, Kashkadarya,and Tashkentcity. "Other qualified" category comprises nurses, feldshers, dentists, and pharmacists;the omitted category compriseshealers, midwifes, and others In the HBS, total household spending on health comprises payments for medical equipment, services, and drug expenditures. Table 12 presents the average nominal household spending on health in the month preceding the survey. In 2003, the average Uzbek household spent 2238 soums per month on health, up from 1422 soums in 2000/01. Spending on medicines constitute 84 percent of these expenditures. Households in the wealthiest quintile spend 1.9 times more on health than the poorest households (3 159soums comparedto 1646amongthe poorest). 109 Table 12. Average householdout-of-pocketpaymentsonhealth. Poorest 2nd 3rd Wealthiest quintile quintile quintile 4th quintile quintile Total Medical equipment 98 145 96 148 153 131 Medical services 74 91 111 176 225 146 2000/01 Drugs 755 1151 1155 1270 1271 1145 Total 926 1386 1362 1594 1648 1422 as YOof total food expenditure 5.1 6.1 5.4 6.2 5.4 5.6 as YOof total expenditure 3.4 3.9 3.4 4.0 3.2 3.5 Medical equipment 134 162 141 138 271 177 Medical services 51 46 96 223 286 157 2002 Drugs 1189 1240 1246 1538 2239 1561 Total 1374 1447 1484 1898 2796 1895 as YOoftotal food expenditure 4.6 3.2 2.9 3.5 5.3 4 as YOoftotal expenditure 3.3 2.4 2.1 2.5 3.5 2.8 Medical equipment 131 176 120 230 209 179 Medical services 93 71 120 252 267 175 2003 Drugs 1422 1256 1653 1883 2683 1884 Total 1646 1503 1893 2365 3159 2238 as YOof total food expenditure 5.5 3 3.5 4.5 5.4 4.5 as YOof total expenditure 3.8 2.3 2.5 3.0 3.4 3.O Source: HBS, own computations basedon food consumption-expenditures quintiles. The variations in health spending are also reflected in the high concentration indexes (see Figure 6). The highest differences are for medical services, followed by drugs and medical equipment, with the inequality in spending on drugs steadily increasing over years: from a concentration index of 0.082 in 2000/01 to 0.142 in 2003. This pattern in spending arises from a combinationof factors includingthe existence of a pro-poor fee-waiver program bearing most of the health costs faced by poor (PHC services are exempted, and the poor might utilize them more); smaller utilization rates among the poorest; and wealthier individuals seeking more expensive (and most probably better quality) health care. Figure 6. Variations inhealthexpenditures 0.6 '6 0.2 --erredicalequiplect--enedcalservices+dngs +tdal 110 ANNEX1.CHILDHEALTH INDICATORSINSELECTED COUNTRIES Births in three Births in five 0-4 years preceding 5-9 years preceding years preceding years preceding the survey the survey the survey Weight- Infant Under-5 Infant Under-5 for-age Height- for- mortality mortality mortality mortality height (W) - ( 5 0 ) (1qO) (5qO) below SD below - 2 SD Armenia 2000 36.1 39.0 50.5 55.0 10.6 2.6 13.0 2.0 23.9 Egypt 2000 43.5 54.3 66.1 84.2 19.3 3.2 18.7 2.5 30.5 Jordan2002 22.1 27.0 26.5 31.5 8.4 2.4 8.4 2.0 34.2 Turkey 1998 42.7 52.1 54.0 67.2 12.0 2.5 16.0 1.9 Source: ORC Macro, 2006. MEASURE DHS STATcompiler. htlu://www.iiieasuredhs.com. May 11 2006. Anemia levels for Uzbekistanare for children 6-35 monthsold. The anemiaprevalencerepresentshouseholdlevelestimates" for Armenia, Kazakhstan, Turkmenistan and Peru 75 For all the other countries where both household and individual level estimates were available, the two estimates were very close (under 0.5 differences). 111 ANNEX2. FEMALES' DECISIONSONHEALTHCARE, RESULTSFROMA PROBIT MODEL Own/joint final say in (at least partially) controls Allowed to go alone to decisions about own health money to buy medicine for self health center wl `work' wlo `work' wl `work' wlo `work' wl wlo `work' variable variable variable variable `work' variable variable Age 0.007 0.012 0.070 0.080 0.034 0.038 (0.95) (1.68)* (8.67)*** (1 0.10)*** (5.99))** (6.84)*** age2 0.002 -0.005 -0.079 -0.094 -0.038 -0.044 (0.19) (0.43) (6.26)*** (7.47)* ** (4,19)*** (4.87)*** PTUISPTULicee 0.106 0.113 0.049 0.064 0.080 0.083 (3.33)*** (3.54)*** (1.54) (2.01)** (3.57)*** (3.71)*** TechnikudCollege 0.033 0.046 0.044 0.071 0.066 0.073 (1.49) (2.09)** (1.83)* (3.06)** * (3.83)*** (4.34)*** Universityhstitute 0.114 0.139 0.022 0.072 0.097 0.110 (3.42)*** (4.27)*** (0.66) (2.23)** (4.05)*** (4.79)* ** Secondquintile 0.043 0.046 0.052 0.056 0.033 0.035 (1.62) (1.70)* (1.85)* (1.99)** (1.68)* (1.80)* Middle quintile 0.062 0.063 0.070 0.072 -0.016 -0.014 (2.21)** (2.27)** (2.38)** (2.47)** (0.75) (0.64) Fourthquintile 0.064 0.062 0.087 0.081 -0.053 -0.054 (2.17)** (2.10)** (2.75)*** (2.56)* * (2.19)** (2.25)** Highest quintile 0.040 0.037 0.077 0.070 -0.130 -0.13 I (1.03) (0.95) (1.86)* (1.69)* (3.85)*** (3.89)* ** 1:currently working, self- 0.066 0.138 0.054 reported (3.52)*** (7.33)*** (3.68)*** Russian 0.401 0.404 0.183 0.193 0.139 0.142 (4.40)* ** (4.47)*** (3.05)*** (3.35)** * (3.45)*** (3.64) *** Karakalpak 0.066 0.060 -0.122 -0.134 0.119 0.116 (1.57) (1.42) (3.01)*** (3.3 5)*** (3.90)* ** (3.74)* ** Tajik -0.002 0.001 -0.055 -0.052 -0.035 -0.035 (0.03) (0.02) (0.92) (0.86) (0.72) (0.72) Kazakh 0.048 0.046 -0.007 -0.010 0.116 0.115 (I.06) (1.01) (0.15) (0.21) (3.63)*** (3.53)*** Other ethnicity 0.155 0.156 0.130 0.128 0.064 0.064 (2.91)*** (2.92)* ** (2.59)*** (2.47)** (1.88)* (1.84)* Share of children 0-5 year olds 0.085 0.069 0.040 0.008 0.201 0.189 (0.99) (0.81) (0.47) (0.10) (2.92)** * (2.74)*** Share of females 7-14 year olds 0.071 0.059 0.019 -0.008 0.006 -0.004 (0.79) (0.66) (0.20) (0.09) (0.08) (0.06) Share ofmales 7-14 year olds 0.024 0.011 -0.041 -0.071 0.102 0.094 (0.26) (0.12) (0.43) (0.75) (1.45) (1.33) Share of males 15-59year olds -0.194 -0.196 -0.147 -0.149 0.007 0.006 (2.1O)** (2.13)** (1.57) (1.59) (0.09) (0.09) Share of females 50+ year olds -0.040 -0.024 0.077 0.120 0.026 0.042 (0.30) (0.18) (0.55) (0.88) (0.26) (0.42) Share of males 601- year olds -0.124 -0.145 -0.240 -0.286 -0.140 -0.159 (0.76) (0.89) (1S O ) (1.79)* (1.14) (1.29) Ln(hhsize) -0.132 -0.131 -0.182 -0.180 -0.023 -0.021 (5.49)*** (5.46)*** (7.24)* ** (7.22)*** (1.21) (1.13) OblasPKarakalpakstan -0.411 -0.408 0.010 0.018 -0.064 -0.061 (12.00)*** (11.87)*** (0.24) (0.45) (1.82)* (1.73)* oblast==Andij an -0.483 -0.483 -0.347 -0.346 -0.008 -0.012 (14.64)*** (14.67)*** (8.08) *** (8.21)*** (0.22) (0.31) oblast==Bukhara -0.428 -0.425 -0.120 -0.105 -0.381 -0.372 (11.87)*** (11.70)*** (2.55)** (2.22) ** (8.46)* ** (8.32)*** oblast==Jizzakh -0.449 -0.451 0.040 0.014 0.030 0.020 (12.98)* ** (I3.14)*** (0.76) (0.28) (0.77) (0.50) oblast=Kashkadarya -0.484 -0.485 -0.194 -0.205 -0.353 -0.360 (14.98)*** (15.02)* ** (4.61)*** (4.87)*** (8.60)*** (8.76)* ** oblasPNavoii -0.346 -0.342 0.084 0.095 -0.206 -0.200 (7.32)*** (7.22)*** (1.28) (1.50) (3.91)*** (3.80)*** oblast==Namangan -0.429 -0.429 -0.280 -0.279 -0.037 -0.038 (11.56)*** (1 1.55)*** (6.39)*** (6.40)*** (0.93) (0.95) 112 Own/joint final say in (at least partially) controls Allowed to go alone to decisions about own health money to buy medicine for self health center oblasP=Samarkand -0.492 -0.492 -0.024 -0.031 0.005 0.002 (15.96)*** (15.96)* ** (0.57) (0.75) (0.15) (0.05) oblast-Surkhadarya -0.466 -0.466 -0.130 -0.143 -0.021 -0.027 (14.07)*** (14.06)*** (2.54)** (2.82)*** (0.49) (0.63) oblast=Sirdarya -0.432 -0.43 1 -0.032 -0.029 0.027 0.026 (10.43)*** (10.49)*** (0.50) (0.44) (0.53) (0.50) oblast===Tashkentregion -0.436 -0.435 -0.040 -0.034 -0.053 -0.050 (12.78)*** (12.68)*** (0.95) (0.82) (1.49) (1.40) oblast-Ferghana -0.437 -0.435 -0.034 -0.028 0.106 0.106 (13.63)*** (13.53)*** (0.95) (0.78) (4.07)*** (4.05)*** oblast=Khorezm -0.417 -0.414 0.227 0.239 -0.227 -0.219 (12.69)*** (12.51)*** (5.48)*** (5.82)*** (6.02)*** (5.85)*** Rural -0.093 -0.091 -0.082 -0.079 -0.073 -0.073 (4.00)*** (3.93)*** (3.20)*** (3.08)*** (3.96)*** (3.89)*** Observations 5393 5393 5452 5452 5451 5451 Robust z statistics inparentheses *significant at 10%; ** significantat 5%;*** significant at 1% 113 CHAPTER 5: SOCIAL PROTECTIONAND THE POOR This chapter reviews the effectiveness and efficiency of the main social protection (SP) cash benefits in Uzbekistan in fighting poverty and reducing income vulnerability. For this purpose, the chapter builds on the analysis presented in the previous Living Standards Assessment report, and updates that analysis using the most recent available rounds of the Household Budgets Survey (2002 and 2003) and the findings o f a qualitativeresearchfieldwork carriedout inthe summer of 2006. The chapter beginswith a review of mainfindings, followed by an overviewof SP programs. Then, it assesses the coverage, targeting incidence, adequacy, and overall effectiveness of these programs. Finally, the chapter concludeswith areview of key issues. MAINFINDINGS 1. Previous analyses of the Uzbekistanmeans-testedbenefits (benefits for low income families and benefits for families with children under 16 years old) praised the social assistance schemes implemented by Mahallas as being "a clear break with the past when benefits were often administered by state-owned enterprises, with entitlement either being universal or linked to households' formal cash income alone". Indeed, the introduction of means-test schemes and delegating their implementationto local communities represented a significant improvement of the social assistance system in Uzbekistan. However, this Poverty Assessment, as well as the previous one (2003), signal that the performance of the social protectionsystem is affected by errors of inclusionand exclusion, as well as by horizontal inequity, which lead to lower targeting performance of the Uzbekistan's social assistance benefits as compared with other countriesin the region. 2. Social protectionspending as percent of GDP decreased in recentyears. This drop in spending is the result of slower growth of social protection expenditures relative to GDP growth, which led to 'a relatively significant reductionof benefits generosity relativeto the average wage. 3. In 2003 about 57 percent of the total population, respectively 62 percent of the poorest quintile, was covered by at leastone social protectionprogram. 4. Despite a decrease in the generosity relative to the average wage, the pension benefits offer a significant protection to the poor, by covering almost half of the population in the poorest quintile and more than 40 percent ofthe poor beneficiaries' consumption. 5. Both coverage and targeting of the Uzbek social assistance programs are rather modestwhen compared to other countries in the region. If we compare only the means-tested programs (all non-contributory programs basedon income-, means-, or proxy means-testmechanisms), ignoringother programsthat form the mix of social assistance, Uzbekistanperformsrelativelywell on coverage, but poor on targeting. 6. Some population groups did not benefit much from economic growth and they started to rely more on social protectionprograms instead: especially the elderly, those who reside in rural areas and in the poorer regionsofthe country. 7. The unemployment benefitfails to offer protectionagainsttemporary loss of income. 8. The targeting performance of means-testedtransfers is low. The means-testedprograms are affected by significant inclusion and exclusion errors. By design, too much space is given for discretion and arbitrariness at local level through the lack of more precise guidelines/methodology regarding the evaluation of household wellbeing and imputation of incomes. Combined with the absence of (proxy) poverty criteria for the distribution of funds between regions and communities, this leads further to 114 horizontal inequity: wealthy regions/communities can receive as much, or even more, funds as poor regions/communities, and, as a consequencethe poor will not receive similar treatment in all regions. 9. The introduction of the cash compensation mechanism for utilities using the eligibility criteria inherited from privilegesdoes not offer protectionto low income groups. OVERVIEW OF THE sOCL4L PROTECTIONSYSTEM In Uzbekistan, as in many other countries in the region, the social protection system comprises two main pillars - social insurance (SI) and social assistance (SA). From a regional perspective, the levels of both social insurance and social assistance spending in Uzbekistan compare well with other ECA countries (Table1). Table 1. Social protectionlevels inselected ECAcountries Yo of GDP GDP per capita SP S I SA (2000 PPP$) Moldova(2004) 10.4 8.7 1.7 1,589 Kyrgyzstan(2003) 7.8 6.0 1.8 1,654 Uzbekistan(2004) 7.9 6.2 I. 7 1,718 Armenia (2002) 4.4 3.3 1.1 3,019 Albania (2004) 6.8 5.9 0.9 4,575 Kazakhstan(2004) 4.1 3.1 1.o 6,838 Bulgaria(2004) 10.7 9.5 1.2 7,424 Romania(2004) 9.0 7.9 1.1 7,793 The main social insurance cash benefits considered in this chapter are pensions and unemployment benefits. The social assistance programs under review include two means-tested benefits (benefits for low income families and benefits for families with childrenunder 16 years old), the child allowance for childrenunder 2 years old, and the compensations for public utilities. The incidence of privatesocial assistance in the form of inter-householdtransfers is also briefly reviewed. Pensions represent the most important social protectionprogram both in terms of expenditure and number of beneficiaries. The pension system is a pay-as-you-go system and comprises old-age, disability and survivor insurancebenefits, beingfinanced in2005 from a 31percentpayrolltax (30 percentto the pensionfund)76paid by employers and 2.5 percent paidby employees. The statutory retirementage is 55 for women, 60 for men, and 54 for redundantworkers. In 2005, in additionto the state system of old-age pensions, a mandatory accumulative system that covers all employees was introduced. The funds of the accumulative pension system are saved in the personal pension accounts with the People's Bank (Halk Bank) and raised from a deduction of 1%of wage. Some 5.5 millions of people are by now coveredwith this system. The unemployment benefit (UB), financed by a contribution of 0.6 percent of the payroll, has the lowest number of beneficiaries and level of expenditure among all social protectionprograms. The UB is providing temporary income support for up to 6 months (4 months in average in 2004). The levelof the unemployment benefit amounts to 50 percent of the last 12 months average salary plus 10 percent for each dependent, but no morethan the averagesalary. Inaddition, the benefitis awardedalsoto new-entrantsinthe labor-market (with no contributionhistory) who do not find ajob. This group representsthe majority of UB beneficiaries, and is entitledto a benefitamount equal to 75% ofthe minimumwage. Giventhe low leveland short durationof the benefit, its annualreplacementrate is less than 5 percentofthe averagewage.77 76 In 2003 the payroll tax was 36 percent (33 percent to pension fund). 77 The replacement rate was computed here by taking into consideration both duration and level of benefit, as the ratio between annual amount of the UB and annual amount of the average wage. 115 The social benefit for families with children under 16 (18) years old78ranks the second, after pensions, in terms of resourcesallocated and number of beneficiaries. It is a devolved means-testedprogram-funded by the central government and implemented by local communities (Mahalla Committees) according to a set of general criteria and rules defined by the Ministry of Labor and Social Protection. Given that the national guidelines are defined at a general level, the MahallaCommittees have a relativelyhigh degree of discretion with respect to eligibility criteria and award of benefits. The income threshold is set following instructions from the Ministry of Labor and Social Protectionand Ministry of Finance and usually corresponds to 1.5 minimumwages per capita. However, these instructionshaverather a guidance role, andthere is evidencethat the threshold is not necessarily uniform across the country, but can vary from regionto region, from rayon to rayon, or even from mahallato mahalla. The benefitamount depends on the number of children, rangingfrom 0.5 of the minimum wage for one child to 1.75 minimum wages for four children or more. The benefit is awarded for 6 months, after which the family should go througha recertificationprocess. The administrative data show that the average monthly benefit per family represents about 1.3 minimum wages (respectively 15 percent of the averagewage in 2005). However, this figure assumes that a family is awardedthe benefittwice a year, which may not be the case for all beneficiaries. The means-test comprisestwo maincomponents-an administrativebureaucratictestingof resources, basedon own statementsand wage/ income certificates,and a verificationof meansprocedure, basedon social inquiriesat claimant'sdomicile(home visits). The benefit for low income families is a means-tested last resort income support, implemented by Mahalla Committees and financed by the central government. The monthly benefit amount can go up to 3 minimum wages for a family, awarded for a period of maximum 3 months. In 2005 the average monthly benefit represented about 1.6 minimum wages. There is evidence that in many cases the Low Income Benefit is awardedto a family once peryear (for about 3 months) which meansthat its adequacy inprotectingthe poor is much lower. The implementationmechanism, includingmeans-test procedures, is similar to the one used for the benefits for family with childrenunder 16 (18), except the documents required(which are less for the low income benefit). Again, the income threshold, issuedat rayon level, has only a guidance role. As in the case of the other social benefits administeredby Mahallas, there are no clear procedures for income imputation,and the most importantcriteriataken into account seem to be the assets identifiedduringthe home visit, as well as the family compositionand characteristics (age, illness, single-memberhousehold, etc.). The benefit amount i s not a differentialone (i-e., the benefitamount does not representthe differencebetween household incomes and a threshold), being at the discretion of Mahalla Committee within the given limits. The low income benefit, like the means-testedchildprotectionprograms, has a strong workfare orientation: in most cases ifthe household headis able to work and not working or not enrolled in active labor market programs, the family is not consideredeligible. The child allowance for childrenunder 2 years old was a universalbenefit until 2002. Startingwith 2002 the benefit became means-tested for non-working mothers, while remaining non means-tested for the working ones. The means-test child allowance is delivered through Mahalla Committees, following the same procedures as the benefit for families with childrenunder 16 (18). In some rayons, MahallaCommittees are instructedto award the means-testedchild allowance only to those families who (would) qualify for the low income benefit. The benefit amount is 2 times the minimumwage, representingabout 23 percent of the 2005 averagewage. The cash compensation mechanism (CCM) for utilities is a program that replaced the privileges system of discounted utilities bills in 2003. Prior to 2003 the Uzbek Government provided a wide range of social privileges to about 12 privileged groups. The most important o f these privileges was a 50 to 100 percent discount on electricity, gas and district heating bills, as well as other housinghtility expenditures. In the spring of2003 a Presidentialdecree, while preservingexistingeligibility criteria, replacedthese discounts with a cash compensation for those recipientsthat were not budget sector employees, while compensating budget sector employees with a salary increase (Stone & Webster, Reforming energy sector social protection in Uzbekistan, 2005). The monthly amount o f the cash compensation payment for most categories o f beneficiaries (privileged) represents 45 percent of the minimum wage, and the estimated the of CCM beneficiariesin 2005 was about 450 thousands (ibid.). 78 Under 18 years old if the child IS in school, otherwise-under 16. 116 Magnitude and trends of Social Protectionprograms Table 2. Uzbekistan- magnitude and trends of Social Protectionprograms 2001 2002 2003 2004 2005 Social Protectionspending (uercent of GDPi SocialProtection 9.1 8.5 8.2 7.9 8.5 PensionFund 6.8 6.3 6.1 6 6.5 EmploymentFund(incl. active measures) 0.3 0.2 0.3 0.2 0.2 SocialAssistance -- Child 2.1 1.9 1.9 1.7 1.8 allowances childrenunder 2 - 0.6 0.6 0.6 0.5 0.6 -- Benefitsfor Benefitsto families with children under 16 (18) 0.8 0.8 0.8 0.7 0.7 low income families 0.06 0.08 0.06 0.06 0.06 Others 0.5 0.4 0.5 0.4 0.4 Reciuientsofmain SP benefits (thousandpersons) Old-age pension 1,857 1,883 1,900 1,914 1,936 Disability pension 585 588 575 564 553 Survivor'spension 229 233 230 228 227 Unemploymentbenefit 73 74 68 62 67 Childallowance-childrenunder 2 542 478 481 504 (thousandapplicationsawarded) Benefitsto families with childrenunder 16(18) 1,658 1,740 1,756 1,822 1,731 Benefits for low-incomefamilies 257 337 255 273 273 Average monthlv benefit amount (soumsl Old-age pension 11,078 14,567 17,707 22,111 31,841 Disability pension 9,933 13,281 16,255 21,082 29,255 Survivor'spension 9,393 12,600 14,987 18,130 26,053 Unemploymentbenefit* 3,670 5,400 7,620 9,345 11,720 Child allowance- childrenunder 2 7,709 10,880 13,060 18,800 Benefitsfor familieswith childrenunder 16 (18) 4,483 6,077 7,211 8,53 1 12,064 Benefitsfor low-incomefamilies 5,83 1 7,907 9,885 10,014 14,795 For reference Minimumwage 3,430 4,535 5,440 6,530 9,400 Minimumpension 6,780 8,970 10,765 12,920 18,605 Average wage 15,856 30,206 40,044 53,201 81,880 Replacementrates -Monthlv benefits as share of the average wage Old-age pension 0.7 0.5 0.4 0.4 0.4 Disability pension 0.6 0.4 0.4 0.4 0.4 Survivor'spension 0.6 0.4 0.4 0.3 0.3 Unemploymentbenefit 0.2 0.2 0.2 0.2 0.1 Child allowance-childrenunder 2 0.3 0.3 0.2 0.2 Benefitsfor families with children under 16 (18) 0.3 0.2 0.2 0.2 0.1 Benefitsfor low-incomefamilies 0.4 0.3 0.2 0.2 0.2 Monthly benefit relativeto the minimum wage Child allowance-childrenunder 2 1.7 2.0 2.0 2.0 Benefitsfor families with childrenunder 16 (18) 1.3 1.3 1.3 1.3 1.3 Benefitsfor low-incomefamilies 1.7 1.7 1.8 1.5 1.6 Source:Based on MLSP and IMF data. Note: Social Developmentand Living Standards inUzbekistan-2005 * 117 The dynamics on SP spending as percent of GDP in the period 2001-2004 years show a decreasing trend, overall, as well as by components. This drop in spending was the result of a slower growth of SP expenditure relative to GDP growth, and resulted in a relatively significant reduction of benefits generosity relative to the average wage. In 2005 the social protection spending as a share of GDP increased to 8.5 percent of GDP, close to the 2002 figure. The main factor contributing to this increase was a sharp increase in the level of pensionbenefits. Inthe case of pensions, despite the increase in the number of beneficiariesby about 2 percent, the spending fell from 6.8 percent of GDP in 2001 to 6 percent in 2004, and increased back to 6.5 percent in 2005. The average pension benefit increased in real terms with about 50 percent since 2001 (Table 2), but this increase was lower than the increase in real wages, conducing to a replacementrate of 40 percent of the average wage (down from 70 percent in 2001). The highest decline in spendingwas experiencedby the employment fund, probably due to some extent to the drop inthe number of UB beneficiariesby 8 percentbetween 2001 and 2005. Table 3. Dynamicsof monthly averagebenefits, constant prices(2001=1) 2001 2002 2003 2004 2005 Old-agepension 1 1.2 1.3 1.5 1.o 1.o Disabilitypension 1 1.2 1.4 1.6 Survivor's pension 1 1.o 1.2 1.2 1.5 Unemploymentbenefit Childallowance - childrenunder 2 (*) 1 1.1 1.5 1.6 1.7 1 1.4 1.4 1.7 Benefits for families with children under 16(18) 1 1.o 1.2 1.2 1.4 Benefits for low-incomefamilies 1 1.o 1.2 1.1 1.4 For reference Minimumwage 1 1.o 1.2 1.2 1.5 I Minimumpension 1 1.o 1.2 1.2 1.5 Average wage 1 1.5 1.9 2.1 2.8 Source: Based on MLSPdata and IMF deflator. Note: (*) 2002=1, Regarding social assistance, some of the decline in spending can be attributed to the termination of part of privileges, as well as to a decrease in the number of beneficiaries of child allowance.However, as in the case of social insurance benefits, the real benefit amounts increased over time (Table 3). In this context, the next section assesses the performance of social protection cash transfers in offering protection against poverty, by assessing their coverage, targeting, and adequacy in time as well as in comparisonwith other countries inthe region. EFFECTIVENESS SOCIALPROTECTIONPROGRAMS OF To determine the extent to which social transfers offer protection to the poorest groups of the population, this section uses 2000/2001; 2002 and 2003 rounds of the Household Budget Survey data. Since the purpose of this chapter is not to estimatepoverty incidencebut to look at the distribution of benefitsover different welfare groups, the analysis presented here uses the consumption aggregate constructed for the ECAPOV database, which allows country comparisons.79 Three main indicators are used to assess the effectiveness of SP programs: coverage (share of population covered by the programs), targeting (share of funds directed to each welfare group of population), and adequacy of benefit (share ofthe benefit inthe consumption of beneficiaries)." 79 A descnptlon of the consumption aggregate can be found In "Growth, Poverty, and Inequality - Eastern Europe and the former Soviet Union", The World Bank, 2005 80 With only Few exceptions, which will be stated, through this paper beneficlanesare considered all the members of a household receiving a beneflt. 118 Inviewofassessingtheincidenceofpensionbenefits,thepopulationwillberankedintoquintilesbasedonthe consumption aggregate. Inthe case of social assistance programs, since one of their implicit objectives is to offer temporary protection against poverty risk (related to temporary loss of income, change in household structure, etc.), the idealapproach would be to identifythe welfare status of a household in the absence of the government intervention and then to rank the individuals/ households on the base of this counterfactual welfare indicator. To obtain unbiased estimates of benefit incidence, one would compute a counterfactual consumption by subtractingthe programbenefits, and addingthen the replacement incomethat the households would generate through their behavioralresponses in the absence of the intervention. A series of studies" seems to suggest that in several countries the share of the replacement income would be around 50 percent of the value of the transfer. In this paper we assume a replacement income of 50 percent and rank households basedon the householdconsumptionminus 50 percent ofthe transfers. While this estimate is not precise, the results presented in the paper are robust to the choice of a different share for the replacement income, especially for transfers that are small comparedto householdconsumption. Overall, the social protection programs covered more than one half of the Uzbek households in 2003, comprisingabout 57 percent of the total population(Table 4). The SP program with the largest coverage is pensions, of which old-age pensions covered 41 percent households (2003). The three main social safety net programs covered together 28 percent of population. As compared with the 2000 survey, the 2003 HBS data show a decrease in coverage of SP programs, mainly due to the decrease in the coverage of social safety nets andprivileges. The figures regardingthe means-testedbenefits (for low incomefamilies, and for families with childrenunder 18) are not consistent with administrativedata; although it is likely that the HBS underestimate the real figures, part of the differences can be explained by the different unit of observation used: the administrativedatareports entitlements, which can includedouble countingof beneficiaryhouseholds. percent of households percent of population 2000 2002 2003 2000 2002 2003 All SP programs 60 63 55 63 66 57 Pensions 41 43 41 44 45 43 Old age pension 33 36 33 36 38 36 Disability pension, for self or children 10 9 9 11 10 10 Loss of breadwinnerpension 2 2 2 2 2 2 Unemployment benefit 0.2 0.1 0.1 0.2 0.1 0.1 Social safety net benefits 27 26 23 32 32 28 Child allowance for children under2 years 14 13 12 18 18 16 Benefitsfor low income families 4 3 2 4 3 2 Benefitsfor families with children under 18 16 15 14 19 18 16 1 Privileges 8 5 2 7 5 2 Inter-householdstransfers 8 10 12 I 6 8 9 Source:HBS 2000-2003 8 1 Ravallton (2000). van de Walle (2001; 2002) 119 protection while benefiting less o f growth, especially those in rural areas, inthe poorer regions o f the country, and elderly. Table 5. Share of beneficiaries' consumption 2000 2002 2003 All SP programs 0.18 0.21 0.25 Pensions 0.17 0.23 0.24 Old age pension 0.17 0.22 0.24 Disability pension, for self or children 0.12 0.14 0.17 Loss ofbreadwinnerpension 0.12 0.14 0.16 Unemploymentbenefit 0.02 0.06 0.05 Social safety net bene@ 0.08 0.09 0.12 Child allowance for children under2 years 0.06 0.07 0.11 Benefits for low income families 0.08 0.08 0.12 Benefitsfor families with children under 18years 0.07 0.07 0.10 Privileges 0.04 0.03 0.04 Inter-householdstransfers 0.10 0.10 0.12 Source HBS 2000-2003. The program with the most significant share in consumption is, as expected, pensions, covering almost a quarter o f their beneficiaries' consumption. The social safety nets (low income and child protection transfers) cover, on average, about 12 percent o f the consumption o f their beneficiaries. Together, all social protection programs cover about half of the poorest quintile consumption and less than a fifth of the wealthiest quintile consumption (Table Al). However, only 16 percent o f the total social protection spending reaches the first quintile, due to the large share of pensions inthe total social protection benefits. Because o f the link with previous contributions from wages, more than half o f the pension benefits go to the wealthiest quintiles - 31 percent and 22 percent o f funds go to the 5`h quintile and 4* quintile respectively, while only 15 percent o f the pension benefits reach the low-income quintile (Table A2). However, pensions offer a relatively good protection to the poor beneficiaries, by covering on average 42 percent o f their consumption (Figure 1). As in the case o f other social protection benefits, the share o f pensions in the consumption o f beneficiaries increased significantly since 2000. Figure 1. Pensions-share of transfers inthe consumptionof beneficiaries from the lst quintile. 50 40 30 ;e 20 10 0 I All pensions Old age Disability LOSS Of breadwinner In 2003, pension benefits covered 97 percent o f the households headed by elderly (60 years old and over), almost 70 percent o f the female headed households, and 46 percent o f the population in the poorest quintile. As expected, the highest coverage is provided by old-age pensions, which cover about 35 to 37 percent of the population in each quintile. Old-age pensions cover 95 percent of all adults aged 60 or over, and 40 percent o f the adults aged 50-59. Disability pensions cover 9 percent of households, making up 10 percent o f the population. They cover about 12 percent o f the population from the poorest quintile (Table A4) and the 120 benefits are distributed more evenly over quintiles than in the case of old-age pensions, which are more regressive. The unemployment benefit has the lowest coverage of all transfers. It reaches only about 0.1 percent of households, and about 0.4 percent of the unemployed looking for a job. Its incidence on registered unemployedis also low-with only about 5 percent of this group receivingunemployment benefits. Most of the beneficiaries are women (70 percent) and those who are between ages 35 and 54 (57 percent). There are several reasons for the low coverage of UB: first, as already mentioned in Chapter 3, only few unemployed lookingfor ajob are registeringwith the UnemploymentOfficesbecausethe services offeredare rather limited and the level of benefits are low. For those who register, the low capacity of the Offices to offer adequate vacancies leadsto the rejectionof the offer by the applicant and thus to the refusal of entitlement. Inaddition, the cost associatedwith entitlement is perceivedto be highby beneficiaries(when comparedto the benefits). ",,, What we can ofer are only low-paid vacancies in the state-run organizations. Butpeople do not want to workfor such salaries and reject thejobs we offer, thus we cannot grant the unemployment benefit to these ayonDepartmentofLabor agree to accept anyjob. ... The social ben the labor exchange. They cannot help me, though I is 7,000 Sums, and I must go to the labor exchange ofice almost evely day to registerfor this money. Woman, 25 years old, recipientofthe unemploymentbenefit Due to the low level of benefits, the adequacy of the UB is also very low, at about 5 percent of the beneficiary's consumption, meaningthat it does-not succeed to offer sufficient protection against temporary loss of income. As described previously, the social assistance component of the Uzbek social protectionsystem is built to a large extent on three benefits accounting for more than three quarters of the socialassistance expenditure (the rest beingspent mostly on the CCM and privileges). Out of these three benefits, two are means-tested,while the third is means-tested for a sub-group of beneficiaries. This is a positive feature of the Uzbek social assistance policy, as compared with other countries in the region(e.g. Moldova, Russia, Belarus, Azerbaijan) that rely on more fragmented systems of categorical benefits and privileges. However, Figures 2 and 3 indicate that the performance of Uzbek social assistance programs is rather modest when compared to other countries in the region. Figure 2 presents the coverage and targeting performance of all non-contributory benefitss2in several ECA countries and it showsthat, despite the strongmeans-testcomponent of Uzbek social assistance system, its targeting performance is lower than that of the majority of countries. From this perspective, Uzbekistan performs better than Tajikistan, GeorgiaandRussia, but worse than all other transition countries analyzed here, includingAlbania, Armenia, Belarus, BulgariaandRomania. 82 Benefits for which the entitlement IS not conditioned by previous contribution (as in the case of social insurance benefits) 121 Figure 2 Performance of non-contributorybenefits,includingprivileges 50 8 -.-.- 0) +I C 3 u40 : U II) 0 p. 5 (u 30 zrc U c s 0 . cm I 20 .-m c z +I a +I 10 0 20 40 60 80 100 coverage of the poorest quintile FA) ECA Targeting Performance, WorlJBank Ifwe compare only the means-testedprograms(all non-contributoryprograms basedon income-, means-,or proxy means-test mechanisms), ignoring other programs that form the mix of social assistance, Uzbekistan performs relatively well on coverage, but poor on targeting (Figure 3). Even if considering only those countries with similar or higher coverageofmeans-testedprograms, Uzbekistan's performanceremainsweak. 122 Figure 3 Performance of means-tested benefits 0 10 20 30 40 50 60 70 coverage of the poorest quintile (%) Source ECA Targeting Performance, WorldBank The three main social assistance benefits (child allowance, benefits for family with children under 18, and benefitsfor low income families) coveredtogether 40 percent of the populationin the poorest quintile in2003 (TableA6). Their aggregatetargetingperformancewas rather low-only 28 percent of the funds reachedthe poorest quintile, while 27 percentwent to the richer4' and 5' quintiles, andthe benefits covered 24 percentof the poorest beneficiaries'consumption. By area of residence, morethan two thirds (71 percent) of funds were distributedto ruralpopulation. All three safety net benefits are progressive, contributingthus to inequalityreduction,but in different degrees. As indicated by Figure 4, all three transfers have similar distributive impact on the first 20 percent of population,to whom they distribute around 28 percent of their funds. However, from that point the programs start to behave different. The low income benefit, contrary to its objectives, provesto be less progressive than the other two transfers. 123 Figure 4. The distributionof social safetynettransfers CO"S.waaO" NIY.., .0611 9ar.g r4.t ,r.n.,.r. ZODD % - os+ 0-0- 0- a IC 0 7- 0- The three social assistanceprogramsoverlap only to a low extent.About 86 percent ofthe beneficiaries benefit from one program, 13.5 percent from two programs, and about 0.5 percent benefit from all three programs. This distribution is slightly different for the poor, disabled, and for those living in the Northernand Southern regions, who are more likely to benefitfrom two programs simultaneously. The funds spent in 2003 by all three programs representedabout 90 percent of the necessary amount to close the povertygapg3for a povertythreshold of 1S minimumwages per capita (the most often usedadministrative threshold). However, the programs altogether covered only about 12 percent of the poverty gap. Taken separately, the child allowance (CA) for children under 2 covered about 5.7 percent, the benefit for families with children under 18 (BFC) covered 5.5 percent, while the low income benefit covered 0.8 percent of the povertygap. Although the primary objectiveofthe two child protectionprograms (CA and BFC) is not necessarily poverty reduction, they are used as benchmarks for a cost-benefit analysis of the low income benefit. This analysis assesses how expensive it is to close the poverty gap by 1 Soum, for each of the three programs under reviewag4Fromthis perspective, the benefitfor families with childrenunder 18 andthe childallowancehave a lower cost: for a reductionwith 1 Soum of the poverty gap, these programs spent about 7.8 Soums, while the low income benefit spent about 8.4 Soums. This high cost is to a large extent a consequence of the poor targetingand ofthe significantleakageof funds towards the high-incomequintiles. Even though between 2000 and 2003 the aggregate targeting performance of the three programs remained about the same (Table A6), the HBS data show that they experienceddifferent dynamics (Figure 6). The child allowance program transferred about 28 percent of the funds to the poorest quintile in 2003, 4 percentage pointsmore than in 2000. The benefitfor families with childrenunder 18 also transferred 28 percent in 2003, but down from 34 percent in 2000. However, it is not clear ifthese changes are the result of improvingthe targeting performance of CA due to the introduction of means-test for non-working mothers, and of a worsening of BFC performance, or rather the result of errors in data c~llection.~~the BFC performance If worsened indeed, the HBS data indicatethat this is due mainly to the entitlementprocess at the Mahallalevel (Table A13). However, even if BFC performance worsened as compared with 2000, it still performs better than the low income benefit, which transfers more than half of the funds tQthe middle and upper quintiles (Table A9, Figure6). 83 The povertygap us computed uungthe counter.factual conswnpuonaggregale (m the absence of half of Uanrfcrs). 84 For this purpare,we wll compute a wunrer-factual pveny gap. usingthe pre-uanrfer conswnpuonandthe 1.5 m m w wager threshold,and then wI1wmpute the dlfierencebetweenthe wuntcr-factual pveny gap andthe actual (pat- uansfeer) pveny gap The resultrepresentsthe "benefit" of the program The " ~ 1 s t "of the program !I consideredthe whale amounl paid to ppulauon inthe form of benefits The ratio "con"i"'benefit"Hilleslimle the amounlspent by the pmgramfor a reducuon wth ISoum of the poverry gap 85 Meha and Rarlud (2W2) suggesthiin2wO the inmctions for interviewersdid not clcarly &runguish between the child allowancefar shldrm underIWO and the benefitfor families wth childrenunder eighteen 124 Figure 5. Coverage (%) 35 , 30 25 20 15 10 5 0 1 1 2 1 3 / 4 / 5 1 / 2 / 3 / 4 / 5 2000 2003 Figure 6. Targeting(%) Figure 7. Adequacy 35 "-.l ~- I 0.25 "- 30 25 20 0 15 15 0 10 10 0.05 5 2 ' 3 ' 4 ' 5 1 0 0 00 4 4 1 1 1 2 1 3 1 4 5 2000 2003 2000 2003 Since 2000 the coverage of both means-testedprograms (the benefit for families with children under 18 and the low income benefit) decreased considerably (Figure 5). This decrease can be largely attributed to a decrease in the number of applications, which was higher in the case of low income benefit. Although both programs are relativelywell known by the poor (Tables A11, A12), the percent of those applyingdropped by about 65 percent in the case of low income benefit and by 30 percent in the case of benefitsfor families with childrenunder 18 since 2000 (Table A13). The mainreasonfor not applyingseems to bethe reluctanceto rely on external sources of support, other than the ones householdmembers can provide. This is most likely due to the incapacityof the system to be pro-activein reaching the poor. The second reasonis the filtering process induced by the documents needed to justify the claims. In theory this process can lead to a good targeting mechanism, if the well-off are discouragedto apply due to the high opportunitycosts involved. However, in the Uzbek case, this filter seems to impose more costs on the poor than the non-poor (Table A15). Finally, a third reasoncontinuesto be the arbitrariness ofMahalla officials. The Uzbek means-testedprograms suffer from high inclusion errors. Tables A11 and A12 show that benefits are awarded to the poor only in a slightly higher proportion as compared to the non-poor: in the case of low income benefits38 percent of the poorest applicants are awarded the benefit, while the corresponding figure for the wealthiest quintile is 31 percent; in the case of the benefit for families with children, 78 percent of the poorest applicants are awarded the benefit, with a corresponding figure of 68 percent for the wealthiest quintile. As several other papers and reports indicatedalready, the targeting performanceseems to be driven by the propensityto apply ratherthan by the entitlementprocess(Le. poor householdshave a higher propensity 125 to apply and consequently represent a larger share of beneficiaries). Inthis case the inclusionerrors can have two possible sources. One of them is the lack of targeting at local level (i.e., Mahallas do not distributethe funds to the poorestmembers ofthe community). The poor targetingperformancewithin Mahallashas several explanations, as identifiedthroughqualitativeresearch: 1) Lack of more precise guidelines/methodologyregarding the evaluation of household wellbeing and imputation of incomes, that leaves room for a high degree of discretion, which, in turn, can generate genuine errors but also arbitrariness Whenin-depthinterviewswere beingconducted, it often seemedthat peoplefeel notablydistrust towards stem. Manypersons interviewedtoldthat the poor families do notreceive andfiends ofthe mahallacommitteestaffmembers. ceivesocial benefits. For example, my neighbor gets a benefit, though her husband is a businessman and earns good money. Z don't know how they can receive the benej2. And somefamilies still cannotget the benefit even ifa husbandin thatfamily can'tfind ajob. " Woman, recipient of social benefit for children 2) Weak administrativecapacity at mahallaleveldue to lack oftraining "We managed to get well with thepresent chairman of our mahalla: he is a very educatedperson, being a former teacher. However, the chairman who had been here before could not record correctly the minutes of the citizens' assembly and the statements of the investigation of economic situation of those families which submitted applications. This all led to errors and violations. Much depend on the skills of the mahalla chairman and secretav, so they need to be trained regularly ....'' Officer ofthe RayonDepartmentofLabor I I 3) Lack o f communication between the local self-government bodies (mahalla) and tax authorities to verify the documents filed by applicants. The second possible source of inclusionerrors is the distribution of funds from the center (Ministry) down to oblast/ rayons, and further, the distribution of funds from the rayons to Mahallas. Even under the assumption of perfect targetingat the local level, errors of inclusion can emerge as a consequence of not distributingthe funds accordingto poverty criteria. As a consequence,two households with similar poverty status but located indifferentregions will nothavethe same probability of entitlement(horizontal inequity). Table 6 shows that, althoughonly 6 percent of the poor are located inTashkent, about 27 percent of the low income benefit funds are allocatedto that region. Table 6. Regionaldistributionof funds. ECAPOVConsumption I Aggregate Threshold: 1.5 minimum Distributionof funds (percent) wages per capita I Benefit for Shareofpoor (%) Share of povertygap Share of LOWincome poor (%) families with (%I benefit childrenunder 18 Ferghana 33 31 22 13 25 Central 18 19 22 18 16 Southern 16 19 21 31 34 Northern 12 14 18 8 13 Mirzachul 9 10 12 3 3 rashkent 12 6 4 27 10 Total 100 100 100 100 100 Source: HBS 2003. Note: We use total consumptionminus halfthe social safety net benefits. 126 The Uzbek means-testedprograms are also affected by significantexclusionerrors. As discussed previously, part of the exclusion errors is due to the lack of pro-active outreach efforts which could bring those discouragedto apply in the system. However, the reasons for applicationsrejectionas they are reported inthe H B S suggest that exclusion errors are also due to the discretion built into the system. The main reason for rejectionas reported by the low income benefit applicants is the refusal of Mahallaofficials to consider their application. This reason is more often mentioned by poor than by non-poor,for both benefits (Table A17). The interviewscarriedout show that inmany cases the refusalto consider an applicationis done onthe ground that the applicant is presumed (based on common knowledge of Mahalla representatives) to be inconsistent with eligibility criteria. In other cases, the refusal (or the rejection of a poor family application) is a consequenceof limitedfunds. "Iwasgivenpermission toget my economic aidmoney whenthisyear began. Whenthe time ofpayment ended, I had written another application, but they refirsed to give me the aid, and said they had other families that must receive the aid, and there is not enoughmoney topayfor everyone. They told me to come again one month or two months later, and then they migh Woman, 69 years old, recipientofeconomicaid The last public program under review in this chapter is the cash compensation mechanism (CCM) for utilities. Since this programwas introducedin 2003, it is not captured by the H B S data that are made available for this report. However, its performance was assessed by Stone 8z Webster (2005)' using a dedicated survey (Uzbekistan HouseholdEnergy Survey - UHES). Unfortunately, the CCM beneficiariessample size is small, and consequentlythe results should be treated with caution. The UHESunderestimatesthe coverage of CCM, as compared with administrativedata. While the administrativedata report a coverage of about 8 percent of households, the survey reports a coverage of 1.3 percent of households. UHES shows that poor population groups are less covered than the wealthy ones, and that the targetingperformance of the transfer is also very weak -more than 50 percent of the funds go to the wealthiest quintile, while the poorest quintile receives only 16percentofthe transfers. In addition to the public social assistance schemes, about 9 percent of Uzbek households benefit from inter- householdtransfers. These transfers are highly regressive (more than 50 percent of transfers are goingto the wealthiest quintile) and seem to be rather specific to urban areas. However, they succeedto protect the poor better than the low income benefit, by coveringabout 6 percent of the populationin the poorest quintile, and makingup morethan 35 percent ofthe poor recipients'consumption. REVIE wOFKEYSOCIALPROTECTION ISSUES 1. Social insurance benefits Uzbekistan made several stepstowards ensuringa highersustainabilityofthe pensions system, while avoiding creating disincentives for employment opportunitiesin the formal sector, by a) adjustingthe "generosity" of pensionbenefits from a replacement rate of 70 percent in 2001 to a replacement rate of 40 percent in 2005; b) reducingthe payroll tax (from 37.5 percent in 2001 to 3 1 percent in 2005, and further to 25 percent in 2006); and c) starting the implementationof the secondpensionpillar (the pensionaccumulation fund). However,the sustainabilityof the Pension Fund is still under question due to the payment of benefits not relatedto pension insurance (care of children under 2, pregnancy and childbirth, temporary inability to work), and low tax compliance. How pension system solvency is affected by the recent changes in the main parameters and its sustainabilityinthe future is worth further investigation. The pensionbenefits offer a significantprotectionto the poor, by covering almost halfofthe populationin the poorest quintile and constitutingmore than 40 percent of the poor beneficiaries' consumption. The second social insurance cash transfer under review, the unemployment benefit, has a very low coverage, and fails to offer protectionagainsttemporary loss of income. 127 2. Social assistancebenefits The design and implementationof the means-testedbenefits have several shortcomings, (highlighted also in the previous Uzbekistan Living StandardsAssessment as well as inother papersandreports).86By design, too much space is given for discretionand arbitrariness at locallevelthroughthe lack of more precise guidelined methodology regardingthe evaluationof householdwellbeingand imputationof incomes. Combined with the absence of (proxy) poverty criteria for the distribution of funds between regions and communities, this leads further to horizontalinequity (wealthyregions/communities can receive as much, or even more, funds as poor regions/communities, and, as a consequence the poor will not receive similar treatment in all regions). To avoidfurther horizontalinequity, incomethresholdscouldbe establishedin atransparentmannerfollowingthe differences inthe priceof livingat the oblasthegionlevel. The lack of moreprecise eligibility guidelines leads to a lack of transparencyand predictabilityof entitlements, and underminesthe possibilityof internalauditand control of implementation, and enforcement. An important constraint for establishing and applying a methodology for evaluationof householdwellbeing is the high degree of earnings informality in Uzbekistan. However, this constraint can be overcome by usingassets checklists. Finally, this chapter finds that the reductionin the number of privileges in 2003 represents an important step towards consolidating the Uzbek social assistance system. However, the introduction of the cash compensation mechanism for utilities using the eligibility criteria inherited from privileges does not offer protectionto low income groups. Mehra and Rashid(2002), Coudouel, Mamie, and Micklewright (1998) 128 ANNEX1. STATISTICSONTHE COVERAGE, TARGETING AND ADEQUACY OFSOCIAL PROTECTION PROGRAMS Al. All SP programs- Coverage, distribution of funds, and adequacy, by household characteristics coverage (YO) distribution of funds share of benefits in (%)' beneficiaries' consumption 2000 2002 2003 2000 2002 2003 2000 2002 2003 quintilesof 1 64 70 62 15 17 16 0.34 0.38 0.48 :onsumption per 2 0.26 0.33 capita 67 68 58 18 17 17 0.25 3 63 65 58 19 18 18 0.20 0.23 0.27 4 63 64 57 22 21 21 0.17 0.21 0.24 5 60 62 52 26 28 28 0.11 0.15 0.17 residence area urban 64 64 57 47 46 45 0.19 0.22 0.24 rural 63 66 58 53 54 55 0.17 0.20 0.26 region Tashkent 66 68 57 25 26 24 0.17 0.20 0.20 Mirzachul 66 57 48 6 5 5 0.17 0.21 0.26 Ferghana 66 66 54 26 23 21 0.17 0.18 0.24 Northern 61 69 61 11 11 11 0.18 0.23 0.32 Central 65 62 55 21 19 20 0.18 0.23 0.28 Southern 54 68 68 12 16 18 0.20 0.25 0.30 sex of HHhead male 60 62 54 73 73 74 0.17 0.21 0.25 female 78 82 71 27 27 26 0.19 0.22 0.25 age ofHHhead 16-29 56 60 56 5 4 5 0.15 0.I 6 0.19 30-39 46 50 43 13 11 11 0.13 0.14 0.18 40-49 49 45 39 15 12 13 0.12 0.13 0.17 50-59 69 73 64 16 18 17 0.16 0.20 0.23 60 and over 93 98 89 51 55 54 0.25 0.29 0.34 educationofHH <4 years 90 96 85 13 9 7 0.21 0.26 0.26 head 5-9 years 80 88 78 19 21 19 0.22 0.26 0.3 1 10-11years 58 63 56 30 30 34 0.17 0.20 0.25 completed vocational/ lyceum 54 54 45 7 7 7 0.I 6 0.17 0.23 incompleteic omplete college 55 54 50 14 14 15 0.16 0.20 0.23 incompleteic omplete higher 61 62 52 17 18 18 0.16 0.20 0.23 Total 63 66 57 100 100 100 0.18 0.21 0.25 129 A2. All pensions-Coverage, distribution of funds, and adequacy, by household characteristics coverage (YO) distribution of funds share of benefitsin ("/I, beneficiaries' consumption 2000 2002 2003 2000 2002 2003 2000 2002 2003 iuintiles of 1 42 48 46 13 15 15 0.3 1 0.38 0.42 :onsumption per 2 0.22 0.27 0.31 :apita 48 46 41 17 16 15 3 45 43 43 19 18 17 0.20 0.25 0.26 4 47 46 43 23 22 22 0.17 0.23 0.24 5 41 43 41 28 29 31 0.12 0.17 0.18 iesidencearea urban 46 47 44 49 49 48 0.19 0.25 0.25 rural 44 44 42 5 1 51 52 0.16 0.21 0.24 region Tashkent 44 50 47 26 28 27 0.19 0.22 0.20 Mirzachul 45 41 38 6 5 5 0.17 0.22 0.26 Ferghana 48 47 46 26 23 20 0.16 0.19 0.20 Northern 52 49 48 11 11 12 0.14 0.25 0.30 Central 46 40 38 22 19 20 0.18 0.26 0.29 Southern 31 42 37 10 14 16 0.19 0.28 0.34 sex ofHHhead male 39 39 37 71 70 71 0.17 0.23 0.26 female 67 72 68 29 30 29 0.17 0.21 0.21 age of HHhead 16-29 20 17 15 3 2 2 0.15 0.19 0.25 30-39 14 14 12 6 5 5 0.I3 0.17 0.19 40-49 23 19 18 10 8 9 0.12 0.16 0.18 50-59 56 58 58 17 19 19 0.14 0.19 0.20 60 and over 95 99 97 64 66 65 0.21 0.26 0.28 education of HH <4 years 92 97 93 15 10 9 0.I 7 0.22 0.21 head 5-9 years 73 81 78 23 25 22 0.19 0.24 0.26 10-11 years 35 38 37 26 27 30 0.16 0.22 0.24 completed vocational/ lyceum 26 24 24 6 6 6 0.18 0.22 0.24 incompletek omplete college 33 32 33 14 13 15 0.18 0.24 0.25 incomplete/c omdete higher 40 41 37 17 19 18 0.17 0.22 0.24 Total 44 45 43 100 100 100 0.17 0.23 0.25 130 A3. Old-agepensions -Coverage, distribution of funds, and adequacy, by householdcharacteristics coverage (YO) distribution of funds share of benefits in ("/I, beneficiaries' consumption 2000 2002 2003 2000 2002 2003 2000 2002 2003 quintiles of 1 34 40 37 13 14 14 0.29 0.36 0.40 consumption per 2 capita 39 39 35 17 16 15 0.22 0.26 0.3 1 3 37 36 36 19 17 17 0.19 0.25 0.26 4 39 39 36 24 22 22 0.17 0.22 0.24 5 34 37 35 28 30 32 0.12 0.17 0.19 residencearea urban 37 40 35 49 49 48 0.19 0.25 0.25 rural 36 37 36 51 51 52 0.15 0.21 0.24 region Tashkent 37 46 41 27 29 29 0.19 0.21 0.20 Mirzachul 36 32 31 6 4 5 0.16 0.22 0.26 Ferghana 40 39 39 26 23 20 0.16 0.19 0.20 Northern 40 40 37 9 10 10 0.13 0.24 0.29 Central 38 33 31 22 18 20 0.18 0.27 0.30 Southern 26 36 31 10 15 16 0.18 0.28 0.34 sex of HHhead male 32 33 31 72 71 72 0.18 0.24 0.26 female 57 63 59 28 29 28 0.15 0.20 0.20 age of HHhead 16-29 14 15 11 2 2 2 0.15 0.19 0.26 30-39 7 8 6 3 3 3 0.12 0.17 0.20 40-49 I O 8 7 5 4 4 0.11 0.15 0.17 50-59 45 47 46 17 17 17 0.13 0.18 0.19 60 andover 92 97 96 73 74 74 0.19 0.25 0.27 educationofHH <4 years 88 93 91 17 11 I O 0.I 6 0.20 0.20 head 5-9 years 67 75 74 24 26 24 0.18 0.23 0.25 10-11years 26 30 28 24 26 28 0.16 0.21 0.25 completed vocational/ lyceum 17 20 18 5 6 6 0.18 0.22 0.25 incompletek omplete college 23 25 26 13 13 14 0.18 0.25 0.25 incompleteic omplete higher 32 34 - . 31 17 _ . 19 19 0.17 0.22 0.25 Total 37 38 36 100 100 100 0.17 0.22 0.24 131 A4. Disability pensions-Coverage, distribution of funds, and adequacy, by householdcharacteristics coverage (YO) distribution of funds share of benefits in ( Y O ) , beneficiaries' consumption 2000 2002 2003 2000 2002 2003 2000 2002 2003 quintilesof 1 12 12 12 17 18 19 0.21 0.25 0.30 consumption per 2 capita 12 11 9 17 17 15 0.14 0.18 0.21 3 11 10 10 21 19 19 0.14 0.16 0.18 4 11 10 10 21 21 22 0.11 0.14 0.16 5 9 9 9 24 24 25 0.08 0.09 0.11 residencearea urban 12 11 11 49 47 52 0.13 0.15 0.17 rural 10 I O 9 51 53 48 0.11 0.14 0.16 region Tashkent 9 10 9 20 21 21 0.12 0.13 0.13 Mirzachul 12 12 9 7 8 6 0.12 0.15 0.16 Ferghana 11 11 I O 26 23 20 0.11 0.11 0.14 Northern 19 13 15 18 16 18 0.12 0.20 0.22 Central 11 10 10 21 21 22 0.12 0.17 0.18 Southern 6 8 7 8 11 13 0.15 0.15 0.22 sex of HHhead male 11 10 10 79 79 79 0.12 0.14 0.16 female 11 11 I O 21 21 21 0.I3 0.15 0.17 age of HHhead 16-29 6 3 6 3 1 3 0.09 0.I 4 0.16 30-39 7 6 5 15 14 12 0.11 0.15 0.17 40-49 12 10 10 30 26 33 0.12 0.15 0.17 50-59 16 17 17 22 31 29 0.11 0.16 0.I 6 60 andover 13 13 I O 30 27 23 0.I3 0.12 0.16 educationof HH <4 years 12 16 12 8 7 5 0.I3 0.13 0.I 7 head 5-9 years 12 14 12 14 17 13 0.13 0.15 0.18 10-11years 11 I O 11 36 33 40 0.11 0.14 0.17 completed vocational/ lyceum 10 7 8 10 7 9 0.13 0.14 0.16 incompletek omplete college 11 9 I O 18 17 17 0.11 0.15 0.16 incomplete/c omplete higher 10 10 8 15 19 16 0.10 0.13 0.15 Total 11 10 I O 100 100 100 0.12 0.14 0.17 132 AS. Loss of breadwinnerpensions-Coverage,distributionof funds, andadequacy, by householdcharacteristics coverage (YO) distribution of funds share of benefits in (%I' beneficiaries' consumption 2000 2002 2003 2000 2002 2003 2000 2002 2003 quintiles of 1 2 2 2 14 28 19 0.23 0.36 0.33 consumptionper 2 capita 2 1 1 19 14 12 0.20 0.22 0.20 3 2 2 1 17 17 17 0.14 0.15 0.21 4 2 1 2 20 14 24 0.12 0.13 0.16 5 2 2 2 30 27 28 0.08 0.08 0.11 residencearea urban 2 2 2 41 49 53 0.12 0.14 0.14 rural 2 2 1 59 51 47 0.13 0.15 0.19 region Tashkent 2 1 2 30 20 16 0.13 0.10 0.08 Mirzachul 3 3 2 10 16 11 0.17 0.23 0.25 Ferghana 2 1 2 21 17 21 0.14 0.09 0.15 Northern 1 2 2 5 12 12 0.07 0.16 0.23 Central 2 1 1 17 15 22 0.09 0.15 0.23 Southern 2 2 2 16 20 18 0.18 0.25 0.21 sex of HHhead male 1 1 1 27 20 23 0.08 0.09 0.14 female 7 7 6 73 80 77 0.15 0.16 0.17 age of HHhead 16-29 2 1 1 6 6 3 0.14 0.15 0.27 30-39 1 1 1 23 25 28 0.19 0.21 0.19 40-49 2 2 2 36 45 41 0.15 0.16 0.18 50-59 2 2 2 12 13 16 0.07 0.09 0.12 60 and over 2 1 1 23 10 12 0.10 0.09 0.13 educationof HH <4 years 3 1 1 11 2 2 0.11 0.07 0.22 head 5-9 years 3 2 2 18 13 14 0.09 0.14 0.15 10-11 years 2 2 2 31 53 51 0.16 0.16 0.18 completed vocational/ lyceum 2 1 1 8 7 4 0.12 0.16 0.17 incomplete/c omplete college 3 1 1 23 14 18 0.14 0.11 0.17 incompletek omplete higher 1 1 1 8 13 11 0.11 0.12 0.11 rota1 2 2 2 100 100 100 0.12 0.14 0.16 133 A6. Social safety net programs-coverage, targeting, and adequacy, by households characteristics Coverage (YO) Targeting (YO) adequacy 2000 2002 2003 2000 2002 2003 2000 2002 2003 quintiles of 1 47 45 40 29 28 28 0.18 0.17 0.24 consumption per 2 39 37 34 24 23 24 0.11 0.11 0.16 capita 3 31 32 29 19 19 21 0.09 0.09 0.12 4 28 27 23 17 17 17 0.06 0.07 0.09 5 18 19 13 12 13 I O 0.04 0.04 0.05 residence area urban 27 23 21 32 27 29 0.08 0.08 0.11 rural 36 37 32 68 73 71 0.09 0.09 0.I3 region Tashkent 24 22 17 14 12 11 0.07 0.07 0.08 Mirzachul 35 26 20 7 5 4 0.08 0.09 0.11 Ferghana 36 34 29 29 27 25 0.08 0.08 0.12 Northern 33 35 30 12 11 12 0.09 0.08 0.14 Central 31 32 26 19 20 20 0.08 0.09 0.13 Southern 35 41 43 20 25 28 0.12 0.11 0.15 sex of HHhead male 33 33 29 83 84 84 0.08 0.09 0.13 female 29 28 23 17 16 16 0.09 0.09 0.12 age of HHhead 16-29 42 47 45 12 13 14 0.11 0.11 0.15 30-39 34 38 33 33 33 32 0.11 0.10 0.14 40-49 25 24 22 18 19 21 0.08 0.08 0.11 50-59 32 33 24 12 14 12 0.06 0.07 0.10 60 and over 36 32 28 24 22 21 0.07 0.08 0.11 educationof HH <4 years 35 36 25 8 6 4 0.08 0.09 0.11 head 5-9 years 37 33 30 13 13 11 0.08 0.08 0.11 10-11 years 35 38 31 44 45 47 0.10 0.09 0.I4 completed 33 32 28 10 12 11 0.09 0.10 0.14 vocational/ lyceum incompletek 28 28 26 14 15 16 0.08 0.09 0.11 omplete college incompleteic 24 21 20 10 I O 11 0.06 0.07 0.10 omplete higher HHhead is no 31 32 28 72 74 75 0.09 0.09 0.13 pensioner Yes 36 33 28 28 26 25 0.07 0.08 0.11 HHhead is no 32 32 28 96 95 95 0.08 0.09 0.12 unemployed, 34 40 29 4 5 5 0.10 0.12 0.15 looking for ajob Yes HHhead is no 32 32 28 99 97 98 0.08 0.09 0.12 unemployed, not 33 38 28 1 3 2 0.11 0.11 0.17 looking for ajob Yes HHheadhasajob no 36 34 29 49 45 42 0.08 0.09 0.12 Yes 30 31 27 51 55 58 0.08 0.09 0.13 HHheadis no 32 32 28 97 96 96 0.08 0.09 0.12 disabled Yes 33 41 31 3 4 4 0.09 0.08 0.11 Total 32 32 28 100 100 100 0.08 0.09 0.12 134 A7. Child allowance for children under 2 -coverage, targeting, and adequacy, by householdscharacteristics Coverage (YO) Targeting (YO) adequacy 2000 2002 2003 2000 2002 2003 2000 2002 2003 quintilesof 1 23 25 23 24 27 28 0.12 0.14 0.21 consumptionper 2 capita 23 20 18 23 22 23 0.08 0.09 0.13 3 18 18 17 19 20 22 0.06 0.08 0.11 4 17 14 12 20 16 16 0.05 0.06 0.08 5 11 11 8 14 14 11 0.03 0.04 0.05 residencearea urban 15 14 12 32 30 29 0.06 0.07 0.10 rural 20 20 18 68 70 71 0.06 0.08 0.11 region Tashkent 14 12 9 15 13 10 0.05 0.06 0.06 Mirzachul 24 17 17 8 5 6 0.05 0.06 0.09 Ferghana 19 19 18 28 28 26 0.06 0.07 0.10 Northern 21 20 16 12 13 12 0.05 0.08 0.13 Central 20 16 17 25 19 24 0.06 0.08 0.12 Southern 15 22 19 13 21 23 0.07 0.09 0.I 4 sex ofHHhead male 19 18 16 83 83 84 0.06 0.07 0.11 female 16 16 13 17 17 16 0.06 0.07 0.I O age of HHhead 16-29 31 32 36 17 16 20 0.08 0.10 0.13 30-39 14 15 13 22 25 24 0.06 0.09 0.I 4 40-49 10 9 8 13 11 12 0.06 0.06 0.09 50-59 25 25 21 19 20 17 0.05 0.06 0.08 60 and over 24 21 20 29 27 26 0.05 0.07 0.10 educationof HH <4 years 21 23 14 9 6 3 0.06 0.07 0.10 head 5-9 years 21 20 20 14 14 13 0.06 0.07 0.09 I O - 11 years 19 19 16 41 41 43 0.07 0.08 0.12 completed vocational/ lyceum 16 14 15 9 I O 10 0.07 0.08 0.11 incomplete/c omplete college 16 16 15 14 16 16 0.05 0.07 0.I O incompleteic omplete higher 17 15 14 14 13 15 0.05 0.06 0.09 HHheadis no 16 16 14 65 67 69 0.06 0.08 0.11 pensioner Yes 24 23 20 35 33 31 0.06 0.07 0.10 HHheadis no 18 18 16 97 97 95 0.06 0.07 0.10 unemployed, looking for ajob Yes 18 16 16 3 3 5 0.07 0.10 0.15 HHheadis no 18 18 16 99 97 98 0.06 0.07 0.11 unemployed,not looking for ajob Yes 15 21 18 1 3 2 0.08 0.09 0.14 HHheadhas ajob no 22 21 19 55 50 47 0.06 0.07 0.10 Yes 15 15 14 45 50 53 0.06 0.07 0.11 HHheadis no 18 18 16 97 97 98 0.06 0.07 0.11 disabled Yes 15 20 12 3 3 2 0.07 0.08 0.07 Total 18 18 16 100 100 100 0.06 0.07 0.11 135 A8. Benefit for families with childrenunder 16 (18) - coverage, targeting, and adequacy, by householdscharacteristics Coverage (YO) Targeting (YO) adequacy 2000 2002 2003 2000 2002 2003 2000 2002 2003 quintilesof 1 33 29 27 34 29 28 0.16 0.14 0.19 consumption per 2 0.09 0.12 capita 23 22 21 23 24 26 0.09 3 17 15 16 19 19 20 0.07 0.07 0.09 4 13 14 12 14 17 18 0.05 0.06 0.07 5 9 9 6 10 12 8 0.03 0.03 0.04 residencearea urban 15 12 12 29 24 26 0.07 0.06 0.08 rural 21 21 19 71 76 74 0.08 0.08 0.10 region Tashkent 12 12 10 11 12 10 0.06 0.06 0.07 Mirzachul 17 14 7 6 5 3 0.06 0.08 0.09 Ferghana 24 20 16 32 29 25 0.07 0.07 0.09 Northern 20 18 20 12 10 13 0.07 0.07 0.10 Central 14 16 13 15 19 16 0.07 0.08 0.10 Southern 23 24 30 24 25 34 0.12 0.09 0.12 sex of HHhead male 19 19 17 84 86 86 0.07 0.07 0.10 female 16 14 13 16 14 14 0.07 0.07 0.08 age ofHHhead 16-29 19 21 19 8 8 8 0.09 0.08 0.10 30-39 25 26 24 43 41 43 0.09 0.08 0.11 40-49 16 16 16 21 25 28 0.07 0.08 0.10 50-59 13 13 9 7 8 6 0.05 0.05 0.07 60 and over 19 14 14 21 17 15 0.06 0.06 0.07 education ofHH <4 years 21 18 14 8 5 4 0.07 0.07 0.08 head 5-9 years 21 17 16 12 11 9 0.07 0.06 0.08 10-11 years 21 22 20 46 47 50 0.08 0.08 0.11 completed vocational/ lyceum 21 21 19 11 15 13 0.07 0.08 0.11 incompleteic omplete college 16 16 15 14 15 16 0.07 0.07 0.09 incompleteic omplete higher 11 9 9 8 8 8 0.06 0.06 0.08 HHheadis no 19 19 18 77 81 83 0.08 0.08 0.10 pensioner Yes 19 15 14 23 19 17 0.06 0.06 0.07 HHhead is no 19 18 16 96 94 94 0.07 0.07 0.10 unemployed, looking for ajob Yes 19 26 18 4 6 6 0.08 0.10 0.12 HHheadis no 19 18 16 99 96 98 0.07 0.07 0.10 unemployed,not looking for ajob Yes 22 27 18 1 4 2 0.07 0.08 0.13 HHheadhas ajob no 20 17 15 44 39 36 0.07 0.07 0.09 Yes 18 18 17 56 61 64 0.08 0.07 0.10 HHheadis no 19 18 16 97 96 96 0.07 0.07 0.10 disabled Yes 18 21 20 3 4 4 0.06 0.07 0.08 Total 19 18 16 100 100 100 0.07 0.07 0.10 136 A9. Benefit for low incomefamilies -coverage, targeting, and adequacy, by householdscharacteristics Coverage (YO) Targeting (YO) adequacy 2000 2002 2003 2000 2002 2003 2000 2002 2003 quintilesof 1 7 5 3 26 28 27 0.16 0.15 0.20 consumptionper 2 capita 4 4 2 24 24 20 0.11 0.10 0.14 3 3 2 2 18 16 25 0.08 0.09 0.13 4 3 3 1 19 18 15 0.07 0.06 0.09 5 2 2 1 12 14 12 0.04 0.04 0.06 residencearea urban 4 2 2 40 23 45 0.08 0.08 0.12 rural 4 4 2 60 77 55 0.08 0.08 0.12 region Tashkent 3 1 2 21 10 27 0.07 0.09 0.14 Mirzachul. 4 3 1 6 6 3 0.08 0.12 0.16 Ferghana 3 1 1 16 8 13 0.07 0.06 0.11 Northern 4 3 1 13 9 8 0.08 0.08 0.12 Central 4 5 2 17 30 18 0.07 0.07 0.11 Southern 6 7 4 26 37 31 0.11 0.09 0.11 sex of HHhead male 4 3 2 79 75 68 0.08 0.08 0.11 female 4 4 2 21 25 32 0.09 0.09 0.13 age of HHhead 16-29 4 5 3 8 14 7 0.08 0.11 0.11 30-39 4 4 2 34 30 26 0.10 0.07 0.10 40-49 5 3 2 29 24 31 0.08 0.08 0.12 50-59 2 2 1 9 11 13 0.07 0.08 0.15 60 andover 3 3 2 21 20 23 0.07 0.08 0.13 education of HH <4 years 3 4 2 8 7 7 0.10 0.11 0.13 head 5-9 years 4 3 1 11 14 9 0.08 0.08 0.12 10-11 years 5 5 2 50 52 54 0.09 0.08 0.13 completed vocational/ lyceum 5 3 2 11 10 7 0.08 0.07 0.11 incomplete/c omplete college 3 1 2 11 11 15 0.06 0.11 0.09 incomplete/c omplete higher 2 1 1 9 6 9 0.08 0.05 0.11 HHheadis no 4 3 2 75 76 75 0.09 0.08 0.12 pensioner Yes 4 3 1 25 24 25 0.07 0.08 0.13 HHhead is no 4 3 2 97 93 97 0.08 0.08 0.12 unemployed, looking for ajob Yes 4 4 2 3 7 3 0.09 0.15 0.07 HHhead is no 4 3 2 96 98 99 0.08 0.08 0.12 unemployed, not looking for ajob Yes 8 2 1 4 2 1 0.15 0.11 0.17 HHheadhasajob no 4 3 2 50 46 52 0.08 0.09 0.12 Yes 4 3 2 50 54 48 0.09 0.07 0.11 HHheadis no 4 3 2 93 95 86 0.08 0.08 0.12 disabled Yes 7 6 8 7 5 14 0.10 0.08 0.10 Total 4 3 2 100 100 100 0.08 0.08 0.12 137 AIO. Overlap o f social safety net programs 2000 2002 2003 Numberof simultaneously receivedtransfers 1 2 3 1 2 3 1 2 3 quintilesof 1 78 21 1 79 19 2 81 18 1 consumption per 2 84 14 1 85 15 1 capita 80 19 1 3 82 17 1 86 13 0 88 11 1 4 84 15 1 88 11 1 91 9 0 5 85 14 1 85 14 1 90 10 0 residencearea urban 81 18 1 84 15 1 88 12 0 rural 81 18 1 84 15 1 85 14 1 region Tashkent 83 16 1 86 14 0 89 11 0 Mirzachul 79 18 3 83 17 1 91 9 0 Ferghana 82 17 1 86 13 1 90 10 0 Northern 75 24 1 87 13 0 79 21 1 Central 83 16 1 87 13 0 88 11 1 Southern 79 19 1 75 22 3 81 18 1 sex of HHhead male 81 18 1 84 15 1 86 13 0 female 82 17 1 84 15 1 85 14 1 age of HHhead 16-29 0 0 0 0 0 0 0 0 0 30-39 75 24 1 82 14 4 80 20 0 40-49 80 20 1 83 16 1 85 15 1 50-59 86 13 1 89 10 1 90 10 0 60 and over 86 12 1 80 19 1 87 12 0 educationof HH <4 years 77 21 1 84 16 1 85 14 1 head 5-9 years 0 0 0 0 0 0 0 0 0 10-11 years 76 22 2 82 18 0 93 5 2 completed vocational/ lyceum 82 16 1 83 16 1 85 15 0 incompleteic omplete college 80 20 1 83 15 2 84 15 1 incomplete/c omplete higher 80 20 0 85 15 1 85 14 HHhead is no 84 14 2 84 15 1 88 12 pensioner Yes 85 14 1 87 13 0 90 10 HHheadis no 82 17 1 85 14 1 87 13 unemployed, looking for ajob Yes 79 20 1 82 17 1 84 15 HHheadis no 81 18 1 84 15 1 86 14 unemployed, not yes looking for ajob 84 16 0 85 15 0 85 15 HHheadhas ajob 81 18 1 84 15 1 86 13 74 21 5 71 28 2 75 25 HHheadis 79 20 1 83 16 1 84 16 disabled 83 16 1 85 14 1 88 12 Total 81 18 1 84 15 1 86 13 138 A I1. Benefit for families with childrenunder 16 (18) - awareness, application, andaward, by householdscharacteristics, 2003 applicants Know Knowled Applicants Waiting Applicants Average duration about geable rejected for awarded benefit the apply for decision awarded program benefit quintilesof 1 87 32 9 13 78 6994 6 consumption per 2 85 27 10 16 74 6940 6 capita 3 87 23 12 15 73 6740 6 4 89 22 16 16 68 6368 6 5 84 12 24 10 67 6109 6 residencearea urban 83 18 19 11 70 6289 6 rural 89 26 10 16 74 6892 6 region Tashkent 84 16 20 8 72 6153 6 Mirzachul 73 21 19 25 56 6998 6 Ferghana 93 24 14 13 73 6676 6 Northern 95 23 6 19 74 6489 6 Central 88 20 14 22 64 7039 6 Southern 78 37 8 9 83 6947 6 sex of HHhead male 87 24 12 15 73 6880 6 female 84 21 16 12 73 5996 6 age of HHhead 16-29 84 27 12 14 73 5980 6 30-39 88 30 11 16 74 7055 6 40-49 87 21 10 13 77 6900 6 50-59 84 15 22 14 65 6260 6 60 and over 87 20 18 14 69 6247 6 education ofHH <4 years 85 22 22 9 69 7056 6 head 5-9 years 87 23 16 14 70 6583 6 10-11 years 87 27 9 15 76 6746 6 completed vocationall lyceum 88 26 16 14 70 6851 6 incompletelc omplete college 86 22 13 13 74 6638 6 incomplete/c omplete higher 85 13 20 15 66 6715 6 HHheadis no 87 24 11 15 74 6852 6 pensioner Yes 86 19 18 13 69 6175 6 HHhead is no 86 23 13 14 73 6713 6 unemployed, looking for ajob Yes 89 28 8 17 75 7007 6 HHheadis no 87 23 13 14 73 6735 6 unemployed,not looking for ajob Yes 78 27 11 18 71 6563 6 HHheadhasajob no 86 23 13 14 73 6519 6 Yes 87 24 13 15 73 6869 6 HHheadis no 87 23 13 15 73 6710 6 disabled yes 90 27 8 12 80 7195 6 Total 87 23 13 14 73 6733 6 139 A12. Low income benefit -awareness, application, and award, by households characteristics, 2003 applicants Know Knowled Applicants Waiting Applicants Average duration about geable rejected for awarded benefit the apply for decision awarded program benefit quintiles of 1 84 8 39 22 38 6394 4 consumptionper 2 44 32 24 6349 4 capita 83 6 3 85 5 39 16 45 5973 5 4 86 6 46 22 32 5467 4 5 81 3 53 16 31 5030 5 residencearea urban 81 5 52 14 34 5819 4 rural 86 6 38 27 35 5956 4 region Tashkent 80 4 45 11 45 5691 5 Mirzachul 70 6 40 35 25 5667 4 Ferghana 90 4 55 11 34 5954 3 Northern 93 4 35 32 33 5503 3 Central 87 7 44 34 23 6233 4 Southern 75 9 36 20 43 5982 5 sex ofHHhead male 85 5 47 22 31 6001 4 female 82 7 34 22 43 5725 4 age of HHhead 16-29 80 8 36 22 41 5299 5 30-39 86 6 44 22 34 5696 4 40-49 84 6 39 27 34 6283 4 50-59 84 3 48 19 33 6173 5 60 and over 83 5 50 17 33 5935 4 education of HH <4 years 82 7 45 22 33 6209 4 head 5-9 years 83 5 50 17 33 6354 4 10-11 years 85 6 38 23 39 5778 4 completed vocational/ lyceum 84 7 5 1 28 20 5115 4 incomplete/c omplete college 84 5 39 22 39 6029 5 incompleteic omplete higher 83 3 62 17 22 6898 3 HHheadis no 84 6 42 24 34 5926 4 pensioner Yes 83 5 49 18 33 5848 4 HHhead is no 84 5 43 22 35 5971 4 unemployed, looking for ajob Yes 87 6 51 23 26 4544 4 HHheadis no 84 5 43 22 34 5898 4 unemployed, not looking for ajob Yes 79 5 52 16 32 6775 4 HHheadhas ajob no 83 6 42 21 37 5936 4 Yes 85 5 45 24 31 5877 5 HHheadis no 84 5 45 23 32 5860 4 disabled yes 87 12 26 17 57 6186 4 Total 84 5 44 22 34 5907 4 140 A13. Awareness, application, and award ofmeans-testedtransfers, by quintiles, 2000 and 2002 Know Knowled Applicants Waiting Applicants Average duration about geable rejected for awarded benefit the apply for decision awarded program benefit Benefitfor families with childrenunder 16(18), 2000 applicants quintiles of 90 46 18 12 71 3091 6 consumptionper 21 capita 88 40 22 13 65 2915 6 3 85 34 19 17 64 3001 6 4 85 29 25 12 63 2741 6 5 84 22 35 11 54 2680 6 Benefitfor families with childrenunder 16(18), 2002 applicants quintiles of 89 33 9 14 77 5348 6 consumptionper 21 capita 88 29 14 13 73 5296 6 3 89 23 18 14 68 5242 6 4 92 21 15 17 68 5275 6 5 90 14 21 11 67 5431 6 Benefitfor families with childrenunder 16 (18), 2003 applicants quintiles of 87 32 9 13 78 6994 6 consumptionper 21 capita 85 27 10 16 74 6940 6 3 87 23 12 15 73 6740 6 4 89 22 16 16 68 6368 6 5 84 12 24 10 67 6109 6 Low income benefit, 2000 applicants quintiles of 1 86 23 43 24 33 2877 4 consumptionper 2 capita 85 19 47 18 34 2903 5 3 83 14 43 23 35 2988 4 4 81 13 50 16 34 2837 5 5 78 9 55 17 28 2546 5 Low income benefit, 2002 applicants quintiles of 87 10 32 22 46 5272 4 consumptionper 21 capita 86 9 37 16 47 5401 4 3 87 6 40 20 40 5786 4 4 88 6 40 13 47 5045 5 5 84 4 44 16 40 5212 4 Low income benefit,2003 applicants quintiles of 1 84 8 39 22 38 6394 4 consumptionper 2 83 6 44 32 24 6349 4 capita 3 85 5 39 16 45 5973 5 4 86 6 46 22 32 5467 4 5 81 3 53 16 31 5030 5 141 A14. Reasons for not applying (%) Benefit for families with childrenunder 18 2000 2002 2003 You haveto fill in various documentsand certificates 23 22 23 We would not like our family's situationto be discussedinpublic 7 6 5 The mahallasdo not distributethe assistancejustly 8 6 7 We are usedto copingourselves 37 46 46 The social assistance paymentsare too low 2 1 1 We are not one of the pooresthouseholds 12 8 7 Our parents,relativeshelp us 6 5 5 Other 7 6 6 Low incomebenefit 2000 2002 2003 You haveto fillinvariousdocumentsandcertificates 17 17 19 We would not likeour family's situationto be discussedin public 8 7 7 The mahallasdo not distributethe assistancejustly 10 7 7 We are usedto copingourselves 39 47 47 The social assistancepayments are too low 1 1 1 We are not one of the pooresthouseholds 15 10 8 Our parents, relativeshelp us 4 5 4 Other 6 7 7 142 A15. Reasonsfor not applying(?!),by quintiles, 2003 Benefit for families with childrenunder 18 quintiles 1 2 3 4 5 You haveto fill in various documentsand certificates 32 24 22 21 21 We would not like our family's situationto be discussedin public 5 6 4 6 6 The mahallas do not distributethe assistance justly 7 6 7 7 6 We are usedto copingourselves 37 48 50 50 46 The social assistancepaymentsare too low 1 0 1 1 2 We are not one ofthe pooresthouseholds 7 5 6 6 10 Our parents,relativeshelp us 4 3 5 6 5 Other 7 8 5 5 5 Low incomebenefit quintiles 1 2 3 4 5 You haveto fill in various documentsand certificates 25 22 19 17 16 We would not like our family's situationto be discussedin public 7 7 6 6 7 The mahallasdo not distributethe assistance justly 8 6 6 7 7 We are usedto coping ourselves 39 46 49 51 47 The social assistancepaymentsare too low 1 0 1 1 1 We are not one of the poorest households 7 7 7 7 13 Our parents,relativeshelp us 3 4 3 5 5 Other 10 9 8 6 5 143 A16. Reasons for rejection o f application (%) Benefit for families with children under 18 2000 2002 2003 applicants applicants applicants our averagemonthly incomeis not low enough; there are others who are more needy 40 40 32 we havetoo muchproperty 7 5 6 that we have aplot which we could make better use of 4 5 2 there are/ is adultmembers ofthe family who couldbe working and shouldfind ajob 13 15 13 The mahallacommitteedoes not want to examine our situation 12 11 17 We hadto fill in various documentsand certificates 7 7 14 Our documentswere not in order (passport, residencepermitetc) 4 3 3 In our mahalla committee these mattersare decided on the basis of connections 5 4 3 Other reasons. 9 9 11 Low income benefit 2000 2002 2003 applicants applicants applicants our average monthly income is not low enough; there are others who are more needy 33 34 29 we havetoo much property 8 4 10 that we have aplot which we could make better use of 4 4 2 there are/ is adult members of the family who could be working and should find ajob 13 12 10 The mahalla committee does not want to examine our situation 20 21 20 We hadto fill in various documentsandcertificates 8 9 12 Our documentswere not in order (passport, residencepermitetc) 3 3 3 In our mahallacommitteethese mattersare decidedon the basis of connections 3 5 5 Other reasons. 8 8 10 144 A17. Reasonsfor rejectionof application(%) by quintiles, 2003 applicants Benefit for families with childrenunder 18 Quintiles 1 2 3 4 5 Total our average monthly income is not low enough; there are others who are more needy 31 23 29 34 44 32 we have too much property 6 1 2 6 5 3 6 that we have aplot which we could make better use of 0 3 2 4 0 2 there are/ is adult members of the family who couldbe working and should find ajob 17 10 11 16 12 13 The mahallacommitteedoes not want to examine our situation 23 18 20 15 10 17 We hadto fill in variousdocumentsandcertificates 5 20 9 13 22 14 Our documentswere not inorder (passport, residencepermit etc) 7 3 3 4 0 3 In our mahallacommitteethese matters are decided on the basis of connections 3 3 5 0 2 3 Other reasons. 8 8 15 13 10 11 Low incomebenefit Quintiles 1 2 3 4 5 Total our average monthly incomeis not low enough; there are others who are more needy 23 28 18 44 29 29 we havetoo much property 4 13 9 10 13 10 that we have aplot which we could make better use of 0 0 3 2 4 2 there are/ is adult membersof the family who couldbe working and should find ajob 7 17 6 9 10 10 The mahallacommitteedoes not want to examineour situation 26 17 29 12 20 20 We hadto fill in variousdocumentsand certificates 15 11 10 6 21 12 Our documentswere not in order (passport, residencepermit etc) 4 6 . 0 2 3 3 Inour mahallacommitteethese mattersare decidedonthe basis of connections 9 0 6 8 0 5 Other reasons. 12 8 19 6 4 10 145 A18. Inter-householdstransfers - coverage, targeting, and adequacy, by households characteristics Coverage (YO) Targeting (YO) adequacy 2000 2002 2003 2000 2002 2003 2000 2002 2003 quintilesof 1 5 6 6 20 10 13 0.40 0.26 0.37 consumptionper 2 4 11 6 7 0.14 0.13 0.15 capita 5 6 3 5 6 7 12 10 9 0.13 0.11 0.11 4 6 8 9 14 19 18 0.08 0.13 0.13 5 11 16 18 44 55 53 0.07 0.09 0.09 residencearea urban 11 16 17 68 81 79 0.10 0.11 0.12 rural 4 4 5 32 19 21 0.10 0.10 0.11 region Tashkent 11 18 21 41 53 58 0.10 0.10 0.12 Mirzachul 12 8 9 12 6 3 0.11 0.13 0.08 Ferghana 7 7 7 21 16 14 0.08 0.09 0.11 Northern 4 4 3 5 4 2 0.08 0.12 0.12 Central 5 6 6 14 17 13 0.08 0.14 0.13 Southern 2 4 8 7 5 10 0.22 0.10 0.13 sex of HHhead male 6 7 8 61 60 65 0.09 0.09 0.11 female 10 13 13 39 40 35 0.12 0.12 0.14 age ofHHhead 16-29 12 15 18 19 22 18 0.12 0.17 0.16 30-39 8 10 11 31 19 22 0.10 0.08 0.10 40-49 6 6 8 25 21 20 0.09 0.11 0.09 50-59 4 7 8 9 11 17 0.07 0.08 0.13 60 and over 5 8 8 15 27 23 0.10 0.10 0.13 education of HH <4 years 3 5 6 2 3 2 0.06 0.12 0.08 head 5-9 years 6 7 9 12 10 10 0.13 0.09 0.12 10-11 years 5 6 8 29 25 32 0.10 0.12 0.13 completed vocational/ lyceum 8 9 8 8 9 7 0.08 0.08 0.11 incompleteic omplete college 9 9 10 19 22 19 0.08 0.12 0.10 incompleteic omplete higher 9 14 13 29 31 29 0.10 0.10 0.11 HHheadis no 7 8 10 81 71 73 0.10 0.10 0.11 pensioner Yes 5 8 8 19 29 27 0.09 0.11 0.13 HHhead is no 6 8 9 92 95 91 0.09 0.10 0.11 unemployed, looking for ajob Yes 7 10 14 8 5 9 0.30 0.19 0.17 HHheadis no 6 8 9 99 99 99 0.10 0.10 0.11 unemployed, not looking for ajob Yes 6 5 9 1 1 1 0.07 0.12 0.18 HHheadhasajob no 6 9 10 50 55 58 0.13 0.12 0.15 Yes 7 8 9 50 45 42 0.08 0.09 0.09 HHheadis no 6 8 9 97 97 93 0.10 0.10 0.11 disabled Yes 9 12 13 3 3 7 0.11 0.08 0.15 Total 6 8 9 100 100 100 0.10 0.10 0.12 146 ANNEX2. OVERVIEW OFTHE SOCIAL PROTECTIONSYSTEM" 1. The LegalFramework The legalframework of the social protectioninUzbekistan consists ofthe Law onthe State-providedPensions for the Citizens of the Republic of Uzbekistan (1993), the Law on Pensions for Military Servicepersons (1992), the Resolutionof the Cabinet of Ministersof the Republic of Uzbekistanon Social Protectionof the Persons Who Suffered from the Consequences of the Chernobyl Catastrophe and Take Residence in the Republicof Uzbekistan(1992, as supplementedin 1994),which ensurethe rightsof personsfor state-provided pensions in the instanceof (i)old agehetirement, (ii)incapabilityto work infull or in part, and (iii) deceaseof a breadwinner. One of the first legally binding documents in Uzbekistan at the very beginningof the independence was the Law on Ensuring Social Protection of Disabled Persons (November 1991) which determines a state policy with regard to disabled persons in order (a) to ensure equal opportunities of enjoyment of their right and freedoms, (b) eliminate the limitations pertaining to their life and activity, (c) provide enabling conditions which should make such persons capable to practice socially inclusive lifestyle, participate in economic and politicalactivitiesandfulfill their civil duties. Protectionof labor and social rightsof the employed populationis codifiedin the Labor Code of the Republic of Uzbekistan (1994), which determinesthe rightsofworkers to labor, leisure, social welfare, guaranteedlabor remunerations standards, labor privilegesfor certain categories of workers, etc. Another importantlegislative enactment in the field of social protection is the Law of the Republic of Uzbekistan on Employment of Population(1992), which providesguaranteesfor social assistance the unemployedpersonsduringa periodof seekingemployment. This Law also establishes additionalguarantees of employment for certaincategories of persons (single parents and parents with many children under 14 years old and disabled children; young persons who graduated from educational establishments; disabled persons whose age is close to the general pensionage, etc.). The Laws on Education(1997) and on Public Healthcare (1996), which determinethe rightsof citizens to free- of-charge education and medicalaid, are directly related to implementingthe state policy of social protection. Social guarantees for housing are provided for in the Law on Privatization of State-Owned Housing Stock (1993). This Law provides for free-of-charge transfer of housing to disabled persons, war and labor veterans and teaching professionals. Budgetary funding of housing construction - as a state support and social protectionof vulnerable people (low-income peopleand familieswith many children, pensioners, war veterans etc.) - is exercised in consistencewith the Decree of President ofthe Republic of Uzbekistan on Measures for EncouragementandDevelopment of HousingConstructioninthe RepublicofUzbekistan(1994). Tax Code of the Republic of Uzbekistan (1995) provides for exemption from or preferentialtaxationof civil society organizations and institutions, associating disabled persons, war veterans and disabled war veterans, disabled persons who acquired disability because of general or occupational diseases, families with many children, families that have lost breadwinners, and single pensioners. The social protectionsystem is regulated also with the Decrees of President of the Republic of Uzbekistan. From 1992 through 2006 the minimum wage, pensions, stipends and social benefits were raised 3 1 times. Until 1994 wages were increased almost for all the workers, and since 1994 wages were raised only for the workers of those organizations, which are funded from the state budget. It is only a minimumwage which is establishedfor self-sustainingorganizations. This section is based on the backgroundnoteproviding an overview of the Uzbekistansocial protectionsystem, written by Olga Nemirovskaya 147 Figure 1. Changes inthe MinimumPensionandWage fiom 1994through2006 (July 1994=1) The social protection system was changed significantly with the decree on Measures for Enhancement of Social Protectionfor Low-incomeFamiliesof 23 August 1994. This bylaw pavedthe way for transition from paying social benefits to the entire population to means-tested targeted social assistance to families. This approach was further developed through the decree on Furtherance of Enhancing State-providedSupport to Families with Children of 10 December 1996, which provides for paying social benefits to low-income families with children, and in the decree on Enhancement of Targeted Support to Socially Vulnerable PopulationGroups of 25 January 2002, which providesfor paying social benefits to non-workermothers who care for children under 2 years old and notable increase of social benefits to children without parents and adoptingparents. The same decree specifies on the categories of the population, which need enhanced social protection, and establishes a system of privilegesandpreferentialbenefitsfor sociallyvulnerablepopulation. The following vulnerable categories and groups of the population are entitled to the right of state-provided socialassistancethroughprovidingsocial benefitsandprivileges andother types of social assistance: 1. Old-age retirementpensioners 2. Pensionerswho need special care 3. Pensionersprivilegedfor outstanding merits 4. Single aged persons who do not have sufficientyears of employment for old-age retirement pensions on the basis of employment years 5. Disabledpersons, belongingto the 1'' and 2"ddisability categories, and disabled children and persons who acquireddisability inchildhood(persons above 16years old,'who do not have sufficientyears of employmentfor disability pension) 6. Veterans of World War I1and personswhose status is equalized to the status ofthese veterans 7. Disabledpersons who acquired disability during World War I1and persons whose status is equalized to the status of these disabledpersons 8. The Heroes of the Soviet Union and persons who are awarded with all the three degrees of the Order of Glory 9. Familiesof deceaseddisabled personswho acquireddisability duringWorld War I1 10. Families of military servicepersons who deceased when fulfilled their duties during military service, includingthose who died duringWorld War I1 11. Persons who participated in eliminating the harmful effects of the accident at the Chernobyl Atomic Power Station 12. Disabledpersons who participatedin eliminatingthe harmful effects of the accident at the Chernobyl Atomic Power Station 13. Familiesof personswho deceasedas a result of the accident at the ChernobylAtomic Power Station 148 14. Parentswho care for a disabledchild, who acquired disability in childhood, livestogether with hidher parentsandneedspermanent special care 15, Children without parents, graduates of the Houses of Children `Mehribonlik' and special boarding schools, who do not have families, relativesor caretakers 16. Familieswho have losttheir breadwinners 17. Low-incomefamilies 18. Low-incomefamilies with children 19. Unemployedwomen who care for childrenunder2 years old 20. Personswho havechronic and severe socially detrimentaldiseasesand haveno disability status 21. Unemployedpersonswho are officially registeredby governmentallabor agencies 22. Childrenunder 18 years old who were evacuatedfrom the alienation areas, extraordinaryresettlement areas and areas where local residentshavethe rightto extraordinaryresettlement 23. Personswho were imprisonedinthe fascist concentrationcampswhen they were underage 24. Personswho survived duringthe blockadeof LeningradduringWorld War I1 25. Persons who were the participants of the `Labor Front' (Le., who worked at the enterprises 1941- 1945). The Structure of Social Protection System in Uzbekistan The social protection system has two major components: social insurance and social assistance. Social insuranceprograms includecertaintypes of socialwelfare paymentswithin the frameworkof mandatory state- led insurance, which are financed from extra-budgetary funds. Social assistance is provided to certain categories of citizens irrespective of their participationin the social insurance system and is financed from the state budget. 1.1. Social Insurance The social insurance system consists of several components: pension (old-agehetirement, disability, deceased breadwinner's family), unemployment insurance, insurance of temporary inability to work (sickness, pregnancy, childbirth, care of childrenunder 2 years old). A key source of fundingof extra-budgetaryfunds is the Single Social Paymentemployers must pay; in 2006 this single social payment amountedfor 25 percent of the total ofthe Wage Fund. A source of funding for unemployment benefits and pensions for the unemployed persons under the general pension age is the funds of the extra-budgetary Employment Fund under the Ministry of Labor and Social Protection. All the other social insurancepayments are paidfrom the extra-budgetary Pension Fund under the Ministry of Finance. The Employment Fundrelies onpaymentsfromemployers, for the amount of 0.8 percent ofthe total of Wage Fund.The PensionFundrelies on several sources: 0 employers' payments inthe amount of 24.2 percentofthe total ofthe Wage Fund employers' paymentsinthe amount of 0.7 percentofthe totalbudget ofwork ~0 employees' payments inthe amount of2.5 percentofthe grosswage 0 the funds from the Employment Fund, which cover expenditures for pensions for the unemployed underthe generalpensionage 0 fundingfromthe state budget, which compensateexpenditures for some social benefits. In 2005, in additionto the state system of old-age pensions, a mandatory accumulative system that covers all the workers was introduced. The People's Bank (Halk Bank) holdsthe accumulative pension system funds as personalpensionaccounts, receivedthroughthe deductions ofwage to the amount of 1percent of wage. Some 5.5 millionsof people are now covered with this system. In accordance with the nationallegislation, pension received through the accumulative system will be paid in additionto the state-provided pensiononce the right to the state-providedpensionwill beeffectuated. Old-age pensions are paid by the social welfare authorities when persons reach the pension age (55 years old for women and 60 years old for men) upon completionofthe minimumyears of employment (20 and 25 years for women and menrespectively). Somecategories of workers (whose employment is basedon hazardous and 149 hardtypes ofwork, agriculturalworkers, teachingprofessionals,some types of artistic professionals,etc.) have the right to obtain old-age pension before reaching the general pension age, if they have required years of employment. Moreover,persons, who were dismisseddue to redundanciesnot earlier than 2 years before the pension age and registeredwith labor exchanges in consistence with the established procedures also have the rightto the pensionbeforethe general pensionage. Old-agepensions havea two levelstructure: 0 the basic pensionwhich amounts to 55 percentof averagewage received duringany 5-years periodof employmentand must notbe lowerthe minimumestablishedby law; 0 the amount to be added to the basic pension, at 1 percent for each employment year exceeding the minimumemployment years, but must not be higherthan the establishedtotal amount linkedwith the minimumwage. In the case of insufficientyears of employment, proportionally lower pensions are paid. Some categories of pensioners receive additional amounts to their pensions. These categories include the disabled persons and veterans of World War 11, families of military servicepersons who died in action and of disabled veterans of World War 11, persons who worked at enterprises and institutions during World War 11, disabled persons, belongingto the 1'' category and some groups of the disabled persons, belongingto the 2ndcategory. The amounts, addedto the pensions, are differentiatedfor different categoriesof pensionrecipientsand rangefrom 25 percent of the minimumwage to 100percent of pensionestablishedfor a person. Disability pensions are paid by social welfare authoritieshodies on the basis of the statement of the Medical Labor Expert Board (MLEB) when disability pension recipients have required years of employment. Requirements for the years of employment are differentiatedby the age of disability pensionrecipient. When years of employment are below the required minimum, disability pension is paid to the amount which is proportionallylowerthanthe standardamount ofthat pension. The followingcategoriesof disabledpersons are establisheddependingonthe incapabilityto work: 0 disabled persons belongingto the lSt category - persons who have lost capabilityto work in full and needcare of another person 0 disabled persons belongingto the 2ndcategory - persons who have lost capabilityto work in full and do notneedcare of another person disabled persons belongingto the 3'd category - persons who have lost capabilityto work in part and are incapableto practicetheir previous occupation. The amount of pension depends on the previous wage (rangingfrom 30 percentto 100 percentof wage), years of employment, degree of reduced capability to work, and reasonfor disability. Disability pension, provided for the disabilityacquiredthroughoccupational trauma and occupational disease, does not depend on the years of employmentand is paidto the amount of 100percent o f the wage. Survivor's pension is providedto the members ofthe family of a deceasedperson, ifthese family members do not have employment. The right to receive such pension is granted to an unemployedwife and children and grandchildren under 18 years old. The amount of pension depends on the previous wage and years of employment ofthe deceasedbreadwinnerandthe number of dependants. Unemploymentbenefit.Unemployment benefit is grantedto the persons who are formally recognizedas unemployedby the governmental labor agencies. Inaccordancewith the Law OnEmployment ofthe Population, unemploymentbenefit is paidduringnot morethan 6 months duringa year to the amount of 75 percentofthe minimumwage andupto the 50 percent ofthe amount ofthe previous wage beforethejob was lost. The amount ofthis socialbenefitmaybe increasedby 10percent, ifthe benefitrecipienthas children. 150 Figure 2. Numberof PeopleRegisteredby Governmental LaborAgencies as Persons SeekingEmploymentand of UnemploymentBenefitRecipients The amount ofthe unemploymentbenefitdepends on the characteristicsofthe unemployedperson. There are severalcategories ofthe unemployed n n persons: i)the unemployedwho seekemployment for the first.time, ii)personswho lostajob and earnings, iii)personswho were dismissedfrom the Armed Forces, iv) personswho seek employment after a longperiodof unemployment,v) persons who graduatedfromeducational establishments. Unemployment benefit is provided through the rayodcity departments of labor, if an appropriate job or professionalretrainingare not providedto an unemployedpersonwithin 10days startingfrom the 0RegirIaredwithlabor apemies aa prsornseekingempioynenl date of application. Includingprrons M o m unemploynentbenefit is grantedio The unemployment benefit is discontinued: i)upon finding a job; ii)upon the expiration of the 6 months period; iii)if it is found that an unemployedperson is inactive in seeking employment. The average periodof receivingunemploymentbenefit is less than 3 months, since 83 percentofthe unemploymentbenefit recipientsobtainemployment before the legally establishedterm expires. Inthe instance of severe shortage of means for living subsistence, the labor exchange may grant economic aid benefitto an unemployedperson in the amount of one minimumwage as an additionalpaymentto the unemploymentbenefit. Social benefits for temporary incapabilityto work are paidthroughthe employer on the basis of the sickness sheet. The amount ofthis social benefitdependson the years of employmentand ranges from 50 percentto 80 percent of the wage. Pursuantto the sickness sheet, only the basic rate of wage is paid, and this socialbenefit i s subject to levying income tax. If the sickness sheet validity term i s extended up to 4 months without interruption within the same year, the sick person is required to have examination with the MLEB for possibility of providing disability status. The same terms are applied to a sickness sheet provided for providingcare to childrenandto family memberswho are incapableto work. Pregnancy and childbirth social benefit is paidtroughthe employer pursuant to the sickness sheet to be issued by the polyclinics responsible for the place of residence of recipient to the amount of 100 percent of wage duringpre-childbirthandpost-childbirthperiods, andtotal durationshall be 126calendar days. Childbirthsocialbenefit is paidthrough the employer to the amount of 2 minimumwages. This socialbenefit maybe receivedby any of employed parents. Childrenallowance for care of childrenunder 2 years old is paidthroughthe employer's organizations by the employer or through the educational establishments by administrations from their own funds. These expenditures are includedin the cost of productionand are not includedin the taxed amount. The amount of this social benefit is 200 percent of the minimumwage. This social benefit may be granted to the father or some other relative of a child, who was formally provided with childcare leave. The employees of the institutions funded from the state budget are provided with this social benefit from the funds of the state budget. Social benefit for funeral is paid to any member of the family of a deceased worker through the employing institutionor socialwelfare office on the basis of a statement of decease providedby the Civil RegistryOffice (ZAGS). This socialbenefit is one-timepayment and amountsto the 200 percentofthe minimumwage. 151 Data on the number of recipients of insurance payments from the Pension Fund (temporary incapability to work, pregnancy and childbirth, care of child under 2 years old, funerals) and average amounst of these payments are unavailable. 1.2. Social Assistance The social assistance system consists of two components: i)means-tested social benefits; ii)other social benefits for vulnerable groups. 1.2.I. Means-tested social benefits A mechanism of providing and paying social benefits for low-income (poor) families is used through self- government bodies, i.e., mahallas. The functions that were previously fulfilled by the state bodies and now delegatedto mahallas include: 0 acceptance of applications from individuals for receivingmoney payments as a from state-provided social assistance 0 identificationof low-income families, which need social assistance and reside within the area of the respectivemahalla 0 accumulating the funds, intended for paying social benefits for low-income families, on the special accountsofthe People's Bank 0 meanstestingo ffamilies, includingfamily money resourcesand other assets 0 makingdecision onthe approval andpaymentof socialbenefits 0 preparingdocuments on payingsocialbenefits (payrolls) and financialreportingdocuments. Decisions on granting (or refusal to grant) means-tested social benefits must be taken by the Assembly of Citizens or the Assembly of Representativesof Houses or Streets of Mahalla. Since 2003 it is admissible to take decisions on grantingmeans-tested social benefits by the Special Commission composed of 20 or more members, who are electedby the Citizens' Assembly. Means-tested social benefits are paid via the offices and branches of the Halk Bank on the basis of payrolls prepared by self-government bodies. State regulation of social assistance to low-income families is implementedthrough the mechanismsof financingand controlover the observance of procedures for granting and paying social benefits, specifiedand approved in the appropriate bylaws (Regulations to be approved by the Resolutionof the Cabinet of Ministers, or to be adopted by the Ministry of Labor and Social Protectionof Populationandthe Ministryof Finance, registeredwith the governmentalbodiesofjustice). The departments of labor and social protection assign their officers for immediate communicationwith the self-government bodies within the process of granting and paying social assistance. Normally every such officer is responsible for 4-5 self-government bodies on the average, with the estimated load of 2,500-3,000 families per responsible officer. These officers are subordinated to the head of department of labor. Accordingto the standard-settingregulatorydocuments, the officer ofthe departmentof labor hasthe rightto: 0 exercise viewing or inspectionof any documents related to granting and paying social benefits and economic aidto low-incomefamilies, includingdataon incomefiled by applicants; 0 request information about income of applicant families from tax authorities, should the data providedby applicant families arise groundedsuspects in regardto veracitythereof; 0 apply to the khokimiyat, ifthe officer would disagreewith the decisions of the self-government bodies on providingsocial benefits and economic aid, reportingon that disagreement and asking the khokimiyat for taking appropriate measures with regard to the executives of the self- government bodies. Important actors of the state-run social assistance process for low-income families are the rayon (city) departments of finance, on which the following responsibilitiesare incumbent, in additionto proper allocating funds to the special accounts of self-government bodies: 0 exercising control over targetedspendingthe allocated funds; 152 checkingthe correctness ofpayrollsand accountingstatements; adjustment and redistributionof funds among self-government bodies in consistence with the needsand spending levelof funds. The means-testedsocial benefits are paidby local departments ofthe People's Bank and their branches, which are located almost in all inhabited localities. At present, more than 5,000 branches of the People's Bank, includingmobileones, operate inthe country. The Rayon and City Khokimiyats bear the followingresponsibilities:organizationalissues of the activitiesof self-government and financial and logistical and technical procurement thereof. A direct responsibility of Khokimiyats is to train the staff of self-government bodies in various fields, inter alia, providing state-run social assistance for low-incomefamilies. The `Mahalla' RepublicanCharitable Fund and its regionalbranches play significant role in organizingthe operation of self-government in the field of providing state-run social assistance for low-income families. Among other functions, the `Mahalla' Fund provide the staff of self-government bodies with services of copying necessary standards-setting and instructions and supplying expandable materials. The Fund also participates in organizingregulartrainingofthe managers andother staff ofthe self-governmentbodies. At the central (republican) and regional levels, the system is managed by the Ministry of Labor and Social Protectionandthe Ministry of Financeandtheir oblast departments. The responsibilitiesofthe centralministriesare: developingstandard-settingdocumentsand instructions; identifyingthe scope of budgetaryfunds and geographic allocationofthese funds; families . exercising control over the observance of granting and paying social benefits for low-income Benefitsfor low-incomefamilies (Economic aidto low-incomefamilies) are paidto the amount of from 1.5 to 3 of the monthly minimum wages. Within the indicatedinterval, self-government bodies may independently establish the amount of economic aid on the basis of real deprivationof low-incomefamilies. This social benefit is paid for a maximumof three months, and such family has the rights to apply for this social benefit uponexpirationofthe three monthperiod. Social benefits to families with underage children. Amount of this social benefit depends on the number of childrenunder 18 years old inthe family: 0 families with the only child under 18 years old receive this social benefitto the amount of 50 percent ofthe minimumwage per month; 0 families with two childrenunder 18 years old receive this social benefitto the amount of 100 percent of the minimumwage per month; 0 families with three childrenunder 18 years old receivethis socialbenefitto the amount of 140percent ofthe minimumwage per month; 0 families with four and more children under 18 years old receive this social benefit to the amount of 175 percentofthe minimumwage per month. This social benefitis paidduring6 months, and uponthe expirationofthe 6 monthsthe recipientfamilies have the right apply for this socialbenefitagain. Social benefit for the unemployed mothers who care of children under 2 years old is paid to the unemployed women, who actually providecare o f a child under 2 years old, to the amount of 200 percent ofthe minimum wage. This social benefit is paid during 12 calendar months and may be extended if economic situation of suchfamily would not improve. 153 1.2.2. Social benefits that are not means-tested Social benefits for the elderly and persons incapableto work and not having sufficient years of employment for old-age pension, are paid in accordance to the legislation on pensions. These social benefits are paid to men who are 65 years old and above and to women who are 60 years old and above, provided that these persons have no close relativeswho are obligedto ensure sustainingtheir agedrelativesinaccordancewith the nationallegislation. The amount ofthis social benefit is 50 percent ofthe minimumwage and is financedfrom the budget. This social benefit is providedby rayon(municipaVcity) departments of socialwelfare pursuantto the application from individuals and on the basis of a statement, which is issued by self-government bodies certifying that the applicant has no relatives who are obliged to ensure sustaining living of their elderly relatives. Social benefits for disabled persons who acquired disability in childhood is paid by social welfare bodies to persons at 18 years old and above, who do not have sufficient years of employmentand formally recognizedas disabled persons, belongingto the 1'' and 2ndcategories, who acquireddisability in childhood, by the Medical Labor Expert Boards (MLEB). The amount ofthis socialbenefit is 100percent of the minimum wage. Social benefit for disabledchildren is approved by the social welfare bodies pursuant to the resolutionof the Medical Control Commissions (MCC) based in polyclinics. The right to this social benefit is granted to childrenunder 16years old. The amount ofthis socialbenefit is 100percent of the minimum wage. Provision of a winter-clothes-kit and free-of-charge school items to children from low-income families is exercised by school administrationspursuant to the resolutionof the Parents' Committee. In2005, the cost of a winter clothes kit was about 90,000 UZS. This type of social assistance covers 15 percent oftotal enrolment of general educationschools. ! Social benefit to childrenwithout parents and children, who are deprivedofparentalcare, is paidat the time of the graduationfrom general educationalestablishments: i)for purchasingclothes, footwear and domesticware to the amount of 100minimum wages, ii)additional socialbenefitto the amount of 5 minimum wages. Social benefit to adoptingparents is granted to the families who are adopted a child as patronage families, to the amount of 3 minimum wages per month per each adopted child until such children reach 18 years old. This socialbenefit is approvedandpaidby the bodiesof parentalcustody. Compensationfor the exemptions of paymentsfor municipal services is paidto all the categories of vulnerable groups inthe amount of40 percent of minimum wage per month. This socialbenefit is paidthroughthe social welfare bodies. In2005, this compensationwas providedto 472,100 persons. In addition to the monetary social benefits, the national legislation allows for several privileges for certain categories of the population, irrespective of their income: tax exemptions; free-of-chargemedicines, dentistry services, prosthetic and orthopedic devices and recreational treatment; transport fare exemptions; compensation for gas for the owners of personally used vehicles; exemptions for payment for telephone communication, etc. Privilegesand exemptions are providedby social welfare bodiesand transport, municipal service and tax authorities pursuant to the documents issued by socialwelfare bodies. More than 2.6 millions of people use these privileges and exemptions (one or several). In 2005, corresponding average amount of these privilegesand exemptions per beneficiarywas about 40,000 UZS. 154