Report No. 44400-GE Georgia Poverty Assessment April 2009 Human Development Sector Unit South Caucasus Country Unit Europe and Central Asia Region Document of the World Bank NAEC NationalAssessment Center NGO NongovernmentalOrganization OECD Organizationfor EconomicCo-operationand Development OOP Out-of-pocket os1 Open Society Institute PAE Per Adult Equivalent PHC Primary HealthCare PPA Programmatic PovertyAssessment PIRLS Progressin InternationalReadingLiteracy Study PPA Programmatic PovertyAssessment (Georgia) PSRO PovertyReductionSupport Operation SA Social Assistance SAESA State Agency for Employmentand Social Assistance SES Supplemental EducationServices SME Small and mediumenterprises SP Social Protection SSBA State Social BenefitsAgency SUSIF State United Social InsuranceFund TSA Targeted Social Assistance TVE Technicaland Vocational Education UBP UniversalBenefit Package UN UnitedNations UNESCO UnitedNationsEducational,Scientific and Cultural Organization UNICEF UnitedNationsChildren's Fund WDI World DevelopmentIndicators RegionalVice President: Shigeo Katsu Country Director: Asad Alam Sector Director:Tamar ManuelyanAtinc Sector Manager: GordonBetcherman Task Team Leader/Co-Task Team Leader: Oleksiy Ivaschenko/AleksandraPosarac CONTENTS Acknowledgements.................................................................................................................................. - 1 - Executivesummary....................................................................................................................................... 1 I. I1. The Reporthas Five Key Messages:................................................................................................ 1 I11. Main Findings.............................................................................................................................. 2 The report offers the following key policy recommendations:.................................................. 12 Chapter 1: MacroeconomicDevelopments in Georgia............................................................................... 15 A. 15 Developmentsof Key Economic Indicators.................................................................................. Introduction.................................................................................................................................... C. B. 16 Key Institutional and Policy Reforms............................................................................................ 23 D. Growth, Productivity,and Employment........................................................................................ 26 E. MacroeconomicChallenges and Growth Sustainability................................................................ 27 F. Conclusionsand Policy Recommendations................................................................................... 29 Chapter 2: Poverty Profile in Georgia......................................................................................................... 30 A. Introduction.................................................................................................................................... 30 B. Changes in Welfare in Georgiaduring2003-2007 ........................................................................ 31 C. D. Consumption-basedPovertyand Inequalityin Georgiain 2007 ................................................... 35 Poverty Profile and Multivariate Analysis of Poverty................................................................... 40 F. E. Conclusions.................................................................................................................................... Other Evidenceof Living Standards.............................................................................................. 46 51 Chapter 3: RuralPoverty in Georgia........................................................................................................... ! 53 A. B. Introduction.................................................................................................................................... 53 C. The Rural Poor in 2007 .................................................................................................................. 58 Constraintsto RuralPoverty Reduction......................................................................................... 66 D . Conclusionsand Recommendations.............................................................................................. 69 Chapter 4: Labor Market Developments..................................................................................................... 72 and Linkagesto Poverty in Georgia............................................................................................................ 72 A. B. Labor Market Dynamics ................................................................................................................ 72 C. Employment,Earnings,and Hoursof Work .................................................................................. 74 83 Labor Marketsand Poverty............................................................................................................ Characteristicsof Unemployment.................................................................................................. D. E. 87 Conclusions.................................................................................................................................... 89 Chapter 5: Social Protectionand Povertyin Georgia................................................................................. 92 A. Introduction.................................................................................................................................... 92 C. B. The Incidenceof SocialTransfers.................................................................................................. Coverage of Social Transfers......................................................................................................... 93 95 D. Poverty Impactof SocialTransfers................................................................................................ 96 E. Conclusionsand Recommendations.............................................................................................. 97 Chapter 6: Health Reform, HealthOutcomes, and Poverty........................................................................ 98 A. Introduction.................................................................................................................................... 98 B. Recent Trends and ReformInitiatives.......................................................................................... 100 i C. HealthOutcomes.......................................................................................................................... 103 D. 109 E. FinancialProtectionand Out-of-pocketPaymentsfor Health..................................................... 113 F. Conclusionand Policy Recommendations................................................................................... PreliminaryEvidence on the Impactof the MedicalAssistance Program (MAP)....................... 116 Chapter 7: EducationReform, EducationOutcomes, and Poverty........................................................... 118 A. Overview of the Education Sector ............................................................................................... 118 B. 124 C. Inequalitiesin Access to Educationand Affordability................................................................. D. Need for Investmentsin Infrastructure........................................................................................ LearningOutcomes...................................................................................................................... 142 144 E. Conclusions and Recommendations ............................................................................................ 147 References................................................................................................................................................. 177 BOXES Box 2.1: Child Povertyin Georgia.............................................................................................................. 44 Box 6.1: Health SatisfactionAround the World ....................................................................................... 104 Box 6.2: MeasuringOOP: How High is Out-of-pocket(OOP) Spending on HealthCare in Georgia?...109 Box 6.3: Issuesin MeasuringFinancialProtectionin Health................................................................... 111 Box 7.1:The EducationSystem in Georgia.............................................................................................. 119 Box 7.2: EnrollmentRates in Basic Educationin Georgia....................................................................... 132 Box 7.3: Perceptionof PrivateTutoringand University Entrance Examinationsin Georgia...................136 FIGURES Figure 1:The ExpectedPoverty Impactof the Conflict with Russiaand GlobalEconomicCrisis..............2 Figure2: MonetaryIncomeper Adult Equivalent by UrbadRural (in 2007 prices) .................................... 4 Figure3: Geographic Distributionof Poverty in Georgia, 2007................................................................... 5 Figure4: Share of In-kindConsumptionand Sales of Agricultural Products in HouseholdDisposable Incomes, Percent........................................................................................................................................... 6 Figure 5: PovertyIncidence, by Type of Employmentand UrbadRural Location, 2007 ............................ 8 Figure6: Adequacy of Pensionsand the TSA across Various Households(pensions and TSA as percent of householdconsumption), 2007 .................................................................................................................. 8 Figure 7: PovertyImpactof Social Transfers, 2007 ..................................................................................... 9 Figure 1.1: ECA: Indexof Real GDP (1990=100)...................................................................................... 15 Figure 1.2: Georgia: Average Real Growth in SelectedSectors................................................................. 17 Figure 1.3: Georgia: FunctionalCompositionofthe State Budget Expenditures (plan), 2008 ..................20 Figure 1.4:Georgia: CPI Inflation(periodaverage), 1998-2007 ............................................................... 21 Figure 1.5: Georgia: Exportsand Importsof Goods, Million US$............................................................. 21 Figure 1.6: Georgia: ExternalCurrentAccount Deficit and Net FDI(percent of GDP) ............................ 22 Figure 1.7: Georgia: EBRDTransition Index in Selected Areas ................................................................ 25 .. 11 productivity)................................................................................................................................................ Figure 1.8: Georgia: RealValue Added (VA) per Employee. by Sector (as proxy measureof labor 27 Figure2.1: Compositionof MonetaryIncomes, 2003 Comparedto 2007.................................................. 32 Figure2.2: MonetaryIncomes per Adult Equivalentby Urban/Rural,(in 2007 prices)............................. 33 Figure2.3: MonetaryIncomesper Adult Equivalentby Quintile, (in 2007 prices) ................................... 33 Figure2.4: Growthand Redistributionof PovertyChanges, 2003-2007 ................................................... 35 Figure2.5: Expenditureand Poverty Levels, by Region............................................................................ 39 Figure2.6: Share of In-kindConsumptionin Disposable Income, Percent................................................ 50 Figure2.7: Compositionof HouseholdDisposable Income-the Poorest vs. the Richest, Percent ...........51 Figure3.1: Georgia- FoodImportsand Exports........................................................................................ 55 Figure3.2: Ruralversus UrbanPoverty Headcount in Georgia, 2007........................................................ 59 Figure3.3: Expenditure,PovertyIncidence, and PovertyConcentrationin Rural Areas, by Region ........63 Figure4.1: Average Monthly Earnings, by Sector (mainjob), 2006.......................................................... 79 Figure4.2: Average Monthly Earnings, by Gender, inthe Private/Public Sector (mainjob), 2006...........80 Figure4.3: Average Monthly Earnings, by Education Level, inthe Private/PublicSector (mainjob), 2006 ..................................................................................................................................................................... 81 Figure5.1 : Coverage of Populationby Targeted Social Assistance ........................................................... 93 Figure5.2: Coverage of Populationby Pensions........................................................................................ 94 Figure 5.3: CumulativeDistribution of Pension,TSA, and other Social Assistance.................................. 95 Figure 5.4: Gini Coefficientsof Populationwith Various Social Transfers ............................................... 96 Figure6.1:Most Common ProblemsFaced by Households, by Quintile................................................... 99 Figure 6.2: Top Prioritiesfor Government Investm'ent Identifiedby Georgian Households...................... 99 Figure 6.3: Self-assessedHealth,by Quintile........................................................................................... 104 Figure 6.4: Tobacco Use and Expenditures,by Quintile .......................................................................... 106 Figure 6.5: PhysicalAccess to Doctor, by Quintile.................................................................................. 106 Figure 6.6: Satisfactionand Trust, by Quintile......................................................................................... 107 Figure 6.7: Utilization of Health Services, by Quintile............................................................................. 108 Figure 6.8: FinancialAccess to HealthCare, by Quintile......................................................................... 108 Figure 6.9: ImpoverishingEffect of OOP for Health................................................................................ 111 Figure 6.10: MAP Coverage: BeneficiariesandNon-Beneficiaries,by Quintile..................................... 114 Figure 6.11: Impactof MAP on Utilization of Urgent Care, 2006 ........................................................... 115 Figure 6.12: Impactof MAP on Probabilityof IncurringCatastrophic Non-DrugOOP.......................... 115 Figure7.1: Gross Primary/Secondary EnrollmentRates in 2005, Georgia in the ECA context] ............. 119 Figure 7.2: School-agepopulationis expectedto shrink rapidly inthe short-to-mediumterm................120 Figure 7.3: Enrollmentrates by age suggest large inequalitiesin access to preschool and tertiary education between (minority) poor and (non-minority)non-poorchildren............................................................... 124 ... 111 Figure 7.4: Tertiary school attainment is significantly lower amongthe poor......................................... 125 Figure7.5: PreschoolNet EnrollmentRates, Georgia in a RegionalContext, 1989-2004 ...................... 127 Figure7.6: Marginal Effect in the Probabilityof a Child Aged 4 to 6 BeingEnrolled in Preschool, Given Observable Characteristics........................................................................................................................ 129 Figure7.7: About 60 percent of all childrenfrom poor households are not enrolled in preschool due to access and affordability constraints .......................................................................................................... 130 Figure 7.8: Comparedto 2003, the share of householdsnot sendingtheir childrento preschool due to lack of facilities has increased in the eastern part of the country..................................................................... 131 children in relationto learningand school environment........................................................................... Figure7.9: Only half of all householdsthink that school conditions meet the educational needs oftheir 133 Figure7.10: Less than 4 percentof all children in Georgiawho are enrolledin basic education attend privateschools........................................................................................................................................... 133 Figure 7.1 1: Children livingin minority householdsare less likely to use privatetutoring..................... 137 Figure 7.12: Reasonsfor Child Absenteeism, Georgia, 2003 and 2006 ................................................... 138 Figure 7.13: Enrollmentrates in higher education in Georgiasurpassthose of other Caucasuscountries. ................................................................................................................................................................... 140 Figure 7.14:Enrollmentby Type of Institutionin Georgia (Individualsaged 16 to 24) .......................... 142 Figure 7.15: Georgia's performance in PIRLS 2006 was lowerthan expected given its levelof income. ................................................................................................................................................................... 143 Figure 7.16: LinkingFacility Quality to EnvironmentalQuality and EducationOutcomes..................... 145 TABLES Table 1:HouseholdTotal Monetary Income (in constant 2007 prices): 2003 HBS Comparedto 2007 LSMS............................................................................................................................................................ 3 Table 2: Consumption-basedPoverty in Georgia. 2007 ............................................................................... 5 Table 3: GeorgiaComparedto the CIS and EU: Selected HealthIndicators. LatestAvailable Year.........10 Table 1.1:Georgia: SelectedEconomicIndicators..................................................................................... 16 Table 2.1: Household Total Monetary Income (in constant 2007 prices): 2003 HBS Comparedto 2007 LSMS.......................................................................................................................................................... 32 Table 2.2: Mean MonetaryIncomesper Adult Equivalentin RealTerms ................................................. 34 Table 2.3: Overall Poverty in Georgia........................................................................................................ 36 Table 2.4: Poverty, by Geographic Region................................................................................................. 37 Table 2.5: Sensitivityof PovertyIncidencewith Respectto the Choice of PovertyLine (percent)...........38 Table 2.6: Decompositionof Inequality, by Region................................................................................... 39 Table 2.7: Inequality in Per Capita ExpenditureDistribution,by Urban and RuralAreas......................... 40 Table 2.8: Inequality in Per CapitaConsumption Distribution,by Urbanand RuralAreas....................... 40 Table 2.9: Poverty, by Employment Status ofthe Household Head(percent)............................................ 41 Table 2.10: Poverty, by EducationLevel of Household Head(percent) .................................................... 42 iv Table 2.11:Poverty. by HouseholdHead's Gender (percent) .................................................................... 43 Table 2.12: Poverty, by DemographicComposition(percent) ................................................................... 43 Table 2.13: The Results of ConsumptionRegressions ............................................................................... 46 Table 2.14: SubjectiveEvaluationof Well-being, 2007 ............................................................................. 47 Table 2.15: Main ProblemFacedby Household,Percent ........................................................................... 48 Table 2.16: ComparingHouseholdConsumptionand IncomeAggregates................................................ 49 Table 2.17: DisposableIncome-basedPoverty........................................................................................... 49 Table 3.1:Trends in Agricultural Sector Performance, Selected Indicators............................................... 54 Table 3.2: Georgia:Trends in Perennial Crop Production('000 tons) ....................................................... 55 Table 3.3: Georgia:Change in MonetaryIncomeof Rural Households over Time ................................... 56 Table 3.4: Changes in RuralHouseholdMonetaryIncomeby Region, 2003-2007 ................................... 57 Table 3.5: Expendituresand Incomesof Rural Households, by Quintile, 2007 ......................................... 59 Table 3.6: Agricultural ResourceBaseof Rural Households in Georgia................................................... 60 Table 3.7: DemographicCharacteristicsof RuralHouseholds in Georgia................................................. 61 Table 3.8: IncomeCompositionof RuralHouseholds by Quintile, Georgia, 2007.................................... 62 Table 3.9: GeorgiaRural HouseholdExpenditureand IncomeComposition,by Region.......................... 65 Table 3.10: Agriculture ResourceLimitations of Poor Rural Households ................................................. 67 Table 3.11: Indicatorsof Vulnerability among RuralHouseholds............................................................. 68 Table 3.12: Subsistence Orientationof Poor Rural Households................................................................. 69 Table 4.1: Labor MarketParticipation,Employment,and Unemployment,2003-2006 ............................ 73 Table 4.2: Participation,Employment,and UnemploymentRates, by Gender .......................................... 74 Table 4.3: EmploymentProfile, 2003 Comparedto 2006 .......................................................................... 75 Table 4.4: EmploymentComposition, by Status of Company's Ownership(mainjob), 2003 Comparedto 2006............................................................................................................................................................. 76 Table 4.5: Employment,by Sector (mainjob), 2003 Comparedto 2006 ................................................... 76 Table 4.6: Real Monthly Earnings, by Sector (mainjob), 2003-2006 ....................................................... 77 Table 4.7a: Inequalityinthe Earnings Distribution, 2003 Comparedto 2006............................................ 78 Table 4.7b: Decompositionof the EarningsInequality, by Public/PrivateSectors, 2003-2006 ................78 Table 4.7~:Decompositionof EarningsInequality, by UrbadRural Areas, 2003 Comparedto 2006.......79 Table 4.8: Hoursof Employmentinthe Main Job Comparedto Havinga Secondary Job ........................ 81 Table 4.9: UnderemploymentProfile, 2003 Comparedto 2006................................................................. 82 Table 4.10: Underemployment,by Sector (mainjob), 2003 Compared to 2006........................................ 83 Table 4.11: UnemploymentProfile,2003 Compared to 2006 .................................................................... 84 Table 4.12: Compositionof Employmentand Unemployment,by EducationLevel, 2003 Comparedto 2006............................................................................................................................................................. 85 V Table 4.13: Unemployment Duration. 2003 Compared to 2006 ................................................................. 86 Table 4.14: Unemployment Duration vs. Various Individual Characteristics. 2006 .................................. 86 Table 4.15a: Distribution of the Working-age Population. by Poverty and Individual Employment Status (shares of total employment. percent)......................................................................................................... 88 Table 4.15b: Poverty Incidence, by Type of Employment and Urban/Rural, 2007.................................... 89 Table 6.1: Georgia: Recent Trends in Health Indicators........................................................................... 100 Table 6.2: Georgia and Region: Selected Health Indicators, Latest Available Year ................................ 101 Table 6.3: Impact of Health OOP on Poverty Indicators, Selected Countries .......................................... 112 Table 6.4: Households Experiencing Catastrophic OOP, Selected Countries .......................................... 113 Table 7.1: Many schools in Georgia are not working at full capacity ...................................................... 121 Table 7.2: Expenditures on general education (primary, basic, and upper secondary) are mainly allocated to pay for teacher wages............................................................................................................................ 122 Table 7.3: Value of Student Vouchers (in GEL)....................................................................................... 123 Table 7.4: There are large inequalities in attainment rates between native Georgian individuals and other minorities living inthe country................................................................................................................. 126 Table 7.5: The number of preschool available places (that is, capacity) has increased from 122,000 in 2003 to 151,000 in 2006............................................................................................................................ 126 Table 7.6: Preschool enrollment rates have increased from 19 percent in 2003 to 23 percent in 2006. Most of the increaseoccurred in the "east" side of the country......................................................................... 128 Table 7.7: Lack of access nearby is the main reason why children from minority groups are not enrolled in preschool............................................................................................................................................... 130 Table 7.8: Enrollment Rates in Basic Education....................................................................................... 131 Table 7.9: School Approval Rates in Georgia........................................................................................... 134 Table 7.10: One-third of all students in basic education participate in additional out-of-class tutoring...135 Table 7.11: Perception of their Socioeconomic Status among Higher Education Students Who Have Used Private Tutors ............................................................................................................................................ 136 Table 7.12: Absenteeism Rates, by Subgroup, Georgia, 2003 and 2006 .................................................. 137 Table 7.13: There has been a significant increase in the percentageof rural children who are absent from school because they need to work............................................................................................................. 139 Table 7.14: Postsecondaryenrollment among individuals in the richest quintile is twice as large as enrollment among individuals inthe poorest quintile............................................................................... 141 Table 7.15: About one-third of all individuals aged 16 to 24 think that their education finished after having attained secondary education........................................................................................................ 142 Table 7.16: More than 60 percent of all school buildings in Georgia need urgent repair......................... 146 Table 7.17: By 2015, if government efforts continue, about 39 percent of the student population would escape the risk of taking classes in a dysfunctional/unsafe building ........................................................ 147 vi ANNEXES ANNEX 1: Povertyand Social Impactof the Crisis................................................................................. 152 ANNEX 2: CumulativeDistribution Functions.2003 Comparedto 2007:National. Urban. and Rural.165 ANNEX 3: Determinantsof UnemploymentProbability.2003 Comparedto 2006................................. 166 ANNEX 4: Statistical Tables .................................................................................................................... 167 ANNEX BOXES Box A.1: SocialTransfers in Georgia....................................................................................................... 174 Box A.2: Targeted SocialAssistance (TSA) is Foundto HaveNo Crowding-outImpacton Private Transfers.................................................................................................................................................... 176 ANNEX TABLES Table A.1: Distribution of SocialTransfers in Georgia. 2007 .................................................................. 167 Table A.2: Impactof Targeted SocialAssistance on Monthly Consumption........................................... 168 Table A.3: PovertyReductionImpactof SocialTransfers for Households that Receive Pensions,TSA, and other SocialAssistance....................................................................................................................... 168 Table A.4: TSA Impacton Transitionacross Quintiles of Per Adult EquivalentConsumption(households that receive TSA) ...................................................................................................................................... 169 Table AS: Poverty Incidence, Severity,and Poverty Gaps beforeand after Pensions, TSA, and other SocialAssistance....................................................................................................................................... 170 Table A.6: Distribution of TSA Beneficiaries, by Household(HH) Size................................................. 171 Table A.7: Coverage of Targeted SocialAssistance (TSA), 2007............................................................ 172 TableA.8: Total BudgetCost of TSA to Cover all Extremeand Overall Poor Population(at current benefit level) ............................................................................................................................................. 173 Table A.9: Total BudgetCosts of Various Social Assistance (SA) Programs.......................................... 173 vii ACKNOWLEDGEMENTS This Poverty Assessment (PA) Report was prepared by the World Bank team as part of the Georgia Programmatic Poverty Assessment (PPA) work. The preparation of the Report was led by Oleksiy Ivaschenko (Task Team Leader, Economist, ECSHD) and Aleksandra Posarac (co-TTL, Lead Economist,ECSHD).The team members included: Diego Angel-Urdinola (Economist, ECSHD), Shiyan Chao (Senior Health Economist, ECSHD), Garry Christensen (Consultant, ECSHD), Mariam Dolidze (Economist, ECSPE), Faruk Khan (Country Economist, ECSHD), Ivan Khilko (Consultant, ECSHD), Nino Kutateladze (Operations Analyst, ECSHD), Michael Lokshin (Senior Economist, DECRG), Nino Moroshkina (Consultant, ECSHD), Rosalinda Quintanilla (Lead Economist, ECSPE), Zurab Sajaia (Economist, DECRG), Afsaneh Sedghi (Economist, ECSPE), Van Roy Southworth (Country Manager, ECCGE) and Owen Smith (Health Economist, ECSHD). The section that provides some preliminary estimates of the poverty and social impact of the recent crisis (conflict with Russia) was prepared by Aleksandra Posarac (Lead Economist, and Country Sector Coordinator for Human Development in the South Caucasus, ECSHD); Nino Moroshkina (Health Specialist, ECSHD); Nino Kutateladze (Education Specialist, ECSHD); Satoshi Ishihara(Social Development Specialist, ECSSD); Joanna De Berry (Social Development Specialist, ECSSD); and Oleksiy Ivaschenko (Economist, ECSHD). Diane Stamm (Consultant, ECSHD) and Carmen Laurente (Senior Program Assistant, ECSHD) provided valuable assistance in editing the Report. The team is also thankful to the Georgia Department of Statistics for sharing the 2003-2006 HBS data, andto MamukaNadareishvilifor sharingthe 2007 LSMS data. The Report has benefited greatly from the peer reviewer comments (on both the Concept Note and the draft Report) provided by Rocio Castro (Lead Economist, CFPVP), Mukesh Chawla (Sector Manager, HDNHE), Jeni Klugman (Lead Economist, AFTP2), Bekzod Shamsiev (Senior Agriculture Economist,ECSSD), and Salman Zaidi (Senior Economist, ECSPE). Very usefulwritten comments have also been received from Anthony Cholst (Lead County Officer, ECCU3), Donna Dowsett-Coirolo (Country Director, ECCU3), Bjom Hamso (Senior Energy Economist, ECSSD), Afsaneh Sedghi (Economist, ECSPE), Van Roy Southworth (Country Manager, ECCGE), and Ruslan Yemtsov (Lead Economist, MNSED). The task was undertaken under the guidance of Tamar Manuelyan Atinc (Sector Director, ECSHD), Arup Banerji (former Sector Manager, ECSHD), Gordon Betcherman (Sector Manager, ECSHD), Donna Dowsett-Coirolo (former Country Director, ECCU3), and Asad Alam (Country Director, ECCU3). - 1 - EXECUTIVE SUMMARY This report presents a comprehensiveanalysis of poverty and its main determinants using the most recent 2007 Living Standards Measurement Survey (LSMS) data. The report provides an in-depth analysis of rural poverty, the linkages between labor markets and poverty, the importance of social transfers for poverty alleviation, and the progress made since 2003 in the health and education sectors. The report also presents some findings on incomes' trends' based on the Household Budget Survey (HBS) from 2003 to 2006. The report aims to inform the development of Georgia's national strategy for growth and poverty reduction. The report also simulates possible poverty impacts from the dual shocks of the August 2008 conflict and the current global financial crisis. After the release of the 2008 data and beyond, it would be possibleto performa detailedanalysis o fthe actualwelfare effects ofthose events. I. TheReporthassixkeymessages: 1. The available data indicate that living standards in Georgia have improved in many dimensions since 2003. This is evidenced by: (a) household monetary incomes during 2003-2007 increased substantially in real terms; and (b) non-income indicators of poverty have improved, including access to and quality of educationand healthcare, and provision of public services (supply of electricity, and so forth). Among the monetary income components, social transfers registeredthe largest increase during 2003-2007. As a result, the share of social transfers in the monetary income of households, and especially poor households, increasednoticeably. 2. Poverty in Georgia continues to be deeply entrenched in rural areas, accounting for 60 percent of the poor. The main reasons for this are: (a) narrowly based economic growth that happened outside of agriculture; (b) agricultural employment, which accounts for 55 percent of total employment, continues to be mostly of a self-subsistence nature; and (c) incomes in the agricultural sector were on aggregate stagnant during 2003-2007, and remainmuch lower than average incomes in the economy. 3. The performance of the labor markets has so far not contributed much to poverty reduction. While the average real earnings in the economy have increasednoticeably since 2003, this did not contributemuchto poverty reduction.The main reasons for this are: (a) comprehensive economic and public sector reforms since 2003 have so far resultedin the shedding of labor-job destruction surpassed job creation; and (b) those sectors that did registeran increase in employment and/or wages (for example, construction, financial services, mining, public sector) account for only a minor share of overall employment. 4. Social assistance became an increasingly important lifeline for Georgia's poor. The main reasons for this are: (a) increased fiscal revenues made it possible to pay off pension arrears and to increase pensions, the main poverty-reducing transfer; and (b) the Targeted Social Assistance (TSA) program introducedat the end of 2006 appearsto be an important source of income for the extreme poor. Social transfers account for about one third of household disposable income for the households in the bottom quintile ofthe distribution,and they reduce both poverty and inequalitysubstantially. ' Theanalysis of trends is based on the income aggregate, since the consumption aggregate is not comparableover time. 1 5. The double shocks of the August 2008 conflict and the global financial crisis risk undermining the poverty reduction effort. Since Georgia has been hit by twin disasters of the August 2008 conflict and still unfoldingglobal economic crisis, the poverty headcount is forecasted to go up from 23.7 percent in 2007 to 27.1 percent in 2009 (Figure 1).2 Were it not for the negative welfare impacts of the twin crisis, the poverty headcount in Georgiacould be expected to decline from 23.7 percent in 2007 to 19.2 percent at the end of 2009. This 2009 gap in the "crisis" poverty headcount versus "no crisis" povertyheadcountrepresents in absolute terms about 350,000 poor people. Figure 1: The ExpectedPoverty Impact of the Conflict with Russiaand GlobalEconomic Crisis 07' 08` 09` -total poverty -m-- ...+...total poverty (no Russianconflict andfinancial crisis), extreme poverty .--8.. extreme poverty (no Russian conflict and financial crisis) Source: World Bank estimatesusing2007 LSMS data. 6. The poverty reduction strategy of the Government of Georgia (GOG) should focus on the following key areas: (a) extendingthe coverage of the TSA to reach more of the poor; that would also reduce poverty in rural areas, which still account for the majority of the poor; however, extended coverage would come at an increased fiscal costs to the Government; (b) promoting investments in infrastructure and creating opportunities for off-farm employment in rural areas; and (c) continuing reforms in the health and education sectors to improve human capital, which is the prerequisite for sustainable economic growth and poverty reduction. Other more detailed policy recommendations across various areas are presented later inthe ExecutiveSummary. 11. Main Findings 7. Since 2003, Georgia has implemented an impressive array of reforms. These reforms are reflected in pronounced political, social, and economic transformationsfollowing the "Rose Revolution" at the end of 2003. The processes since the start of the reforms can be qualified as unique in terms of speed of reforms, degree of innovations, and extent of institutional restructuring. The reforms are recognized to have noticeably improved the institutional environment, provided a basis for more sustainable economic growth and human capital accumulation, and increased multi-fold foreign direct investments. 8. Unfortunately, the absence of comparable household surveys over time precludes a proper analysis of the impact of these reforms on growth and poverty reduction. However, the data do The no-crisis scenario is based on the assumption of the economic growth at an average of 5% per annum during 2008-09. The actual (crisis) scenario is based on the simulated impact of the August 2008 conflict (see Annex l), andthe projectionsofthe 2.5% decline in per capitaconsumptionin 2009 dueto the impact ofthe financial crisis. 2 permit a full welfare analysis for 2007, plus the analysis of some trends since 2003. The mainfindings are presentedbelow. Werfare dynamics during 20034007 9. A direct comparison of consumption-basedpoverty during 2003-2007 is not possible due to non-comparable consumption aggregates between the HBS and the 2007 LSMS. This non- comparability is driven by the differences in the recall period and the list of consumption items. That is the main reason why the analysis of the welfare dynamics is based on monetary incomes, which are comparable across the two surveys. Unfortunately, there is evidence to suggest that the quality of the consumptiondata gathered through the H B S has deterioratedsince 2003, as the technical capacity of the Department of Statistics to administer the survey at the same level of quality has deteriorated. In this regard, the World Bank believes that the Governmentneeds to implement some measures to restore the technical capacity within the Departmentof Statistics. That would allow the Governmentto use the HBS as a useful and reliabletool for continuouspovertymonitoring. 10. The analysis of monetary incomes indicatessubstantial improvementssince 2003. Household real monetary incomes during 2003-2007 increased on average by 3 1.6 percent, or 7.9 percent per year. This rate of growth is cpnsistent with the rate of growth in national income reported in the national accounts (NA). Table 1: HouseholdTotal Monetary Income (in constant 2007 prices): 2003 HBS Compared to 2007 LSMS of which Total household -income Total household self-employed waae income Total household soc transfers income (incl pensions) Total household private (domestic) transfers Total household foreian transfers Total household rental income Source World Bank estimates using 2003 HBS and 2007 LSMS data. 11. Among the monetary income components, social transfers registered the largest increase during 2003-2007. Socialtransfers increasedin real terms from an average of 12.9GEL per household in 2003 to 42.6 GEL per household in 2007, or by 2.3 times. The importance of social transfers has also increased in relative terms. Their share in total monetary incomes increased from 6.8 percent in 2003 to 17.2 percent in 2007. It is worth notingthat privatetransfers (both domestic and foreign) decreasedat the same time. However,the analysis of the 2007 LSMSdata indicates that Targeted Social Assistance (TSA) had no crowding-out effect on private transfers. It appears that a decline in private domestic transfers since 2003 has been driven by crack-downof corruption, and a decline in remittances has been driven by Russia's sanctions on migrants from Georgia (deportation of migrants, cancellation of flights between Russia and Georgia, etc.) 12. Mostly as a result of pro-poor Targeted Social Assistance, poor rural areas registered a more significant increasein monetary incomescompared to urban ones. During2003-2007, per adult equivalent monetary income increased by an average of 29.9 GEL, or 81.3 percent, in rural areas (from a very low base) and 13.1 GEL, or 11.4 percent, in urban areas (Figure 2). However, average monetary incomes in rural areas are still at only about 50 percent of the average monetary incomes in urban areas. 13. Across quintiles o f the monetary income per adult equivalent (PAE) distribution, the largest rate of increase was registered by the poorest and the richest quintiles of the distribution. The structure o f monetary incomes indicates that the bottom quintiles have benefited mostly from increased social transfers, while the top quintile took advantage o f rising real salaries and income from own business. Figure 2: Monetary Income per Adult Equivalent by Urbanmural (in 2007 prices) 140 0 1200 ~ 1 I 1000 1 II 800 II I 600 I 400 1I 2 0 0 1 0 0 I Urban Rural Total Source: World Bank estimates using2003 HBS and 2007 LSMS data. 14. If poverty dynamics were to be measured by using monetary income PAE, they would definitely show a decline in poverty during 2003-2007. The analysis o f the cumulative distribution functions for 2003 and 2007 indicates that at any level o f the poverty line, the poverty incidence would have declined. At the total poverty line o f 71.6 GEL, the decline in total poverty is 4.4 percentage points, and at the extreme poverty line o f 47.1 GEL, the decline is 7.4 percentage points. The poverty decline is more pronounced in rural areas. 15. The poverty numbers released recently by the Department o f Statistics also indicate a decline in poverty. These numbers show a decline in the poverty headcount from 24.6 percent in 2004 to 21.3 percent in 2007. These poverty estimates are based on the 2004-2007 HBS data and the relative poverty line o f 60 percent o f median per capita consumption. Hence, the World Bank and the Department o f Statistics find broadly similar trends. However, the poverty numbers obtained by the Department o f Statistics are not directly comparable to the poverty estimates produced by the World Bank (WB) due to the differences in: (a) methodology (absolute poverty line and per adult equivalent consumption used by the WB versus relative line and per capita consumption used by GDS); and (b) data sources with different sampling and questionnaires (2007 LSMS used by the WB versus 2007 HBS used by the GDS). Poverty profile (200 7) 16. The analysis o f consumption-based poverty using the 2007 L S M S data and the World Bank methodology indicates that 23.6 percent o f the Georgian population is poor, and 9.3 percent is extreme poor. The poverty headcount is 29.7 percent in rural areas and 18.3 percent in urban areas. The extreme poverty headcount i s 12.4 percent and 6.7 percent, respectively (Table 2). As a result, rural areas account for 59 percent o f total poor, and 62 percent o f extreme poor. 4 Table 2: Consumption-based Poverty in Georgia, 2007 - I3eadcou"t Poverty Squared Poverty Paver+L i m e - R>,te(PO > G *, p cP 1) GZ%p 2007 2007 2007 72.6 L r r r i (IotaIpovcrty) Urban 18.3 5.3 2.3 Standard Error 1.2 0.5 0.3 Rural 29.7 9.2 4.1 Standard Error 1.7 0.7 0.4 Total 23.6 7.2 3.1 Standard Error 1 .o 0.4 0.2 Poverty Li,ze =47.2 Lari (exframe pover-0.1 percent o f GDP. 29. In terms of both coverage and benefit adequacy, pensions and TSA play the most dominant roles. Fifty-five percent o f households receive pensions and 8.4 percent receive TSA. In the 1st consumption quintile, 60.8 percent o f households receive pensions, and 15.4 percent receive TSA. For the extreme poor households, pensions and TSA account on average for 37.7 percent and 12.7 percent, respectively, o f total household consumption (Figure 6). Figure 6: Adequacy o f Pensions and the TSA across Various Households (pensions and TSA as ___ percent of household consumption), 2007 I TSA I I 0 10 20 30 40 50 60 Source. World Bank estimates using 2007 LSMS data. 8 30. Without social transfers, and especially pensions, poverty incidence and the poverty gap would be much higher. Under the scenario o f no pension payments, the poverty incidence would increase from 23.6 percent to 32.9 percent, and the poverty gap would increase from 7.2 percent to 14.3 percent. TSA (with its current coverage) and other social transfers further reduce the poverty incidence and the poverty gap by another 2 percentage points (Figure 7). Figure7: Poverty Impact of SocialTransfers, 2007 25 0 Source: World Bank estimates using2007 LSMSdata 31. Without social transfers, and especiallypensions, inequality would also be much higher. The simulations indicate that without pensions and other social transfers (including TSA), the Gini coefficient o f per adult equivalent consumption would be 40.9 percent, compared to the actual o f 36.3 percent. These simulations assume no behavioral changes on the part o f households, such as the search for alternative income sources, etc. 32. TSA reaches the poor well, and substantially increases their disposable income. The targeting o f the TSA is good-7 1 percent o f the TSA beneficiaries are among the consumption poor. TSA significantly increases the incomes o f poor households-by 72 percent for all poor households, and by 105 percent for extreme poor households (10 percent o f the population). 33. But its coverage even among extreme poor is still limited. As a result, the impact of the TSA on overall poverty remains marginal. As o f September 2007, TSA covered only 19 percent o f all (consumption) poor individuals and about 30 percent o f extreme poor individuals. The overall poverty incidence with no TSA would be 25 percent (compared to 23.6 percent actual, with TSA), so 54,000 people escape poverty because o f TSA. Reduction comes mostly from the rural poor (rural poverty incidence is 31.E percent with no TSA compared to 29.7 percent with TSA). Sixty-six percent o f current beneficiaries are located in rural areas, reflecting the concentration o f poor in rural areas. 34. The analysis indicates that the TSA does not crowd out private transfers. Econometric results using 2007 data fail to detect any statistically significant differences in the amounts o f the private transfers between the TSA-recipient households and the households with the similar TSA score that receive no assistance. In particular, the average amount o f private transfers among the non-recipients was 33.5 Lari, while the average amount o f private transfers for the households receiving the TSA was 30.2 Lari. This difference is statistically insignificant. Similar conclusions are made based on the multivariate analysis that controls for the differences in the observable characteristics o f households from the two groups. 9 35. Extending the coverage of the TSA program even in its current format holds much promise for further poverty reduction. The simulation analysis indicates that extending TSA coverage to all extreme poor would result in a decline of the extreme poverty incidence from 9.3 percent to 2.9 percent, and extending the TSA coverage to all poor would result in a decline of the overall poverty incidence from 23.6 percent to 12 percent.However,the extension of coverage will come with fiscal costs-another 0.4 percent of the GDP to extend coverage to all extreme poor, and 1.1 percentto extend coverage to all poor. Health sector reforms and outcomes 36. Health issues loom large in the lives of Georgia's poor. When asked to identify the main problems faced by their family, two-thirds of the poor mentionedthe purchase of medicines, while over half noted access to medical services. These responses place health concerns on a par with food security and employment as the most important problems faced by the poor. Similarly, when the Life in Transition Survey asked Georgian households to name the top two priorities for government investment, the health sector was the most common answer, includingamongthe poor. 37. Georgia's health indicators are generally better than those prevailing elsewhere in the Commonwealth of Independent States (CIS), but fall short of those in the new European Union (EU) member states. With the exception of infant mortality, which remains relatively high, all other health outcomes fall between those achieved in the CIS and EU-12 (Table 3). With respect to health system performance, however, Georgia lags far behind all regional groupings, with a low outpatient contact rate and high out-of-pocket(OOP) spending. Table 3: Georgia Compared to the CIS and EU: Selected Health Indicators, Latest Available Year Indicator Georgia EU-15 EU-12 CIS (W. Europe) (E. Europe) Lifeexpectancy 73.1 79.7 74.0 67.0 Infant mortality rate 19.7 4.3 8.3 13.4 (per 1,000 live births) Maternalmortality 23.0 5.3 9.0 28.2 (per 100,000 live births) Mortalitydue to diseases 545.1 238.7 523.4 773.0 of the circulatory system (per 100,000) Outpatient contacts 2.2 6.5 7.8 8.6 (per person per year) Source: WHO/Europe "Health for All" database, except *=Georgia National Health Accounts, 2006. 38. An ambitious health reform plan was launched in 2006. The measures include: (a) the introduction of the Medical Assistance Program (MAP) for the poor, benefiting about 700,000 people with a proxy means score below 70,000; MAP offers generous coverage and no co-payments, but does not cover drugs; (b) a voucher scheme is being expanded during 2007-2008 whereby the government pays private insurers to purchase MAP services on behalfof the poor; (c) publicly funded healthcoverage is beingstrengthenedfor about 400,000 civil servants includingteachers, police, and the military; and (d) reforms are underwayto strengthen private involvementin health care provision at both the hospital and primary care levels. In view of this rapidly evolving reform environment, more time and data will be requiredto reachclear conclusionsabout policy impact. 39. Reforms are being introduced in a context of significant inequality in health outcomes between the rich and poor in Georgia. The factors that play a role in the "health production function" are complex, but the evidence suggeststhat a significant component of this inequality can be attributedto 10 differences in utilization of care and the underlyingfinancial access considerations.Other factors, such as physical access (proximity) to care and non-clinical quality (as proxied by indicators of satisfaction and trust), do not appear to reflect significantinequality. 40. Preliminary evidence on MAP performance suggests good potential and room for improvement. A prerequisite for its success in improving both health outcomes and financial protection of the poor will be good coverage, and there is scope for greater progress in this regard because the majority of the poor are not yet benefiting from the program. Among those who do benefit, there is evidence that MAP is havinga positive impact on healthcare utilization and therefore has good potential to improve health outcomes among the poor. However, early evidence suggests that MAP is not havinga significant impact on OOP for health, in large part due to the exclusion of drugs from the MAP benefit package. Drugs account for about two-thirdsof household spending on healthamongthe poor. Education sector reforms and outcomes 41. Since 2003, the Government has undertaken a set of comprehensive reforms in the education sector. Those reforms include: (a) the introductionof a per capita funding scheme for schools, (b) an administrative school consolidation program; and (c) efforts to increase teacher salaries and improve the quality of postsecondary education. Furthermore,the government has made important efforts to improve the school learning environment through addressing the needs of deteriorating school infrastructure, the result of a lack of maintenanceover many years. 42. These efforts have contributed to improvements in the efficiency of using public funds for education. The achievements here since 2003 include: (a) the consolidation process has reduced the number of inefficienthnderusedschools-the number of schools was reduced from 3,154 to 2,331; and (b) the new per capita financing scheme has increased transparency and fairness in financial resources a1locations. 43. The reform efforts have also contributed to improvements in access to education. The achievements here since 2003 include: (a) preschoolenrollment rates have increased from 19 percent to 23 percent; (b) the number of preschoolplaces (that is, capacity) has increased from 122,000 to 151,000; (c) gross enrollmentrates in basic education are almost universalat 96.4 percent; and (d) enrollmentrates in postsecondary education increased from 45 percent in 2003 to 48 percent in 2006. Indeed, Georgia currentlyenjoysthe highest enrollmentrates for vocational and universityeducation inthe Caucasus. 44. The reforms have also contributed to improvements in the quality of education. The achievements here since 2003 include: (a) the share of childrenwho were absent from school due to lack of heatingat school decreased significantly from 37 percent to 15 percent; (b) in 2003, only 35 percent of students between9 and 10 years old mastered tasks that requirethe process of retrieving implicitly stated information from a certain part of a text, and interpreting and integrating ideas and information; this ranking increased to 45 percent; and (c) school approval rates for basic education improved from 53 percent to 57 percent. 45. Despiteall this progress, there are several issues that still need to be addressed in the sector in relation to education financing. They include: (a) the spending on education in Georgia (at 2.6 percent of GDP) remains very low by Europe and Central Asia (ECA) standards (at 4.4 percent of GDP on average); (b) most of the general education budget in Georgia is spent on current needs, such as wages and social security contributions for teachers; (c) the new per capita financing scheme in its current format may be contributing to increased education inequality, because small schools in rural areas (generally servingthe poor) are underfunded (comparedto larger schools in urban areas) despite receiving additional funds from the Ministry of Education; and (d) while the current government plan for capital 11 investments will be able to contain the rapid school depreciation, it will do so only partly. The expected sharp decrease in demand for education services due to a shrinking school-age population provides additional grounds for further development strategies to rationalize the already underutilized school system which, as of today, operates with greater-than-neededstocks of schools and teachers. 46. There are also several issues that still need to be addressed in the education sector in relation to access. They include: (a) the still large inequalitiesin enrollmentrates between poor and non- poor individuals, particularly at the preschool and postsecondary levels; (b) ethnic minorities display lower enrollment and attainment rates than native Georgians; (c) in rural areas, lack of access to facilities nearby is still the main reason why parents do not send their children to preschool; (d) urban non-poor children broadly use Supplemental EducationServices, such as private tutoring, to prepare for university entrance examinations; poor children cannot afford such services; and (e) drop-out rates in secondary education are high in rural areas, amongthe poor, and amongethnic minority children. 47. Additional issues that still need to be addressed in the education sector in relation to quality include: (a) poor rural households perceive that the quality of education received by their children has deterioratedsince 2003; (b) student absenteeism reaches up to 10 percent in poor regions, such as Kakheti and Imereti; and (c) Georgia's performance in PIRLS 2006 was lower than expected given its level of economic development. 111. The report offers the followingkey policyrecommendations: 48. Growth andpoverty. Despitethe rapidtransformations in Georgiasince 2003, the country's rural regions remain poor and underdeveloped. Hence, Georgia cannot rely solely on rapid national growth to generate broad-based poverty reduction. Policies are needed to integrate the rural poor into the growth process.The recommendations here include: Continued support for expansion and rehabilitation of physical infrastructure, especially in remote rural and mountainous areas. However, investments in physical infrastructure would not be enough-they need to be complemented by the continued investments in human capital accumulation through ongoing reforms in the health and education sectors. This will create a strong foundation for sustainable growth and further poverty reduction. Integrating the rural poor into the growth process, which can be achieved through: (a) expandingopportunities for off-farm employment in rural areas, including support for small and medium enterprise development; (b) increasing farm productivity and agricultural production in regions with high agricultural potential; and (c) exploring new markets for agriculturalproducts; this is especially importantgiven the lack of access to some traditional (for example, Russia) markets. Using the pledged donor resources with a view of the longer-term agenda focused on building Georgia's economic competitiveness. One of the main impacts of the twin crises is the expected slowdown in foreign direct investments (FDI), which are expected to decline from an estimated US$ 1.1 billion in 2008 to US$0.4 billion in 2009. At the same time, the donors pledged resources of US$ 4.5 billion over three years (which is US$ 1.5 billion per year, or one quarter of Georgia's GDP), and if used wisely, those resources could at least partly counterbalance the negative impact of the crisis on the economy and household welfare. 12 49. Social transfers andpoverty. Because increased employment and integration of the rural poor into the national growth process would require a longer time period given that reforms started relatively recently, and the reduction of poverty in the short to medium term may need to rely on redistributing the benefits of growth (and increased fiscal revenues) through social transfers, the recommendations here include: 9 Extending the coverage of Targeted Social Assistance holds much promise for poverty reduction, especially in terms of reachingthe rural poor. There is evidence that the current targeting formula works quite well in identifying needy households, so no substantial revisionsto the formula are warranted at this stage. However, more fiscal resources would be neededto extendthe coverage. The size of the TSA benefit should ideally take into account the extent of household poverty, to become a more efficient tool of reducing poverty. In its current design, the TSA benefit is not adjusted to reflect the distance to the cutoff proxy means score. As a result, 40 percent of the beneficiaries stay poor after the TSA benefit is received, while 12 percent of the beneficiaries move from the second to the third quintile of the per capita consumption distribution.The size of the TSA benefit should be periodicallyadjusted to take into account the risingcost of living, includingfood prices. Consider the consolidation of other fragmented social protectionprogramsinto the TSA benefit. There currently still exist a number of social protection programs that try to target specific groups of population, but effectively are aimed at reachingthe poor.These programs include the Internally Displaced Person (IDP) allowance and gas and electricity vouchers. The performance of these programs in terms of the poverty-reducing impact is difficult to assess due to the lack of data. However, giventhe evidence that the TSA targetingof the poor is good, there might be no need for havingthose other programs "on the side." The budget resources used for those programs could be alternatively redirectedto fund the extension of . TSA coverage. Addressing vulnerabilities arising from both the August 2008 conflict and the current global downturn. It is important to have the policy instruments in place to mitigate the economic and social impact of the twin crises, and the TSA could be an efficient tool to shield the poor. However, in times of the economic crisis the needs for expanding social protection would need to be balanced against the reality of constrained public fiscal resources. Job creation through targeted public investments or public works, especially in rural areas, could be a way to go. 50. Education. Georgia has been on a path of introducing far-reaching reforms in the education sector. However, there are still many outstanding issues that need to be addressed. A number of educationpolicy interventionscan be suggestedin relation tofinancing, access, andquality. . I n relation tofinancing, these include: (a) revising the per capita financing formula so that small schools in rural areas (generally serving the poor) receive enough funding for operation; and (b) promotingincentives for private sector provision of educationservices, especiallyat the preschooland technicaland vocational education (TVE) levels. I n relation to access, these include: (a) addressing lack of access to preschool education in rural areas; (b) sustaining investments in heating provision to decrease student absenteeism; (c) encouraging school boards to better monitor teacher absenteeism, especially in poor regions; and (d) considering targeted programs through conditional 13 cash transfers (CCTs) in order to contain large drop-out rates among the poor after secondary education. I n relation to quality, these include: (a) introducing a private school accreditation system, in order to reach a better balance between quantity and quality of private schools; (b) subsidizing supplementary education services to prepare students for university entrance exams in rural areas and in regions with a high incidence of poverty and/or a high concentration of minority children; (c) continuing participation in international testing schemes (like PIFUS) and engaging in more frequent national assessments to monitor progress in test scores; and (d) sustaining investments in physical infrastructure for education to assure a better learning environment that promotes attendance, student health, and quality ofteaching. 5 1. Health. Georgia has been implementingcomprehensive reforms in the health sector since 2006. The study findings suggest several implicationsfor current policy initiatives. The "all or nothing" character of both MAP eligibility (below or above 70,000) and benefit package definition (health services or drugs) may warrant reconsideration as programs are revised and fine-tuned. Intermediate benefit packages with moderate co- payments would be one possible approach to balance budget considerations with achieving more widespread access to health care and financial protectionamongvulnerablehouseholds. The vulnerability of non-MAP beneficiaries,if and when the Universal Benefit Package (UBP) is eliminated, should be an area of policy focus. Currently, there are many poor households that are not covered by MAP. High out-of-pocket expenses among non-MAP beneficiaries are correlated with elderly households and those with non-working household heads, and these groups will havedifficulty obtainingcoverage in private insurance markets. The shift toward greater private sector involvementin the health sector is more likely to achieve its full potential if supported by strengthened oversight capacities. Reform design has been mindful of the need for safeguards to protect access and equity, but careful implementationwill be required to ensurethese outcomes are achieved. 52. Poverty Monitoring. It is very important that the Government has reliable poverty monitoring data. The capacity to monitor poverty is especially important given the high pace of reforms undertaken in Georgia. As has often been the case in other transitional countries, the needs of the Department of Statistics appear to be somewhat neglected. It is clear that the Department of Statistics cannot function properly with its current technical capacity level. The Government needs to take measuresto restore this capacity, which would include proper levels of funding and staffing. The independenceofthe Department from the Ministry of Economic Development may also bejustified in this regard. 53. To sum up, the Report has six key messages: (1) The living standards in Georgiahave improved in many dimensions since 2003; (2) Poverty in Georgiacontinues to be deeply entrenched in rural areas, accountingfor 60 percent of the poor; (3) The performance ofthe labor markets has so far not contributed much to poverty reduction; (4) Social assistance became an increasingly important lifeline for Georgia's poor - the Targeted Social Assistance (TSA) program introduced at the end of 2006 appears to be an important source of income for the poor; (5) The double shocks of the August 2008 conflict and the global financial crisis risk underminingthe poverty reductioneffort; (6) The poverty reductionstrategy of the Government of Georgiashould focus on extendingthe coverage of the TSA to reachmore of the poor and promoting investments in infrastructure and creatingopportunities for off-farm employment in rural areas. 14 CHAPTER 1:MACROECONOMIC DEVELOPMENTS IN GEORGIA A. Introduction3 A.I. Economic DevelopmentsbetweenIndependence and the Rose Revolution 1. Georgia's economic contraction following the transition after the breakup of the Soviet Union was among the most severe in the Europe and Central Asia (ECA) region and within the Commonwealth of IndependentStates (CIS). The contractionwas not followed by strongand sustained growth until 2003. As a result, the level of output in Georgia in 2007 still stands at less than 70 percent of its 1990 levelof output, while the majority of CIS countriesfully recovered by 2007 (Figure 1.1). 2. The initial stage of the transition from a centrally planned to a market economy that started in the early 1990s implied changes in the basic institutions of the economy and loss of output. It decentralized economic decision-makingprocesses, liberalized prices and wages, and exposed enterprises to competition. These major changes in the rules of the game, along with disintegration of traditional economic links and other major obstacles, including civil war and ethnic conflicts, resulted in a substantialfall in output in the early phases of the transition until 1994. 3. In the subsequent phase of recovery from 1995 to 2003, Georgia did not achieve strong and sustained growth. Furthermore,growth was often dependent on either single large investment projects, such as the Supsa and Baku-Ceyhan pipeline construction, or followed the patterns of the volatile agricultural sector. As a result, many people movedto low-productivityjobs in the agriculture, rural, and informal sectors. Georgia had an increasingly weak governance framework, corruption was rampant, selective reform efforts were not being sustained, and the country was drifting rapidly toward "failed state" status. These conditions created a fertile environment for the peaceful Rose Revolution in late 2003. z o ' @ ' E ' B I' c ' B I' 1 'I' 1 'I'I 'I' R 'I'1 ' 1' 1'I' c ' Source: WB Regional Tables, March 2008 for ECA and CIS; Georgia LDB for Georgia. 3 This chapter was preparedby Mariam Dolidze (Economist, ECSPE), Faruk Khan (Country Economist, ECSHD), RosalindaQuintanilla(LeadEconomist,ECSPE), andAfsaneh Sedghi (Economist,ECSPE). 15 A.2. Economic Developmentsafter the Rose Revolution 4. The new government's aggressive institutional and structural reform agenda since the Rose Revolution in 2003 benefited economic development. It has targeted widespread corruption, and created economic incentives for a large segment of the economy to go formal via substantially reduced tax rates on one hand and enhanced tax administrationon the other. In a very short time the government managed to establish fiscal discipline and tax compliance and offered businessesa liberal and functional legislative framework, improved infrastructure, good governance, security, and economic openness. These improvements provided the foreign investment community with confidence in the new administration's commitmentto build a competitiveenvironmentfor the private sector. 5. The progress on reforms since the Rose Revolution is reflected in a number of assessments and ratings of Georgia during the last five years. The "Doing Business" 2007 Report named Georgia the top reformer and rated it among the top 20 performers a year later. Georgia received a positive sovereign ratingby S&P and Fitch in 2005 for the first time, and maintaineda BB rating despite external political and economic sanctions and tensions with a major economic partner and neighbor. Georgiaalso has a continuously improvingcorruption perceptionindex (CPI) and a European Bank for Reconstruction and Development (EBRD) transition index, and other positive survey results. Those signals have been interpretedby foreign investors as an opportunity in the region to invest resources.Annual foreign direct investment (FDI) flows have increased fivefold in nominaldollar terms since 2003 and have become one ofthe strongest drivers ofthe recent high growthtrend (Table 1.1). Table 1.1: Georgia: Selected EconomicIndicators 2001 2002 2003 2004 2005 2006 2007 GNI Per Capita(US$,Atlas method) 680 730 860 1,060 1,330 1,560 1,990 UnemploymentRate, Average 11.1 12.6 11.5 12.6 13.8 13.9 13.6 RealGDP Growth(YOchange) 4.8 5.5 11.1 5.9 9.6 9.4 12.4 CPI (year-on-year,% change) 3.4 5.4 7.0 7.5 6.2 8.8 11.0 RevenuesandGrants (% of GDP) 16.1 15.9 16.5 22.7 24.7 25.8 28.3 ExpenditureandNet Lending(% ofGDP) 18.5 17.3 18.3 19.7 25.1 27.9 33.8 Overall FiscalBalance(% of GDP)a -2.1 -2.1 -1.5 -0.3 -2.2 -2.4 -4.3 Current Account Balance (% ofGDP) -6.6 -6.3 -9.6 -6.9 -11.1 -15.1 19.7 Gross DomesticInvestments(YOof GDP) 21.9 22.1 24.4 28.3 26.8 27.4 28.3 FDIInflow(% of GDP) 2.5 3.6 8.4 9.8 7.1 13.7 17.0 ExternalDebt (%df GDP) 49.6 51.6 46.4 34.5 26.7 23.1 16.8 Note: a. The budget deficit is presented on a cash basis, including accumulationirepaymentof arrears, while expenditures are on a commitment basis. Source: Departmentfor Statisticsof the Ministry of Economic Development; World Bank staff estimates. B. Developmentsof Key Economic Indicators B.I. Evolution of GDP and Sourcesof Growth 6. GDP growth averaged 9.7 percentduring 2003-2007, but was led by selected sectors. Strong economic growth during 2003-2007 has been led by several sectors, includingtrade services (77 percent growth), construction (182 percent growth), financial intermediation (220 percent growth), and manufacturing (90 percent growth). Agriculture has experienced volatile growth, as a result of its sensitivity to weather conditions and Russian trade sanctions. The growth pattern across sectors is discussed in a greater detail below. 16 7. Agriculture has been an important but volatile sector of the Georgian economy. Real agricultural output has been stagnant since 1997. After independence in 1991, the agricultural sector underwenta severe crisis, partly due to the civil war, which resulted in the destruction of the productive ability of collective and state farms. A process of land privatization was started in 1992, with the state agricultural land beingdistributedto private households. Over the years, the agriculture and rural sectors have been hit by a number of adverse factors, including frequent floods and droughts, outages of watedenergy supply, and a deteriorating road network. Furthermore, plot sizes shrank, leading to decliningproductivity. As a result, the sector has witnessed little growth, including in recent years, when the economy in general has rebounded strongly. The share of agriculture in GDP fell from around 50 percent in 1990 to around 17 percent in real terms in 2007. Besides unfavorable initial conditions and institutional constraints, the sector is highly dependent on weather conditions and, recently, on Russian trade sanctions. The loss of this major market for wine, fruits (apple), citrus, and vegetables, which represent almost 35 percent of total value added in agriculture, contributedto price dampening on banned products and negative growth of real income. Real agricultural output increased by a mere 3 percent during 1997-2002, and declinedby 1 percent during 2003-2007 (Figure 1.2). Figure 1.2: Georgia: Average Real Growth in Selected Sectors 1 1997-2002 (aw growth 4 8%) I2003-2007(a- growth 9 7%) 40 30 1 I 20 10 0 I Source: Department for Statistics ofthe Ministry of Economy of Georgia and World Bank staff calculations 8. The construction sector registered impressive growth. The growth pattern of the construction sector has been largely determined by strategic energy projects in the country, particularly during the period of low performance of the economy-the Supsa pipeline led the growth of 52 and 35 percent during 1997 and 1998, respectively.Then, after two years of stagnation, the sector recovered again, as the Baku-Ceyhanoil pipeline project started in late 2001, and has been leading the sector together with the Shah-Denis gas pipeline project, representing over 50 and 30 percent of output, respectively, over the last several years until 2006. Since the Rose Revolution, the growth has also been shared by active private sector initiatives in the construction of housing and business facilities, complemented by public participation in the building and reconstruction of state-owned health, education, and other facilities, reducing dependence of the sector on single projects and introducing a higher degree of sustainability. The 14 percent growth has been fully attributed to the construction of buildings in 2007, while the pipelinecapitalworks representedonly 12percent oftotal production. 17 9. Financial services have been the most successful area of the Georgian economy in terms of growth rates. The performance of the sector had been dramatically deteriorating prior to the Rose Revolution.However, despite the negative impact of the memory of inflation of the 199Os, the financial and particularly the bankingsector hadbeen experiencingextraordinarilystronggrowththroughout 1996- 2007, increasingthe share of financial intermediation in GDP as much as sevenfold by 2007. In recent years, the financial sector expandedparticularly rapidly, with the credit-to-GDPratio increasingfrom 8.3 percent in 2002 to 26 percent, and total assets of the banking system reaching 42.6 percent of GDP in 2007-comparable to European Union accession countries at the point when their accession was considered. 10. The manufacturing sector experienced a sharp contractionin output during the initial phase of the transition. However, aggressive privatization by state authorities created a favorable environment for the restoration of industrial production. Under the Soviet central planning system, Georgiawas allocatedcertain sectors of industrialproduction.Therefore, a large share of industrialgoods had been produced by state-owned enterprises, such as energy productionand distribution, water supply, oil refinery, ferrous metallurgical manufacturing, mining of manganese ore deposits, mechanical engineering, chemicals, and a great part of the food enterprises. With independence, these industrieswere not financially viable, and were too much of a burden on the government. This made the industrial sector vulnerable and productivity low. After the Rose Revolution, the new government started a competitive privatization process, successfully attracting domestic and foreign investors. Those investments have already contributedto about 15 percent growth of industry since 2003, and many are still to recover under privateownership.The performance of the manufacturingindustry,which currently represents 70 percent of industrial production, has been led by food, beverage, and tobacco manufacturing with quite impressivegrowth of over 70 percent since 2003 though all other subsectors were performing also quite well. 11. High domestic demand has largely contributed to the growth of domestic trade services. For the last five years, the share of wholesale transactions has been permanently growing and reached 60 percent in 2007, while retailtrade made a decreasing contributionto trade services production. Because a majority of locally consumed products are imported, both wholesale and retail trades are largely dependent on external trade transactions that have intensified through an extremely liberalized tariff regime, with a weighted average rate of below 1 percent.The real growth of national absorption4by 78 percent has had a positive impact on trade transactions as well. The contribution of trade and other business activities to GDP increased to 14 percent of GDP. The sector has contributed to overall economic growth. 12. The value added of social services increased since 2003. Backed by growing state budget resources, social services such as health, education, cultural, and community services reversed their decliningtrend and started to improveperformance, with average growth rates of 7 to 8 percent per year during 2003-2007. Public administrationand defense activitieshave had somewhat low performance, but lately have shown relatively strong improvements, driven by increased government spending on defense. Other services sectors with a relatively low share of GDP, like communication, real estate and rental, and restaurants and hotels, have in the last three to five years benefitedfrom increasingFDI resources. As a result, the growth inthose sectors has been above the nationalaverage growth rate. 13. production and to continuously expand imports volume^.^ Therefore, demand has been driven largely External capital and financial-inflow-generated incomes allowed the country to absorb local 4Definedas gross nationalincomeminusbalance oftrade; all components deflatedby GDP deflator (WorldBank staff calculations). This couldbe a source of macroeconomicvulnerability inthe longerterm. 18 by privateconsumption, given its dominant share in GDP, though growing at 8 percent per year, which is slower than GDP-the average national growth rate.National absorption increased by 78 percent during 2003-2007, compared to 58 percent of GDP growth itself for the same period, mostly driven by imports on one hand and by rapidly expandingcredit to the private sector on the other. Governmentconsumption has beengrowingby an average of over 30 percent per year in real terms, which has morethan tripled the real value of current state spendingS6Budgetary capital spending also reflects a significant government contributionto the improved infrastructure, including uninterruptedenergy/gas supply, high standard for roads, better water supply, and rehabilitation and construction of education and health facilities. At the same time, there has been a limited response of domestic private investment,which has been reflectedin a volatile and downwardtrend of gross national savings. 14. Strong economic growth and an appreciation of the national currency vis-a-vis Special Drawing Rights (SDR) and the US dollar resulted in a substantial increase of dollar-denominated pei capita income. The gross national income (GNI) per capita increased from US$680 in 2001 to US$1,990 in 2007 (Table 1.1). The estimated negative populationgrowth' also contributedto the increase of GNI per capita.This, together with positive creditworthinessassessment, resulted in the conversionof Georgia into a blendcountry eligible for InternationalBank for Reconstructionand Development(IBRD) resources. B.2. Fiscal Sector: TaxAccumulation and Management of Public Resources 15. The increase in tax revenues and proper management of government expenditures have become possible as a result of tax reform and deregulation (which reduced the informal economy), and particularly of measures to curb corruption.Tax revenue collection has increased from 14 percent of GDP in 2003 to almost 22 percent in 2006, and is estimated to have increased further, to over 25 percent in 2007. This major achievement allowed the government to start addressing critical social and infrastructureneeds. Social spending has increased by about 5 percentage points of GDP to date and now represents more than one-thirdof the budget. Socialassistance policy has been targeted to assist the most vulnerable in society.* 16. Current expenditure has been directed by the government to clear previously accumulated wage and pension arrears, to increasethe value of social transfers, and to raise salaries for public employees in real terms. Pensions were increased from 14 GEL in 2003 to 70 GEL in April 2008, and the Targeted Social Assistance (TSA) program was introduced at the end of 2006. Social protection currently accounts for almost 20 percent of state budget spending (Figure 1.3). Capital expenditure increased from almost zero to 7 percent of GDP by 2007, and translated into strong growth performance facilitatedby better infrastructure.Still, the government maintainedan acceptable level of fiscal deficit, in the range of 2 to 3 percent of GDP until 2006; however, a spike in expenditures in 2007 resulted in a higher deficit of4.3 percent of GDP. '*ThePoverty The constant series for GDP consumption is obtainedusingGDP deflator for all components. UnitedNations DevelopmentProgram(UNDP) population data. ReductionSupportOperation04 (PRS04)programdocument(p.20). 19 Figure 1.3: Georgia: Functional Compositionof the State Budget Expenditures (plan), 2008 9 . 5 14.4 0 General public services D efense 0 Order & Security E c o n o m i c A f f a i r s H e a l t h 0 Education Social p r o t e c t i o n 0 Other Source: Ministry o f Financeof Georgia. 17. A reduced financing gap over the last five years, and strong depreciation of the US dollar have resulted in improving external debt burden indicators.External debt declined from 46.4 percent at the end o f 2003 to 17 percent o f GDP by the end o f 2007. Therefore, the government holds some leverage in terms o f fiscal sustainability. That was the reason behind the issuance o f commercial debt of around 4 percent o f GDP (US$500 million worth o f Eurobonds), which is directed to the extra-budgetary reserve funds o f Georgia-the Future Generation Fund and the Stable Development Fund. It is important that the government manages those resources in accordance with budgetary needs under tighter fiscal policy. B.3. Monetary Policy and Inflation 18. Large capital inflowsand the expansion of private credits, coupledwith rising global energy and food prices, have challenged the ability of monetary authorities to maintain stable inflation. Increased confidence on the part o f businesses and consumers resulted in broad measures o f money supply to increase by an average o f 50 percent over the last year, and the credit-to-GDP ratio surged from 8.3 percent in 2002 to 26 percent in 2007. On the other side, higher global food and energy prices added pressure on the local price level, and year-on-year inflation has reached 11 percent by end December 2007 (Table 1.4). However, core inflation has been still maintained in single digits all throughout 2004- 2007. To strengthen the National Bank's authority and monetary function, the Parliament recently adopted a financial reform package, which supports the efforts to change institutional arrangements for supervision o f financial sector participants, commercial banks, insurance companies, capital markets, and microfinance institutions. The package also introduces transparent extra-budgetary funds as a tool for restraining budgetary pressures on money markets. As a medium-term measure, the authorities intend to tighten fiscal and monetary policies and move to inflation targeting. 20 Figure 1.4: Georgia: CPI Inflation (period average), 1998-2007 :: 20 CPI Inflation (avg, percent) 5 Source: National Bank of Georgia. B.4. External Sector Developments: Trade and Current Account 19. The increased openness' and solvency of the Georgian markets led to a steeper upward trend of import flows for the last five years. Both the composition and quality of imported products have improvedas well, which resultedin the higher cost of imports, exclusiveof the world price spikes in food and energy. As a result, importshave grown by an average of 20 percent in realterms and almost 50 percent in nominal US dollar terms per year since 2003. At the same time, exports have experienced a negative impact from the Russian ban imposed on export products since March 2006. Despite the trade restrictions,however, exports growth rates were restored to double-digitgrowth rates by mid-2007due to exports diversification to other markets. Although the economy showed a remarkable ability to expand export access to other markets, the restrictionstook their toll on rural exporters in 2006 and 2007.'' This is not surprisinggiven that exports to markets other than Russia and Turkmenistan,while startingfrom a small base, have exhibited growth in excess of 30 percent since the second quarter of 2006. Overall, imports had been rising much faster than exports, leading to the rapidly soaring current account deficit (Figure 1S), Figure 1.5: Georgia: Exports and Imports of Goods, Million US$ 6000 - 5000 -. 4000 -~ 3000 -- 2000 -. 1000 -~ ___.-- I--- --Y --- * - A -- 0 , -exports - .c imorts Source: Department for Statisticsof the Ministry o f Economic Development of Georgia. Averageweighted import tariffcurrently is below 1 percentin Georgia. loThis is discussedin greaterdetail inthe chapter on rural poverty. 21 20. Trends of merchandise trade have been consistent with the growing external trade and current account deficits, while other components of the external current account have been generally stable. The deficit has been financed by large inflows of FDI, which reached 13.7 percent of GDP in 2006. Notwithstanding the political uncertainties in the last quarter of 2007 and the expected negative impact on capital flows, FDIremainedhigh at 17 percent of GDP (Figure 1.6), even larger than anticipated,and have fully financedthe external currentaccount deficit, in additionto the accumulationof internationalreserves. Figure 1.6: Georgia: External Current Account Deficit and Net FDI (percent of GDP) 1 s E3 CAD 63 FDI 1 0 S 0 Source: National Bank of Georgia. B.5. External and Internal Shocks and Policy Responses 21. Over the last three years, Georgia has faced several negative external and internal shocks. Those are discussedin detail below. 22. The Russian ban on major export commodities particularly affected the wine industry. In the absence of its largest traditional market for its most profitable agricultural business, Georgian winemakers, with government support, took major steps to diversify the trade directions, improve production quality, and reduce dependence on the Russian market. Still, the immediate impact was devastating for export revenues throughout 2006, and the grape cultivating industry suffered as sales prices dropped significantly. 23. Sharp energy price increases, with the price of imported gas rising from US$65 in 2005 to US$l10 in 2006 to US$235 per 1,000 cubic meters from January 2007. The government put in place an electricity lifeline tariff system to protect vulnerable groups and a gas subsidy for the two last winters. The negativeimpact of these shocks on the extreme poor has been reduced by one-time financial support to all pensioners and other vulnerable groups in the last quarter of 2007 to ease the impact of the higher electricity and gas prices. 24. Sharp international food price increases were partially offset by strong appreciation of the Georgian Lari (GEL) vis-a-vistrading partner currencies, but still have negatively affected inflation rates in Georgia, reaching its highest points since the Russian financial crisis of 1998. Higher prices of importedproducts and strong growth of domestic demand also pushed up prices of domestically produced goods." The poorest, who spend a large share of their income on food, have been particularly under I'GlobalEconomicProspects2008: InflationandCommodityMarkets. 22 pressure. The Governmentof Georgia has been relying on a targeted safety net programas an instrument to provide assistance to the poorest people and to mitigatethe negativeimpact of price shocks, rather than on using direct interventionsand controls over food prices, which may send a negative message to the privatesector. 25. The results of these shocks were mitigated by effectively implementingeconomic reforms and safeguarding macroeconomic stability. C. Key Institutional and Policy Reforms 26. Corruption was identified by the new government as a major obstacle to development and growth. A number of institutional reforms have been carried out since 2003 to curb corruption. In order to optimize the public-to-private-sector ratio, a process of downsizing of government apparatus has been initiated.The number of executiveministrieswas reduced from 18 to 13, and many departmentsand units were unified and optimized. This had undesirable and painful initial effects on employment as the number of public servants has been almost halved and replaced.The reorganization, however, allowed substantialimprovementof the functioning of government institutions, and realwages in the public sector rose above the subsistenceminimum, which was notthe case previously. 27. The very first and remarkable reform, which almost eliminated corruption deals in the system, was traffic police restructuring. Police reform was identified by the Georgianpopulationas one of the most successful projects of the new government. It facilitated not only the lives of local citizens, but also has had an extremely positive effect on turnover of transit traffic, making Georgia the safest, most secure, and cheapestroute for traders and travelers. 28. One of the most successful economic reforms with apparent positive results has been tax reform. It aimed at lifting the tax burden on business and at the same time, reducing incentives for making illegal deals with tax officers and introducing a culture of tax compliance. The effect was more encouraging than expected. The government had to make a number of supplements (in contrast to previous years' tradition of sequestrations) to the budget during the fiscal year to reflect continuously rising tax proceeds. During 2005-2007, the number of taxes was cut from 23 to 7, a progressive income tax of 22 percent was replaced by a flat 12 percent tax, and the value-added tax was reduced by 2 percentage points to 18 percent. The social tax underwent major changes as it was reduced from 33 percent to 20 percent initially, and later was incorporated into an income tax that totaled 25 percent, substantially reducing labor costs. Recently, the profit tax rate was cut from 20 percent to 15 percent. Import tariff rates experienced a serious transformationin support of trade facilitation, particularly with regardto the import of investmentgoods and new technologies. 29. Together with improved collaboration between tax authorities and taxpayers, those legislative changes had a dual impact. On the one hand, they have improved the business environment by putting all market players under equal and fair competitive conditions, and on the other hand, they fueled substantial resources to the government budget, enhancing its redistribution function, which translated into threefold increased social-related spending, and into large investment projects in roads, energy, gas and water supply, educationand healthfacilities, and other important public goods. 30. Heavy and ineffective regulation of the labor market has also been considered by the government as a constraint to doing business,and an incentiveto have nontransparent employment relations.A new labor code was developed and adopted in 2006 with some quite liberal elements and provisions-the government took a financial burden of maternity leave payment of private sector employees, allowed written and verbal contracts, and simplified hiring-firing regulations. The new code 23 was evaluated by the experts as one of the most liberal labor laws in the world, in contrast to the Soviet system code Georgiahaduntil 2006. 31. A number of important institutional and policy reforms have been identified by the government that have been endorsed by the World Bank, and support provided through the four- staged Poverty Reduction Support Operation (PRSO). The policy agenda under this lendingprogram covered four importantareas.I2 (i) Strengtheningpublic sector accountability, efficiency, and transparency. This pillar has focused on continued improvements in the public sector that cut across public expenditure management, fiduciary activities, and public administration. Specific policy actions under this component have aimed at: strengthening public finances, budgeting, and expenditure management; increasing transparency and efficiency in the public procurement and financial management systems; strengthening public administrationand intergovernmentalfiscal relations; and enhancingthe transparency and accountability of the public sector. (ii) Improvingelectricityandgassectorservices. Electricityandgas serviceshaveimproved significantly.The government has updated and is implementinga medium-termstrategy plan that outlines the reforms. Specifically, the reforms have sought to address electricity sector debt, improve payment collections, strengthen the monitoringand reportingof electricity services, and diversify supply sources in the gas sector. The government has completed the privatization of assets in power generation and distribution. The reforms in the electricity and gas sectors have surpassed what was originally envisioned in the program, which aimed at reachinga collection rate of 65 percent and gradually improving service and reducing blackouts. Today, nearly all paying consumers have electricity service 24 hours a day, 7 days a week, collection rates are above 90 percent, and blackouts are infrequent. Prudent investments over the last several years and significant improvementinthe managementof electricity units were centralto the turnaround of the sector. In addition, the government implemented large tariff increases in 2006, which was a critical step on the road to a financially sustainable energy sector, while well-targeted electricity and gas subsidies helped protectthe most vulnerable. (iii) Improving the environmentfor private sector development. This pillar recognizes that the development of the private sector is critical to sustaining high growth and creating employment opportunities for the poor. The government's policy measures under this component have focused on reducing the costs to business activity imposed by the public sector and improving the investment climate. It has supported the government reform program in further simplifying tax legislation, strengthening institutional capacity, and streamlining the regulatory framework and reform measures to improve standardization and quality assurance to enhance internationalcompetitiveness. (iv) Improving socialprotection, education, and health care services. This pillar has focused on deepening the ongoing reforms to further improve the access to, and accountability and affordability of, services in the social sectors and social protection for the vulnerable. The government has introduced a fiscally sustainable targeted social assistance program aimed at improvingthe welfare ofthe extreme poor, the coverageof which expanded rapidly by the second year of implementation. The reform program has also included improving the efficiency and efficacy of resources in education by: (a) strengthening the capitation financing model; (b) developing the systematic testing and monitoringof results and strengthening quality-enhancing dimensions, including educational materials and curricula for students, and mechanisms for '* PRSO4 program document. 24 accreditation of schools and certification of teachers; and (c) participating in international assessment programs such as the Trends in International Mathematics and Science Study (TIMSS) and the Progress in InternationalReading Literacy Study (PIRLS). In the health sector, the program has supported government efforts to adopt a revised basic benefits package and a primary healthcare planaimed at improvingequity of access to healthservices. 32. Other reforms affecting institutional setup of the country cover: (a) regulation of entrepreneurship and monopolies through the newly established Agency for Free Trade and Competitiveness, which replaced two state institutions-antimonopoly service and price inspection; (b) business registration procedures have been simplified through the introduction of one-stop-shop registration windows and by designating one responsible agency for registration, which is the tax department; (c) the Parliament recentlyadopted a financial sector reformpackagethat supports the efforts to change institutional arrangements for the supervision of financial sector participants, commercial banks, insurance companies, capital market, and microfinance institutions; the package also introduces transparent extra-budgetary funds as a tool for restrainingbudgetary pressures on the money market; and (d) legislationhas been developed and adopted on the creation of a free trade and industrial zone in Poti, which is expectedto attract FDIandto create a favorableinvestment climate for the whole country. 33. An overall assessment of reforms in Georgia has been undertaken by a number of international institutions, and Georgia's progress on a number of fronts has been widely acknowledged. Those assessments include: WB/IFC-Doing Business reports 2006-2008, EBRD- transit index 1990-2007, American Chamber of Commerce-Regional Investment Climate, Transport & Trade Facilitation survey 2008, Corruption Perception Index by Transparency International 2003-2007, sovereign ratingby Standard & Poor's (S&P) 2005-2007, and Individual Banks' ratingby Fitch for three Georgiancommercial banks, 2007. All have reportedremarkable progress made by the country in terms of doing business, reducingcorruption, improving state-performed services, and improvingthe country's creditworthiness(Figure 1.7). Figure 1.7: Georgia: EBRD Transition Index in Selected Areas 5.0 -2003 P2005 2007 4.0 3 .O 2.0 1.o 0.0 Source: EBRD transition report 2003, 2005, 2007. 25 D. Growth, Productivity, and Employment 34. The rapid economic and political changes in Georgia at the beginningof the transition gave rise to specific problems relating to human capital. Huge structuralshifts in the demand for goods and services resulted in structural shocks in the demand for labor. The wage structure was changed considerably; some skills and behavioral patterns began to be valued much lower, whereas other skills and patterns were assigned a high value. When the Soviet Union collapsed, Georgia, like many other CIS countries, suddenly had thousands of unemployed engineers and scientists, but no industry that could employ them. The market responded by large numbers of skilled people withdrawing from the market, and others resortingto the informal sector or migratingto other countries. 35. As a result of the sharp contraction of output following transition and the subsequent hesitant recovery through 2003, large numbers of people became attached to low-productivity activities in agricultural and rural areas. Consequently, the Georgian economy ended up with a populationof which more than 50 percent were self-employedin agricultural production.A large portion of the hired labor force then accumulated inthe public sector. As a result,the labor force has concentrated in the agricultural and public sectors, and physically has not been able to follow the fast-developingand new requirements of the labor market, creating a skills mismatch in the market. Such labor distribution itself implied extremely low labor productivity and some degree of mismatch in capital and labor compatibility in the productioncycle due to the high speed of reforms and a slower capacity of the labor force to adjust. 36. Georgia used to have the highest share of informal economy in the ECA region and the Former Soviet Republics until the Rose Revolution, because the cost of going formal was much higher than the benefits. A heavy tax burden on labor, inadequate labor codehegulation, and non- functioningtrade unions have conditionedcreation of a large share of informaljobs. As expected, those jobs have been of low productivity, low-paid,and without social coverage. Such an environmentcreated a layer of working populationthat remained under the poverty line.I3 37. Poor initial conditions after the collapse of industrial production, and lack of innovations and new technologies at the beginning of transition for long period of time, have been the main reasons why productivity has been far below the potential level all across the economy. The new government carried out reforms, which had an immediate positive impact on formalizing employment relationsthrough a new labor code and a substantial reductionof taxes on wages and salaries. Increasing openness of trade and innovations through a new customs code and customs reform suggested rapidly improving productivity in the industrial and service sectors. However, the opposite process has been observed in the agricultural sector, which accumulated an excessive labor force while at the same time showing a stagnant growth pattern.Since 1998, growth in productivity has been observed in industry and services, but not in agriculture(Figure 1.8). 13The issue ofthe workingpoor in Georgiais discussed in greater detail inthe chapter on labor markets and linkagesto poverty. 26 Figure 1.8: Georgia: RealValue Added (VA) per Employee,by Sector (as proxy measureof labor productivity) 25 ~ i II 22.2 I ~ !I 20 ~ I I 5 ' I I i ~ 04- I 1998 2003 2004 2005 2006 2007 Agriculture Services Industry +Average VA per employee Note: Value-added numbersare based on national accounts Source; Departmentfor Statistics o f Georgia. 38. Growth in productivity has resulted in rising real wages almost across all sectors of the economy.14 The service sector cost o f labor has grown particularly rapidly. The largest increase in earnings was registered in the financial intermediation and information technology (IT) sectors. A very high rate o f wage increase has been also observed in public administration, as civil service reform resulted in lower employment and higher compensation. However, real wages o f those employed in agriculture increased very little since 2003. Along with growing productivity, there is still a lot o f potential for the increase o f labor participation and for further enhancing economic growth and its sustainability . E. Macroeconomic Challenges and Growth Sustainability 39. Georgia faces significant but manageable macroeconomic risks and vulnerabilities. On the domestic front, these include: 40. Budgetary pressures and other economic impacts deriving from external developments. As a result o f strong tax performance, the fiscal deficit on a cash basis was in the range o f 2 to 3 percent o f GDP during 2005-2006. However, the budget resources available for financing country needs are still limited despite substantially improved revenue collection. Recent energy and food price shocks and Russian sanctions created the necessity for additional support to socially vulnerable layers o f the population through increased social allowances and pensions. The support also included one-time energy and gas subsidies to households and to the private sector through emergency credits to gas distribution companies, low-interest business credits, and support to the wine industry and agricultural sector. State employment programs were also introduced to facilitate retraining o f the unemployed labor force. Those budgetary necessities have contributed to a higher fiscal deficit o f 4.3 percent o f GDP for 2007, largely 14A detailed analysisof the labor market in Georgia is provided inthe chapter on labor markets and linkages to poverty. 27 financed by.external privatization proceeds and other external concessional borrowing. To place control over expansionary fiscal policy, the authoritiesintendto moveto a balancedbudget in the mediumterm. 41. Rapid growth of credit to the private sector. Georgia's private sector credit is rising at its fastest annual rates in the last four years. Long-maturity foreign exchange consumer credits increased nine fold and business credits sevenfold from end-2003 to end-2007. The share of consumer credit increased to about one-third of the total credit amount. Such a sharp growth of credit is driven by increasing confidence on the part of foreign investors,who are willing to provide resourcesto Georgiancommercial banks due to relatively high interest rates in the local money market. The banking sector in 2007 borrowed over US$700 million from external sources as long-term loans. Such outstanding and unusual growth could be partially attributedto poor initial conditions prior to the Rose Revolution, when banking credits were expensive and not easily accessible in the absenceof sufficient collateral. On one hand, these developments indicate the income growth of the population and improved solvency, but on the other hand, such a rapid growth of money supply creates access liquidity on the market, putting demand-driven pressure on inflation in additionto the FDI and other external capital inflows. Continued sharp increases in energy prices and further increases in food prices represented supply-sidepressure on local prices. To overcome the pressure, a tight monetary and fiscal policy is needed, and the government is committed to movingto a programof targetinginflation. E.1. External risks and vulnerabilities include: 42. Disruptions in some traditional trade patterns and slow progress in trade diversification. Russian sanctions have changed the commodity structure of Georgian export substantially. Georgian producers reorientedtheir resourcesto fit requirements of European and other large markets, like Turkey. However, this process requires considerable effort, quality improvement, infrastructuredevelopment, and widespread product marketing. After a sharp drop in export revenue growth in 2006, there has been a moderaterecoverystartingin mid-2007, reflectingsome trade diversificationtrends. In the longer run, the normalizationof economic relationswith the closest and biggesttradingpartners is crucial for sustainable growth, as is aggressively pursuingthe diversificationprocess. 43. Shocks to the sustainability of large capital inflows tofinance the growing external current account deficit in light of the vulnerabilities in the international financial markets and the slowing growth of the world economy. Substantial capital inflows have fueled domestic demand and generated a large and growing external current account deficit, and pressures for inflation and local currency appreciation. The authorities have responded by intervening heavily in the foreign exchange market to dampen appreciation of the GEL, and they have sterilized part of the inflows through domestic open market operations. While the policy response thus far has been sound, the large and widening external current account deficit representsa risk of the need for adjustment to any shortfall in capital inflows in the medium term. Factors that could generate such a shortfall include a fall in oil prices (since much of the inflows are from energy-richcountries), negative developments in real estate markets, and a deterioration in political or regional stability. The policy response to any reversal in capital inflows would likely require fiscal adjustment, which would be particularly difficult to achieve during an election year and with the continuedsignificantsocialand infrastructureexpenditureneeds.These risks are manageableand they are mitigated by the government's strong commitment to safeguarding growth and stability and its demonstratedtrack recordof pursuinga sound mix ofmacroeconomic policies. 44. Other risks include possible deterioration in the regional political security environment, and reform fatigue among stakeholders in response to the broad range and rapid pace of substantial changes being undertaken by the government. While the former set of risks are largely outside the mandate of the Bank and its sphere of influence, factors mitigating the reform-relatedrisks include the determinationof the government, with support from the Bank, to ensure that the reforms provide tangible 28 benefits to the population, and reinforced efforts by the President and the government to explain the rationale of reform initiatives and the anticipated improvements in governance and living standards that these will bring. Continued support from the international community for broad-based consultation and consensus building is critical. Continuingtensions in the regions of Abkhazia and South Ossetia raise the risk of security issues.It is hoped that international efforts to mediate these conflicts will be successful in mitigatingthis risk.I5 F. Conclusionsand Policy Recommendations 45. An extensive reform agenda launched after the Rose Revolution in 2003 resulted in reductionof the most severe forms of corruption,improvementsin tax collection,increased salaries and pensions, and prudent macroeconomic performance. Economic growth has been quite impressive as well. To ensure growth sustainability, the government should maintain macroeconomic stability, which would facilitate trade, foreign direct investment, and economic development. It is essential for Georgia to develop clear policies in those areas, which represents potential sources for growth on one hand, and potentialrisks on the other.Policies should include: (i) Diligent management of capital inflows using existing monetary and fiscal instruments. It is important to combat inflation through a combination of tightened monetary policy with a flexible exchange rate policy and tightened fiscal policy. The scope and scale of the macro-policy mix should avoid policy shocks in order to test the market for necessary adjustments. (ii) Supportdevelopmentof a healthyfinancial system. Sincethe capitalmarketsare still at an early stage of development, having a sound banking system is vital. Together with strengthening financial supervisioncapacity, there is a need to improve risk management, apply internationalauditing and accountingstandards, and enhance human resources capacity. (iii) Diversijy exportproducts and strengthen non-price competitiveness, enhance quality control of Georgian products, and support popularizationof those products in potential markets abroad. That way, current external revenue would be ensured and the current account deficit narrowed. (iv) Keep attracting FDI and maximizing its benefits. To do this, improving the overall business environment is critical. Apart from managing external uncertaintiesand externalities of capital inflows and fast development, the government needs to identify constraints for domestic investments. Special attentionmust be given to the small and mediumenterprise (SME) sector as far as labor and capital productivity is concerned. Improvedperformance of domestic SME would positively affect national savings, thus lessening the country's current high dependence on external resources. Government policy actions should be directed toward: (a) strengthening the rule of law and enhancedproperty and human rights protection,(b) improvements in the business climate and public sector institutionalization, (c) strengthening of implementation of the legislative framework of initiated reforms, and (d) further support of market economy reforms. The Government of Georgia has declared its determinationto addressthese challenges. l5The section draws from the Georgia PRS04 programdocument, May 2008, p. 7. 29 CHAPTERPOVERTYPROFILEINGEORGIA 2: A. Introduction 46. This chapterI6presents a profile of the extent of poverty in Georgia in 2007, and the changes in monetary incomes" that have occurred during 2003-2007.'8 The analysis is based on the 2007 Living Standards Measurement Survey (LSMS)data, and for comparisons over time, the 2003 Household Budget Survey (HBS) data are used. The chapter consists of six sections and is organized as follows. After a brief introduction (Section l), Section 2 presents changes in monetary incomes (and income-based poverty) during 2003-2007 in Georgia as a whole and across regions. Section 3 describes consumption- based poverty and inequality in Georgia using the most recent 2007 data. Section 4 presents poverty profile for 2007 and conducts a multivariate analysis of poverty. Section 5 analyzes other evidence of living standards, including subjective poverty, sources of income, and income-based poverty. Section 6 provides concluding remarks. 47. Georgia has maintained a record of strong growth and macroeconomic stability since early 2004, which it has achieved through an adequate mix of sound fiscal and monetary policies and ambitious structural reforms. The real gross domestic product (GDP) growth rate in Georgia averaged 8.9 percent during 2003-2006, led by construction, financial intermediation, and communications, while agriculture and manufacturing were generally stagnating. Since 2004, the drivers of growth started to shift toward the higher productive sectors owing to the active and growing intervention of external private capital, which allowed industrial production to revive, particularly in 2005 and 2006. At the same time, inflation was kept at about 8 percent through most of the period." This good economic performance has been achieved against several external shocks, which included especially sharp energy price increases, severe flooding, and a Russian ban on Georgian exports. 48. While economic growth has been on the right track, it has not been supported by increased employment. Similar to other countries undergoing deep structural changes, employment in Georgia has declined, mirrored by increased unemployment and decreased labor market participation. Since 2003, the absolute number of employed declined by an estimated 178,000 people.*' A significant part of this decline i s explained by the downsizing of the public administration sector following public sector reform, and a decline in employment in the agricultural sector. 49. Although there are indications that the overall poverty rate has begun to decline, a solid trend toward declining poverty has yet to be established. The analysis of trends in monetary incomes during 2003-2007 indicates that poverty is on the decline. However, the improvements in monetary incomes, especially in rural areas, have been mostly driven by increased social transfers. The 16 This chapter was preparedby Oleksiy Ivaschenko(Economist, ECSHD). The analysis of changes between 2003 and 2007 is limited to monetary incomes since consumption aggregatesare not comparable between 2003 and 2007. For the same reason, we cannot make comparison based on the more comprehensive measure of disposable incomes (which, in addition to monetary incomes, include in-kind consumption). '* The poverty line for Georgia has been derived on the basis of the 2007 LSMS of Georgia using the Food ExpenditureMethod(World Bank, 2002). The extreme (or food) poverty line is basedon the cost of the typical food basket providing2,260 calories per personper day. The non-foodexpenses are then added based on the consumption pattern of the 2"dto 4'h quintiles of the consumption distribution to derive the total poverty line. Changes in the cost o f living over time have been taken into account using Consumer Price indexes provided by the Department for Statistics of Georgia (DS). 19 Itjumped above 10 percent in 2007. 2o A detailed analysis of employment is presentedin Chapter 4 ofthe Report. 30 consumption data comparable over time would be needed to establish future poverty trends with more certainty. 50. The main challenge for the poverty reduction efforts in Georgia is the deeply entrenched poverty in rural areas, where most of the poor live. While the incomes of those employed in selective sectors of the economy (public administration,finance, information technology (IT), construction, and so forth) have increased significantly in real terms, the incomes in rural households were stagnant. Agriculture in Georgia continues to be dominated by low productivity self-subsistence agriculture, makingeradicationof poverty in rural areas extremely challenging.2' 51. The introduction of targeted social assistance for the extreme poor since end-2006 is a step in the right direction. As reforms are being implemented, business opportunities are expanding and foreign direct investments increased, the tax revenues have also expanded in real terms. Improved fiscal revenues allowed the Government to increase social spending and introduce the Targeted Social Assistance (TSA) programfor the extreme poor. The analysis indicatesthat the TSA program has already contributed to poverty reduction and will bring further poverty reduction if the program's coverage is expanded to reach more of the poor. As a result of extending the TSA, the trend toward the gradual alleviation of poverty will also become more resilientto shocks. B. Changes in Welfare in Georgia during2003-2007 52. The direct comparison of consumption-basedpoverty during 2003-2007 is not possible due to incomparable consumptionaggregates between the 2003 HBS and 2007 LSMS. However, a look at comparable monetary incomes indicatesthat householdwelfare improved during this period.22Indeed, householdmonetary incomes during 2003-2007 increasedon average by 3 1.6 percent, or 7.9 percent per year (Table 2.1). 53. If poverty dynamics were to be measured by using monetary income per adult equivalent (PAE) as a measure of welfare, then one would definitely find a decline in poverty during 2003- 2007. The analysis of the cumulative distribution functions for 2003 and 2007 indicatesthat at any level of the poverty line, poverty headcount and incidence declined (Annex 2, Figure A.l).23The decline in monetaryincomepoverty is especiallyevident in rural areas. 54. The increase in monetary incomes during 2003-2007 was mostly driven by an increase in social transfers. Among the monetary income components, social transfers registered the largest increase-from an average of 12.9Lari (GEL) per household in 2003 to 42.6 Lari (GEL) per household in 2007, or by 2.3 times (Table 2.1). At the same time, private transfers and remittances declined. The main reason for increasedsocialtransfers is the introductionof the TSA at the end of 2006. 21Giventhe predominantlyruralnatureofpoverty in Georgia, ruralpoverty is analyzed in great detail in Chapter3. 22A more comprehensivecomparison o f householdincomeand expenditure was not possiblebecause data for farm incomeand in-kindconsumptionare not comparable. 23The levels of poverty are not presentedhere because monetary income is substantially lower than consumption, which would result in overestimationofthe poverty levels. 31 Table 2.1: Household Total Monetary Income (in constant 2007 prices): 2003 HBS Compared to 2007 L S M S Total Annual 2003 (HBS) ZOO? (LSMSJ Change, % Change,% Total household income (GEL, monthly, in 2007 prices) 188.1 247"5 31.6 7.9 of which: Total householdw i n c o m e 83 9 53.3 13.3 Total householdself-employedwage income 47 0 13.7 3.4 Total household SOC. transfers income (incl. pensions) 12 9 231.O 57.8 Total householdprivate (domestic)transfers 24 9 -73.1 -18.3 Total household foreigntransfers 15 1 -41.7 -10.4 Total household rental income 2 6 -6.0 -1,5 Source: World Bank estimates using2003 Georgia HBS and 2007 LSMS data 55. The importance of social transfers has also increased in relative terms. The share o f social transfers in total monetary incomes increased from 6.8 percent in 2003 to 17.2 percent in 2007. The share o f monetary incomes from self-employment somewhat declined-from 25 percent to 21.6 percent (Figure 2.1). 56. Remittances play a relatively modest role in the total composition of incomes. Moreover, the share o f remittances in the household monetary incomes declined from 8 percent in 2003 to 3.6 percent in 2007 (Figure 2.1).24 It is likely that this decline is due to the measures imposed by the Russian authorities against Georgia's migrants (seasonal workers). Figure 2.1: Composition of Monetary Incomes, 2003 Compared to 2007 1 1 0 Other monetary income 60 IRemittances D Pnbate domestic transfers II Social transfers 4O Self-employed income 0 Wages II I 20 i lL o 03 07' 1 Source. World Bank estimates using 2003 Georgia HBS and 2007 LSMS data. 24Note that according to the macro data, total remittances in Georgia (as captured through the Balance of Payment data) account for about 6 percent of the GDP. Given a relatively modest role of remittances in Georgia, they are not analyzed in detail in this report. 32 57. Mostly as a result of pro-poor TSA, poor rural areas registered a more significant increase in monetary incomes. During2003-2007, PAE monetary incomes increased (from a very low base) by an average of 29.9 Lari, or 8I.3 percent in rural areas and 13.1 Lari, or 11.4 percent in urban areas (Figure 2.2). Figure 2.2: Monetary Incomes per Adult Equivalent by Urbanmural, (in 2007 prices) Source: World Bank estimates using 2003 Georgia HBS and 2007 LSMS data. 58. Across quintiles of the monetary income per adult equivalent (PAE) distribution, the largest rate of increase was registered by the bottom two quintiles and the top 20 percent of the distribution. However, even after a substantial rate o f increase, the average monetary incomes for the 1'` and 2"d quintiles of the distribution were 7 and 29 Lari per month, respectively, compared to 304 Lari for the top quintile (Figure 2.3). The structure o f monetary incomes indicates that the bottom quintiles have benefited mostly from increased social transfers, while the top quintile took advantage o f rising real salaries and incomes from own business. Figure 2.3: Monetary Incomes per Adult Equivalent by Quintile, (in 2007 prices) 1 IIIj250 *0° 150 100 50 I Q1 Q2 0 3 Q4 Q5 Total I Source: World Bank estimates using 2003 GeorgiaHouseholdBudget Survey (HBS) and 2007 LSMS data 33 59. The gains in monetary incomes have been mostly pro-poor across regions. Regions with the lowest monetaryincomes seem to have benefitedmost, in both absolute (size of the increase) and relative (rate of increase) terms (Table 2.2). For example, in Samtskhe-Javakheti and Samegrelo, monetary incomes PAE increased by about 30 Lari, representingan increase of 90 percent. Tbilisi benefited by about the same in absolute terms, but by 29 percent in relative terms. The main reason for increased monetary incomes in the poor regions is the pro-poor focus of the TSA program. However, in the three regions(ShidaKartli, Ajara, and Mtskheta-Mtianeti), monetary incomes declinedduring2003-2007. Table 2.2: Mean Monetary Incomesper Adult Equivalent in RealTerms Mean Mean Income Income PAE, Real PAE, Real 2007 2007 2003 2007 Absolute Change change% Urban 115.1 128.3 13.1 11.4 Rural 36.8 66.7 29.9 81.4 Household's affiliated region Kakheti 49.0 54.9 5.8 11.9 Tbilisi 136.1 175.5 39.4 29.0 Shida Kartli 60.4 45.8 -14.6 -24.2 Kvemo Kartli 52.7 73.8 21.1 40.0 Samtskhe-Javakheti 34.8 65.0 30.2 86.5 Ajara 92.4 86.8 -5.7 -6.I Guria 38.9 58.1 19.2 49.4 Samegrelo 35.0 68.9 33.8 96.6 Imereti 56.0 77.6 21.7 38.7 Mtskheta-Mtianeti 75.5 70.2 -5.3 -7.0 Lowest quintile 0.6 6.9 6.3 1,043.0 2 15.0 29.2 14.2 94.3 3 45.6 57.0 11.4 24.9 4 88.8 100.3 11.6 13.0 Highest quintile 225.8 303.7 78.0 34.5 Total 75.1 99.4 24.3 32.3 Source; World Bank estimates using 2003 Georgia HBS and 2007 LSMS data. 60. The analysis also indicates that during 2004-2005, there was a positive welfare impact from the payment of pension and public wage arrears. According to International Monetary Fund (IMF) data, in 2004, about GEL 97 billion and GEL 162 billion were paid to pensioners and employees in the public administration, respectively, due to elimination of arrears in pensions and public wages. Most of the pension arrears were to pensioners in rural arrears because the payment of pensions was current in larger urban settlements. In 2005, amounts paidto these two groups were around GEL 21 billion and GEL 84 billion, respectively.The conducted simulations indicatethat hadthe arrears not been paid, the poverty incidencewould be up to 2 percentagepointshigher. 61. The (monetary income) poverty decomposition indicates that growth has been key in poverty reduction. It is customary to decompose the change in headcount poverty into "growth" and "redistribution" components (Datt and Ravallion2002). The growth component is the difference between the two poverty indexes, keeping the welfare distributions constant. The redistribution component is the 34 change i when the mean of the two distributions remains constant. (The third component in this decomposition, the residual component, shows the change in poverty as a result of the interaction of growth and inequality). This poverty decompositionanalysis indicates that given the growth component alone, the incidence of extreme and total poverty would decline by 6.3 and 9.1 percentage points, respectively. However, the positiveeffect from growth was counterbalanced by the negative effect from rising inequality inmonetary incomes (Figure 2.4). Figure 2.4: Growth and Redistributionof Poverty Changes, 2003-2007 I Source. World Bank estimates using2003 Georgia HBS and 2007 LSMSdata. C. Consumption-based Poverty and Inequality in Georgia in 2007 C.1. Poverty Headcount, Depth, and Severity 62. This section presents a picture of consumption-based poverty in Georgia using the 2007 LSMS data. The welfare measure used here is the consumption PAE. Two poverty lines have been derived from the data: (a) the food (extreme) poverty line equal to 47.1 Lari is based on the cost of the typical food basket producing2,260 calories per day, and (b) the total poverty line equal to 71.6 Lari is equal to the food poverty line plus the allowance for basic non-food expenditures (based on the consumptionpattern found inthe 2ndthrough 4t"deciles ofthe consumptiondistribution). 63. Poverty in Georgia is widespread. According to the welfare measure used and the poverty lines, 23.6 percent of the population is poor and 9.3 percent is extreme poor. The average consumptionof the poor falls short of the overallpoverty line by 7.2 percent (Table 2.3). 35 Table 2.3: Overall Poverty in Georgia Headcount Poverty Squared RawPo) Gap(P1) Ga (p2 Poverty 2007 2007 2007 Poverty Line = 71.6 Lari (totaCpoverty) Urban 18.3 5.3 2.3 Standard Error 1.2 0.5 0.3 Rural 29.7 9.2 4.1 Standard Error 1.7 0.7 0.4 Total 23.6 7.2 3.1 Standard Error 1.o 0.4 0.2 Poverty Line = 47.I Lari (extreme poverty) Urban 6.7 1.8 0.8 Standard Error 0.8 0.3 0.1 Rural 12.4 3.2 1.3 Standard Error 1.1 0.3 0.2 Total 9.3 2.4 1.o Standard Error 0.7 0.2 0.1 Source: World Bank estimates using 2007 LSMS data. C.2. Regional Dimensions of Poverty 64. Poverty in Georgia is entrenched in rural areas. In terms of the distribution of the poor across urban and rural areas, the concentration of the poor in rural areas is 59 percent. Moreover, rural areas account for 62 percent of the extreme poor. Higher concentrationof poverty in rural areas is driven by a muchhigher incidenceof poverty in rural areas compared to urban areas-29.7 percent compared to 18.3 percent, respectively.The share of population living in rural areas is actually somewhat lower than that living in urban areas. The poverty depth is also more profound in rural areas (Table 2.4). 36 Table 2.4: Poverty, by Geographic Region Poverty Headcount Distribution Distribution Rate of the Poor of Population 2007 2007 2007 Poverty Line = 72.6 Lari (totalpoverty) Urban 18.3 41.1 53.1 Rural 29.7 58.9 46.9 Region Kakheti 46.3 15.7 8.0 Tbilisi 12.9 15.6 28.5 Shida Kartli 59.4 18.9 7.5 Kvemo Kartli 17.3 7.6 10.4 Samtskhe-Javakheti 18.1 3.3 4.3 Ajara 27.4 9.5 8.2 Guria 33.2 5.3 3.8 Samegrelo 14.4 5.7 9.3 Imereti 19.1 14.1 17.5 Mtskheta-Mtianeti 40.6 4.2 2.5 Total 23.6 100.0 100.0 Poverty Line =47.2 Lari (extremepoverty) Urban 6.7 37.9 53.1 Rural 12.4 62.1 46.9 Region Kakheti 20.8 17.9 8.0 Tbilisi 4.8 14.7 28.5 Shida Kartli 32.2 25.9 7.5 Kvemo Kartli 4.8 5.4 10.4 Samtskhe-Javakheti 5.8 2.7 4.3 Ajara 7.5 6.6 8.2 Guria 12.4 5.0 3.8 Samegrelo 2.9 2.9 9.3 Imereti 7.5 14.1 17.5 Mtskheta-Mtianeti 18.5 4.9 2.5 Source: World Bank estimates using2007 LSMS data. 65. The incidence of poverty varies considerably across different parts of the country. The highest incidence of poverty is observed in Shida Kartli (59.4 percent), followed by Kakheti (46.3 percent), and Mtskheta-Mtianeti (40.6 percent). Those regions also have the highest rates of extreme poverty (Table 2.4). Harsh terrain, physical isolation, and political uncertainty combine to make living conditions extremely hard in these regions. There is somewhat more potential in the southern areas of Shida Kartli and Mtskheta-Mtianeti,because land is better and there is reasonable road access to Tbilisi. But the winters are severe and irrigation is an importantrequirementfor agriculture.The western areas of Khaketi have become important for wine production and have grown accordingly, but low rainfall, limited scope for irrigation, and physical isolation limit the capacity to raise incomes in the northernand eastern areas of this region. Livestock production remains the main source of income for rural people in these areas. 37 66. The lowest incidenceof poverty is found in Tbilisi (12.9 percent), Samegrelo (14.4 percent), and Kvemo Kartli (17.3 percent). The wealthiest regions lie in a continuous arc running from Samegrelo in the northwestto Kvemo Kartli in the southeast. The western regions in this arc (Samegrelo and Imereti) benefit from favorableagricultural conditions and good access to urban markets and seaports on the Black Sea. A large river plain affords fertile soils, rainfall is high, and winters are mild. Inthe east, Kvemo Kartli is favored by good agricultural land in the south, between Tbilisi and the border with Azerbaijan, and its proximity to Tbilisi and Rustavi. 67. The intermediate level of poverty is observed in Ajara (27.4 percent) and Guria (33.2 percent). These are the regions in which political instability and the collapse of tea production have resultedin high poverty levels-despite favorable underlying conditions. Both regions benefit from good agro-climatic conditions and their proximity to the urban centers and ports on Georgia's Black Sea coast. But both have suffered from the political situation in Ajara and the collapse of tea production.However, prospects for growth and poverty reductionhave improved now that the political climate in Ajara is more stable, and the economies of these regions should benefit from development o f the ports in Batumi and Poti, and recoveryof the Black Sea tourist industry. C.3. Sensitivity and Robustness of Poverty Estimates 68. Poverty estimates presented above depend critically on: (a) the way the poverty line is defined and updated; and (b) the choice of welfare measure. In this section, robustness of poverty is examined by reviewing the changes in the poverty headcount when the poverty line is increased and decreased by 5, 10, and 20 percent (Table 2.5). 69. There is some clustering of the population around the poverty line. Having increased or decreased the poverty line, it is observed that poverty rates increased or decreased by a correspondingly higher percentage, indicating population clustering around the poverty line. For instance, increasingthe total poverty line by 5 percent would lead to the poverty headcount rising by 2.7 percentage points. Another 5 percent increase in the poverty line would result in a further rise in the poverty headcount of 2.6 percentage points. Increase in the extreme poverty line of 10 percent would pushthe extreme poverty headcount by 2.5 percentage points(Table 2.5). Table 2.5: Sensitivity of Poverty Incidencewith Respect to the Choice of Poverty Line (percent) 7nn7 Poverty Change Incidence(P0) Actual(%) from Poverty Line = 71.6 Lori (totalpoverty) Actual 23.6 0.00 +5% +10% 26.3 11.41 28.9 22.44 +20% --510% 33.8 42.84 %o 21.4 -9.30 19.2 -1 8.94 -20% 14.8 -37.44 Poverty Line =47.Z Lnri (extreme poverty) Actual 9.3 0.00 +5'/0 10.4 11.59 +10% 11.8 26.38 --+210% 0% 14.3 52.68 5%o 8.2 - 1 2.66 6.7 -27.84 -20% 5.1 -44.92 Source: World Bank estimates using 2007 LSMS data. 38 C.4. Inequality 70. Expenditure and poverty levels vary widely by region. These differences are apparent in Figure 2.5, which shows the strong relationship between poverty levels and household PAE expenditure-and the location o f each region along this continuum. The decomposition o f inequality indicates that inequality among regions contributes about 4.5 percent o f the overall inequality (Table 2.6). The differences across and within regions result in the Gini coefficient for the overall distribution o f PAE expenditure beingequal to 36.3 percent (Table 2.7). Table 2.6: Decompositionof Inequality, by Region GE(0) GE(1) GE(2) Overall Inequality 2007 22.4 22.6 29.1 Urban 23.0 23.2 29.6 Rural 19.5 19.1 23.3 Within Group Inequality 2007 21.3 21.6 28.0 Between Group Inequality 2007 1.o 1.o 1.o Between Group Inequality as O b o f Overall Inequality 2007 4.6 4.5 3.5 Source World Bank estimates using 2007 LSMS data. 71. Inequality was higher in urban areas. The Gini coefficients for urban and rural areas are 36.8 percent and 33.7 percent, respectively. The 9Ot"/lOth percentile ratio is 5.4 in urban areas, compared to 4.9 in rural areas (Table 2.7). Inequality between urban and rural areas is higher in the upper compared to the lower part o f the PAE expenditure distribution (Table 2.8). Figure2.5: Expenditureand Poverty Levels, by Region Kakheti, 46 3 Mtskheta-Mtianeti, Tbilisi, 12.9 40 60 80 100 120 140 160 180 200 220 241 Mean consumption PAE, GEL Note: The size ofthe bubble reflects the region`s share ofthe total number of poor Source. World Bank estimates using 2007 LSMS data. 39 Table 2.7: Inequality inPer Capita Expenditure Distribution, by Urban and Rural Areas Bottom Half of the Upper Half of the Interquartile Distribution Distribution Range Tails p25lp10 p50lp25 ~ 7 5 1 ~ 5 0 p901p50 ~ 7 5 1 ~ 2 5 p90/p10 Gini Total 2007 1.51 1.53 1.55 2.31 2.37 5.33 36.25 Urban 2007 1.47 1.56 1.55 2.38 2.42 5.44 36.82 Rural 2007 1.48 1.53 1.53 2.14 2.34 4.86 33.72 Note: p90/75/50/25 refer to the percentiles of the expendituredistribution. Source: World Bank estimates using 2007 LSMS data. Table 2.8: Inequality in Per Capita ConsumptionDistribution, by Urban and Rural Areas p10 p25 p50 p75 p90 2007 1.25 1.25 1.27 1.29 1.40 Source: World Bank estimates using2007 LSMS data. D. Poverty Profile and Multivariate Analysis of Poverty 72. The 2007 LSMS contains extensive modules on various characteristicsof households-place of residence, demographic composition, housingsituation, access to facilities, sector of employmentof adult householdmembers, education attainments, and so forth. This section uses data from these modules to estimate poverty rates across households with different characteristics. D.1. Poverty Profile 73. A poverty profile describes the poor by indicating the risk of being poor according to various characteristics, such as type of employment, level of education of the household head, demographic composition of a household (that is, gender, household size, number of children, ethnic status), and so forth. This section provides a profile of consumption-based poverty with respect to those characteristics. A poverty profile is presented using2007 LSMSdata.*' 0.2. Typeof Employment of the HouseholdHead 74. Employment status is strongly correlated with poverty. Households headed by wage earners experiencethe lowestrates of overalland extreme poverty-12.7 percent and 4.4 percent, respectively.In contrast, households headed by the unemployed or those out of the labor force were poorer on average than households headed by any employment category (both in terms of overall and extreme poverty), indicatingthat the lack of employment is one of the main causes ofpoverty in Georgia(Table 2.9). 75. Among employed, households headed by self-employed in agriculture and in "other" employment activities had the highest poverty incidence. The poverty headcounts for those types of householdsare 23 percent and 24.2 percent, respectively,which are at or above the nationalaverage (23.6 percent). Self-employment in agriculture also has the highest incidence of extreme poverty among employmentcategories-9.1 percent (Table 2.9). This is especiallyworrisome given that agriculturalself- 25Tables with standard errors for overall and extreme poverty incidence are available, but not presentedhere to save on space. 40 employment is the most represented employment sector category, with 20 percent of all poor (and 40 percent of working poor) falling into this category. 76. Self-employment outside of agriculture is associatedwith a somewhat higher risk of poverty than wage employment, but this risk is much lower than the national average. Households with heads engaged in non-agriculturebusiness face a poverty headcount of 15.4 percent--compared to 12.7 percent in wage employment (Table 2.9), and a 23.6 national average. Interestingly, the incidence of extreme poverty for those self-employed outside of agriculture is the same as for wage employment (4.4 percent), and is much lowerthan the nationalaverage (9.3 percent).26 Table 2.9: Poverty, by Employment Status of the HouseholdHead (percent) Distribution Headcount Poverty Distribution o f Rate o f the Poor Population 2007 2007 2007 Poverty Line = 71.6 Lari (totalpoverty) Type of employment (with not employed) Wage empl. 12.7 10.5 19.6 Self-empl. (agr) 23.0 19.6 20.1 Own business (non-agr.) 15.4 5.7 8.7 Other empl. 24.2 0.9 0.8 Not empl. 29.5 63.3 50.7 Poverty Line = 4 7.Z Lari (extreme poverty) Type of employment (with not employed) Wage empl. 4.3 8.9 19.6 Self-empl.(agr) 9.1 19.7 20.1 Own business (non-agr.) 4.4 4.1 8.7 Other empl. 6.6 0.6 0.8 Not empl. 12.3 66.7 50.7 Total 9.3 100.0 100.0 Note: "Distribution of the Poor" and "Distribution of Population" represent the share of individuals or the share of poor individuals living in specific types of households (for example, in a householdwith a head self-employed in agriculture). Source: World Bank estimates using 2007 LSMS data. 0.3. Education of the Household Head 77. Householdswith more educated household heads are less likely to be poor. Differences in educationalattainment of heads of householdsare reflectedin considerablydifferent poverty rates (Table 2.10). Households with heads that have incomplete secondary education have a poverty rate of 31.8 percent, which is the highest rate among all education groups. Even complete general secondary education does not reduce the risk of poverty by much. Nevertheless, the poverty rate significantly declines as the level of education by a household head increases to technical and vocational education (TVE) or university. HavingTVE brings down the probability of being in poverty to 21.7 percent, while 26 This section only considers the type of employment, and not the specific employment sector, since data on the latter are not available in the 2007 LSMS. However,the relationshipbetweenpovertyand the sector of employment i s analyzedin Chapter 6 on labor markets,usingHBS data. 41 havinga universitydegree brings it down to 12.1percent.While this indicates that education pays off, the poverty rates even among households with highly educated heads are still high. Similar trends across educationlevelsare observed for the incidenceof extreme poverty. Table 2.10: Poverty, by Education Level of HouseholdHead (percent) Poverty Distribution Distribution Rate of the Poor of Population 2007 2007 2007 Poverty Line = 71.6 Lari (totalpoverty) Incomplete secondaryhelow 31.8 20.0 14.9 Secondary 29.0 48.1 39.3 TVEhpecial secondary 21.7 18.1 19.7 Incomplete university 21.9 0.9 1.o University 12.1 12.9 25.2 Poverty Line =47.I Lari (extreme poverty) Incomplete secondaryhelow 12.9 20.5 14.9 Secondary 12.0 50.6 39.3 TVE/special secondary 8.3 17.4 19.7 Incomplete university 2.9 0.3 1.o University 4.1 11.2 25.2 Total 9.3 100.0 100.0 Source; World Bank estimates using 2007 LSMS data. 0.4. Demographics 78. Female-headed householdsdo not face a significantly higher incidence of total poverty, but do face a higher risk of extreme poverty. In terms of total poverty, the incidence of poverty among female-headed households (25 percent) is just somewhat higher than among male-headed households (23.1 percent). However, female-headed households face a statistically significantly higher risk of extreme poverty-1 1.3 percent compared to 8.6 percent among male-headed households (Table 2.11). The high share of female-headedhouseholds (29 percent) could be explainedby seasonal labor migration and its patterns-it is usually a main male breadwinnerwho heads abroad in search of better employment opportunities. That may also explain why female-headed households do not face a significantly higher risk of overall poverty. 79. Both an increase in the number of small children and an increase in the number of household members are related to an increase in poverty. Households with no children face a poverty risk of 19.4 percent, while those with 3 or more children experience the risk of poverty of 37.5 percent. About 48 percent of households have 1 or 2 children, which is associatedwith a poverty risk of 24 to 26 percent (Table 2.12). Large households with 5 or more members have an incidenceof overall and extreme poverty above the national average. Moreover,this group is very populous, comprisingabout one-halfof all households and 55 percent of all poor (Table 2.12). The importantfactor in explainingpoverty in large families is dependency ratio. The dependency ratio measures the number of householdmembers of non- 42 working age (children and elderly) that have to be supported by the household's working members. Larger households have more childrenand, thus, a lower ratio of income earners than smaller households, which causes lower levels of consumption. The child poverty in Georgia is highlighted in larger detail in Box 2.1 below. Table 2.11: Poverty, by HouseholdHead's Gender (percent) Headcount Poverty Distribution Distribution Rate of the Poor o f Population 2007 2007 2007 Poverty Line = 71.6 Lari (total poverty) Female 25.0 29.0 27.4 Standard Error 1.4 0.0 0.8 Male 23.1 71.O 72.6 Standard Error 1.2 0.0 0.8 Total 23.6 100.0 100.0 Standard Error 1 .o 0.0 0.0 Poverty Line =47.Z Lari (extreme poverty) Female 11.3 33.1 27.4 Standard Error 1 .o 0.0 0.8 Male 8.6 66.9 72.6 Standard Error 0.8 0.0 0.8 Total 9.3 100.0 100.0 Standard Error 0.7 0.0 0.0 Source: World Bank estimates using 2007 LSMS data. Table 2.12: Poverty, by DemographicComposition(percent) Poverty Distribution Distribution Rate of the Poor of Population 2007 2007 2007 Poverty Line = 71.6 Lari (totalpoverty) N of children age 0-15 N o children 19.4 37.5 45.8 1 23.9 22.8 22.6 2 26.3 24.6 22.2 3 or more children 37.5 15.0 9.4 Household size 1 16.0 3.2 4.7 2 17.9 7.9 10.5 3 18.6 10.6 13.5 4 22.7 22.8 23.7 5 24.9 19.8 18.8 6 27.1 15.9 13.9 7 or more 3 1.4 19.8 14.9 Total 23.6 100.0 100.0 Poverty Line = 4 7.I Lari (extremepoverty) Nof children age 0-15 No children 7.0 34.4 45.8 1 10.3 24.9 22.6 2 8.6 20.5 22.2 3 or more children 20.1 20.3 9.4 Household slze 1 6.3 3.2 4.7 2 6.2 7.0 10.5 3 7.8 11.3 13.5 4 8.1 20.7 23.7 5 9.2 18.5 18.8 6 11.1 16.5 13.9 7 or more 14.3 22.8 14.9 Total 9.3 100.0 100.0 Source; World Bank estimates using2007 LSMS data. 43 Box 2.1: Child Poverty in Georgia27 Children in Georgia are at a slightly higher risk of poverty than the general population. Twelve percent of children compared to 9 percent of the general population live in extreme poverty, while 28 percent of children compared to 24 percent of the general populationlive under the total poverty line. Similar to the overall poverty, child poverty is concentrated in rural areas. Children living in rural areas comprise less than half of all children in the sample (46 percent), but represent 61 percent of all poor children. This reflectsthe fact that subsistencefarming remains a major source of income in most rural areas, and opportunitiesfor off-farm employment continue to be very limited. Child poverty is often reflected in poor housing conditions. Poor children are much more likely to find themselves disadvantagedcompared to the non-poor children on various measures of the housing deprivation (see Table below). For children,housing problems can pose risks to their health and safety, but also may impact on their education and social development. Table: Housing Deprivation Housing Item Percent Percent Percent Children Poor Non-poor Children Children House is in a very bad condition 6 12 4 Househas no fbrniture/self-made 6 14 3 Part of dwelling is destroyedor walls are very crackeddamp 17 27 13 Windowslparts of windows have no glass 26 45 19 Earth floor or floor needsto be renewedrepairedcompletely 17 31 11 Electric lighting is absent or some rooms have no bulbs 15 27 11 Note: Base: all children. Source: UNICEF and World Bank estimatesusing 2007 LSMS data. The risk of poverty for children is strongly related to the parents' employment status and receipt of social transfers. Lack of local employment opportunities, the number of children in the household, having children or adults with disabilities or chronic diseases in the household who need special care and may incur higher medical expenses, and parents' lack of education are all factors that are associatedwith a higher povertyrisk for children. Parents' unemployment is one of the strongest predictors for child poverty. Having at least one employed person in the household reducesa child's risk o f extreme poverty from 22 percentto 8 percent, and the total poverty risk from 44 percent to 25 percent. Social transfers, and especially pensions, help reduce child poverty rates. Almost one-half of all children (47 percent) are living in households with at least one pension recipient, which indicates the prevalence of three- generational households.Among children living in pension-recipient households, pensions amountedto 63 GEL per month, on average. Ifpensions are not counted as household income, the child extreme poverty rate is 6 percentage points higher (18 percentcomparedto 12 percent) and the total povertyrate is 5 percentagepointshigher (33 percent compared to 28 percent).28Targeted Social Assistance also reduces the child poverty rates by another 2 percentage points. "ThisBoxisbasedonthebackgroundpaperonchildpovertypreparedjointlywith UNICEF.Themainauthorsof &hebackground paper were Jonathan Bradshaw (University of York, UK), Yekaterina Chzhen (University of York, UK), Dimitri Gugushvili(UNICEF, Georgia), and PetraHoelscher (UNICEF Regional Office, CEEKIS). The work on the UNICEF side was coordinated by Gordon Alexander (Senior Advisor on Economic and Social Policy, UNICEF RegionalOffice, CEEICIS). 28These simulations assume intra-householddistributiono f income from pensions. 44 D.5. MultivariateAnalysis of the Poverty Correlates 80. A poverty profile is a useful descriptive method, but it cannot be usedto gauge the net association between household characteristics and the probability of a given household being in poverty. For this purpose, regression analysis that accounts for the correlation among various household characteristics is most useful.Regression analysis helps to purge out the multivariatecorrelationacross these factors to find the net effect of each of them. The associations among background characteristics could be different in urban and rural areas, so they are examined for urban and rural areas separately. (For example, the economic return on educational investment might be higher in urban areas because of the larger number of productive opportunities to be found there.) It is also more efficient to analyze a linear relationship between a log of consumption PAE and a set of household characteristics than a relationship between a categorical outcome defined as "poorhon-poor" and a set of determinants. The results of the regression analysis are presented and discussed below. 81. Regional factor is a significant correlate of poverty. In urban areas, compared to households livingin Kakheti, mean consumption varies from being on average about 40 percent higher in Tbilisi and Mtskheta-Mtianeti to being 18 percent lower in Shida-Kartli. In rural areas, compared to households living in Kakheti, mean consumption varies from being on average about 40 to 45 percent higher in Kvemo Kartli and Samegrelo to being25 percent lower in Shida-Kartli(Table2.13). 82. Higher educational attainment significantly increases household consumption. In urban areas, households with university-taught household heads enjoy 37.6 percent higher consumption compared to households where the household head has incomplete secondary education. As can be expected, returns on all levels of education are higher in urban areas (Table 2.12). 83. Type of employment is a significant determinant of household consumption. Clearly, in both urban and rural areas, households with no employed heads experience significantly lower levels of consumption. Agricultural self-employment in rural areas is associated with 6.7 percent lower consumption compared to wage employment. Agricultural self-employment in urban areas is associated with higher consumption, likely because in urban areas, this type of employment serves as a supplement to more formal wage employment. 84. Demographics of the household do matter. Male-headed households are somewhat better off than female-headedhouseholds in urban areas, but not in rural ones. Finally, larger households experience lower levels of per capita consumption. 45 Table 2.13: The Results of ConsumptionRegressions 2007 2007 Urban Rural Coefficient S.E. Coefficient S.E. Household characteristics Logo f household size -0.158** 0.07 -0.150*** 0.05 Log o f household size squared -0.008 0.03 -0.040 0.03 Region Kakheti (dropped) (dropped) Tbilisi 0.396* ** 0.07 0.759*** 0.11 Shida Kartli -0.178** 0.09 -0.244* ** 0.04 K v e m o Kartli 0.174** 0.08 0.446* ** 0.04 Samtskhe-Javakheti 0.277*** 0.09 0.380*** 0.04 Ajara 0.070 0.08 0.063 0.05 Guria -0.098 0.10 0.223*** 0.04 Samegre1o 0.122 0.08 0.3 93*** 0.04 Imereti 0.193** 0.08 0.309*** 0.04 Mtskheta-Mtianeti 0.436*** 0.10 0.13 1*** 0.05 Characteristics of household head Log o f household head's age -0.085 0.05 0.377*** 0.05 Gender Female (dropped) (dropped) M a l e 0.092*** 0.03 0.033 0.02 Education Incomplete secondary/below (dropped) (dropped) Secondary 0.114** 0.06 0.089*** 0.03 TVEIspecial secondary 0.201*** 0.06 0.173*** 0.03 Incomplete university 0.62 1 *** 0.I 3 0.178* 0.11 University 0.376*** 0.06 0.353*** 0.04 Type of employment (with not employed) Wage empl. (dropped) (dropped) Self-empl. (agr) 0.1 17* 0.07 -0.067 * 0.04 Own business (non-agr .) 0.045 0.05 0.089 0.06 Other empl. -0.027 0.15 -0.196* 0.I1 Not empl. -0.200*** 0.03 -0.324*** 0.04 Constant 4.955*** 0.23 3.171*** 0.21 Number o f observations 2,227 3,029 Adiusted R2 0.177 0.250 Note: Levels of statistical significance:***p60 years) 0.9 0.8 0.9 HouseholdHead YOFemaleheaded 33.7 29.3 34.1 Age 62.9 59.5 60.8 Education Illiterate 3.1% 3.9% 3.2% Primary and vocational 17.0% 14.4% 15.7% Incompletesecondary 22.3 25.8 19.4 Secondary education 45.4 49.6 55.3 HigherEducation 11.2 5.8 5.1 Number 2,242 569 217 Source; World Bank estimates using 2007 LSMS data. 123. The modest observed difference in family size between poor and non-poor rural households may be more important than it seems. While family size is relatively small for both poor and non-poor households, the percentage increase inthe number of adults and children in poor households is quite large relative to non-poor households, and the resources that poor households have to support these extra household members are substantially less. B.3. Household Income Sources 124. Subsistence agriculture remains an integral part of the rural economy. Farm income (farm produce sales plus the value of in-kind (own) consumption) is the major income source for all but the poorest 20 percent of rural households (for whom social transfers dominate other sources of income) (Table 3.8). Farm income is derived predominantly from in-kind consumption. Less than 20 percent of total farm income derives from cash sales of farm products.This reliance on subsistence agriculture is a marked contrast to Georgia's former status as a major source of high-valueexport cash crops (tea, citrus, grape, wine) for the former Soviet Union. 61 --- -Table--3.8:- --- Income Compositionof Rural Householdsby Quintile, Georgia, 2007 - --- - - - --- - -- __ - - _ _ _ _ _ _ _ _ -- - I I_ - ___ All Q1 a 2 63- Q4 Q5 GELMonth TotalIncome 260.9 98.0 138.6 195.4 259.2 468.6 In-Kind Income 91.5 23.9 47.0 66.3 97.8 166.7 Salary and Wage Income 53.0 7.0 16.8 30.9 49.1 118.3 Self-Employment 34.2 6.7 13.9 20.6 33.6 71.0 Farm Sales 14.6 2.7 4.4 9.0 12.3 32.8 Property Rental 0.2 0.0 0.0 0.0 0.4 0.3 RemittanceIncome 3.O 0.4 1.1 2.6 3.6 5.2 Internal Private Transfers 3.5 1.6 2.4 3.5 4.0 4.9 Pensions 39.9 42.7 38.8 44.7 40.0 35.4 Social Assistance 8.3 11.4 10.3 8.2 8.7 5.0 Other Incomea 12.8 1.5 3.8 9.6 9.8 28.8 % Composition Total Income 100.0 100.0 100.0 100.0 100.0 100.0 In-Kind Income 35.1 24.4 33.9 33.9 37.7 35.6 Salary and Wage Income 20.3 7.1 12.1 15.8 18.9 25.2 Self-Employment 13.1 6.8 10.0 10.5 13.0 15.2 Farm Sales 5.6 2.8 3.2 4.6 4.7 7.0 Property Rental 0.1 0.0 0.0 0.0 0.2 0.1 RemittanceIncome 1.1 0.4 0.8 1.3 1.4 1.1 InternalPrivateTransfers 1.3 1.6 1.7 1.8 1.5 1.o Pensions 15.3 43.6 28.0 22.9 15.4 7.6 Social Assistance 3.2 11.6 7.4 4.2 3.4 1.1 Other Incomea 4.9 1.5 2.7 4.9 3.8 6.1 % of HouseholdsReporting Income Sources In-KindIncome 87.2 75.4 89.1 88.0 89.9 93.6 Salary and Wage Income 19.6 8.3 13.7 19.8 21.5 35.0 Self-Employment 17.2 8.6 10.9 15.3 21.8 29.4 Farm Sales 27.7 12.7 19.5 27.6 32.8 46.0 Property Rental 0.4 0.2 0.2 0.3 0.7 0.5 RemittanceIncome 3.7 2.1 2.5 3.8 4.1 5.8 InternalPrivate Transfers 9.1 8.9 7.4 10.1 10.2 9.1 Pensions 63.8 73.7 65.3 64.4 60.1 55.4 Social Assistance 22.6 34.0 29.8 20.5 17.0 11.7 Other Incomea 12.9 4.0 6.9 14.4 16.2 23.3 Note: a. Income from stipends, alimony, and other (irregular)monetary income. Source: World Bank estimates using 2007 LSMS data. 125. Public transfers, derived mostlyfrom pensions, are the major source of income for the rural poor, accounting for 55.2 percent of income in Q1 and 35.4 percent of total income in Q2 (Table 3.9). Average pension incomes are fairly uniform across expenditure groups, and pension payments appear to be reaching a high proportion of poor rural households. The data also suggest that social assistance payments are being targeted reasonably Both the average payment and the incidence of social assistancepayments increaseas poverty increases. 126. Salaries and self-employment are an important income source for higher-income rural households - those in the top two expenditure quintiles (Q4-Q5). However, only 20 percent of rural households earn income from either of these two sources, and most of these are in the upper-expenditure 4'The targeting of social assistance is discussedin detail in the Chapter on social assistance. 62 quintiles. Salary incomes tend to be higher than income from self-employment, but access to these two income sources appearsto be similar, based on the frequency with which they are reported. 127. Remittance incomes are reported by a low 3.7 percent of rural households, and account for only 1.2 percent of total household income.42Higher-income households (Q5) benefited most from remittancesin 2007 and low-income households (Q142) the least. Privatetransfers from people within Georgia were a more important source of rural household income than remittances. Income from propertyhandrental is negligible. B.4. Regional Characteristics 128. Rural poverty levels, and the characteristicsof rural poverty, vary widely by region. These differences are apparent in Figure 3-3, which shows the strong relationship between poverty levels and rural household expenditure, as well as the location o f each region along this continuum.43 Figure 3.3: Expenditure,Poverty Incidence, and Poverty Concentrationin Rural Areas, by Region 70.0 60.0 s 50.0 Mtskheta-Mtianeti, 4 2 40.0 U cI 9 30.0 0 n$ 20.0 10.0 KIverno Kartli, 13 0.0 1 0 20 40 60 80 100 120 140 160 180 20( Mean consum ption PAE, G B Note: The size of the bubble reflects the region's share in the total number of rural poor. The number next to the region's name indicate incidenceof total poverty. Source: World Bank estimates using 2007 LSMS data. 129. Differences in geography, proximity to major markets and transport routes, and political stability result in two broad zones of economic well-being in rural areas. The wealthiest regions, with the lowest levels of rural poverty, lie in a continuous arc running from Samegrelo in the northwest to Kvemo Kartli in the southeast (Table 3.9).44The western regions in this arc (Samegrelo and Imereti) 42The share of householdsreporting incomes from remittancesin the 2007 LSMS is much lower compared to that reportedinthe 2003-2006 HBS.It is not very clear why it is the case. 43A similar graphicalrelationshipis observed for extreme poverty. 44 Rural households in the area near Tbilisi exhibit even higher incomes and lower levels of rural poverty. But because they are very few in number, and derive most of their incomes from wages and salaries, they are not consideredinthis analysis. 63 benefitfrom favorableagricultural conditionsand good access to urban markets and seaports on the Black Sea. A large river plain affords fertile soils, rainfall is high, and winters are mild. In the east, Kvemo Kartli is favored by good agricultural land in the south, between Tbilisi and the border with Azerbaijan, and its proximity to Tbilisi and Rustavi. The region of Samtskhe-Javakheti is mountainous, less well endowed for agriculture, and roads are poor. But it forms a southern corridor linking Ajara and Guria to Tbilisi and benefits from its proximity to northern Turkey, which creates opportunities for small-scale trade and commerce. Average income from self-employmentis highest in this region, as is the number of households engaged in self-employmentactivity. 130. The poorest regions form a second continuous arc along the fringes of the mountains that form the northern border with Russia. Poverty rates are very high here, ranging from 43.4 percent in Mtskheta-Mtianetito 64.2 percent in Shida Kartli. Extreme poverty is also highest in these regions.Harsh terrain, physical isolation, and political uncertainty combine to make living conditions extremely hard in these regions. There is somewhat more potential in the southern areas of Shida Kartli and Mitskheta- Mtianeti, because land is better and there is reasonable road access to Tbilisi. But the winters are severe and irrigation is an important requirement for agriculture. The western areas of Khaketi have become important for wine productionand have grown accordingly, but low rainfall, limited scope for irrigation, and physical isolation limit the capacity to raise incomes in the northern and eastern areas of this region. Livestock productionremains the main source of income for ruralpeople in these areas. 131. Ajara and Guria form an intermediate group by poverty level. I n these regions political instability and the collapse of tea production have resulted in high poverty levels-despite favorable underlying conditions. Both regions benefit from good agro-climatic conditions and their proximity to the urban centers and ports on Georgia's Black Sea coast. But both have suffered from the political situation in Ajara and the collapse of tea production. Prospects for rural growth and poverty reduction have improved now that the political situation in Ajara is more stable, and the rural economies of these regions should benefit from development of the ports in Batumi and Poti, and recovery of the Black Sea tourist industry. 64 Q Z Q N N N d 132. A comparison of regional differences in household income composition shows that the poorest two regions (Shida Kartli and Kakheti) have much lower levels of farm income than the other regions (Table 3.9). Non-farm incomes from salaries and self-employment are also lower in these regions-indicative of the depressed local economies they rely on for non-farmemployment and income generation, and their relative isolation from Tbilisi. Public transfers (including pensions) are thus the main income source. 133. Both farm and non-farm incomes are higher in the other regions, due to more favorable agro-climatic conditions, stronger local economies, and better access to major urban centers. Total farm incomes are highest and most market(cash) oriented in western Georgiawhere farmers benefit from good agro-climatic conditions and access to urban and export markets on the Black Sea coast (Table 3.9). For non-farm income, salaries are more important than self-employment in most regions. This is consistent with the generally weak development of small-scale private enterprise in Georgia, particularly outside Tbilisi. 134. Comparison of public transfers suggests that while average pension payments per household are fairly uniform across regions, the level of social assistance is not always consistent with the level of rural poverty. Relatively high average social assistance payments per household are observed in Imeretiand Samegrel-two of the wealthiest rural regions.Higher levels of social assistance are reportedin Shida Kartli, however, the regionwhere it is most needed. C. Constraints to Rural Poverty Reduction 135. The preceding analysis indicates that the major constraints to rural poverty reduction are as follows: The harsh conditions in mountainareas Limited access to agriculturalresources 0 Low incomediversificationand capital reserves in the face of high income variability 0 Limited opportunitiesto increase household income. 136. These factors are discussed in greater detail below to provide a clearer sense of how to achieve rural poverty reduction in Georgia. Where appropriate, data from the preceding analysis are combined with information from the 2004 Agricultural Census to indicate the number of poor farm households affectedby these constraints. C.1. Harsh Conditionsin Mountain Areas 137. Rural poverty is deepest and most pervasive in the mountain areas, particularly in the northern and eastern regions of Georgia. Scarce agricultural land and harsh winters limit the ability of mountain people to achieve even subsistence production, and poor roads leave them isolatedfrom urban centers and essential public services. The local and regional economies in which these people live are depressed as a result of these conditions, with little consequent opportunity to generate non-farm income from salaries and self-employment.Out-migration has occurred widely in response to these conditions (Faiez and others 2004), but not all rural people are in a position to leave and create a better life elsewhere. Those left behindare often the least able to cope in such a harsh environment. 138. I t is difficult to achieve a sustainable increase in the incomes of poor households living in such conditions.Public transfers may be the most effective vehicle for poverty reductionfor many rural mountain people, especially those living in extreme poverty. Wider coverage of Targeted Social 66 Assistance programs among the extreme poor may provide a fiscally acceptable response, but it is only a partial solution. Measures to reduce isolation, improve local governance, and strengthen civil society are also needed to improve the environment for economic activity. The progress made on these issues under the EconomicDevelopmentand PovertyReductionProgram (EDPRP) shouldbe continuedin this regard. C.2. Limited Access to Agricultural Resources 139. The small size of farms that emerged in responseto land reform severely limits the ability of rural people to earn an adequate living from agriculture alone. The 2004 Agricultural Census reports that 71 percent of farm householdshave farms less than 1 hectare, 22 percent have farms of 1 hectare to 5 hectares, and only 1.4 percent have farms greater than 5 hectares.A further 5 percent are reportedwith no land at all. This land scarcity also limits the number of livestock that can be carried, and the ability to increase and diversify farm incomes. Poor access to credit and farm inputs further limit the ability of farmers to increase production from the land, livestock, and other resources they do have. The limited supply of farm machinery is less of a constraint given the predominance of small, subsistence-oriented farms and the abundant supply of rural labor. 140. These resourcelimitations are especially pronounced among the rural poor. More than one- third (36 percent) of poor rural households and one-halfof the extreme poor reportedno land use in 2007 (Table 3.10). Similarly, 53 percent of poor rural households and 69 percent of the extreme poor report no livestockownership. Table 3.10: Agriculture ResourceLimitations of Poor Rural Households Poor Extremely Poor `YORural Householdsnot Using Land 36.0 50.2 `YORural Households with No Livestock 52.5 69.1 Number of households 135,000 55,158 Source: World Bank estimates using 2007 LSMS data. 141. Without these basic agricultural resources, rural people inevitably struggle to survive. Non- farm income from salaries and self-employment can provide a viable alternative source of income, but these opportunitiesare typically less accessible for lower-incomehouseholds in rural areas, as discussed below. 142. A well-functioning land market is a double-edged sword when land is scarce, and access to land is a determinant of poverty. There is a clear need for farm size to increase in Georgia if agriculture is to provide a reasonable standard of living for farm families, and land sales are an obvious means to this end. But land sales by poor farmers can also result in a landless class of rural people, with even higher levels of poverty. These people lose their homes and their livelihoods, and are unable to find acceptable replacements for either. For older rural people an effective pension system together with a good price for land can reduce this risk, while facilitating farm enlargement. But in Georgia it will be some time before these preconditionsare in place.45 143. A well-functioning rental market for rural land is the best alternative under these circumstances,because it responds to the needs of both "buyers" and "sellers." The survey provides no evidence that such a market exists. Indeed the low reported levels of property income suggest the 45The Government plans to provide social assistance to all poor people, irrespective of age. The current old age pension system will be gradually phased out and replacedby employer/employeecontributions and savings schemes for the self-employed. 67 contrary-although this is unlikely in Georgiagiventhe scarcity of rural landand the lack of any obvious impedimentto land rentals.Informal landrental markets are generally ubiquitous in conditions like these. Further study of the links between landownership,land use, land markets, and rural poverty is neededto clarify this issue. C.3. Vulnerability to Price and Income Shocks 144. Agriculture provides an inherently variable source of income due to the vagaries of climate and prices. Rural people are typically adept at managing this variability, however, by diversifying household incomes (farm and non-farm income) and by using food reserves and livestock as a form of self-insurance.The effectivenessof these strategies varies according to the magnitude and frequency of income shocks, the extent to which householdincomes are diversified, and the availability of reservesfor use as self-insurance. 145. Rural households in Georgia face high levels of income variability, from diverse sources, and have a limited capacity to withstand this variability. Climate and market risks are both high in rural areas. Political instability adds a further source of risk as a result of deteriorating relations with Russia, and the political uncertainty faced by regions close to Abkhazia and South Ossetia. These influences are reflectedin the marked inter-annualvariation of agriculturalGDP. The capacityto manage this risk is severely weakened by low levels of food production(inadequate for food self-sufficiency)and the needto rely heavily on purchased food, the limited diversificationof farm and non-farmincomes, and low incomes from internal private transfers and remittances (Table 3.11). The high frequency and magnitude of income shocks also makes it difficult for poor rural households to accumulate reserves, increasing the likelihood that there are no reserves at all. For both poor and extremely poor rural households, the major source of income stability is publictransfers. Table 3.11: Indicators of Vulnerability amongRural Households Poor Extremely Poor Income Diversification 'YOhouseholds with livestock 47.5 30.9 % households with salary income 14.8 15.2 'YOhouseholds with self-employment income 13.5 5.5 % households with internal private transfers 3.3 2.8 'YOhouseholds with property income 0.4 0.5 %households with remittance income 1.8 0.5 'YOhouseholds with public transfers 71.9 69.6 Dependence on Purchased Food %of food purchased (household average) 73.7 75.2 Capital Reserves 'YOhouseholds with livestock 47.5 30.9 Source: World Bank estimates using2007 LSMS data. C.4. Limited Opportunities to Increase Household Incomes 146. While there is clear potential to increase farm production and incomes in the higher potential regions, many rural households lack the resources or incentivesto do so. Low-input, low- output subsistence production systems currently predominate (Table 3.12), because they are the least costly and least risky (but also the least rewarding).A shift to more commercial production systems, with a higher marketed output, would require farmers to increase the purchase of fertilizer and seed-with an increased consequent requirement for seasonal finance and increased exposure to risk. Commercial 68 production also requires increased participation in agricultural input and output markets-markets that remainweak due to Georgia's poor physicaland institutionalinfrastructure. Table 3.12: Subsistence Orientation of Poor Rural Households Poor Extremely Poor Percent HouseholdsSellingFarmProducts 13.9 10.1 FarmSales as Percentof fotal Farm 8.0 10.3 Income _..___..._.__.I_._...._...___..._.___._ " .- - -. - -,- Source: World Bank estimates using 2007 LSMS data 147. Opportunities to earn non-farm income are also restricted, as is its capacity to lift poor households out of poverty. Salaries remainthe principlesource of non-farm income, but they are earned by only 15 percent of poor households. The opportunities for self-employmentare even more restricted, especially for the extreme poor (Table3.12).This low level of SME activity is an economy-wideproblem that is now receiving increased attentionfrom donors and government. The impact of increased non-farm income on total household income and poverty may also be limited, given the low levels of non-farm income actually earned. Poor rural households reported average monthly non-farmincome earnings of 42 GEL per householdand extremelypoor rural householdsreported 22 GEL per household. D. Conclusionsand Recommendations D.1. Current Realities 148. Continued high levels of rural poverty in Georgia reflect the deep-seated problems that must be overcome if rural living standards are to improve. Most of Georgia's recent economic growth has been driven by construction, banking, and mining, and the benefits of this growth have accrued to Tbilisi and other major urban centers. Much less progress has been made in rural areas. A higher priority for rural development is critical, together with a realistic medium-termtime frame for achievingtangible results. The constraints to rural poverty reductionare not amenableto quick fixes. 149. Without a substantial, sustained program to reduce rural poverty, it will remain the major source of poverty in Georgia, and rural people will continueto be left behindas the economy grows. Measuresto reducerural poverty should respondto the following realities: The mountain areas where rural poverty is deepest and most pervasive are also the most difficult areas in which to achieve a sustainable increase in rural householdincomes. Increased agricultural output can make a significant difference to rural incomes and reduce poverty rates, but it will not be enough to lift all rural people out of poverty. Small farm size restricts the contribution that agriculture can make to household incomes, in all regions; the agricultural potential in mountain regions is very low; and more than a third of poor rural households use no agriculturalland at all. Measures to increase non-farm incomes are most likely to make a difference in areas where local economies are more vibrant and local governance is effective. The principal beneficiariesare thus likely to be the non-poor and the ruralpoor who are close to the poverty line. The impact of increased non-farmincome on rural povertymay be limited, nevertheless, even for these people. Opportunitiesto earn non-farm income are restrictedand the incomes earned are low. 69 0 Public transfers are likely to be the most effective means to raise and stabilizethe incomes of rural people in extreme poverty-particularly those inmountain areas. 0 The continuation of programs to strengthen rural land markets, improve rural governance, and strengthen agricultural markets will all contributeto rural povertyreduction. Further, economy-widemeasures to improve physical infrastructure, strengthen civil society and the rule of law, and promote SME development are also essential for rural poverty reduction. 0 Rural people pay a high price for Georgia's political instability through unexpected losses of markets and remittance income, the low quality of governance and civil society in areas adjacent to disputed territories, and the disincentivesthat these conditions create for business activity and investment. 0.2. Prioritiesfor Rural Poverty Reduction 150. These realities suggest that a multifaceted response to rural poverty is required, based on (a) measures targeted to different levels and sources of rural poverty, and (b) additional sector and economy- wide measuresthat will benefit all ruralpeople. D.2.1. FocusedInitiatives 0 Expand TargetedSocial Assistance (TSA) and ensure thatpeople in rural mountain areas are able to appb 15 1. The highestconcentrationof extreme poverty in rural areas occurs in the mountain regions of Shida Kartli, Kakheti, and Mtskheta-Mtianeti. These regions account for 63 percent of all extreme poor in rural areas. The households there are frequently landless, with few sources of income, and too poor to escape the physical isolation, harsh winters, and political uncertainty that depress the local economies in these regions. Income transfers targeted to these people on the basis of landownership and location could make a substantial difference to the level and stability of household incomes. Extended coverage of the current TSA program may provide a viable solution by reaching extreme poor concentrated in those areas. 0 Increase agricultural production in regions with higher agriculturalpotential 152. I n those areas where agro-climatic conditions are favorable for agriculture, there is considerablescope to raise productivity and output. Even with current productionsystems, substantial yield responses are possible with a modest increase in the use of good-quality seed, fertilizer and agricultural chemicals. This will require better access to credit and more active, efficient markets for agricultural inputs. Where infrastructure is poor and agricultural markets are weak, farmers are likely to use these inputs to raise the productionof staple food crops. With better, more accessible market outlets, producerswill be more likely to shift productionto higher-valuecash crops, at least in part. 153. Measures of this nature are likely to have the greatest impact on rural poverty for households close to the poverty line, particularly in western Georgia where rural poverty is high but underlying agricultural conditions for crop productionare favorable. The feasibility and potential benefits of equivalent measures to promote livestock production in eastern and northern Kakheti should also be examined. Ruralpoverty is very high in this traditional livestock-producingarea. 70 D.2.2. Cross-cutting Initiatives Supportfor SME development 154. Non-farm income is currently a small but valuable component of rural household revenue, which enhances and stabilizes overall household income. Salaries and wages account for most of this income, however, due to the weakness of SME activity. This weakness, despite the acknowledged potential of SME activity to generate employment and raise incomes, is a problem throughout Georgia and has been accorded a high priority in current donor programs. Rural areas should be included in initiatives to promote SME development, especially in regions where local economies are vibrant and there is evident demand from rural people for such support. Continuedsupportfor the expansion and rehabilitation of physical infrastructure 155. The ongoing rehabilitation and expansion of Georgia's road and port system is a precondition for sustainable growth in most commercial activity. These investments will help to link and integrate domestic markets, and improve access to export markets in Europe and the Middle East. Rural people will benefit through better links to domestic and export markets, increased opportunitiesfor non-farm income, and better access to public services in urban centers. 71 CHAPTERLABORMARKET 4: DEVELOPMENTS AND LINKAGESTO POVERTY IN GEORGIA A. Labor Market Dynamics 156. The analysis of the labor markets trends presented in this chapter46is based on the period 2003-2006, while the linkages between labor markets and poverty are explored using the most recent 2007 Living Standards Measurement Survey (LSMS) data to make the findings comparable with the key numbers presentedin the poverty profile (which is also based on the 2007 data). Years 2003-2006 is the period for which the comparable Household Budget Survey data are available. This is also the period that witnessed substantial reforms in the public administration and economic agenda following the "Rose Revolution" at the end of 2003. Hence, the 2003 data can be used effectively as a benchmark for the pre-reform state of the labor market. The 2007 LSMS data are used to explore the linkages betweenthe labor markets and poverty.47 157. The economic growth in Georgia during 2003-2006 happened against the background of declined employment.The economy of Georgia registered an average annual growth rate of more than 8 percent during 2003-2006.48 However, as is often the case in countries undergoing deep economic restructuring, many labor market indicators have worsened in Georgia. The absolute number of employed declined from the pre-reform level of 1.89 million people in 2003 to 1.71 million in 2006, or by 178,000 people (Table 4.1). A significant part of this decline is explained by downsizing of the public administration sector following public sector reform, and the declined employment in the agricultural sector. The absolute number ofjobs lost is quite evenly distributed between urban and rural areas. 158. Reforms since 2003 have pushed up the absolute number of unemployed. Consistent with the public administration reform and other economic developments, the absolute number of unemployed increasedby about 17 percent during 2003-2004, or 50,000 people (Table 4. However, during 2004- 2006, the number of unemployed declined by 10,000 people -a sign of the economy recovering from the immediate "adjustment" impact of the reform. Nevertheless, the number of unemployed in 2006 still exceeds that in 2003. The unemployment rate increased from 9.6 percent in 2003 to 12.4 percent in 2006. Taking into consideration individuals who have got discouraged looking for ajob, the unemployment rate in 2006 was 16 percent, compared to 12.4 percent in 2003. It is worth noting that the estimated 46 This chapter was prepared by Oleksiy Ivaschenko(Economist, ECSHD). 47 The 2007 LSMS data are much less rich for the labor market analysis than 2003-06 HBS data. For instance, the 2007 data do not provide the sectors of employment (only type of employment, such as wage employment, self- employed [in agriculture], and own business [non-agriculture]).Due to the differences in the reference period and the structure of the questionnaire (the Household Budget Survey [HBS] questionnaire is more detailed), the 2007 employmentiunemploymentaggregate numbers are also not comparable with the 2003-06 numbers. Hence, for the analysis ofthe labor markets dynamics, only 2003-2006 HBS data are used. 48 It is believed that not an insignificant portion o f gross domestic product (GDP) growth in Georgia over the last three years has been driven by the grey sector of the economy going legal as a result of many policy measures that have been implemented. 49 Unemploymentnumbers presentedin this chapter are basedon the two criteria. Unemployed (1) (strong criterion) = A personjobless in the last seven days, who has been lookingfor ajob duringthe last four weeks, and is ready to start working in the next seven days. Unemployed (2) (weak criterion) = Unemployed (strong criterion) + not looking for a job during the last four weeks because: (a) expectingjob interview result; (b) expecting to hear from employment service; (c) lost hope of finding ajob, but is ready to start working in the next seven days if offered a job. 72 unemploymentrate in Georgia is about 5.2 percentage points higher than the Organizationfor Economic Co-operation and Development (OECD) country average, and 2.9 percentage points higher than the European Union (EU)-19 average (Table 4.2). Interestingly, in Georgia, the unemploymentrate is higher among men.50 Table 4.1: Labor Market Participation, Employment, and Unemployment, 2003-2006 2003 2004 2005 2006 Total population age 16+ 3,112,062 3,175,986 3,078,237 3,055,045 Employed during last 3 months 1,885,872 1,856,394 1,764,035 1,707,095 Not employed during last 3 months 1,226,190 1,319,592 1,313,445 1,347,950 Unemployed (1) 200,385 250,807 242,350 241,273 Unemployed (2) 268,008 318,175 301,967 326,267 Total labor force (1) 2,086,257 2,107,201 2,006,385 1,948,368 Total labor force (2) 2,153,880 2,174,569 2,066,002 2,033,362 Participation rate (I), % 67.0 66.3 65.2 63.8 Participation rate (2), YO 69.2 68.5 67.1 66.6 Employment rate , % 60.6 58.5 57.3 55.9 Unemployment rate (I), % 9.6 11.9 12.1 12.4 Unemployment rate (2), % 12.4 14.6 14.6 16.0 Average monthly labor earnings (in 2006 prices) 165.1 174.1 192.3 209.6 Notes: Unemployed (1) (strong criterion) = A personjobless during the last 7 days, who has been looking for ajob during the last 4 weeks, and is ready to start working in the next 7 days. Unemployed (2) (weak criterion) = Unemployed (strong criterion) + not looking for ajob during the last 4 weeks because: (a) expectingjob interview result; (b) expecting to hear from employment service; (c) lost hope of finding ajob, but is ready to start working in the next 7 days ifoffered ajob. Source: World Bank estimates using2003-2006 Household Budget Survey data. 159. The downward adjustment in the number of employed has been followed by the simultaneous increase in average monthly earnings. While the employment numbers declined during 2003-2006, average real monthly earnings increased by an aggregate of 27 percent during 2003-2006 (or 9 percent per year) (Table 4.1). This increase, however, has been driven by a few sectors employingonly a fraction of the employed, including public sector employment. Monthly salaries of those in public administration (5 percent of total employment) who retained their job status after the public sector downsizing have doubled in real terms. Most of this increase happened during 2004-2005, when the public administration reformtook place. In contrast, average earnings in agriculture, the largest employer in the country (54 percent of total employment), increased by only 12 percent on aggregate during 2003- 2006, which is miniscule given a very low base-earnings in wage agriculture are 60 percent of the national average earnings, and earnings in self-subsistence agriculture5' are only 20 percent of the nationalaverage. 160. The labor force participation rate has shown a slight decline over the past few years. The data indicatethat the participationrate, a summary measureof labor supply, has declined from 67 percent in 2003 to 63.8 percent in 2006. This trend has been driven mostly by the decline in public administration 50Unemploymentand its major determinantsare discussed in a greaterdetail later inthe chapter. 51This takes into accountthe value of in-kindconsumptionand sale of agriculturalproducts. 73 employment.Inabsolute numbers, the total labor force in Georgiahas been about 2 million people (Table 4.1).52The participation rate in Georgia was about 5 percentage points lower than the average rates in OECD and EU-19 countries (Table 4.2). 161. There is a gap in the participation rates between men and women. The participation rate for men is 75 percent, while that one for women is 57 percent. This gap of 18 percentage points is comparable to the gap observed in OECD countries (Table 4.2). Despite significantly lower participation rates, women account for about 47 percent of the total labor force in Georgia, since the absolute number of working-age women is much higher relativeto that of working-age men-1.7 million compared to 1.5 million, respectively. Table 4.2: Participation, Employment, and Unemployment Rates, by Gender Georgia OECD EU-19 Participation rate (%) Female 56.6 60.1 62.2 Male 75.1 80.3 77.7 Total 65.2 70.1 69.9 Employment rate (%) Female 50.0 55.8 55.9 Male 65.8 75.0 71.2 Total 57.3 65.3 63.5 Unemployment rate (%) Female 11.7 7.2 10.1 Male 12.4 6.7 8.5 Total 12.1 6.9 9.2 Source: World Bank estimates using 2005 Household Budget Survey data. OECD and EU-19 numbers are from OECD Economic Outlook, 2005. B. Employment, Earnings, and Hours of Work 162. Rural areas account for almost two-thirds (65 percent) of the total employment in Georgia. The absolute numbers of employed in rural and urban areas are 1.1 million and 0.6 million, respectively. In terms of the distribution across administrative regions, Imereti and the capital city of Tbilisi each account for about one-fifth of the total employment. The employment profile across main geographic areas and individual characteristics did not change in any significant way during 2003-2006 (Table 4.3). Hence, in discussingthe employment profile, we use the most recent 2006 data.53 52The referencepopulationare those age 16 and above. 53The 2007 LSMS data also show the composition of employment to be very similar to the 2006 composition, but with a somewhat smaller share of rural employment. 74 Table 4.3: Employment Profile, 2003 Compared to 2006 Characteristic Employed Empoyment Employed Eempoyment Location Urban 637,878 36.2 597,199 35.3 Rural 1,125,023 63.8 1,093,588 64.7 Region Kakheti 186,947 10.6 203,290 12.0 Tbilisi 303,650 17.2 290,134 17.2 Inner Kartli 141,778 8.0 129,251 7.6 Lower Kartli 198,735 11.3 205,480 12.2 Samtskhe Javakheti 103,018 5.8 109,140 6.5 Adjara 157,787 9.0 121,229 7.2 Guria 81,965 4.6 76,196 4.5 Samegrelo 178,565 10.1 168,058 9.9 lmereti 367,257 20.8 346,329 20.5 Mtskheta Mtianeti 43,199 2.5 41,680 2.5 Gender Female 832,867 47.2 795,738 47.1 Male 930,034 52.8 895,049 52.9 Age group 15-20 62,501 3.5 54,984 3.3 21-25 115,507 6.6 105,366 6.2 26-30 163,972 9.3 147,278 8.7 31-35 172,815 9.8 139,745 8.3 36-40 190,491 10.8 191,702 11.3 41-45 222,427 12.6 195,186 11.5 46-50 191,298 10.9 199,299 11.8 51-55 167,633 9.5 161,941 9.6 56-60 99,927 5.7 141,972 8.4 60+ 376,330 21.3 353,314 20.9 Education 4Secondary general 228,635 13.0 175,399 10.4 Secondary general 697,027 39.5 702,348 41.5 Vocational 145,454 8.3 155,384 9.2 Secondary special 224,717 12.7 217,629 12.9 University 460,431 26.1 438,228 25.9 Total 1,762,901 100.0 1,690,787 100.0 Source' World Bank estimates using 2003 and 2006 Household Budget Survey data. 163. Males have a higher share in total employment. Males account for 53 percent of total employment.The gender composition of employment did not change during 2003-2006 (Table 4.3). 164. The elderly population accounts for a substantial share of total employment. While the prime working-age population(26 to 60) contributes 69 percent to total employment, the population age 60 and over also accounts for a noticeable share-2 1 percent (Table 4.3). The substantial contribution of the elderly population to total employment is explained by the high life expectancy in Georgia combined with economic circumstancesthat force many peopleto work beyondtheir retirement age. 165. The bulk of employment is accounted for by people with secondary general education, who account for 42 percent of the total employment in Georgia. The next-largest group is people with university-level education, which accounts for 26 percent of the total employment. People with vocational and secondary special education follow closely with 22 percent of total employment. Those with below secondary general education make up only 10 percentoftotal employment(Table 4.3). 166. The private sector accounts for three-quarters of total employment. Most of the privatization process in Georgia, similar to other Commonwealth of Independent States (CIS) countries, happened in the 1990s.The share of the private sector has been around 75 percent(Table 4.4). 75 Table 4.4: Employment Composition,by Status of Company's Ownership (main job), 2003 Compared to 2006 Status of Company's Ownership 2003 2006 Public organization financed from budget 17.4 18.3 Public organization with private funding 6.0 6.1 Private organization without foreign share 74.2 74.3 Private organization with foreign share 2.2 1.2 Other 0.1 0.1 Total 100.0 100.0 Source: World Bank estimates using 2003 and 2006 Household Budget Survey data. 167. Agriculture remains the dominant sector of employment. The sector has historically been the major contributor to total employment, and continues to account for 54 percent of total employment. While the share of agricultural employment did not change during 2003-2006, the absolute number of employed in this sector declined by 95,000 people. Services account for about 17 percent of total employment. The public administration, education, and health sectors combined contribute about 14 percent to total employment.They lost about 34,000 jobs during 2003-2006. Manufacturingaccounts for about 5 percent of the total, which lost 12,000 jobs since 2003. Construction gained 26,000 jobs during 2003-2006, and accounts for 4 percent in total employment(Table4.5). Table 4.5: Employment, by Sector (main job), 2003 Compared to 2006 2003 2006 2003-2006 change N of N of N of % Sectorof employment Employed % Employed % Employed Change Agriculture,forestry, fishing 1,013,750 53.7 918,519 53.7 -95,231 -9.4 Mining and quarrying 21,051 1.1 28,213 1.7 7,162 34.0 Manufacturing 98,306 5.2 86,459 5.1 -11,847 -12,l Electricity, gas, water supply 17,522 0.9 18,304 1.1 782 4.5 Construction 46,919 2.5 73,257 4.3 26,338 56.1 Wholesaleand retailtrade, repairs 205,603 10.9 161,432 9.4 -44,171 -21.5 Hotelsand restaurants 19,901 1.1 15,808 0.9 -4,093 -20.6 Transport,storage and communications 75,525 4.0 72,888 4.3 -2,637 -3.5 Financialintermediation 12,497 0.7 13,943 0.8 1,446 11.6 R&D, Computer and relatedactivities 34,988 1.9 20,352 1.2 -14,636 -41.8 Public administration and defence;compulsory social security 97,889 5.2 74,745 4.4 -23,144 -23.6 Education 136,300 7.2 128,346 7.5 -7,954 -5.8 Healthand social work 46,824 2.5 43,634 2.6 -3,190 -6.8 Community and socialservices 51,601 2.7 41,712 2.4 -9,889 -19.2 Private households as employers 4,568 0.2 10,112 0.6 5,544 121.4 Extraterritoria organizat ons 4,412 0.2 1,232 0.1 -3,180 -72.1 Total 1,887,656 100.0 1,708,956 100.0 -178,700 -9.5 Source; World Bank estimates using 2003 and 2006 Houset d Budget Survey ( ta. 168. Nationwide, average real monthly earnings increased significantly-by about 27 percent- during 2003-2006. The largest increase in earnings was registered in the financial intermediation and information technology (IT) sectors-by 162 percent and 110 percent, respectively (Table 4.6). A 76 substantial increase in real earnings has also been witnessed by those employed in public administration, education, and health, as the wages of public employees have risen as part of the reform of public administration. However, wages of those employed in agriculture increased by only 12.4 percent during 2003-2006. In addition to low productivity, wages in agriculture have been affected by adverse weather shocks (2004 drought) and by trade sanctions imposedby Russianauthorities (2006). Table 4.6: Real Monthly Earnings, by Sector (main job), 2003-2006 Average Total Annual Change, Growth 2003 2006 YO Rate, % Sector of employment Agriculture, forestry, fishing 112.3 126.2 12.4 4.1 Mining and quarrying 145.6 126.7 -13.0 -4.3 Manufacturing 172.1 188.0 9.2 3.1 Electricity,gas, water supply 162.8 197.7 21.4 7.1 Construction 290.6 245.9 -15.4 -5.1 Wholesale and retail trade, repairs 186.9 205.5 10.0 3.3 Hotels and restaurants 239.5 145.2 -39.4 -13.1 Transport, storage and communications 226.3 248.2 9.7 3.2 Financial intermediation 216.4 566.9 162.0 54.0 R&D, Computer and related activities 114.5 240.4 110.0 36.7 Public administration and defense: compulsory social security 145.8 296.7 103.6 34.5 Education 70.1 101.9 45.3 15.1 Health and socialwork 77.7 125.3 61.2 20.4 Community and social services 130.8 161.4 23.4 7.8 Private householdsas emdovers 137.2 162.2 18.2 6.I Extraterritorialorganizations 667.4 635.3 -4.8 -1.6 Total 159.9 197.6 23.6 7.9 Note; All earnings are expressedin 2006 prices. Source: World Bank estimates using 2003 and 2006 HouseholdBudget Survey data. 169. Earnings inequality appears to be quite high. The estimated national Gini coefficient of earnings is 44 percent (Table 4.7a), which places Georgia among high-inequalitycountries in transiti01-1.~~ This level of earnings inequality appears to have been stable over the period of analysis. There is a substantialgap betweenthe top decile of income earners and the bottom decile.The poorest person in the top decile earns almost 10 times as muchas the richestperson in the poorest decile. The richest person in the bottom decile also earns only one-thirdof the medianearnings. 170. There is a substantial differential in earnings between the private and public sectors, but the sector gap narrowed during 2003-2006. In 2003, average monthly earnings in the private sector were 50 percent higher than those in the public sector. However, as the result of the proportionally higher increase in wages in the public sector driven by public administration reform, the gap between average earnings in the private sector and in the public sector narrowedto 29 percent (Table4.7a). As a reflection of the public sector catching up with the private sector in the level of earnings, the importance of the between inequality in the explanation of the overall inequality declined during 2003-2006 (Table 4.7b). Inequality in earnings within the public sector seems to be comparable to that in the private sector. In 2006 the ratiosofthe top to bottomdeciles were 8 and 10 in the privateand public sectors, respectively. 54See Rutkowski and others (2005) for inequality estimates for other countries. 77 171. There is a noticeable gap in earnings between the urban and rural areas. Income earners in urban areas are getting paid on average 43 percent more than those in rural areas (Table 4.7a). This gap in earnings between the urban and rural areas remained unchanged during2003-2006. The actual gap would be even higher considering that many people in rural areas depend on subsistence agriculture, and do not have any cash income. Despite significant differences in the mean earnings between urban and rural areas, most o f the overall inequality is still explained by the inequality within urban and rural areas (Table 4.7c). Table 4.7a: Inequality in the Earnings Distribution, 2003 Compared to 2006 Mean sfd. err. Gini std. err. p90/p10 p90/p50 pIO/p50 2003 National 166.0 3.83 0.444 0,008 9.67 2.90 0.30 Public 129.5 5.04 0.450 0.017 7.14 2.78 0.39 Private 194.7 5.50 0.414 0.012 7.50 2.50 0.33 Urban 183.2 5.43 0.439 0.012 8.57 3.00 0.35 Rural 128.2 4.65 0.427 0.017 6.67 2.56 0.39 2006 National 209.5 5.16 0.435 0.012 8.00 2.67 0.33 Public 177.3 6.53 0.422 0.014 10.00 2.90 0.29 Private 229.0 7.24 0.435 0.016 9.00 2.81 0.31 Urban 233.5 7.80 0.435 0.014 9.20 2.88 0.31 Rural 163.2 5.24 0.405 0.012 7.50 2.31 0.31 Note: All earnings are expressed in 2006 prices; p9Oip10, p90ip50 and pIOip50 refer to percentiles ofthe earnings' distribution Source: World Bank estimates using 2003 and 2006 Household Budget Survey data. Table 4.7b: Decompositionof the Earnings Inequality, by PublicD'rivate Sectors, 2003-2006 GE(0) GE(1) GE(2) Overall Inequality 2003 0.348 0.350 0.517 Urban 0.349 0.376 0.655 Rural 0.302 0.301 0.412 2006 0.343 0.346 0.555 Urban 0.322 0.315 0.465 Within Group Inequality 2003 0.324 0.328 0.495 Between Group Inequality 2003 0.020 0.020 0.019 2006 0.008 0.007 0.007 Between Group Inequality as % of Overall Inequality 2003 6.2 6.0 3.9 2006 0.02 0.02 0.02 Note: All earnings are expressed in 2006 prices; GE refersto the General Entropy inequality index. Source: World Bank estimates using 2003 and 2006 HouseholdBudget Survey data. 78 Table 4.7~:Decompositionof EarningsInequality,by UrbanmuralAreas, 2003 Comparedto 2006 GE(0) GE(1) GE(21 Overall Inequality 2003 0.348 0.350 0.517 Urban 0.343 0.339 0.479 Rural 0.317 0.335 0.559 2006 0.343 0.346 0.555 Urban 0.344 0.346 0.553 Within Group Inequality 2003 0.335 0.338 0.505 2006 0.329 0.333 0.542 Between Group Inequality 2003 0.013 0.012 0.012 Between Group Inequality as % of Overall Inequality 2003 3.9 3.7 2.3 Note: All earnings are expressed in 2006 prices. Source: World Bank estimates using 2003 and 2006 Household Budget Survey data. a 172. There are significant differences in earnings across sectors of employment. The highest earnings are in the financial services sector, followed by construction and transportation (Table 4.5). Workers employed in construction and transportation earn twice as much as those employed in agriculture. In terms of average monthly earnings these sectors are followed closely by public administration.Of note, there is a large gap between the earnings in public administrationand those in the health and education sectors-average monthly earnings in health and education are only about 50 percent of earnings in public administration. Manufacturingis not as attractive in terms of pay as many other sectors, but it still pays about 40 percent more than agriculture(Figure4.1). Figure4.1: Average Monthly Earnings,by Sector (mainjob), 2006 0 25 50 75 100 125 150 175 200 225 250 275 300 Lari I Source: World Bank estimates using 2006 Household Budget Survey data. 79 173. The gender pay gap in Georgia is significant in both the private and public sectors. The regression analysis indicates that, controlling for other characteristics, males earn on average 84 percent more than females in the private sector, and 88 percent more than females in the public sector. Nevertheless, this finding has to be interpretedwith caution since the dependent variable in the model is monthly rather than hourly earning^.^' Because men tend to work longer hours, the identified gender pay gap would also capture the impact of this factor. Controlling for other characteristics, females in the private sector earn 19 percent more than females in the public sector. Males in the private sector earn 16 percent more than males in the public sector (Figure 4.2). Figure4.2: Average Monthly Earnings, by Gender, in the PrivateD'ublic Sector (main job), 2006 I "i , j:: 8 174. The educational premium in earnings is especially pronounced in the private sector, where people with university education earn on average 78 percent more compared to those with below secondary general education. In the public sector the premium on having a university education over below secondary general education is only 22 percent (Figure 4.3). Those with vocational and higher education earn on average 50 percent more in the private sector. The premium of having vocational education over below secondary general education is 20 percent in the private sector, while it is negative 18 percent in the public sector. The estimated premium in the private sector for each additional year of schooling of around 9 percent is generally consistent with the premium for higher education found in many other transitionalcountries (Yemtsov, Cnobloch, and Mete2006). 55Since the number of working hours per week is a categorical variable, it is impossible to construct a reliable measure of the hourly wage. 80 Figure 4.3: Average Monthly Earnings, by Education Level, in the PrivatePublic Sector (main job), - 2006 I I @ secondary pnvate sector vocational private sector lllpunivemity pnvatesector I I II< secondary public sector vocational public sector IuniversitvDublicsector Source; World1Bank estimates using 2006 Household Budget Survey data (based on the estimatedmonthly earnings equation). 175. More than a quarter of Georgia's workers are employed more than 41 hours a week in their main job. The actual share is even higher consideringthat 15 percent of workers report working varying hours dependingon a specific week (Table 4.8). About 40 percent of respondentsreport working 21 to 40 hours a week, and 13 percent report working less than 20 hours a week. 176. About one-tenth of Georgian workers are employed in at least two jobs. The analysis of the prevalence of the secondjob dependingon the hours worked in the mainjob indicates that it is the people working 21 to 41 hours per week that are more likely to have the secondary job (Table 4.8). The prevalence of the second job among those working less than 20 hours a week is lower than the average prevalence. This indicates that this group indeed consists mostly of people looking for part-timejobs (such as students), while those working 21 to 41 hours in their main job are likely to be involuntary underemployed, and thus are looking for supplementary jobs. The regression analysis indicates that the following categories of workers are more likely to have a secondary job: (a) age group 36 to 60, (b) males, (c) more educated, and (d) rural residents. Table 4.8: Hoursof Employment in the Main Job Compared to Having a SecondaryJob Number of Work Hours per YOof Total % Employed in % of Total % Employed in Week Employment Secondary Job Employment Secondary Job 20 15.7 5.1 12.9 6.0 21-30 16.6 11.1 12.3 8.4 31-35 12.7 8.6 10.3 12.1 36-41 17.8 10.8 17.8 11.5 42-50 14.9 8.8 15.6 6.4 51-60 5.8 5.8 6.4 2.5 60+ 5.9 4.7 4.4 5.1 varying hours 10.5 3.7 20.5 4.0 total 100.0 100.0 100.0 100.0 Source; World Bank estimates using 2003 and 2006 Household Budget Survey data. 81 177. The analysis of underemployment indicates that it is more educated people, females, and older people who are more likely to be underemployed.Rural residents also seem to face higher odds of being underemployed (Table 4.9). These underemployment risk factors are also evident from the regression analysis. Looking across the sectors of employment, the underemployment odds are highest in public education and health, financial services, and public utilities (Table 4.10). Underemployment chances are lowest for those employed in agriculture. Unfortunately, the proper analysis of underemployment in Georgia is constrained by the fact that the respondents in the survey have not been asked if they work more/less than normal working hours on a voluntary or involuntary basis.j6 The findings presented here are based on definingunderemployed (in the mainjob) as those working 21 to 41 hours inthe mainjob, and havinga secondaryjob. Table 4.9: Underemployment Profile, 2003 Compared to 2006 2003 2006 Number of Under- % of Total INumber of Under- % of Total Under- employment Under- Under- employment Under- Variable employed Rate, % employed employed Rate, % employed Location Urban 27,385 4.3 31.2 17,675 3.0 24.1 Rural 60,435 5.4 68.8 55,773 5.1 75.9 Region Kakheti 12,694 6.8 14.5 13,062 6.4 17.8 Tbilisi 7,761 2.6 8.8 4,509 1.6 6.1 Inner Kartli 5,016 3.5 5.7 4,002 3.1 5.4 Lower Kartli 6,459 3.3 7.4 4,439 2.2 6.0 Samtskhe Javakheti 6,786 6.6 7.7 1,489 1,4 2.0 Adjara 4,206 2.7 4.8 1,837 1.5 2.5 Guria 8,456 10.3 9.6 4,531 5.9 6.2 Samegrelo 6,957 3.9 7.9 14,339 8.5 19.5 lmereti 25,772 7.0 29.3 22,638 6.5 30.8 Mtskheta Mtianeti 3,713 8.6 4.2 2,602 6.2 3.5 Gender Female 45,261 5.4 51.5 43,998 5.5 59.9 Male 42,559 4.6 48.5 29,450 3.3 40.1 Age group 15-20 0.0 0.0 0.0 21-25 1,761 1.5 2.0 1,620 1.5 2.2 26-30 5,456 3.3 6.2 4,120 2.8 5.6 31-35 7,765 4.5 8.8 4,710 3.4 6.4 36-40 9,344 4.9 10.6 8,857 4.6 12.1 41-45 15,510 7.0 17.7 12,694 6.5 17.3 46-50 14,482 7.6 16.5 13,951 7.0 19.0 51-55 12,983 7.7 14.8 9,815 6.1 13.4 56-60 4,786 4.8 5.4 9,267 6.5 12.6 60t 15,733 4.2 17.9 8,414 2.4 11.5 Education < Secondarygeneral 3,243 1.4 3.7 970 0.6 1.3 Secondarygeneral 18,451 2.6 21.0 17,718 2.5 24.1 Vocational 9,917 6.8 11.3 9,094 5.9 12.4 Secondaryspecial 14,291 6.4 16.3 14,936 6.9 20.3 University 41,718 9.1 47.51 30,730 7.0 41.8 Total 87,820 5.0 100.01 73,448 4.3 100.0 Note: Underemployed (in mainjob) are defined as people working 21 to 35 hours a week in the mainjob and reporting having a secondaryjob (hence, in most cases it is not by choice that they do not work full time in their mainjob). Source: World Bank estimates using 2003 and 2006 Household Budget Survey data. 56People can be forced to work less than normal working hours because the alternative (more hours)jobs are not availableor are costly to find. Peoplecan be forced to work longerhoursbecauseof economic necessity. 82 Table 4.10: Underemployment,by Sector (mainjob), 2003 Comparedto 2006 2003 2006 Number of Under- % of Total Number of Under- % of Total Under- employment Under- Under- employment Under- employed Rate, % employed employed Rate, % employed Sector of Employment Agriculture, forestry, fishing 12,407 1.3 14.1 10,215 1. I 13.9 Mining and quarrying 817 5.1 0.9 1,393 5.2 1.9 Manufacturing 3,616 4.5 4.1 5,846 7.6 8.0 Electricity, gas, water supply 4,112 20.6 4.7 1,538 8.0 2.1 Construction 1,217 2.9 1.4 1,891 3.8 2.6 Wholesale and retail trade, repairs 9,059 4.7 10.3 7,520 4.8 10.2 Hotels and restaurants 999 6.2 1.1 440 2.7 0.6 Transport, storage, and communications 5,692 7.5 6.5 2,412 3.3 3.3 Financial intermediation 840 10.0 1.0 2,052 14.7 2.8 R&D, computer, and related activities 1,944 6.2 2.2 1,579 6.3 2.1 Public administration and defense; compulsory social security 10,137 11.2 11.5 4,287 5.5 5.8 Education 24,973 17.9 28.4 23,048 16.8 31.4 Health and social work 7,471 15.6 8.5 6,918 13.8 9.4 Community and social services 4,186 10.0 4.8 3,274 8.6 4.5 Private households as employers 0.0 0.0 Extraterritorial organizations 0.01 0.0 Total 87,820 5.0 1OO.Ol 73,448 5.0 100.0 Note: Underemployed (in mainjob) are defined as people working 21 to 35 hours a week in the mainjob and reporting having a secondaryjob (hence, in most cases it is not by choice that they do not work full time in their main.job). Source; World Bank estimates using 2003 and 2006 Household Budget Survey data. C. Characteristicsof Unemployment C.1. Who is Facing a Higher Risk of Unemploymentin Georgia? 178. Unemploymentrisk in Georgia is predominantlyaffectingthe youth. The age group 21 to 25 faces the highest rate of unemployment-30.8 percent (in 2006) compared to the average rate of 12.4 percent (Table 4.11). Interestingly,the unemploymentrate among those aged 15 to 20 is lower, reflecting the fact that in this age group there are many students who have a temporaryjob. In other words, the highest unemploymentrate among those aged 21 to 25 indicates difficulties in entering the labor market for those who have very limited or no work experience but who look for jobs appropriate to their educational background. Controlling for other individual characteristics, the risk of unemployment for those aged 21 to 25 is about 22 percent. The unemploymentrisk declines to about 17 percent for those aged 26 to 30, and to 11 percent for those aged 3 1 to 35. The risk of unemploymenttapers off further as age increases. 83 Table 4.11: Unemployment Profile, 2003 Compared to 2006 2003 I 2006 Numberof % of Total Number of % of Total Unemployed Unemployment Unemployed Unemployed Unemployment Unemployed Variable Rate,% Rate, % Location Urban 161,680 18.9 80.i 187,831 23.8 77.8 Rural 38,705 3.1 19.: 53 442 4.6 22.2 Region Kakheti 10,357 4.8 5.2 13,367 6.4 5.5 Tbilisi 100,788 23.6 50.2 117,200 28.3 48.6 Inner Kartli 8,109 5.1 4.1 13,116 8.3 5.4 Lower Kartli 10,241 4.5 5.1 17,072 8.0 7.1 Samtskhe Javakheti 1,944 1.8 1.1 2,548 2.3 1.1 Adjara 23,109 12.9 11,: 27,083 17.3 11.2 Guria 2,099 2.4 1.1 2,813 3.4 1.2 Samegrelo 9,384 4.5 4.i 6,122 3.4 2.5 lmereti 27,122 6.6 13.: 33,922 9.2 14.1 Mtskheta Mtianeti 7,232 12.3 3.6 8,030 14.3 3.3 Gender Female 101,314 10.3 50.6 104,268 11.5 43.2 Male 99,071 9.0 49.4 137,005 13.1 56.8 Age group 15-20 17,609 17.7 8.8 22,539 24.6 9.3 21-25 35,883 21.7 17.9 50,784 30.8 21.0 26-30 29,446 14.9 14.7 40,477 20.7 16.8 31-35 23,407 11.3 11.7 26,547 15.6 11.0 36-40 20,904 9.1 10.4 26,354 12.1 10.9 41-45 23,130 9.3 11.5 26,624 12.3 11.0 46-50 18,626 8.6 9.3 23,293 10.7 9.7 51-55 14,007 7.6 7.0 12,845 7.1 5.3 56-60 7,604 6.7 3.8 6,897 4.7 2.9 60+ 9,769 2 3 4 9 4,913 1 4 2 0 Education Secondary general 6,870 2 7 3 4 13,710 6 8 5 7 Secondarygeneral 66,463 8 1 332 71,917 9 1 29 8 Vocationallsecondary special 42,285 9 6 21 1 55,313 12 6 22 9 University 84,101 15 0 420 99,149 19 2 41 1 Total 200,385 9 6 1000 241,273 12 4 1000 179. Gender does not appear to be a crucial factor in defining the risk of unemployment. The unemployment rates among men and women are 13.1 percent and 11.5 percent, respectively(Table 4.11). Controlling for other individual characteristics, men and women face similar unemployment probabilities (Figure 4.4b). However, as discussed later, women are more likely to find themselves among unemployed ifthey havenopreviouswork experience or ifthey are long-termunemployed. 180. Unemployment rates are higher among more educated. The unemployment rate in Georgia increases proportionally with the education level-from 6.8 percent for people with below-secondary education, to 12.6 percent for people with vocational education, to 19.2 percent for people with higher education degrees (Table 4.11). However, controlling for other characteristics, the risk of unemployment does not vary much with the level of education (Annex 3, Table A.l). Women with a higher education education face a somewhat smaller risk of unemploymentcompared to the same category of men. Women 84 with vocational education face a somewhat higher risk of unemployment compared to men with similar education. 181. The "educational attainment" mismatch is a significant cause of unemployment, The educationalstructure of the employed seems to differ considerably from that of the pool of unemployed. In other words, there is a substantialmismatch in the education level between what the labor market is looking for and what those looking for jobs are able to offer (Table 4.12). The calculated index of educationmismatchwas equal to 16.7 percent in 2003, and it increasedto 17.6 percent in 2006.j7This is substantially higher than for other countries in the region, including neighboring Armenia (World Bank 2006). The estimated index indicates that 2.2 percentage points of the total unemployment rate of 12.4 percent in 2006 is attributableto the education mismatch. Table 4.12: Compositionof Employment and Unemployment,by Education Level, 2003 Compared to 2006 2003 % of Total N of % of Total Excess N of Employed Employed unemployed Unemployed supply (YO) Education < Secondary general 246,651 13.1 6,870 3.4 -9.7 Secondary general 758,151 40.3 66,463 33.3 -7.0 Vocationallsecondary ! 397,541 21.I 42,285 21.2 0.03 University 477,998 25.4 84,101 42.1 16.7 2006 % of Total N of % of Total Excess N of Employed Employed unemployed Unemployed supply (%) Education c Secondary general 188,332 11.0 13,710 5.7 -5.3 Secondary general 721,744 42.2 71,917 30.0 -12.3 Vocationallsecondary ! 383,611 22.4 55,313 23.0 0.6 University 416,052 24.3 99,149 41.3 17.0 Source: World Bank estimates using2003 and 2006 HouseholdBudget Survey data. 182. Geography is a substantial determinant of unemployment.The unemployment rate in urban areas is five times higherthan in rural areas-23.8 percent compared to 4.6 percent. Urbanunemployment accounts for 78 percent of the total unemploymentin the country. Of note, the capital city of Tbilisi has the highest unemploymentrate in the country and accounts for 49 percent of total unemployment.Adjara has the second-highest unemploymentrate. The lowest rates are observed in Samtskhe Javakheti, Guria, and Samegrelo (Table 4.11). Controlling for other variables, residence in Tbilisi is associated with a 22 percent risk of being unemployedcompared to a 16 percent risk on average in other urban areas. The risk of unemployment in rural areas is less than 3 percent. The low unemployment rate in rural areas is explainedby the rural population's relianceon subsistenceagriculture. Substantialregional dispersions in the unemploymentrate indicatelow labor mobility across regions of the country. C.2. Duration of Unemployment 183. The duration of unemployment in Georgia has a bipolar shape. The bulk of unemploymentis accounted for by those who try to enter the labor market for the first time, and by those who could not reenter the labor market for a long time. In 2006, individuals who never worked before accounted for about 30 percent of total unemploymentin Georgia.At the same time, 27 percent of people report having 57The educationmismatch(which can also be consideredthe skill mismatchifeducation is used as a proxy for skill) is the sum o f the "excess supply" for each level o f educational attainment, where "excess supply" is a positive number.Naturally,the sum of all positive andnegativenumbersshouldadd upto zero. 85 been unemployed (at least in the formal sector) for more than three years. The composition of unemploymentin terms of unemploymentduration did not change much during 2003-2006 despite solid rates of economic growth (Table 4.13). Table 4.13: UnemploymentDuration, 2003 Compared to 2006 2003 2006 N of o/o of 1otal N of % of Total Duration of Unemployment Unemployed Unemployed Unemployed Unemployed Im. 7,280 3.6 7,477 3.1 1-3 m. 10,703 5.3 16,684 6.9 3-12 m. 27,881 13.9 25,733 IO. 7 1-2 years 30,531 15.2 35,217 14.6 2-3 years 18,620 9.3 19,165 7.9 > 3 years 48,362 24.1 64,445 26.7 Never worked 57,008 28.4 72,552 30.I Total 200,385 100.0 241,273 100.0 Source; World Bank estimates using 2003 and 2006 Household Budget Survey data. 184. Younger people and/or people with unfinished education and/or no labor market specialization are dominating the group trying to enter the labor market for the first time. Of the total unemployed aged 21 to 25, about 66 percent try to enter the labor market for the first time. This statistic drops to 29 percent for those aged 3 1 to 35. Among unemployedwith nojob specialization,about 38 percent are those who seek to enter the labor market for the first time. The prevalence of no previous job experience is significantly higher among females-39 percent compared to 24 percent among males. There is also a noticeabledifferential between urban and rural residents in the percentage of unemployed with no previouswork experience-26 and 42 percent, respectively(Table 4.14). Table 4.14: Unemployment Durationvs. Various Individual Characteristics,2006 Number of Never Unemployed Unemployed 1 Unemployed Unemployed Worked, % < 12 m., % to 3 yrs., % > 3 yrs., % Location Urban 187.831 26.6 20.3 23.8 29.3 Rural 53,442 42.2 22.0 18.1 17.7 Gender Female 104,268 38.6 12.0 20.7 28.7 Male 137,005 23.6 27.3 24.0 25.2 Age group 15-20 22,539 66.4 22.4 6 9 4.2 21-25 50,784 63.2 10.1 17.5 9.2 26-30 40,477 34.3 19.3 28.2 18.2 31-35 26,547 29.5 20.5 27.4 22.5 36-40 26,354 9.0 25.0 19.4 46.6 41-50 49,917 2.8 28.0 28.7 40.6 51-60 19,742 0.0 27.1 25.4 47.5 60+ 4,913 0.0 10.8 16.6 72.5 Education < Secondarygeneral 13,710 44.4 27.6 16.8 11.2 Secondarygeneral 71,917 35.9 22.2 19.6 22.3 Vocational 16,638 21.3 30.9 16.4 31.5 Secondaryspecial University.. 38,675 18.1 13.8 24.4 43.7 99,149 30.3 18.6 26.1 24.9 Specialization No specialization 82,159 38.3 22.5 19.1 20.I Professional 96,716 29.8 18.5 26.7 25.0 Technician 44,705 22.5 14.6 25.7 37.3 Clerk, service worker 5,475 13.5 32.9 10.9 42.7 Craft worker 7,384 7.8 46.4 8.2 37.6 Plant/machineoperator 4,834 17.5 37.0 3.6 41.8 Total 241,273 30.I 20.7 22.5 26.7 Source; World Bank estimates using 2006 Household Budget Survey data. 86 185. Urban residents, women, older people, and individuals with specialized skills are more likely to report longer spells of unemployment.Twenty-nine percent of urban unemployedcomparedto 18 percent of rural unemployedreport havingbeenunemployedfor morethan three years, and 29 percent of women compared to 25 percent of men report being unemployed for more than three years. The chances of long-term unemployment increase noticeably after age 35. Among unemployed with secondary special education, 44 percent report a spell of unemploymentof more than three years, which is much higher than any other education group. For instance, this compares to only 22 percent among unemployedwith secondary general education (Table 4.14). D. Labor Markets and Poverty 186. Labor markets transmit growth to the poor when unemploymentand/or underemployment are reduced, and/or the labor earnings of the poor increase. The unemployed poor benefit from growth through increased employment, and the working poor, whose share in total unemployment is usually significant in low-income countries, gain from rising productivity and real earnings. Since the poverty profile is discussed in great detail in a separate chapter, we present here only a few key facts relatingpoverty and labormarkets. 187. The labor markets in Georgia since 2003 have seen many developments that are expected to be related to poverty. Real earnings generally increased during this period; however, the absolute number of employed declined by about 170,000 and the number of unemployed increased. The largest increase in realearnings happened in sectorsthat employ only a small fraction ofthe total employed, such as public administration (5 percent of total employment), financial services (0.8 percent), information technologies(1.5 percent), and healthand education (1 1 percent combined).At the same time, 54 percent of the employedare concentrated in low-productivity, mostly self-subsistence agriculture,where earnings (including the value of in-kind consumption) remain extremely low. Given these characterizationsof the labor market, it is not surprisingthat poverty in Georgia, especially in rural areas (where most of the poor are) remainsdeeply entrenched. 188. Employment in public administration/public services is associated with a lower risk of poverty now compared to three to four years ago. As mentioned, during 2003-2006 the highest increase in realearnings happened for those employed in public administratiodpublic services, while this sector also experienced substantial downsizing. Consistent with this increase in real earnings for those (still) employed in public administration after the public sector reform, we find that residence in households where the household heads are employed in public administration/educatiodhealth is associated with about a 3-percentage-pointlower risk of poverty in 2006 compared to 2003. 189. Employment status is significantly correlated with the incidence of poverty. Among the working-age poor population, 66 percent are inactive or unemployed. This indicates that the lack of employment is one of the maincauses of poverty.At the same time, working poor represent 34 percent of the working-age poor population.The share of working poor is much higher in rural areas41 percent compared to 26 percent in urbanareas (Table 4.15a).'* 190. A significant contribution to the total number of the poor is coming from those who are inactive in the labor market rather than unemployed. Inactive working-age individuals represent 28 58The profile of poverty by the employment status presented here is based on the labor data from the 2007 LSMS. In terms of the employmentiunemployment numbers these data are not comparable with the 2003-2006 HBS due to the differences in the questionnaire design. 87 percent of the total working-age poor, with respective numbers in urbanand rural areas of 33 percent and 25 percent (Table 4.15a). 191. Employment in self-subsistence agriculture is a poor remedy against the risk of poverty. Earnings in self-subsistence agriculture (accounting for the value of in-kind consumption) are extremely low-about 20 percent of median earnings in wage employment. Yet, self-subsistence agriculture accounts for 73 percent of the total employedin rural areas, As a result of low earnings in self-subsistence agriculture and a high concentration of labor there, the poverty incidence among individuals working in self-subsistenceagriculture is 22.6 percent, compared to 11.4percent for wage earners (Table 4.15b). Table 4.15a: Distribution of the Working-age Population, by Poverty and IndividualEmployment Status (shares of total employment, percent) Poor Non-Poor 2007 2007 Total Working-age Population Employed 34.5 50.9 Standard Error 0.9 0.6 Unemployed 29.4 19.1 Standard Error 0.9 0.5 Inactive 36.0 30.1 Standard Error 1.0 0.5 Working-age Population Living in Urban Areas Employed 25.9 42.3 Standard Error 1.5 0.8 Unemployed 34.8 22.2 Standard Error 1.6 0.7 Inactive 39.3 35.5 Standard Error 1.6 0.8 Working-Age Population Living in Rural Areas Employed 41.O 63.O Standard Error 1.2 0.8 Unemployed 25.4 14.6 Standard Error 1.0 0.6 Inactive 33.6 22.4 Standard Error 1.I 0.7 'ole: Poverty Line = 71.6 Lari (2007 prices). Source: World Bank estimates using the 2007 LSMS data. 88 Table 4.15b: Poverty Incidence, by Type of Employment and Urbanmural, 2007 2007 Wage empl. Urban 9.7 Standard Error 0.8Z Rural Standard Error z16.3 . 44 Total 11.4 Standard Error 0.7 1 Self-Empl (agr) Urban 12.5 Standard Error 2.89 Rural 22.6 Standard Error 0.85 Total 21.9 Standard Error 0.82 Own business (non-agr.) Urban 16.5 Standard Error 1.75 Rural 14.7 Standard Error 2.20 Total 16.0 Standard Error I.40 Other empl. Urban 18.4 Standard Error 4.98 Rural 25.4 Standard Error 5.87 Total 21.3 Standard Error 3.81 Total working age Urban 17.6 Standard Error 0.54 Rural 28.7 Standard Error 0.59 Total 22.6 Standard Error 0.40 Source: World Bank estimates usingthe 2007 LSMS data. E. Conclusions 192. During 2003-2006, real monthly earnings in Georgia increased in line with economic growth rates. Following the "Rose Revolution" at the end of 2003, and subsequent political and economic transformations in Georgia, average real monthly earnings increased at about 9 percent per year. Much of the observed increase in earnings has been driven by the increase in wages of public servants as part of the reform of the public apparatus. During 2004-2005 alone, the wages in public administrationincreasedby 77 percent-the largest rate of increaseacross all sectors. 193. The period of political and economic transformation that started in Georgia at the end of 2003 has yet to be accompanied by increasedemployment and/or reduction in unemployment.The absolute number of employed declined from the pre-reform level of 1.88 million people in 2003 to 1.71 million people in 2006, or by 178,000. This decline in absolute employment was driven mostly by the downsizing of the public administration sector following the public administration reform, and the reduction in agricultural employment. Reforms have also pushed up the absolute number of 89 unemployed-from 200,000 in 2003 to 240,000 in 2006. However, after unemploymentreached its peak in 2004, there registereda slight decline. 194. A gender gap is evident in many aspects of the Georgia labor market. Women's participation in the labor market is much lower than men's-57 percent compared to 75 percent. There is also a substantialestimated gap in monthly earnings between men and women. Unfortunately, the data do not allow analysis of to what extent this gap is attributableto the differences in hourly wages comparedto the differences in labor supply. Women are also much more likely than men to find themselves among unemployed if they have no previous work experience and/or if they have been unemployed for a prolongedperiodoftime. 195. There is a good deal of evidence suggesting that economic realities make people stay in the labor force longer and/or work longer hours. The elderly population (age 60 and above) accounts for 21 percent of total employment. Economic necessity coupled with longevity is the clear factor here. About one-tenth of Georgia's workers are employed in at least two jobs, and at least one-quarter of Georgia's workers are employed morethan 41 hours a week. 196. Earnings inequality is estimated to be quite high, and there is a substantial divide in the level of earnings between urban/rural areas and between the private/public sectors. The estimated Gini coefficient of monthly earnings is 45 percent. There are substantial differentials in earnings between urbanand rural areas and the private and public sectors. Incomeearners in urban areas are getting paid on average 43 percent more than those in rural areas. This gap in earnings betweenthe urban and rural areas remained unchanged during 2003-2006. The actual gap would be even higher considering that many people in rural areas depend on subsistence agriculture, and do not have any cash income. As a result of the proportionally higher increase in wages in the public sector driven by the public administration reform, the gap between average earnings in the private and public sectors narrowed from 50 percent in 2003 to 29 percent in 2006. There is an educationalpremium in earnings, which is especially pronounced in the privatesector. While agriculturecontinues to be the dominant sector of employment, it is also the sector with one of the lowestrates of pay. 197. Youth unemploymentseems to be of particular concern.The age group 21 to 25 faces the risk of unemployment in excess of 20 percent, even when other individual characteristicsare controlled for. Analysis of the duration of unemployment indicates that entering the labor market for the first time represents a particular problem for youth-66 percent of the total unemployed aged 21 to 25 are those who have no previous labor market experience. There appears to be no significant gender gap in the overall risk of unemployment. However, women are much more likely than men to find themselves unemployed if they have no previous work experience or are long-term unemployed. The risk of unemployment appears to be much higher in urban areas. Urban residents, women, older people, and individuals with specializedskills are more likely to report longer spells of unemployment.The chances of long-termunemploymentincreasenoticeablyafter age 35. 198. Unemployment rates in Georgia are higher among the more educated labor force. While in most transitional and developed countries more-educated people are generally better positionedto locate jobs, in Georgia this is not the case. The unemployment rate is 19 percent among higher education graduates, and 7 percent among individuals with less than secondary general education. The analysis indicates that a large pool of highly educated people in Georgia seems to contribute to the mismatch between the skills/educationthat the employers are looking for and the skills that the potential employees can offer. The estimates indicatethat 2.2 percentagepointsof the total unemployment rate of 12.4 percent in 2006 was attributableto the education mismatch. If not properly addressed, the existing imbalance in 90 the labor market will continueto contributeto high unemployment,degradationof skills, and incentives for labor to migrates5' 199. The labor markets in Georgia since 2003 have seen many developmentsthat are expected to be related to poverty. Real earnings generally increased during this period; however, the absolute number of employed declined by about 170,000 and the number of unemployed increased. The largest increase in realearnings happened in sectors that employ only a small fraction of the total employed, such as public administration(5 percent of total employment), financial services (0.8 percent), IT (1.5 percent), and health and education (1 1 percent combined). At the same time, 54 percent of the employed are concentrated in low-productivity, mostly self-subsistence agriculture,where earnings (including the value of in-kind consumption)remainextremely low. Given these characterizationsof the labor market, it is not surprisingthat poverty in Georgia, especially in rural areas (where most of the poor are) remains deeply entrenched. 200. Employment in public administration/public services is associated with a lower risk of poverty now compared with three to four years ago. As mentioned, during 2003-2006 the highest increase in realearnings happenedfor those employedin public administration/publicservices, while this sector also experienced substantial downsizing. Consistent with this increase in real earning for those (still) employed in public administration after the public sector reform, we find that residence in households where the household heads are employed in public administration/education/health is associatedwith about a 3-percentage-pointlower risk of poverty in 2006 comparedto 2003. 201. Low earnings and high concentration of labor in self-subsistence agriculture are the dominant drivers of poverty in rural areas. Earnings in self-subsistence agriculture(accountingfor the value of in-kind consumption) are extremely low-about 20 percent of median earnings ,in wage employment.Yet, self-subsistence agricultureaccounts for 73 percent of the total employedin rural areas. Those factors combinedexplain why the poverty incidenceamong individualsworking in self-subsistence agriculture is 22.6 percent, comparedto 11.4percent for wage earners. 59The analysis,based on the 2005 survey of return migrants in Georgia, indicatesthat: (a) most of the migration is circulatory rather then permanent; (b) there is a clear distinction in the migrationdestinationflows to CIS compared to non-CIS countries, driven by the level of educationhkills of the migrant; (c) most return migrants report the acquisitionof new skills while abroad;and (d) returnmigrantsgain in real earningscompared to their pre-migration level o f earnings(Ivaschenko2006). 91 CHAPTER 5: SOCIAL PROTECTIONAND POVERTY IN GEORGIA^^ A. Introduction 202. Georgia's system of social transfers reaches 57.8 percent of the population and includes pensions; assistance to internally displaced persons (IDPs); targeted social assistance (TSA); and subsidized energy consumption provided to certain categories of the population-pensioners, teachers, farmers, and so forth-on an ad hoc basis. In 2007, accordingto budget executiondata, Georgia allocated 4.1 percent of gross domestic product (GDP) to social transfers (for survey-based estimates see Annex 4, Table A.9). Their share of total budget spending was 13.2 percent. Relative to other countries in transition, and in particularto European countries, Georgia does not spend much on social protection.On average, in 2005, EU-15, EU-25, and EU-27 countries spent about 20 percent of their GDP on social protectionprograms. In the regionalcontext, Georgia's spending on social protectionis similar to that of Armenia, which in 2007 spent 4.8 percent of GDP on such programs. (For an overview of the social transfers in Georgia, see Annex 4, Box A.1). 203. Pensions are the largest social transfer in Georgia. About three-quarters (72 percent) of public spending on social transfers is allocated to pensions, and they are received by close to 850,000 individuals. Of all Georgian households, 55 percent reported pensions as their source of income. In April 2008, the average pension amounted to GEL 73 per month (about US$46), an 83 percent increase in nominalterms relative to April 2007. Despite this increase, pensions make up less than 15 percent of the average wage. In terms of the pensions' share in GDP, at 2.9 percent of GDP, Georgia is among the Europe and CentralAsia (ECA) countries with the lowest pensions/GDPspendingratios. 204. Social assistance to poor and vulnerable households comprises targeted social assistance (TSA) and several categorical benefits (to poor single pensioners, poor pensioner couples, orphans, disabled children, blind people, and families with seven or more children). Categoricalbenefits are closed for new entries and are expected to be gradually replaced by TSA. 205. TSA is aimed at providing income support and consumption smoothing among the very poor households in Georgia. It was launched in July 2006, after 18 months of intense preparation, including developing and testing a proxy means targeting mechanism, designing implementation procedures, establishing an agency, hiring and training of staff, developing an automated management informationsystem (MIS), and receivingapplicationsand collectingand processing informationon more than 200,000 applicant households from all over Georgia. The proxy means-testing mechanism was chosen as suitable for Georgia because of the high level of informality of economic activities, which makes income from formal sources an inaccurate indicator of household welfare. All Georgian households are entitled to apply for TSA. At the end of February 2008, there were 467,749 households (40 percent of the households) with 1,486,281 members (35 percent of the population) registered with the databaseofthe poor and vulnerablepopulation. 206. TSA is expected to gradually replace all categorical benefits. When TSA was launched, householdsreceivingcategoricalbenefits were given an option to register with the database, be tested and scored, and if they qualified, receive TSA, or to continue receivingtheir categoricalassistance. Since the launch of the TSA, about half the households receiving categorical benefits have chosen to make a transition to the new system. Administrative data indicate that, currently, about 3 1,000 Georgian householdsreceive categoricalsocial assistance. 60This chapter was prepared by Aleksandra Posarac (Lead Economist, ECSHD), Oleksiy Ivaschenko(Economist, ECSHD),and Ivan Khilko(Consultant, ECSHD). 92 207. Assistance to internally displaced people (IDPs): The civil strife at the beginningof the 1990s internally displaced more than 300,000 Georgians. Over time, some of the IDPs have left Georgia, some have died, and some have changed their status by marrying. Consequently, their number seems to have declined. According to the census conducted by the United Nations, there were about 220,000 IDPs in Georgia in 2003. All IDPs are entitled to receive some kind of assistance, irrespectiveof their well-being. In 2007, the Governmentspent GEL 32.5 million (0.2 percent of GDP) on benefitsto IDPs. 208. The energy subsidies are aimed at helping householdscope with the cost of energy. They are distributed on an ad hoc basis, usually at the beginningof winter, to certain categories of the population, including teachers, pensioners, and farmers. In 2007, Georgia spent GEL 79 million (almost 0.5 percent of GDP) on energy subsidies. The program is envisioned to continue in 2008, although with a somewhat decreasedbudget allocation. 209. This chapter is divided into five sections. Section 2 provides details about coverage of social protection (SP) programs, section 3 provides information about the targeting of SP programs, section 4 discusses the poverty impact of transfers, and section 5 provides conclusionsand recommendations. An annex provides tables detailing the results of calculations carried out to produce the results discussed. It also includesa box providing a detaileddescriptionofthe SP programs in Georgia. B. Coverage of Social Transfers 210. Social transfers reach a majority of the Georgian population. Almost 60 percent of the population report living in households receiving at least one social transfer. Pensions reach most of the population; 53.4 percent of individuals lived in households receivingpensions. They are followed by targeted social assistance(6.7 percent) and other social assistance programs (4.2 percent) (Annex 4, Table A.1). Figure 5.1: Coverage of Population by Targeted Social Assistance (By deciles; individuals are ranked by per adult equivalent pre-TSA consumption) I 1 2 3 4 5 6 7 8 9 10 Source: World Bank estimatesusing2007 LSMS. 211. Coverage of most programsis higher in the lower (poorer) deciles, although it varies across programs. For instance, in the case of TSA, 3 1.6 percent of the bottom pre-TSA consumptiondecile and 11.9 percent of individuals in the second-lowest pre-TSA consumptiondecile receive TSA; the coverage 93 then falls gradually, to 0.61 percent for the top decile (Figure 5.1). Pensions reach 88.3 percent of the first decile and 67.7 percent of the second decile of individuals ranked by their pre-pension consumption. Coverage of the top decile is much lower, at 33.6 percent (Figure 5.2). Figure5.2: Coverageof Population by Pensions (Individuals by deciles; ranked by per adult equivalent pre-pension consumption) 1 2 3 4 5 6 7 8 9 10 Source World Bank estimates using 2007 LSMS. 212. TSA is targeted well, but its coverage is limited. Relative to pensions, TSA covers a much smaller number of individuals-6.7 percent compared to 53.4 percent. Looking at the coverage of the poor and extreme poor population, TSA covers 30.4 percent of the extreme poor and 19 percent of poor population. Therefore, it falls short of reaching its objective of assisting all people living in extreme poverty. TSA is, however, very well targeted, as 70 percent of its recipients are pre-TSA consumption poor. In other words, the error of inclusion is 30 percent. This targeting performance puts Georgia's TSA amongthe best-performingsimilar programs in the world (Annex4, Table A.7). 213. TSA coverage is much better when one looks only at households that have applied to be registered with the database on poor and vulnerable households and to have their eligibility to receive TSA tested.6' All Georgian household are entitled to register with the database on poor and vulnerable householdsand to be tested. So far, 40 percent of all households representing about 35 percent of the population have registered with the database. TSA coverage of the population registered with the database was 17 percent. The coverage of extreme poor registered population was 42.7 percent and of poor registered population3 1 percent. 6'An intense public informationcampaign preceded introductionof TSA. TSA is a demand-basedprogram; that is, households self-select as potentially eligible. The self-selection process seems to perform quite well because the poverty incidence is much higher among the registered than the non-registered population. Overall poverty incidence among the registered population was 36.4 percent compared to 17.9 percent in the non-registered population.Respectivepercentages for extreme poverty were 16 percent and 5.5 percent. 94 C. The Incidence of Social Transfers 214. What share of program benefits accrue to poor and low-income households? Across all socialtransfers, 25.0 percent of program benefitsare received by the bottom 20 percentof the population; at the other end, 16.0 percent of transfers are received by the top 20 percent of the population(Annex 4, Table A.1). 215. TSA distribution is strongly pro-poor (Figure 5.3 and Annex 4, Table A.1). Two-thirds of TSA resources accrue to the bottom quintile. Therefore, 20 percent of the population receives 65 percent of resources allocated to TSA, resulting in a ratio between the population quintile and its share in the distributionof the TSA resources of almost 3.2, which is one of the highest among similar programs. The cumulativeTSA distributionshare reaches 81.1 percent for two bottompre-TSA consumptionquintiles. 216. The analysis indicates that the TSA does not crowd out private transfers (Annex 4, Box A.2). The results of regression discontinuity design (RDD) using2007 data fail to detect any statistically significant differences in the amounts of the private transfers between the TSA-recipient households and the householdswith the similar TSA score that receive no assistance. In particular,the average amount of private transfers among the non-recipients was 33.5 (with a standard error of 3.91) while the average amount of private transfers for the households receiving TSA was 30.2 (standard error of 6.85). This difference is statistically insignificantwith a t-ration of 0.46. The similar conclusions are made based on the multivariate analysis that controls for the differences in the observable characteristics of households from the two groups. 217. Distribution of other social transfers is less pro-poor, with other social assistance performing better than pensions. The bottom pre-social assistance quintile gets 40 percent of resources allocatedto other social assistance. For the bottomtwo quintiles, this percentage is 60 percent. While this i s not a poor performance, it could be improved to be more comparable to TSA. As far as distribution of pensions is concerned, respective bottom quintiles shares are 30 and 50 percent. Figure 5.3: Cumulative Distribution of Pension, TSA, and other Social Assistance 100 90 80 70 60 50 40 30 20 10 0 1 2 3 4 5 6 7 8 9 7 0 Deciles /+TSA ---I--- Pensions Other ~ Source: World Bank estimates using2007 LSMS 218. The pro-poor distribution of social transfers makes them an important source of livelihood for the poor. Across all programs and all beneficiaries, mean social protectioncash transfers comprise 12 percent of household consumption.However, this figure jumps to 37 percent for overall poor and to 55 95 percent for extreme poor, thus highlighting the importance of these transfers for the poorest segments of society. Pensionsand TSA both constitutea high share of householdconsumptionacross distribution,and particularly for the poor and extreme poor: pensions make up 28 percent of consumptionfor the poor and 38 percent for the extreme poor, while TSA makes up 7 percent and 13 percent, respectively (Annex 4, Table A.2). D. Poverty Impact of Social Transfers 219. Without social transfers, the extent of poverty in Georgia would be much higher. Without socialtransfers, the overall poverty incidence would increase from 23.6 to 35 percent (almost 50 percent); the poverty gap would more than double-from 7.2 percent to 16.9 percent-and the severity of poverty would increase almost five times-from 3.1 percent to 15 percent. The impact on extreme poverty would be even more dramatic-the incidence of extremely poor population would more than double, jumping from 9.3 percent to 20.3 percent; the extreme poverty gap would increase almost five times (from 2.4 percent to 16.0percent); and the poverty severity would increase 16 times-from 1 percent to 16 percent (Annex 4, Tables A.3-AS). 220. Pensions account for the lion's share of poverty reduction. The payments of pensions reduce poverty by about 9 percentage points in both cases, whereas TSA and other social assistance account for about a 1 percentage point reduction each. The impact of TSA, however, would be much higher ifthe program was expanded to cover more individuals, because coverage currently is relatively low. If TSA were to be expanded, the additional cost of bringing all of the extreme poor to the extreme poverty line would be about GEL 60 million, or between 0.4 and 0.6 percent of GDP. The lower estimate assumes perfect targeting, while the higher estimate allows for some error of inclusion and leakage of resources. The additional cost of bringing all poor to the overall poverty line would be about GEL 250 million, or 1.5 percent of GDP, assumingperfect targeting. With an error of inclusionof about 30 percent, additional resources requiredto eliminate poverty increase to 2 percent of GDP. In another scenario, to extend the current amount of TSA to all extreme poor and poor, the cost estimate ranges from 0.4 percent to 1.1 percent of GDP (Annex 4, Table A.8). 221. Inequality would be significantly higher in the absence of social transfers. The Gini coefficient of the per adult equivalent consumption increases from 36.3 to 40.9 when social transfers are removed from household consumption (Figure 5.4). Adding pensions decreases the Gini to 37.7 and further to 36.3 when all other transfers are added. Figure 5.4: Gini Coefficientsof PoDulationwith Various Social Transfers 42 0 41 0 40 0 39 0 38 0 37 0 36 0 35 0 34 0 33 0 No social transfers (STJ Pensions, but no other ST All social transfers Source; World Bank estimates using2007 LSMS 96 E. Conclusionsand Recommendations 222. Georgia's social protection system is extremely important for poverty reduction. Without social transfers, the overall poverty incidence would increase by almost 50 percent and the poverty would become very deep and severe. The impact on extreme povertywould be even more dramatic: one-fifth of the populationwould become extremely poor, the extreme poverty gap would increase almost five times, and poverty severitywould increase 16times. 223. Good poverty reduction performance indicates effective use of resources. For each percentage of GDP allocated to social transfers, the poverty incidence falls by 2.8 percentage points, the poverty gap decreases by 2.4 points, and poverty severitydeclines by 2.9 points. 224. Pensions deliver most of the poverty reduction performance of social transfers. Targeted social assistance (TSA) contributes to poverty reduction as well, but because of the small size of the program(0.4 percent of GDP, coveringonly one-fifth of the poor), the impact is much smaller. However, relative to its budget, TSA performs better than pensions in poverty reduction; for each point of GDP spent on TSA, poverty incidence declines by 3.5 points compared to 2.5 pointsfor pensions. 225. While Georgia's TSA program is among the best-performing similar programs in the world, its coverage is limited. Seventy percent of TSA beneficiaries are pre-TSA poor, indicating a reasonable error of inclusion of 30 percent. Almost two-thirds of all resources allocated to TSA are received by the bottom 20 percent of the population, ranked by their pre-TSA consumption. Yet, the program covers only 34 percent of its target group-extreme poor population.This low coverage is to a large extent driven by the modest resources allocatedto TSA. 226. During the next several years, Georgia should focus on eliminating extreme poverty. Covering the extreme poverty gap would require an additional 0.4 to 0.6 percent of GDP (to allow for some error of inclusion and leakage). The existing fiscal constraint could be relaxed by consolidating social assistanceto internally displaced persons (IDPs) into TSA, completingthe transition of other social assistance program into TSA more rapidly, and by integratingenergy subsidies with the TSA program.In addition to allocating more resources to TSA, the following is recommended: (a) fine-tune the proxy- means formula used to target TSA based on the findings of the 2007 Living Standards Measurement Survey, (b) recertify households registered with the database for more than two years, (c) launch an information campaign and outreach efforts to encourage more households to register with the database, (d) consider increasing the cutoff score, and (e) differentiate the amount of TSA based on the poverty depth (that is, households whose proxy score is further down from the cutoff score are awarded higher benefits). 227. Overall, given the importance of social transfers for the well-being of the Georgian population, it is important to keep resources allocated to finance them at least constant in real terms (eventual consolidation of programs should not have a negative impact on resources). Preferably, their share in GDP would remain at least constant, or even increased, so that less-fortunate individualscan benefitfrom Georgia's strong growth, as well. 97 CHAPTERHEALTH 6: REFORM, HEALTH OUTCOMES, AND POVERTY A. Introduction62 228. Health issues loom large in the lives of Georgia's poor. When asked to identify the main problemsfaced by their family, two-thirds of the poor mentionedthe purchase of medicine, and over half notedmedical services (Figure 6.1). These rates are substantially higher than those reportedby the better- off, and they place health concerns on a par with food security and employment as the most important problems faced by Georgia's poor. Similarly, when the Life in Transition Survey asked Georgian households to name the top two priorities for government investment, the health sector was the most common answer, including among the poor (Figure 6.2). Health should thus be front and center in discussions about the scope for public policy to improvethe lives ofthe poor in Georgia. 229. The relationship between health and poverty is varied and complex. Poverty makes people more vulnerable to ill health, and leaves them less able to cope and get better if they do fall sick. Causality also runs in the opposite direction, as poor health can lead to impoverishmentif it causes lost income or high medical bills. In keepingwith the World Bank's recently updated Health, Nutrition, and Populationstrategy, this chapter will focus on issues relatedto two strategic objectives: (a) improvingthe level and distribution of health outcomes among the population, especially the vulnerable; and (b) improving financial protection (preventing poverty due to ill health).63Policy issues related to both will also be discussed, with special reference to the recently introduced Medical Assistance Program (MAP) for the poor. 230. I n general the chapter offers a "mid-reform snapshot" of health and poverty in Georgia in 2007. It is a "snapshot" because it draws largely on two household surveys conducted in the middle of that year, and in the absence of longitudinal data, it is not possibleto disaggregate nationwide trends in recent years by socioeconomic status.64It is "mid-reform" because the timing of the surveys implies that they capture a period after MAP was introduced (albeit with two different eligibility criteria and benefits packages), before the MAP voucher program was launched (in late 2007), before the Universal Benefit Package (UBP) is eliminated (expected around 2009), and during which all stakeholders are adjustingto the new policy regime (population knowledge is growing, the private sector response shifting, and government stewardship evolving). These reforms will be described further in the next section. In the midst of all these moving parts, more time and data will be requiredto reachclear conclusions. 23 1. This chapter is structured as follows. Section 2 providesan overview of recent trends in health outcomes in Georgia in an international context, and a summary of the rapid reforms to the health sector currently underway in the country. Section 3 focuses on the distribution of health outcomes and their determinants amongthe poor and non-poor in Georgia, includingindicatorsrelatedto self-assessed health status, access, utilization, and satisfaction with health services. Section 4 takes a closer look at financial protection, and in particularthe link between out-of-pocketspendingand poverty. Section 5 discussesthe 62This chapter was preparedby Owen Smith(Economist, ECSHD). 63 See World Bank (2007). Two other strategic objectives identified in this document, the health sector's contributionto fiscal sustainabilityand goodgovernance, are not addressed here. 64The surveys are the LivingStandardsMeasurement Survey (LSMS)usedelsewhere inthis report,and the Health Utilizationand ExpenditureSurvey (HUES) describedbelow. The Lifein TransitionSurvey (LiTS) is also used. 98 potential contributionof MAP to the respective objectives. Section6 summarizes and suggests key issues for policy consideration. Figure6.1: Most CommonProblems Faced by Households, by Quintile 80% 70% 60% v) 50% % 0 'f 40% v) m ap .5 30% 20% 10% 0% Source:LSMS Purchase Of edicine Medicalservices Hungerlunderr irishrnent Employment , I p r e 6.2: Top Prioritiesfor GovernmentInvestment Identifiedby GeorgianHouseholds (% households listing sector as one of top two priorities, by quintile) 80% 70% 60% 50% 8 40% 30% 20% 10% 0% jource L,TS Education Health Care Housing Pensions Environment Infrastructure Othei 99 B. RecentTrends and Reform Initiatives B.1. Recent Trends 232. Recent trends in Georgia's health indicators point to steady improvements. Table 6.1 shows the evolution of key health outcomes during 2000-2006. Life expectancy has increased and infant and maternal morality rates have declined. The number of maternal deaths in particular fell sharply between 2004 and 2006, but since it is a rare event, these numbers are very small (correspondingto a decline from 21 deaths to 11 deaths) and subject to measurement challenges, and thus it is difficult to automatically attribute this decline to specific changes in poverty or health system performance. In general, health outcomes reflect a wide array of determinants and tend to evolve more slowly than economic indicators, but the trend in Georgiais inthe right direction. Future improvementsmay also be forthcoming in view of the increased rates of healthcare utilization, also shown inthe table. 233. Although the international focus is on maternal and child health and the Millennium Development Goals (MDGs), over 95 percent of mortality in Georgia is due to non-communicable disease and injuries. Cardiovasculardisease is by far the leadingcause of death. Key risk factors include high blood pressurefor both genders, tobacco use for men, and high body mass index for women (WHO- Euro 2005). An alternative measure of the disease burden is to calculate "excess deaths," that is, the number of deaths from different causes that could be avoided if Georgia had the same mortality rates as those that prevail in Western Europe (EU-15). For example, the correspondingfigures in Georgia are 9 avoidable deaths each year due to maternalmortality, about 650 due to infantmortality, and nearly 15,000 due to diseases of the circulatory system. This provides a strong indication of where the major gains in health status are to be found in Georgia. Table 6.1: Georgia: Recent Trends in Health Indicators -Indicator _ _ _- . . - 2000 2001 2002 2003 2005 2006 I 2004 71.3 71.5 71.5 72.0 7i.4- 73.1 n.a. Health Infant Mortality Rate 22.6 22.9 23.8 24.8 23.8 19.7 n.a. outcome (per 1,000 live births) indicators MaternalMortality 49.2 58.7 45.1 49.8 45.3 23.4 23.0 (per 100,000 live births) Health Outpatient Contacts 1.4 1.5 1.6 1.8 2.0 2.1 2.2 system (per personper year) indicators Source WHOiEurope Health for AI1 database 234. Georgia's health outcome indicatorsare generally better than those prevailing elsewhere in the Commonwealth of Independent States (CIS) but fall short of those in the new European Union (EU) member states. Table 6.2 compares health outcomes and health system performanceindicators in Georgia and other European countries. With the exception of infant mortality, which remains relatively high, all other healthoutcomes fall betweenthose achieved in the CIS and EU-12. With respect to health system performance, however, Georgia lags far behind all regional groupings, with a low outpatient contact rate and high out-of-pocketspending. These issues will be discussed extensively in later sections. Overall, the combination of better health outcomes than in the CIS but worse system performance indicators is likely at least in part to be a reflection of the critical role of behavioral factors (for example, alcohol consumption or diet) in determining population health. It should not cast doubt on the considerablescope for further healthsystem strengthening. 100 Table 6.2: Georgia and Region: Selected Health..... __-_ . ..... Indicators, Latest Available Year . ........ ._._....._..__..._...-..-..........-..........I. Indicator Georgia EU-15 EU-12 CIS .... Lifeexpectancy................. ................. _..............I .......73.1.... iY!.-E!!!Y!Pel....... iE.:.E!!O.Pl.-, ............. 79.7 74.0 67.0 Infantmortality rate 19.7 4.3 8.3 13.4 Health (per 1,000 live births) outcome Maternalmortality 23.0 5.3 9.0 28.2 indicators (per 100,000 live births) Mortality due to diseases of 545.1 238.7 523.4 773.0 the circulatory system (per 100,000) Outpatientcontacts 2.2 6.5 7.8 8.6 Health (per personper year) system Out-of-pocket payment on 72.1" 16.0 27.2 37.4 indicators health (% of total health expenditure) Source: WHOEurope Health for All database, except = GeorgiaNational Health Accounts, 2006. * B.2. Recent Reform Initiatives 235. From the end of the Soviet Semashko model of state health care in 1990 until 2005, the Georgian health sector was characterized by severe underfunding by the state, resulting in high out-of-pocket payments for health care and deteriorating infrastructure. A number of attempts at reform during this period (including experimentation with payroll-funded social insurance) did not fundamentally alter this picture. The general population is eligible for the Universal or Basic Benefits Package (UBP or BBP), which offers a limited benefitspackage with relatively high co-payments. In its current form it provides limited coverage for primary health care (PHC) services, a co-pay of 25 percent on emergency care, no coverage of planned inpatientcare, and six days' coverage of urgent inpatientcare. 236. An ambitious health reform plan was launched in 2006. In the context of broader efforts to reduce the overall tax burden, and hence with limited public finances to spend on health, the thrust of the new reform initiative has beento promote greater private sector involvement in both finance and delivery and to target public fundingto the poor. 237. The most significant reform in terms of poverty implications is the implementation of the Medical Assistance Program (MAP)for the poor, introduced in June 2006. The MAP offers a free, extensive benefits package financed through general tax revenues and targeted to the poorest segment of the population. The target group is identified by the same proxy means test carried out by the State Agency for Employment and Social Assistance (SAESA) for the Targeted Social Assistance (TSA) program, whereby each applicant is given a score based on a list of variables related to the household's welfare. There are currently about 690,000 individuals eligible to receive MAP benefits. A total MAP budget of GEL 80 million (about 0.4 percent of gross domestic product [GDP]) was plannedfor 2008. 238. Despite its short life span, the MAP has already rapidly evolved. Three different benefits packages have been offered (an initial one from June to December 2006, a less generous one from January 2007 to September 2007, and a new, most generous package from September 2007 to the present). The most recent version covers all primary care and urgent inpatient care (without limit), and planned inpatient care up to 15,000 GEL annually. Pharmaceuticals and dental care are excluded, in addition to a short negative list of other excluded services. Also, two different thresholds for eligibility have been implemented(a cutoff score of 100,000 up until December 2006, and a score of 70,000 since then, with the exceptionof Adjara regionthat has maintaineda cutoff of 100,000 throughout). 101 239. The most significant change in MAP since its launch has been the new role of private insurance companies. Until September 2007, MAP was administered by the State United Social Insurance Fund(SUSIF), now renamedthe Healthand Social Programs Agency (HeSPA). These agencies reimbursed providers on a contracted fee-for-service basis. Beginning in September 2007, the responsibility for provider payment in the Tbilisi and Imereti regions shifted to private insurers, with other regions following suit between April and October 2008. Insurance companies are being contracted by HeSPA to purchase health care from providers on behalf of eligible households based on a capitation system (insurers in turn generally use case-mix methods for contracting providers). These capitation payments are beingrevisedupwardin 2008 to GEL 9.2 per monthfor ages 0 to 64 and GEL 15 per month for those aged 65 and over. It is hoped that private insurers will be able to achieve greater efficiency in health care purchasing than a public agency. In effect, the reform entails a transition from a single, passive state purchaserto multiple, active payers inthe healthservices market. 240. A key concern with private insurance competition is risk selection to attract a healthier and therefore less expensive risk pool, including by denial of coverage to less healthy individuals. However, a number of design features are intendedto avoid this outcome. Insurers are prohibited from turning away any households that qualify for MAP, and they are required to offer the publicly defined benefits package across a minimum network of providers.The risk adjustment for the elderly should also help in this regard. As always, however, proactive implementation of these safeguards will be key to success. Four insurance companies were initially active in contractingfor MAP, but the largest recently signaled it may withdraw due to the difficulty of making a profit. It will be important to monitor these developments closely. 241. Those who do not qualify for MAP fall into two broad categories. First, the government is in the process of expanding state funding to provide health benefits to about 400,000 civil servants, includingteachers, police, and military personnel. Private insurers are also beingcontractedto participate in this plan, but under different terms from MAP. The rest of the population (about 3.3 million) are potential beneficiaries of the narrower UBP,65financed through the state budget, but there are plans to eliminate this funding in the near future, leavingthese individualsto pursue coverage independentlyfrom the private insurancemarket.This would generate substantial budgetary savings, but at the risk of leaving a large share of the Georgian populationwith no coverage at all. Policy issues relatedto the full package of healthfinancing reformswill be discussed further in a later section. 242. Reforms to privatize health care provision are also under way. The hospital sector suffers from excesscapacityand is in need of reinvestment, and thus a hospitalmaster planwas developed which envisions 100 new privatizedhospitalswith a bed capacity of 7,800 (compared to 17,000 now, but with a 32 percent occupancy rate). The tendering and contracting process of hospital investment packages has begun, albeit slowly. Investors are required to follow the master plan with regard to the various departments and services to be available in each hospital, with only minor changes allowed. There are significant potential risks including substandard quality and/or delays in construction, and weak accountabilitymechanisms, especially because a secondary market for these hospital investmentpackages is expected to emerge (as original investors look to subcontract construction). The government expects that new hospitalswill be completewithin two to three years. About five hospitals are planned to remain in public hands. 243. The Government also made a decision to privatize the primary health care (PHC) facilities beginning in late 2007. The approach to PHC privatization is divided into urban and rural components. In urban areas all PHC facilities will be sold, and it is expected that market competition will help reduce the current overcapacity of urban PHC facilities identified in the PHC master plan. For the rural 65There are also some small, universal disease-specific programs(for example, HIVIAIDS). 102 component, where private investment is unlikely and could pose a risk to population access, the government has decided to provide rural physicianswith grants. These one-time payments can be worth up to GEL 7,000 per practice and are intended for rehabilitation of facilities, purchase of medical equipment, and training in family medicine.Originally there had been a plan requiringprovidersto sign a contract agreeing to provide PHC services in the stipulated location for at least seven years, but this has yet to be implemented.The grant program has been rapidly rolled out in 2008. About 900 rural practices nationwideare foreseen accordingto the master plan. 244. I n sum, Georgia is undergoinga period of rapid market-oriented health sector reform. It is too early to reachany clear verdict on this ongoingprocess, but it will clearly have important implications for poverty and health. Careful implementationof safeguards to protect equity, and close monitoring of results, will improvechances for success. C. Health Outcomes 245. The main objective of any health system is to improve health outcomes among the population, and since we are interested in equity, this means not only the level but also the distribution of outcomes. This is also the first strategic objective of the World Bank's recently updated Health,Nutrition and Populationstrategy. Healthoutcomes can of course be measuredin many ways. The previous section showed recent trends in infant and maternal mortality (both MDGs), but unfortunately only nationwide averages are available for these indicators, because the most commonly used survey for analyzingequity of MDG outcomes, the Demographic and Health Survey (DHS), has not been conducted in Georgia. Information on differences in life expectancy between the poor and non-poor is also lacking. But we care about morbidity as well as mortality, and here there is evidence of a clear health gradient in Georgia.This sectionoffers a snapshot of equity in healthoutcomes and their determinants. 246. There are stark differences in self-assessed general health status by rich and poor in Georgia (Figure 6.3). Respondents in the richest quintile are over three times more likely to report being in good health than those in the poorest quintile, while the poor are over five times more likely to report bad healththan the rich. These responsesreflect subjectiveperceptionsof well-being, and while they may include some bias, the health literaturesuggests that this is a significantpredictor of mortality and a more reliable metric for gauging health status than current morbidity or the use of medical care (Gertler and others 2000). We can be confident of the existence of a steep gradient, if not the exact magnitude, and as such these results point to an important area for improvement in the Georgian health sector. Box 6.1 compares patterns of self-reportedhealth satisfactionin Georgiawith the rest ofthe world. 103 Figure 6.3: Self-assessed Health, by Quintile 70% 60% 50% 40% s 30% 20% 10% 0% 1 2 3 4 5 Source LlTS [E?Good or very good =Bad or very bad 1 Box 6.1: Health Satisfaction Around the World In 2006, nationally representative samples of 132 countries were asked about their "health satisfaction" and "life satisfaction" by the Gallup polling company. Descriptive analysis was carried out by Deaton (2007). The figure below shows the relationship between health satisfaction and national GDP per capita. Each circle representsa country, and the plotted regression lines provide a disaggregation by age of the country averages represented by the circles. In general, there is a clear positive correlation betweenhealth satisfaction and nationalincome among older cohorts, but no significant relationshipamong the young. Other findings (not shown) suggest a general decline in health satisfaction with age in most low- and middle-income countries, and this profile is particularly steep in the Europe and Central Asia (ECA) region. With respect to life satisfaction, the survey revealed a positive relationshipwith national GDP per capitathat was generally consistent across all age groups. Georgia is an outlier in the survey results. It reportedthefourth lowest average rate of health satisfaction in the world (after Ukraine, Haiti, and Russia), despite having much better "objective" health indicators such as life expectancy than many other countries (for example, in Africa). Interestingly, low rates o f satisfaction with general health in Georgia relative to other countries does not appear to translate into low satisfaction with health care services, since Georgia has one of the higher rates of satisfaction in this regard amongtransition countries, as reportedin the LiTS and discussed in Sundaram and Zaidi (2007). Georgia is also somewhat unusual on the life satisfaction scale. Along with Armenia, it is the only country among the bottom20 in terms of average reportedlife satisfaction that is not amongthe world's poorest nations. 104 A 9 i - - 2 n + ` 60-'0 I 8 0 10000 20000 30000 40000 Pei capita GDP in 2003, 2000 cliained PPP do1l:iis Source: Deaton(2007). 247. Identifying the causes underlying these inequalities in health status can provide important information to inform policies. Untangling the "health production function" is complicated, however, because in addition to income, health status can reflect a wide array of considerationsincluding genetics, behavior, education levels, exposure to pollution, access to medical care (physical and financial), utilization of services, the quality of care received(clinical and non-clinical), and so on. In the absenceof completedata on each of these factors, here we can only provide some general speculationof the relative importanceof various factors in determiningthese inequities. 248. Behavioral issues such as tobacco use can have an important impact on health equity outcomes independent of health system performance. Tobacco use is a key risk factor in Georgia's disease burden, particularly for men, Figure 6.4 shows adult smoking prevalence by quintile. Although overall tobacco use is somewhat more common among richer quintiles,the poor are more likely to smoke non-filteredtobacco products. As the figure also indicates, tobacco expenditures account for a larger share of total household consumptionamongthe poor, reaching 13 percent inthe poorest quintile. Furtherwork on behavioral issues such as diet, exercise, and alcohol use may also provide useful insights. 105 Figure 6.4: Tobacco Use and Expenditures, by Quintile 30% 25% 20% 1 15% 0 4 .5 10% 5% 0% Smoking prevalence (non- Smoking prevalence Smoking prevalence Tobacco expenditures (% filtered products) (filtered products) (total) of total spending) Source LSMS 249. Health system indicators that do not reveal a significant difference between the poor and non-poor include physical access to care and satisfaction with health care provided. Figure 6.5 shows physical access (proximity) to a doctor. Despite the greater proportion of poor in Georgia living in rural areas, there is no significant difference across quintiles in physical proximity to health care. Of course, the poor may face greater costs (direct and indirect) in reachingthe nearest provider, but distance is not a key constraint. Figure 6.6 shows indicators of satisfactionand trust by quintile. This i s a possible proxy indicator for the non-clinical quality of health care received (for example, waiting times, doctor attitudes, and so forth). Due to informational asymmetries, its reliability as an indicator of clinical quality of care is less certain. Again, there is no distinguishable gradient across quintiles. This is a good sign, since in other country contexts the poor may suffer from systematic discrimination. There is also little sign of a gradient with respect to the frequency with which informal payments are reported to be required by providers. Figure 6.5: PhysicalAccess to Doctor, by Quintile 100% 90% 80% 70% 60% S 50% 40% 30% 20% 10% 0% 1 2 3 4 5 Quintile Source HUES /@8Accesswithin 15 minutes .Access within 30 minutes] 106 Figure 6.6: Satisfactionand Trust, by Quintile 80% 70% 60% 50% be 40% 30% 20% 10% 0% 1 2 3 4 5 Source LiTS /QUnsatisfied with care =Satisfied with care OTmst nearest facility 1 250. Although there is only a mild pro-rich gradient with respect to utilization of health care services, this pattern suggests significantly less equality than may appear at first glance. Figure 6.7 shows utilization of health care services over a six-month period, with use of about 27 percent by the poorest quintile and 35 percent by the richest quintile. However, in view of the generally better state of health of the non-poor noted above, an "equal" pattern of health care utilization conditional on need would entail significantly higher rates of utilization by the poor.66 25 1. The (self-reported) share of the adult populationaged 30 and over who have had their blood pressure taken during the past year is almost 50 percent higher in the richest quintile than the poorest (Figure 6.7). This suggests several observations: (a) there is inequalitynot only in the decisionto seek care, but also in the intensity of treatment received; (b) the clinical quality of care received by the poor may be worse than that provided to the rich; and (c) it is likely that there are higher rates of undiagnosed chronic disease (for example, hypertension) among the poor than among the rich. As noted in the previous section, high blood pressure is the most important risk factor for poor healthamong both men and women in Georgia. 66 The self-assessed health indicator and utilization data are drawn from different surveys, and therefore a calculation of utilization conditional on need cannot be easily made with available sources. 107 Figure6.7: Utilizationof Health Services, by Quintile 4 50% 45% 40% 35% 30% S 25% 20% 15% 10% 5% 0% Used services in last 6 months Had blood pressure taken in last year iourse HUES 252. A major reason underlying inequality in utilization is financial access to care. As a result, the poorest quintile is approximately twice as likely as the richest to forego care due to financial barriers. Figure 6.8 shows the reported frequency of financial barriers causing households to forego a medical consultationwhen sick, the purchase of prescribed drugs, and hospitalizationwhen advised by a doctor. Although there is variability in the middle quintiles, the gap between the richest and poorest quintiles for all three indicators is clear, as the poorestquintile is approximatelytwice as likely as the rich to forego care due to financial barriers.Out-of-pocketpayments for health care, which as noted are among the highest inthe Europeanregion,will be examined in more detail inthe next ~ection.~' Figure 6.8: Financial Access to Health Care, by Quintile Reporred frequency of financial barriers causing households to forego. Medical wnsultalion when sick Purchase of drugs Hospilali~ation 67A common exercise in measuring equity in health finance is benefit incidence analysis. This, however, cannot be undertaken in conjunctionwith available survey data until the 2007 National Health Accounts exercise is complete. But given the highrates of out-of-pocketspendingon health and rapidly changingbudget priorities in the sector, the value of such an exercise will be greater once the reformdust settles. 108 253. In sum, there is evidence of significant inequality in health outcomes between the rich and poor in Georgia. The factors that play a role in the "health production function" are complex, but the evidence suggests that a significant component of this inequality can be attributed to differences in utilization of care and the underlying financial access considerations. Section 5 will address preliminary evidence on the scope for MAP to improve these indicators.Other factors, such as physicalaccess to care and non-clinical quality as proxied by indicators of satisfaction and trust, do not appear to reflect significant inequality.As noted earlier, it bears repeating that the snapshot of indicatorsprovidedhere is taken in the midst of far-reaching reforms to the Georgian health sector, and more time and data will be requiredfor a clearer picture to emerge. D. FinancialProtectionand Out-of-pocketPaymentsfor Health 254. While health system objectives typically emphasize improving health, the topic of the previoussection, another important objective is financial protection. The need for health care is often unpredictable and costly, and public policy has a potentially important role to play in improving household welfare in the face of this uncertainty.Unfortunately,it often does not live up to this potential. This section looks at Georgia's recordin providingfinancial protectionin health, with special reference to the poor. 255. A key reason for emphasizing financial protection is because health expenditures are qualitatively different from most other items in a household consumption basket. This is because the spending is usually not voluntary (for example, if arisingdue to an unwanted health shock), and may not be associated with an improvement in household well-being to the same extent as the purchase of other items.A householdforced to make high health expenditures would not have these resources availablefor spending on necessities such as food and shelter. Also, the uncertainty and potentially high cost associated with health expenditures make them amenable to prepayment and risk-pooling arrangements. For all these reasons, a more desirable counterfactualto high out-of-pocketspending on health would be some form of prepayment mechanism (whether throughgeneral taxes or a contributory insurance scheme) to providefinancial protectionagainst healthshocks. This has been achieved in many countries, but not in Georgia. 256. The main indicator used to evaluate financial protection in health is out-of-pocket payments (OOP), and in Georgia these are quite significant by any metric. Although estimates vary widely in absolute (GEL) terms across different surveys (see Box 6.2), they are clearly substantial bothas a share of total household consumption (about 7 to 10 percent), and by international standards. According to the most recent national health accounts exercise conducted in 2006, OOP represents 72 ercent of total health expenditure in Georgia, which would rank it highest out of 53 Europeancountries.6H) Box 6.2: Measuring OOP: How High is Out-of-pocket (OOP) Spending on Health Care in Georgia? Estimating health care utilization and out-of-pocket payments through household surveys can be challenging. A common approach is to ask individuals about their last visit to a health care provider, often restrictedto a certain time frame such as the past month. However, those who are sick, especially with a serious andor chronic illness, may make several visits during that period, and often to more than one provider. The survey results in this case will tend to underestimate both utilization and expenditures. Moreover, rare events such as hospitalizations are more easily captured through longer recall periods, whereas more "regular" occurrences suchas drugpurchasesor clinic visits are better addressed over shorter periods to reduce recall bias. For these reasons, guidelines for health modules tend to recommend more WHO-EuropeHealthfor All database.The mostrecent data for other countries is 2004 109 thorough survey instruments to capture all visits, and multiple recall periods depending on the type of service (for example, one month for outpatient care and 12 months for inpatient care) (Gertler and others 2000). But an extensive health module may not bt?feasible in the context of a general consumption expenditure survey. Existing surveys in Georgia may be subject to some of these shortcomings in estimation. The regular Household Budget Survey (HBS) uses a four-week recall period for utilization that can capture only one visit for general curative care, and does not ask about utilization at all for the chronically ill.Expenditures are captured using a three-month recall period. The LSMS does not ask about utilization, and adopts a 12- month recall period for health OOP (and most other non-food expenditures). On the other hand, the 2007 Georgia HealthUtilization and Expenditure Survey (HUES) was able to focus exclusively on health issues andthus employeda survey instrumentthat more closely followedthe recommendedapproach.aResultsfor healthcare utilizationwere broadly comparableto available administrativedata on service use. As a result, there is considerable variation across surveys in their estimates of OOP for health in Georgia. While both the HBS and LSMS surveys estimatedaverage health expenditure o f about GEL 70 per personannually, the HUES estimate was three times higher. Although the true figure is unknown, it is likely that the HBS and LSMS results are significantly underestimated. However, these issues do not findamentallyalter the storyline in this section. Since the consumptionaggregate (particularly for non-food expenditure) may also be underestimated,and this serves as the denominator in calculations of catastrophic OOP, the resulting ratio should be more reliable. With respect to international comparisons of impoverishmentdue to OOP, Georgia's indicators may be even higher, but its near-worst relative ranking would remainunchanged.Nevertheless,future survey work inGeorgia shouldbe mindfulofthese issues. a. The HUES sampled households that had participated in the 2006 HBS, and thus could use the consumptionaggregate from that survey. The sample size was 3,2 18 households. 257. A decompositionof OOP into subcategories also provides useful insights. Estimatesare quite consistent across surveys, with medicines accounting for slightly over half of all OOP, hospital services about one-quarter to one-third, and ambulatory services representing about 20 percent of the total. Although drugs represent the largest category of spending, the variability of OOP on health services is larger. In addition, the shares vary significantly by socioeconomic status, with poorer households spending over two-thirds of their OOP on drugs, while the correspondingfigure for the better-off is closer to 40 percent. The share of spending devoted to hospital and surgical services is quite low by the poorest quintile, but much higher by richer quintiles. While the new Medical Assistance Program (MAP) for the poor provides coverage for most health services but excludes drugs, the higher drug share amongthe poor i s also observed among those poor who are not beneficiaries. The issue of drug spending will be discussed further inthe next section. 258. A common approach for evaluating out-of-pocket (OOP) expenses for health is to measure the extent to which they are "impoverishing." That is, if a household has total consumption expenditures (pre-OOP) above the national poverty line, but their total non-medical spending (post-OOP) is below the poverty line, they could be considered to have suffered impoverishment due to OOP for health. Whether this is an accurate way to evaluate the true poverty impact of OOP is a matter for debate, and is discussed in Box 6.3. For now, impoverishment due to OOP for health in Georgia is illustrated graphically in Figure 6.9. Households are ranked along the horizontal axis by total consumption. The vertical "drip" lines represent OOP for health, and the poverty line is indicated by the horizontal line at GEL 71.6. When total householdconsumption places a household above the poverty line but health OOP drops them below, it can be argued that impoverishment due to health spending has occurred. Although the thickness of the vertical lines may exaggerate the incidence of impoverishment, it is nevertheless a frequent occurrence. 110 Box 6.3: Issues in Measuring Financial Protection in Health The discussion in this section about impoverishing and catastrophic OOP for health raises several issues that warrant closer scrutiny.* On one levelthey may underestimate the disruptive impact o f illness on households. Most obviously, they do not account for income losses arising from a health shock (for example, time away from work), which in some contexts have been shown to be even more importantthan the direct impact of medical bills (Gertler and Gruber 2002). However, this is arguably an issue to be addressed by the social protection system more broadly, not just by health financing arrangements. Moreover, a full analysis of this would require more sophisticated survey instruments than those currently available for Georgia. A second reason is that high OOP at the point of service may oblige families to forego necessary treatment altogether, and this would not be captured in the data discussed in this section. While such a scenario is indeed worrisome, it concerns financial access as an instrument to achieve better health outcomes (as discussed in the previous section), rather than considering financial protection as an importantobjective ofthe health system itself, as it is addressed here. On the other hand, the approach discussed here may also overestimatethe disruptive impact of OOP. First, it presumes OOP is involuntary. In some instances this is surely not the case, but to assume the opposite, that health spending is entirely discretionary, seems even less plausible, so as a first approximationthis seems reasonable.Second, and more important, the discussion ignores the issue of how householdsactually cope with high OOP. We have assumed that in the absence of OOP, total consumption would have been the same, but the household could have afforded spending on "better" things. In reality households are likely to draw on several possible coping mechanisms that would allow for consumption smoothing, such as drawing down savings, borrowing, or selling assets. It has been assumed here that a costly illness episode in one period has an immediate and commensurate impact on total consumption in the same period, which is surely not right. However, a full analysis of this would require detailed longitudinal surveys on illness episodes, health spending, and other household decisions, which are not currently available. Although coping mechanisms allow for the possibility of a "softer landing" in the aftermath of a health shock than the results here suggest, recourse to these channels still implies a significant negative impact on inter-temporalwell-being, and is therefore less desirable than the existence of appropriateprepaymentand risk-poolingmechanismsthat would truly providefinancial protectionto the population. a. The discussionhereborrows from Wagstaff(2008). I Figure 6.9: Impoverishing Effect of OOP for Health Effect of OOP health payments on Pen`s Parade of Household ConsumptionDistribution - w L1 E tE I B8 .L B iJ L 1 501 1001 1501 2001 2501 3001 3501 4001 4501 5001 Households ranked in ascending order of total consumption [per adultequivalent) Source LSMS 111 259. High OOP for health has a significant impact on poverty among Georgian households. Using the concept of impoverishment we can recalculatekey poverty indicators in Georgia by "netting out" OOP. Indeed, becausehealthspending does not necessarily capture an increase in household welfare in the same manner as other goods, a case is sometimes made to exclude it altogether from the consumption aggregate and thus from calculations of poverty statistics (Deaton and Zaidi 2002). As presented earlier in this report, the poverty headcount in Georgia is 23.6 percent, a figure that is based on a consumption aggregate that includes OOP for health. If we calculate the poverty headcount without OOP, it rises to 26.5 percent. Thus, an additional 3 percent of Georgianhouseholds are classified as poor if we subtract their (potentially involuntary and non-welfare improving) health expenditures. The incidence of poverty does not capture the severity, and thus we can make the same adjustment for calculating the poverty gap. This rises from 7.2 percent to 8.3 percent of the poverty line when we account for OOP. The meanpositivegap is 30.3 percent ofthe poverty line. 260. Georgia fares quite poorly relative to other countries in terms of providing financial protection against impoverishment due to OOP. Table 6.3 compares Georgia with several Asian countries, both low and middle income, for which similar measures have been calculated. Only Vietnam has a higher indicator. Table 6.3: Impact of Health OOP on Poverty Indicators, Selected Countries Country Percentage Change in Percentage Change in (normalized) Poverty (normalized) Poverty Gap Headcount Due to OOP Due to OOP Bangladesh 4.9% 9.4% China 4.1% 7.1% Georgia 12.7% 16.1O/o India 2.6% 6.0% Indonesia 2.9% 4.7% KyrgyzRep. 6.0% 8.0% Malaysia 2.1% 3.0% Nepal 1.6% 3.4% Philippines 2.1% 2.8% Sri Lanka 4.3% 5.3% __Thailand 2.8% 4.2% Vietnam __ __ - -_-. 12.1% Note Based on national poverty line-in Georgia, US$Z/day elsewhere - - _ _ 18.3% Source Georgia LSMS and van Doorslaer and others (2006) 261. An alternative approach for highlighting the impact of OOP on households is to measure the extent to which they are "catastrophic." Impoverishing OOP puts the emphasis on crossing the poverty line irrespective of the size of payments. Catastrophic health expenditures occur when they exceed some threshold of either total or non-food expenditure. The choice of threshold is somewhat arbitrary, but here we will follow a common practice in recent literature and use 10 percent of total consumption expenditure and 25 percent of nonfood expenditure. Again, the idea is that these expenditures displace spending on other goods and services, and would not be incurred if appropriate prepayment mechanismswere in place.The extent to which this metric accurately captures the disruptive nature of high OOP is also a subject of debate, as discussed in Box 6.3. 262. The incidence of catastrophic OOP for health in Georgia is also high relative to other countries.The share of households with OOP exceeding 10 percent oftotal expenditure is estimatedto be 17.6 percent in Georgia, with a mean positive overshoot among those households of 13.1 percent of expenditure.The share of householdsfor which OOP exceeds 25 percent of non-foodexpenditureis 25.7 percent, with a mean positiveovershoot of 18.5 percent. For comparison purposes, these rates exceedany 112 o f those that prevail in 14 Asian countries as shown in Table 6.4. An alternative definition o f catastrophic OOP, 40 percent o f "capacity to pay," was used to calculate results for 59 other countries, and Georgia has a higher incidence than all but 6 o f these (and all but 1 o f 13 transition economies in that dataset, as shown in Table 6.4). Table 6.4: HouseholdsExperiencing Catastrophic OOP, Selected Countries Country 10% of Total 25% of Non-food Country 40% of Capacity to Pay Bangladesh 15.6% 14.7% Azerbaijan 5.8% China 12.6% 11.2% Bulgaria 2.0% Georgia 17.6% 25.7% Croatia 0.2% Hong Kong 5.9% 2.4% Czech Rep. 0.0% India 10.8% 9.8% Estonia 0.3% Indonesia 4.4% 4.4% Georgia 5.1% Korea 10.4% 4.8% Hungary 0.2% Kyrgyz Rep. 5.8% 9.3% Kyrgyz Rep. 0.6% Malaysia 2.0% 0.8% Latvia 2.7% Nepal 5.9% 9.2% Lithuania 1.3% Philippines 4.6% 3.8% Romania 0.1% Sri Lanka 3.0% 3.4% Slovakia 0.0% Taiwan 6.4% 1.5% Slovenia 0.1% Thailand 3.5% 1.8% Ukraine 3.9% Vietnam 15.1% 15.1% West. Europeavg 0.6% Source: Author calculations for Georgia (LSMS); van Doorslaer and others (2007); Xu and others (2003). 263. I n sum, although more ideal indicators of the disruptive impact of OOP on Georgian households are elusive, there is strongevidence pointing to a lack of financial protectionin health in Georgia. OOP for health is high whether measured relative to total consumption or by international standards. The next section considers whether M A P is likely to help in this regard. E. Preliminary Evidenceon the Impact of the Medical Assistance Program (MAP) 264. The most important ongoing health reform in terms of equity implicationsis the MAP.This section provides some preliminary evidence on the ability of M A P to improve on the findings related to health outcomes and financial protection noted above. 265. A prerequisite for the success of MAP in improving both health outcomes and financial protection for the poor will be good program coverage and targeting. The evidence in this regard is mixed. Figure 6.10 indicates the number o f M A P beneficiaries and non-beneficiaries by consumption quintile, and suggests there is a significant coverage shortfall, with only about 25 percent o f the poor receiving M A P benefits. This i s in part a result o f budget limitations. About 44 percent o f M A P beneficiaries are poor, which is significantly lower than the TSA (70 percent), but M A P has a higher eligibility threshold (70,000 instead of 52,000) and larger number of beneficiaries (about double). Reflecting the poverty profile presented in earlier chapters, and the extent to which M A P does reach the poor, it is noteworthy that M A P beneficiaries are older, more rural, and have a household head that is less likely to be in the labor force than non-beneficiaries. 113 Figure 6.10: MAP Coverage: Beneficiariesand Non-Beneficiaries,by Quintile 900,000 800.000 700,000 600,000 500,000 400,000 300,000 200,000 100,000 1Source LSMS Beneficiaries Non-beneficlanes 266. There is evidence that MAP is having a positive impact on utilization of health services by beneficiariesand therefore has good potential to improve health outcomes among the poor. Figure 6.11 shows utilization of "urgent care" by beneficiaries and non-beneficiariesduring the first six months of the program in late 2006 (when the eligibility thresholdwas a score of 100,000) (Hou and Chao 2008). It draws on administrative data and adopts a regression discontinuity design for conducting impact evaluation.It indicatesthat MAP cardholders in the region of the thresholdwere ninetimes more likely to use urgent care than households that were not eligible (from about 1 urgent care utilization per 1,000 people every six months to nearly I O per I,OOO). Utilization has also increased among the poorest of the poor. While there are some important caveats about the magnitude of this increase in utili~ation,~~the positive directionof the impact (if not the size) is certain, and is suggestive of good program potential to improve health outcomes. It should be reiterated, however, that MAP beneficiaries cannot always be equated with the poor in view of the targeting performance noted earlier, and this will need to be improvedifMAP is to achieve its potentialto reduce inequalities. 69 These arise due to data issues (dropped observations where no score was available) and supply-side noise (different provider reimbursement rates). There should also be concern that illness episodes were beingreclassified "urgent" in order to gain coverage when this was not the case. Data are not available to compare total health care utilization by MAP and non-MAPpopulations, which may be more similar. 114 Figure6.11: Impact of MAP on Utilization of Urgent Care, 2006 I I \ - -----4-------/ * ; - 3- -*-* -- p ~ `\ \ (Y 5 8 - --=- --- -- --*- -_- -*-< +#-- **--- *--+--Y.= - - 0 - 5 -> '* 267. Early evidence suggests that MAP is not having a significant impact on OOP for health, most likely due to the exclusion of drugs from the MAP benefit package. Using a sample of householdsdrawn from the LSMS in the neighborhoodof the new threshold eligibility score of 70,000, a regression discontinuity analysis suggested no impact of MAP on lowering total out-of-pocket payments for health. There is, however, some weak evidence that MAP is lowering non-drug OOP (that is, expenditures on healthcare services only)." As indicated in Figure 6.12, the probability that a household faces out-of-pocket spending on non-drug health bills that exceeds 10 percent of total consumption is loweredby about 5 percentagepointsat the thresholdscore if it receivesthe MAP benefit. Figure6.12: Impact of MAP on Probability of IncurringCatastrophic Non-Drug OOP e ( 60000 65000 eligibili score 7Q 0 75000 80000 Source: Author calculations, based on LSMS 70The evidence is weak in the sense that significance of the dummy variable for beneficiaries is not always robust to changes in functional form (linear or quadratic) or to variation of the bandwidth on which the regression discontinuity design is estimated. Potential reasons for a weak response of OOP to MAP could include the following: (a) persistent informal payments; (b) the survey recall period includes early 2007, when a less comprehensive benefits package was offered; (c) the survey recall period includes the early stages of MAP implementation, and therefore householdsidentified as beneficiariesmay not have been members for the full twelve months for which expenditure data was collected. 115 268. Although the introduction of insurance has the potential to reduce the incidence of catastrophic health expenditures, international evidence from other low- and middle-income countries provides a mixed picture. While such comparisons can only be indicative given wide variation in pre-reform (if any) and post-reformbenefits packages, significant declines in OOP spending following insurance expansion cannot be guaranteed. For example, while programs to improve coverage inMexico and Thailand in 2001 did help reducethe incidenceof catastrophichealthspending, reforms in Vietnam and China had more mixed result^.^' The reasons why insurance reforms may not reduce the frequency of high OOP vary, but often arise due to increased utilization coupled with less than full coverage of services. Supplier response can also play a role. A plannedimpact evaluationof MAP should provide insights on this issue inthe Georgian context. F. Conclusionand Policy Recommendations 269. The preceding sections highlighted the significant inequalities in health status and health care use betweenthe rich and poor in Georgia, and a lack of financial protection from high out-of- pocket expenditures. The potential of the MAP programto address these challenges was also discussed. This section summarizes some implications for current policy reform initiatives. As noted earlier, it is important to underline that the policy environment is rapidly evolving-benefits packages have been redefined more than once, eligibility criteria changed, financing mechanisms reformed, and all stakeholdersare on a steep learningcurve-and as such more time and data will be requiredto reach clear conclusionsabout policy impact.Herewe provideonly some initial speculationon key issues. 270. Although the impact of MAP on utilization suggests it can help improve health outcomes among beneficiaries, its overall effect on the poor will be diluted unless coverage is strengthened. Reducingthe large number of poor who are not currently benefitingfrom MAP should be a key priority. 271. The broad evidence on MAP coverage and its impact on key indicators also suggest policy implicationswith respect to non-beneficiaries. In particular,the reality that three-quarters of Georgia's poor are not currently benefitingfrom MAP coverage, while those who are covered have responded with much higher utilization rates of healthcare than similar householdsjust above the eligibility score (as in Figure 6.1l), implies a certain inequality of its own. This will become even starker if plans to eliminate the UBP for the non-MAP populationare carriedout. Policiesthat soften this "all-or-nothing" approachto insurance coverage-for example, by offering intermediate coverage in the form of co-payments or alternative less generous benefits packages-might be considered after the initial stage of MAP implementationis achieved. 272. The characteristics of households that face high out-of-pocket health bills but do not have MAP coverage also points to the potential consequences of UBP elimination. A regression of the probability of these households suffering catastrophic OOP (defined as exceeding 10 percent of total consumption)on various household characteristics suggests that the most vulnerable are those households with elderly (over 65), young children (under 5), and a household head who is not employed.'* These households are also the least likely to successfully obtain health insurance coverage from the private market due to risk selection by insurers and the absence of employer-based risk-pooling options. Although UBP does not offer extensive service coverage and includes significant co-payments, it does offer some financial protection,and the fate of these households once it is eliminatedshould be the focus of increasedpolicy discussion. 71These examples are cited in Wagstaff(2008). 72The education and gender of the household head and urbanhral location are not significant correlates of high OOP. 116 273. The limited impact of MAP on OOP for health underlines the implications for financial protection of excluding drug costs from the benefit package. As noted, drug costs typically represent about two-thirds of total OOP for health by the poor, whether they are MAP beneficiaries or not. The variability of drug spending is much lower than that for healthservices, and thus the "insurance value" of covering drugs may be less. But to the extent that medication plays a key role in improving health outcomes, their exclusion will be important. Indeed, given the likelihood that chronic care is less commonly diagnosed among the poor, MAP may lead to higher OOP as more service use by individuals identifies their previously unknown need for prescription medicationto manage chronic disease. In this event, MAP would help healthoutcomes but financial protection could deteriorate. Of course, expanding the benefit packageto include medicationwould have a significant impact on program cost and therefore fiscal sustainability. Partial drug coverage by MAP (with relatively high co-payment rates) could help balance these objectives. Alternatively, a closer look at the pharmaceutical sector and the scope for achieving lower drug prices could reveal possible reformsthat would have as large an impact on OOP as the introductionof MAP. 274. I n addition to MAP, the reforms aimed at promoting a larger role for private providers (both PHC and hospitals) will also affect health equity in Georgia. Whether these effects are positive or negative will depend to a large extent on the enforcement of safeguards to protect both access and quality of care. Strengthening the oversight and stewardship functions of the Ministry of Health to overseethe privatehealth sector should be a top priority in this regard. 275. I n sum, the rapidly changing policy environment in Georgia offers the potential to address some of the inequalitiesin the health sector identified earlier. However, much remainsto be done. The coverage of MAP is a particularly important area for improvement. While early evidence on MAP suggests a positive impact on utilization, there has been less progress in providing financial protection. The vulnerability of non-MAP beneficiaries,ifand when the UBP is eliminated, should also be an area of policy focus. Broadly speaking, the "all or nothing" character of both MAP eligibility (below or above 70,000) and benefit package definition (health services or drugs) may warrant reconsideration as programsare revisedand fine-tuned. 117 CHAPTER 7: EDUCATION REFORM, EDUCATION OUTCOMES, AND POVERTY A. Overview of the Education Sector73 276. Since the Rose Revolution, the Government of Georgia has undertaken a comprehensive (and impressive) reform process in the education sector. The focus of the reform has been to redesign the old education system through institutional change and infrastructure improvements (Box 7.1). The process included "finance and governance" reform whereby management of the education system was decentralized.All educational institutionswere established as public legal entities. Now, each school is governed by a Board of Trustees empowered by a financial management authority. The administrative structure of the education system has also been adapted. Former Regional and District Departments of Education have been replaced by a network of 72 Education Resource Centers (ERCs) responsible for facilitating the work of schools through collecting data, organizing training, conducting research, and monitoring accounting. The finance reform of the education sector reflects the promise made by the government (since the Rose Revolution)to fight corruption. The government in January 2006 introduced a per capitafunding formula nationwide at the general secondary education level.Under the new scheme, schools receive a direct transfer of funds from the Ministry of Education and Science (MoES) based on the number of students enrolled for a given year. The voucher covers current school expenditures, of which teacher salaries are the main component. Ongoing reforms in Higher Education in Georgia are primarily carried out in compliancewith requirements of the BolognaProcess.The BolognaProcess aims to create a European Higher Education Area by 2010, in which students can choose from a wide and transparent range of high quality courses and benefit from smooth recognition procedures. The Bologna Declaration of June 1999 has put in motion a series of reforms needed to make European Higher Educationmore compatibleand comparable, more competitive and more attractivefor Europeans and for students and scholars from other continents. 277. Georgia's education system has achieved internationally acceptable levels of net enrollment and school survival rates despite relatively modest levels of public expenditure. Public expenditure on education in Georgia, at approximately 2.7 percent of gross domestic product (GDP) in 2007, is very low relative to the Europe and Central Asia (ECA) and the Organization for Economic Co-operationand Development (OECD) country average (at approximately 4.4 percent and 5.O percent of GDP, respectively).Since 2000, Georgia has spent roughly 2.2 to 2.7 percent of its GDP on education.In 2004, expenditure on education showed a sharp increase (to 2.9 percent of GDP) as the government fully paid the accumulated arrears on teacher salaries. Despite the fact that Georgia spends on education half of what is spent on average within the region, indicators such as gross primary and secondary enrollment rates do not differ from those regionally(Figure 7.1). 73This chapter was preparedby Diego F. Angel-Urdinola (Economist, ECSHD), Nino Kutateladze (ETC, ECSHD), and Ivan Khilko (STC, ECSHD). 118 Figure 7.1: Gross Primary/Secondary Enrollment Rates in 2005, Georgia in the ECA context] I 4 O1 140 1 60 0 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8OW 10000 12OW GDP Per Capita GDP Per Capita Source: EDSTATS(2005). Box 7.1: The Education System in Georgia Georgia's education system is structured as follows: noncompulsory preschool education (2 years), compulsory education (9 years: 6 years for primary education and 3 years for lower-secondary education), upper secondary (3 years), and tertiary education (4 years for bachelor's degree and 2 years for master's- level education). In addition, the system offers programs for initial vocationaleducation after the completionof compulsory education, and higher professional education and training that can be obtained after completion of the full cycle of secondary education, and offerseducation for students with special needs. I 278. Georgia has a large school system serving a shrinking school-age population. According to data provided by the 2005 Education Management Information System (EMIS), Georgia's compulsory education system comprises approximately 620,000 students, 66,000 teachers, and 3,100 school buildings (approx. 2,600 schools). The student-teacher ratio is approximately 14.4 at the primary education level and 9.2 at the secondary education level (the ECA averages in 2005 were 16 and 12 percent for primary and secondary, respectively). This low ratio (given Georgia's low levels of expenditures) is the heritage o f a large education structure previously financed under the Former Soviet Union (FSU). Also, Georgia had a larger student population in the early 1990s than it does today.74A difficult political economy of downsizing the teacher force is compounded by the fact that the current system provides strong incentives for teachers over retirement age to stay in the job. This phenomenon arises mainly because the average teacher pension in Georgia (currently at GEL 70 per month) is lower than the average teacher salary (currently GEL 200 per month) despite recent efforts to close this gap (pensions have almost doubled since 2006). Furthermore, labor market opportunities for older teachers outside the school system are still not very promising and teachers over retirement age (60 for women and 65 for men) are not forced into retirement, partly because of a general lack of qualified younger/entry-level teachers. 74 During 1991-1994, following a collapse of the economic output by 72 percent and a correspondingcollapse of the tax base, public expenditure on education in Georgia decreased from 7 percent to 1 percent of GDP. This was one of the worst declines in output in the ECA Region-the most severe cut in the Region in public spending on education and unique in the history of education worldwide (World Bank 2001). Despite the shock, the country managed to retain a large share of the teachers employed in the system by freezing wages and accumulatingarrears. Similar to other countries in transition and other sectors of the labor market, adjustment took place through lower wages and arrears rather than through firing people. Public expenditures on education increased back to 2.4 percent of GDP by 1998 and have remained roughly unchangedsince. 119 279. Demand for general education services is projected to decrease rapidly in the near future. Due partly to migration and partly to declining fertility rates, the population of Georgia has been decreasing in recent years (a drop that has been noticeable for the education system), and the trend is expected to continue in the near future. Estimates indicate that during 2005-2050, Georgia's school-age population will contract by 45 percent (at a rate of 1.7 percent per year). This trend will inevitably have an impact on the future demand for education services. While currently Georgia's education system serves approximately604,000 children aged 5 to 14, it will serve only 270,000 childrenof similar age in 2050. This sharp decrease in demand provides additional grounds for further development strategies to rationalizethe already underutilizedschool system which, as of today, operates with greater-than-needed stocks of schools andteachers (Figure 7.2). Figure 7.2: School-age populationis expected to shrink rapidly in the short-to-medium term. Population 5 to 14 as a % of Total Population 16 - 14 - between2000and 2015 12 - Moderate Decrease 10 - between2015and2050 8 - 6~ 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Source; UN population database. 280. Rationalization is a very sensitive issue, with major political constraints. Public general education in Georgia is predominantly funded from the central budget. As part of the reform, greater financial and management autonomy has been granted to schools, which have been converted from local- government budget organizationsto autonomous Legal Entities of Public Law (LEPLS), and are funded directly from the Ministry of Education and Science (MoES) via per-studentfinancing scheme. Schools are governed by the recently elected School Boards of Trustees, composed of elected teachers and parents, a student representative, and in some cases, a local-government representative. The Board appoints the school director, approves the budget, and oversees and advises school management. The lump-sum amount (which is generally equivalent to a student voucher times the number of student) (Table 7.3) is transferredfrom the MoES into the school's own bank account and can be spent in any way that the school management, approved by the Board, decides. However, spending is subject only to a minimum salary for teachers, depending on their qualifications, experience, and the size of their class. Local departments of education have been replaced by a network of 72 Educational Resource Centers that facilitate (rather than control) a school's work through collecting data, organizing trainings, conductingresearch, and monitoringfinancialaccounting. 281. Georgia has more teachers and schools are becominglarger than what the system requires to operate adequately. With the exception of schools in more populated regions (Tbilisi, Mcheta Mtianeti, and Kakheti) school facilities are beingunderutilized. In regions like Guria, Achara, and Lower Qartli, school utilization rates vary between 82 and 89 percent. Although utilization rates higher than 70 percent are generally considered acceptable, maintenance of some large schools (some of which are 120 largely in disrepair) is quite co~tly.'~Moreover, given that the student population is shrinking rapidly, demand for schools is likely to decrease further in the near future. Although the government has engaged in an aggressive processof school administrativeconsolidation(more details below), the process is not an easy one. School consolidationmeansthat some school principals will lose their status. Also, in mountain areas, where access to schools is limited and the population is scattered (such as in Guria and Mcheta Mtianeni), schools can be neither consolidated nor closed, because students would not have access to other schools within a reasonable distance. Furthermore,the main challenge of consolidation is reducing the number of teachers. As presented in Table 7.1, in order for the education system to achieve a teacher- student ratio aligned with the ECA average (16 students to 1 teacher), about 28,000 teachers would be made redundant. Dueto social and political implicationsof cuttingteacher jobs, this is likely to be a slow and far-reaching process.76Also, in mountainous regions, where the student population is small and largelypoor, it may not be possible to achieve higher student-teacher ratios. Table 7.1: Many schools in Georgia are not working at full capacity Average Student-teacher Actual Number Number of UtilizationRatea Ratio of Teachers Excess Teachers InnerQartli 94.27 9.22 5,269 2,233 Tbilisi 99.20 12.55 11,883 2,560 Achara 87.47 8.42 7,425 3,519 Guria 82.42 7.54 2,592 1,37 1 Imereti 90.97 8.68 10,704 4,897 Kakheti 96.54 8.99 6,422 2,8 I 3 MchetaMtianeti 101.82 7.02 2,570 1,442 Lower Qartli 88.88 10.30 7,394 2,632 Samtskhe- Javakheti 94.72 7.54 4,624 2,446 Samegre1o 92.34 7.45 7,690 4,110 Total - 9.27 66,573 28,023 Note: a. Utilization rate is defined as the number of classrooms used compared to the number of classrooms available for academic purposes. Source: World Bank using EMIS 2005. 282. Most of the general education budget in Georgia is spent on teacher wages and wage-related taxes, leaving very little space for capital investments. The quality of the schools built during the Soviet times was generally poor. Emphasison school maintenancewas scarce during these times, and the collapse of education spending at the beginning of the transition resulted in large capital disinvestments. As a consequence, the current stock of school infrastructureis deterioratingquickly. Yet, about 92 percent of the overall budget for general education is spent on teacher salaries and utilities (mainly school heating, water, and electricity). As indicated in Table 7.2, capital expenditures for general education are quite low (at about 4 to 5 percent of overall education spending), given the rapid deterioration of the nation's large stock of school infra~tructure.'~ 283. There is obviously a tradeoff the Georgian government has to make between investing in their teachers or in infrastructure. Investing in teachers has proven internationally to be one of the most effective ways to improve education quality. Nevertheless, partly due to significant increases in teacher wages and expenditures on utilities during 2004-2005 (wages increasedfrom 62.9to 71.2 percent 7sIn some cases, given the precarious state of some school buildings, it is cheaper to consolidate schools and build a new buildingthan to rehabilitatethe existing infrastructure(World Bank 2006). 76The Education Management Information System of MoES indicates that the national student-teacher ratio based on full-time equivalent for the academic year 2006-2007 was 14 in cities, 10.5 in villages, and 7 in mountainous regions. 77Recent energy price increases may have negatively affected Georgia's recurrent budget for financing public education. 121 of the overall budget), capital expenditures decreasedfrom 5.5 percent of the total budget in 2003 to 2.5 percent in 2004 and to 1.9 percent in 2005.78Teacher salaries in Georgia are defined by the recently adopted teacher pay scheme (MoES Decree No. 576 dated October 21, 2005), which provides a basis to remunerateteachers above the minimum threshold accountingfor their qualificationsand experience. The relative position of teachers has improved recently; the average salary for teachers is now between GEL 195 and GEL 283 per month depending on teacher workload, years of experience, and educational qualifications. While investing in teachers is a policy decisionthat should be encouraged and praised, the government needs to be aware that deteriorating schools contribute to lower quality of learning (by promotingstudentheacher absenteeism and a deficient learningenvironment), and that there is an urgent needto mainstreammore capital investments into the system. Table 7.2: Expenditureson general education(primary, basic, and upper secondary) are mainly allocated to pay for teacher wages. Item as a Share of the FY2000 FY2001 FY2002 FY2003 FY2004 FY2005 Total Budget % salaries 69.00 66.81 63.61 64.42 62.90 71.23 %contributions to social 19.60 22.33 22.58 21.83 24.91 14.45 security YOutilities 3.97 3.51 4.92 5.47 3.93 6.53 YOcapitalexpenditures 3.24 3.53 5.31 5.44 2.48 1.89 % other expenses 4.19 3.82 3.57 2.84 5.78 5.90 Total nominal expenditure(inthousands o f GEL) 58,501 82,053 95,525 91,219 168,425 131,433 Source; Ministryof Educationof Georgia. 284. An impressive set of reforms of finance and governance of educational institutions have been prepared since 2001 and launched since 2005. The finance reform in Georgia was largely designed to improve efficiency and transparency in resource allocation through addressing the shortcomings of the old system inherited from Soviet times. Under the old system, the largest share of funding for secondary educationwas allocatedby the localgovernment.The budget of the rayon (district) came from two main sources of revenue: a transfer from the central Ministry of Financeand local taxes. The funds for education were transferred to the district offices, which then distributed them among schools in an unclear manner. While this practiceworks very well in many other countries(and especially in those with developed decentralized systems), inequities in capacity and control mechanisms across districts in Georgia was a major constraint to this choice of financing system. Transfer amounts were iteratively negotiated without clearhtandardparameters and left much room for corruption. As a result, the previous system was characterized by large disparities in per-student expenditures and lack of transparency in resource allocation in the absence of adequate criteria. One of the key targets of the government was to take away education funds from local governments to avoid misappropriation of funds. Today, the Ministry of Education is fully responsible for managing resource allocation. The government launched the implementation of the new per-capita-based funding scheme nationwide in January 2006. 78Since the introductionof per capitafinancing, consolidateddata are no longer availableby type of spendingitem. One hypothesisfor this is that teacher salaries after 2005 account for an even higher share of total spending, given the 2006 reform of teacher wages. Compilingthis information from individual school spendingreports may not be easy, but it is necessary in order to monitor progress and financial performance. The MoES should design mechanismsto collect, compile, andprocessthe sector's consolidatedfinancial information. 122 285. Under the new scheme, schools receive direct transfers of funds from the Ministry of Education based on the number of students enrolled for a given year. The voucher covers current expenditures of schools including teacher salaries. The per-student voucher levels were defined by GovernmentDecree No. 182 dated October 14, 2005. The formula is relatively simple and differentiates costs per student only by geographic location (highest for those in remote regions of the country, lowest for those in cities). Coefficientsare set as 100 percent of the voucher for urban schools, 140 percent for rural schools, and 170 percent for mountainareas. Voucher levels have since increased several times by government decree (Table 7.3). Table 7.3: Value of StudentVouchers (in GEL) 2006 2007 2007 since 2007 since October December City Voucher 220 235 250 300 Village Voucher 330 350 350 420 MountainVoucher 396 425 425 510 Source: Government Decree No. 246 dated November 13, 2007; Government Decree No. 191 dated September 12, 2007; and Government Decree No. 182 dated October 14, 2005. 286. Despite significant progress, there are still a number of financing issues to be resolved. Under the current allocation scheme, small schools (generally in rural areas) have complained to the Ministry claiming they are underfunded, despite receiving additional funds. Such schools represent 34 percent of the total number of schools. Roughly 53,388 students are enrolled in these schools (9 percent of the total public secondary school student population). At the same time, large schools in the city are in a far more advantageous situation and some receive more funds than needed for regular operation. The Ministry has yet to find solid data supporting these claims and, if verified, address the problem through fine-tuning the current allocation formula. Also, while the new financing system aims at improving efficiency and at reducingcorruption,there is no assessmentof whether this has beenthe case. 287. Sufficient financial resources and well-defined maintenance responsibilitiesare pivotal to school sustainability.The lack of school maintenance(as will be exploredin detail in section 4) has been a serious problem in Georgia. To address this issue, Georgia decentralized school ownership (without selling rights) to autonomous school boards (Andrews 2005), which, in theory would assume responsibilityfor capital development and maintenanceof school buildings, reducing(but still retaining) legal dependence on local governments for this purpose. The main problem with this initiative is that it poses intrinsic disadvantages for poor districts with low executing capacity.Not all districts in Georgia enjoy the same fiscal capacity, and thus the ability to maintain schools may differ substantially across districts. The initial per capita financing scheme prohibited the use of school-voucherfunds for capital investment. An amendment to this was made by the Law on Secondary Education in 2006. However, funds receivedthrough per capita are not always enough to undertakethe needed maintenanceand repairs in some large and highly deterioratedschools. Currently, the law is unclear about how to finance major capitalrepairs,and the government is developinga strategy to address this issue. 288. The government also launched a two-stage school consolidation process. The initial phase was carried out prior to the establishment of schools as public legal entities (2005), and involved administrative mergers of primary and basic schools with neighboring complete secondary education institutions.As a result, the number of schools was reduced from the existing3,154 to 2,500. The second stage was launched in parallel to the introductionof a new per capita financing system at the secondary education level, during which the number of small schools was optimized, further reducingthe number of schools to 2,33 1. School consolidationhas not contributedto lowering the needs for capital investments and maintenance, since most school buildings remain open and functional after consolidation (World Bank 2006). 123 B. Inequalitiesin Access to Educationand Affordability 289. While at the basic mandatory education level (nine years of schooling) no major differences in enrollment rates are observed between poor and non-poor populations, they are large at the preschool and postsecondary levels. As illustrated in Figure 7.3, differences in enrollment rates do not differ much between richer and poorer between ages 7 and 15 (normal ages to be in basic education). However, enrollment rates for individuals between years 4 and 7 (normally the age to be at preschool) in the poorest quintiles are 20 to 30 percent lower than those for individuals in the richestquintile.This large difference may be explained by the fact that preschool (not being compulsory) is mainly paid out-of- pocket. At age 7, about 68 percent of all the children in the richest quintile are enrolled in school compared to only 53 percent in the poorest quintile, suggesting late enrollment in basic education, especially among the poor. Enrollment rates increase rapidly after age 7, reachingalmost 100 percent at age 9 for poor and non-poor children. Between ages 15 and 17 (the ages that correspond to upper- secondary school), enrollment rates drop rapidly among the poor. At age 17, presumably after having finished upper-secondary education, enrollment rates drop rapidly both among rich and poor individuals. At age 18, enrollment rates among children in the richest quintile are more than twice as high (70 percent compared to 30 percent) as those for children in the poorest quintile. Enrollment rates decrease steadily for poor and non-poor individuals after age 20. Preschool and postsecondary enrollment rates are significantly lower among childrenfrom the Armenian and Azeri minoritie~.'~ Figure 7.3: Enrollment rates by age suggest large inequalities in access to preschooland tertiary educationbetween (minority) poor and (non-minority) non-poor children. 100 0 80.0 U e 600 Lu E * 40.0 20.0 0 - - , , , , , , , , , , , , , , , , , , , , , , 0.0 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 2425 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 2021 22 23 24 25 Age Age Source: World Bank using Georgia 2006 HBS data. 290. There are some differences in school attainment between poor and non-poor children and between urban and rural areas. Figure 7.4 displays completion rates in Georgia among individuals in the age 16 to 29 cohort. Almost everybody included in this population has finished primary school and has attained some incompletesecondary (although some individuals in the poorest quintile drop out after having attained primary school). Differences in attainment between the poorest and richest individuals (and between urban and rural areas) become larger at the secondary education level. Secondary 79Unfortunately, HBS data do not ask households about the year in which students are enrolled by education level (that is, we know whether students are enrolled in primary education but not whether they are in their first, second, or third year). As such, the data do not allow computation o f net enrollment rates by education level. 124 completionrates are roughly 85 percent among individuals in the highest quintile compared to 70 percent among those in the poorest quintiles. A similar difference holds among individuals living in urban areas compared to those living in rural areas. Note than a 70 percent secondary completionrate amongthe poor (ages 16 to 29) is higher than that of countries of a similar income level in Latin America (at less that 40 percent in countries like Nicaragua or Guatemala; see Angel-Urdinola and Laguna [2008]) and seems aligned with that of other countries in the Caucasus (in Armenia, for example, the similar rate is at 75 percent) (Angel-Urdinola, Jain, and Prina 2006). However, after upper-secondary education, drop-out rates are high at all socioeconomic levels. As such, tertiary education attainment is low, particularly among the poorest individuals in rural areas. While 30 to 40 percent of all individuals in the richest quintile and in urban areas have attained some form of tertiary school (technical education, college, or university), only 10 to 20 percent among individuals in the poorest quintiles and in rural areas have done so* Figure7.4: Tertiary school attainment is significantlylower amongthe poor. 100.0 100.0 80.0 80.0 U -0 0 .E 60.0 aJ B 'zc 60.0 U 5 U S 40.0 S 40.0 20.0 20.0 Poorest Males ma'eS L ~ 0.0 0.0 Some Inc. Complete Tertiary UniRrsity Some Inc. Complete Tertiaty Unimity primary secondary secondary technical primary secondary secondary technical Source: World Bank usingGeorgia 2006 HBS data. 291. Attainment rates among minorities (particularly Azeri) are substantially lower than among native Georgians. As indicated in Table 7.4, drop-out rates after primary education are substantially larger for minority individuals compared to those for native Georgians, especially among the Azeri population.Estimates indicatethat less than 2 percent of all Azeri and less than 8 percent of all Armenian individuals aged 16 to 29, most of whom reside in rural areas, attained some form of tertiary education comparedto 20 to 27 percent amongnativeGeorgians.80 80In Georgia, the Azeri population is generally poorer than average (Azeris face a 1.5 times higher risk of poverty than the general population [World Bank 20081). Low attainment rates among the Azeri community are not only a consequence of their ethnicity but also of their socioeconomic status. Nevertheless, as will be shown later in this chapter, even after controlling for socioeconomic status, children from minority groups (especiallyArmenians and Azeris) display less-favorable education outcomes related to enrollment and attainment at all education levels compared to native Georgians. 125 Table 7.4: There are large inequalitiesin attainment rates betweennative Georgian individualsand other minoritiesliving in the country. .-_____--___I____.._ _. .......... _ - -Attainment Rates Georgian Azeri Armenian Other Individuals Aged 16 to 29 Ethnicity Some primary 100.0 100.0 100.0 100.0 Inc.secondary 99.5 91.4 99.8 98.8 Completesecondary 79.0 56.8 71.2 77.9 Tertiarytechnical 26.8 2.0 8.0 23.2 Highereducation ___20.5 1.3 l__-___-____.-. 4.0 _____~-_.II--- 11.5 Source: World Bank using Georgia2006 HBS data. B.1. Preschool 292. Preschool net enrollment rates in Georgia, while aligned with those in other Caucasus countries, are low by European standards. Prior to transition, participation in the preschooleducation sector in Georgia, similar to what prevailed in many other post-Soviet countries, was quite high. During the mid-1990s transition, public investments in preschool education suffered the most. As a result, the number of preschool institutions and children enrolled almost halved during 1989-1 994 (Table 7.5). While there is limited information related to why parents prefer preschool home care compared to institutional care (whether this is because of low institutional quality or because home care is socially preferred), about 25 percent of all children in Georgiawho are not enrolled in preschooleducation live in households that opt for home care (HBS 2006). Access to preschooleducation in preparationfor school constitutes an important proxy for education quality. As such, the government needs to oversee (and perhaps finance) children's educationindependentlyof the parent's choice of care. Internationalevidence (MIT 2006) suggests that high-quality early childhoodeducation helps prepareyoung children to succeed in school and eventually in life. Investments in early education have been shown to provide great economic returns because they are associated with lower repetition and drop-out rates throughout a student's lifetime. Also, a developed early care education industry could be economically important because it creates jobs and allows parents to be economically active. In terms of access to preschool education, Figure 7.5 shows that the ECA region is split into two groups, with Georgia among those countrieswith low enrollmentrates. Table 7.5: The number of preschoolavailable places (that is, capacity) has increased from 122,000 ............... in 2003 to 151,000 in 2006. I.__ ........ ..... .......... Total Number of Total Number of Number of Places Number of Students Preschool Students per 100 Places Institutions 1990 2,454 199,982 - - I996 1,253 83,081 - - 2002 1,213 73,26 1 122,949 60 2003 1,225 74,309 122,278 61 2004 1,247 75,361 123,178 61 2005 1,214 76,4 16 128,338 60 2006 1,197 77,922 15 1,337 51 Note: "-" means the data are not available. Source: Ministry of Education and Science 126 Figure7.5: PreschoolNet Enrollment Rates, Georgia in a Regional Context, 1989-2004 65 55 /+EUCEE 1 45 Georgia *WClS *Armenia 35 I-Georgia-Azerbaijan ~ 25 -Central Asia 15 5 .I98919901991 d ' ~ " " ' " ' ' " " , 1992199319941995199619971998l99920002001200220032004 Note: EU CEE = Central and Eastern European accession states. WCIS = Commonwealth of Independent States. Source: World Bank (2007) using UNICEF data. 293. Estimateshighlight important differencesin gross preschool enrollment rates by region, strata, ethnicity, and socioeconomiccondition.For instance, preschoolenrollment is higher (at 30 to 34 percent) amongchildren in urbanareas, of Georgianorigin, and from the richest quintiles, and lower (at 7 to 13 percent) among children in rural areas, among minority children, and amongthose living in Samtskhe-Javakheti, Inner Kartli, Lower Kartli, and Guria. 294. Nevertheless, significant progress was achieved during 2003-06 in improving preschool enrollment rates, especially among the poor, Table 7.6 presents estimates of preschool gross enrollment rates for children aged 4 to 6 by subgroup for 2003 and 2006 using HJ3S data. Estimates indicate that important improvements in preschool enrollment rates were achieved during 2003-2006, especially among children from the poorest households, living in the east side of the country, and from rural areas. Due to data limitations, we cannot assert whether such progress was the result of education policy or due to the fact that more women with children entered the labor force. However, there has been no important change in preschoolpolicy (or investments in new preschoolfacilities) since 2003 that may explain the progress achieved. Indeed, the number of preschool facilities has been decreasing steadily since the 1990saccordingto the data providedby the MoES. 127 Table 7.6: Preschoolenrollment rates have increasedfrom 19 percentin 2003 to 23 percent in 2006. Most of the increaseoccurred in the "east" side of the country. Preschool Enrollment 2003 2006 Preschool Enrollment 2003 2006 Rates; Rates; Children aged 4 to 6 Children aged 4 to 6 Strata YO. YO Location YO YO Rural 7.8 13.6 East 19.0 25.0 Urban 29.1 34.0 West 19.1 20.3 Region Quintile Kakheti 17.7 20.9 PoorestQuintile 11.3 16.1 Tbilisi 25.9 34.8 Q2 13.6 21.4 InnerKartli 12.9 10.5 Q3 20.1 24.5 LowerKartli 12.4 22.5 44 26.1 21.9 Samtskhe-Javakheti 7.9 13.8 RichestQuintile 29.6 30.4 Adjara 14.4 13.5 Ethnicity Guria 7.4 11.6 Georgian - 25.5 Samegrelo 22.6 33.7 Azeri - 6.2 Imereti 22.0 17.9 Armenian - 9.0 Mtskheta-Mtianeti 20.0 27.3 Other - 38.7 Total 19.1 23.1 Total 19.1 23.1 Note. "-" means the data are not available. Source: World Bank using Georgia2003 and 2006 HBS data. 295. Geographic location, ethnicity of the household head, and education characteristicsof the mother are important determinants of preschool enrollment. The previous analysis provides a descriptivepicture of preschool enrollment without controllingfor the effect of other variables.Figure 7.6 illustrates the conditional probability (as estimated by a basic probit regression model) of preschool enrollment for children aged 4 to 6. The bold circle represents the "zero" effect line. Characteristics associated with a higher probability of children being enrolled in preschool are plotted above the "zero" effect line, and those associated with a lower probability are plotted below the "zero" effect line. As in any probit model, the conditional probability of a given characteristic is evaluated at the mean of the characteristic's distributionand interpretedrelative to an omitted variable, as specified by the category in brackets includedin Figure7.6. 296. Estimates indicate that "locality" is the main factor affecting the probability of children aged 4 to 6 being in preschool. Controlling for other factors, children living in urban areas as well as those living in Samegrelo, Kakheti, and Lower Kartli are about 28 percent more likely to be enrolled in preschoolthan children in rural areas and those living in Tbilisi (the omitted category). Children living in households with a head of Azeri and Armenian origin are 10 to 30 percent less likely to be enrolled in preschoolcompared with those living in households of native Georgian origin. Interestingly, whether or not the children's mother has a job and/or works full time does not influence the likelihood of a child attendingpreschool, contrary to what one would expect. 297. Children from householdswith a headhpouseself-employed in the agriculture sector are 10 to 15 percent less likely to be enrolled in preschoolcomparedto children living in householdswith a headhpouse working as a wage earner in non-agriculture-related activities. The education of the mother is an important determinant of preschool enrollment. Estimates indicate that children living in households with a spouse who has attained tertiary education are 20 percent more likely to be enrolled in preschool compared to otherwise similar households with a spouse who has attained primary education and below. The education of the head, on the contrary, does not display a significant impact on hidher children's probabilityto be enrolled in preschool. 128 Figure7.6: Marginal Effect in the Probability of a Child Aged 4 to 6 Being Enrolled in Preschool, Given ObservableCharacteristics [vs. Head is Georgian or Other Ethnicity] [vs. Quintiie 11 Quintile ,-& Head is Azeri is Armenian Quintile 4, ,Head works in ag. self-employment Head of household has a job Head is employed full time Spouse works in ag self-emp Spouse of household has a job Spouse employed full or part time Mtskheta-Mtianeti Gur!tdgra' ' Spouse has some sec. edu. Spouse has some tertiary edu. 1 `Inner dad11 - - __ - -..I ~ [vs Tibilisi] amtskhe-Javakheti [vs Spouse has some pnmary ed. or below] Source: World Bank using Georgia 2006 HBS data. The model also controls for the age of the head/spouse, householdsize, and education of the head. 298. Lack of access, preference for home care, and affordability are the main reasons why parents do not send their children to preschool (Figure 7.7). Estimates usingHBS 2006 data allow for identifying the main reason why parents do not enroll their children in preschool.Urban households are more likely than rural ones to not send their children to preschool due to money shortages (30 percent in urbancompared to 8 percent in rural areas), which seems surprisingsince urban households are generally less likely to be poor than rural ones. This occurs because urban households have (for the most part) access to preschool facilities, while rural ones do not. Indeed, the majority or rural households (about 57 percent) do not have access to a preschool facility nearby. Preference for home-basedcare is also higher in urban areas than in rural areas (45 percent compared to 15 percent). Due to limited information provided by the survey, we cannot assert if such a preference relates to quality and taste or to opting for home-basedcare to not spend money on preschool care. 299. In rural areas, lack of access to facilities nearby is the main reason why parentsdo not send their children to preschool (57 percent in rural compared to only 1 percent in urban areas) (Table 7.7). Lack of access is also the main reason why children from minority groups (generally concentrated in rural areas) are not enrolled in preschool.There is also some variation in the causes for non-enrollmentacross socioeconomic groups. While money shortages, as expected, affect more households in the poorest consumption quintiles, preference for home care (maybe because of low quality of preschool institutions and/or because mothers from richer households can afford to stay at home) is more frequent among households in the richestconsumptionquintiles. Finally, note that about 7 of every 100 children aged 4 to 6 are not enrolled becausetheir parentsconsider them too youngto start their education. This may explain why, as mentioned, estimates by age suggest late enrollment into primary education (1 to 2 years later than when the education system stipulates). 129 Figure7.7: About 60 percentof all children from poor householdsare not enrolled in preschool due to access and affordability constraints. 70 1 =Rural +Urban 701 Too young Prefers Health Money No access Other Poorest Q 2 Q3 Q4 Richest home care problems shortages nearby Quintile Quintile Source: World Bank using Georgia 2006 HBS data. Table 7.7: Lack of access nearby is the main reasonwhy children from minority groups are not enrolledin preschool. Why is the Child Not Rural Urban Georgian Azeri Armenian Total Enrolled in Preschool? YOtoo young 6.7 10.45 9.7 0.7 2.5 7.9 % prefershome care 15.49 45.17 27.8 12.9 17.5 25.2 % healthproblems 2.16 5.58 3.9 0.0 0.0 3.3 YOmoney shortages 7.79 29.78 15.6 5.8 20.6 15.0 %no access nearby 57.3 1 0.73 33.1 69.1 53.4 38.7 % other 10.55 8.29 10.0 11.5 6.0 9.8 Total 100 100 100.0 100.0 100.0 100.0 Source: World Bank using Georgia 2006 HBS data. 300. Compared to 2003, affordability constraintsto preschooleducation have decreased in rural areas and in the western side of the country. Figure 7.8 presentsestimates of the share of households in Georgia that claim to not send their children to preschooldue to access and affordability constraintsfor 2003 and 2006. While the national shares do not display a significant change duringthe period, estimates indicate importanturbadrural and east/west variations.The share of households in the eastern side of the country claimingto not have access to preschool facilities nearby increasedfrom 28 percent in 2003 to 36 percent in 2006. Since 97 percent of all preschools in Georgia are publicly owned, this may indicatethat some preschools may have been shut down. Indeed, the number of state-owned preschool institutions drastically decreased during 1990-2006-from 2,454 institutionsto 1,197. On the contrary, affordability constraints (generally related to non-tuition out-of-pocket expenditures in transport, food, and school supplies) displayeda significantreductionin rural areas and in the eastern part of the country. This result may be explained by the fact that overall income levels in poor rural areas increased significantly during 2003-2006 due to economic growth and expansion in coverage of targeted social assistance transfers (World Bank 2008). 130 Figure 7.8: Compared to 2003, the share of households not sendingtheir children to preschooldue to lack of facilities has increased in the eastern part of the country. IChlldren not In preschool due to lack of access Children not in preschooldue to lack of money ' O % l 30% 5 27% T o 2003 2006 60% 50% 45% 40% 30% 20% 10% 0% Rural Urban East West Rural Urban East West Note: Lines in bars display confidence intervals (statistical significance is assured when confidence intervals do not overlap). Source; World Bank using Georgia 2003-2006 HBS data. B.2. BasicEducation 301. While there are not large differences in access to basic education across subgroups, perceptions of quality vary significantly across regions, strata, and ethnic groups (Table 7.8). According to HBS (2006) data, enrollment rates for basic education in Georgia oscillate between 95 and 100 percent for children in different strata, regions, and consumption quintiles (Box 7.2). HBS estimates indicate that 76 percent of all children aged 7 to 17 who claimed not to be enrolled in basic education were so due to health problems. Access and affordability constraints account for less that 10 percent of why students claim not to be enrolled in basic education. Table 7.8: Enrollment Rates in Basic Education Enrollment Rates in Basic 2003 2006 Enrollment Rates in Basic 2003 2006 Education, Aged 7 to 17 Education,Aged 7 to 16 Strata YO YO Location YO YO Rural 95.2 East 96.1 Urban 97.6 West 96.7 Region Quintile Kakheti 95.6 Poorest Quintile 96.3 Tbilisi 98.3 Q2 95.4 Inner Kartli 95.7 Q3 96.4 Lower Kartli 94.2 Q4 96.5 Samtskhe-Javakheti 97.1 Richest Quintile 97.5 Adjara 95.1 Ethnicity Guria 97.2 Georgian 96.6 Samegrelo 96.9 Azeri 93.5 Imereti 95.5 Armenian 95.9 Mtskheta-Mtianeti 100 Other 100 Total 96.4-_ _Total - - 96.4 * -_._ ~ -- - -. - -- Note Due to data constraints we could not calculate enroirmGnt rates by level - - __-- - - - - Source World Bank using Georgia 2003-2006 HBS data 131 Box 7.2: Enrollment Rates in Basic Education in Georgia As documented in Godfrey (2007), confusion still remains over the actual levels of net enrollment rates in basic education and primary completionrates in Georgia. Net enrollment rates: According to data provided in the UNICEF TransMONEE (2007) database, net enrollment rates in basic education in Georgia oscillatedbetween 95 and 97 percent during 1996-2002. This made Georgia the best-performingcountry in the Caucasus regionin this respect until 2002. After 2002, data become unavailable since calculations submitted to the UNICEF Innocenti Research Center in Florence, which maintains the TransMONEE database, were found to be inconsistent. Since then, a new data series on net enrollmentrates has been compiledby UNESCO.According to UNESCO(2006), net primary enrollmentrates decreased in Georgia by 5 percentagepoints during 2003-05-fiom 92 percent in 2003 to 87 percent in 2005-a decrease that has not been acknowledged (or accepted) by the Government of Georgia, which claims that net primary enrollment rates are close to universal accordingto data providedby its Education ManagementInformationSystem (EMIS). Gross enrollment rates: HBS 2006 data indicate that gross enrollment rates in basic school are 96.4 percent for children aged 7 to 17 (95 percent in rural areas and 98 percent in rural areas). Differencesacross regions are mild, with Lower Kartlibeingthe only regionwith a gross enrollment rate below 95 percent. Primary completion rates: Data from the World Development Indicators dataset indicate an important deteriorationin primary completionrates since 2004. However, according to the World Bank (2008a), Georgia is on the list of ECA countries that will likely achieve Millennium Development Goal 2 (MDG 2) (universal primary education). Data published by the Ministry of Education and Science based on the EMIS indicate that primary completionrates in 2005 were 98 percent (MoES 2007). To move toward achievingthe MDGs, Georgia became an FTI (Fast Track Initiative) partner in July 2007 after endorsement of Georgia's 2007-201 1 Education Sector Strategy and Action Plan.aGross enrollment rates at the lower-secondary-educationlevel increasedfrom 91 percent in 2003 to 95 percent in 2005, catching up with Armenia and Azerbaijan. At the upper-secondary level, gross enrollment rates fell sharply from 75 percent in 1989 to 50 percent in 1994, and the trend has gradually reversed since 1995. a. Despite the fact that Georgia was not eligible for financial support from the FTI catalytic fund, the process was very helpful in strengthening donor coordination in the country. The strategy document provides a useful tool for monitoring progress in the sector and providing coordinated support. 302. Household perceptions in regard to school performance differ largely across regions, suggesting inequalities in education quality. Despite equal access, important differences in relation to quality still remain, especially across different strata and regions. As Figure 7.9 illustrates, only 39 percent of all households in rural areas and 66 percent of all households in urban areas feel that their children attend schools that have acceptable learning and school environments. Unfortunately, HBS data do not go deeper into the reasons why households make such statements, although important differences in quality perceptions exist across regions. While the majority of households in Imerti, Samagrelo, Kekheti, M.-Mtineti, and Inner Qartli are not satisfied with their children's school conditions, the majority of all households (50 to 77 percent) in Lower Qartli, Adjara, Guria, Samtskhe-Javakheti, and Tbilisi are satisfied. Households of Azeri origin (mainly living in Inner Qartli and Kakheti) seem less satisfied with their children's school conditions compared to households of Armenian origin (mainly living in Lower Qartli and M.-Mtineti). It is surprising that differences in perception of school conditions do not differ much across income quintiles (approval rates on school conditions vary between 50 and 57 percent across all quintiles). While school conditions may be perceived similarly across socioeconomic conditions, quality of teaching and expectations (both very important learning factors) may also differ across different socioeconomic groups. Unfortunately, HBS data do not collect information in this regard. 132 Figure 7.9: Only half of all householdsthink that school conditionsmeet the educationalneedsof their children in relationto learning and school environment. 90 3 C ' E E C P o ' z p f $ $ . ? % 2 Y - B a ; S a ? $ a % + v) z .EY, 5-1 7 v) Note: Sample: children aged 7 to 17 enrolled in basic education. Source: World Bank using Georgia 2006 HBS data. 303. Most basic education services in Georgia are supplied by the State, with only a small share of children (generally the better off) being enrolled in private schools. Less than 4 percent of all children in Georgia who are enrolled in basic education are enrolled in private schools. Private schools, generally more accessible to households from higher socioeconomic groups, are often associated with better learning outcomes. As indicated in Figure 7.10, urban children and especially those from richer households and of non-Georgiadnon-Azerihon-Armenian origin (probably European expatriates) are more likely to be enrolled in private education. PGents of children enrolled in private education seem to have rather similar levels of satisfaction with school conditions (with acceptancerates varying between 53 and 54 percent) compared to parents of children enrolled in private sector schools. This may occur because public and private schools have similar levels of overall quality and/or parents of children in private schools may be more critical (and/or have different expectation levels) about their education, since most parents pay for it out-of-pocket. Unfortunately, there are no data available for assessing differences in quality (as proxied by test scores) between public and private schools in Georgia. Figure 7.10: Less than 4 percentof all children in Georgiawho are enrolled in basic education attend private schools. School conditions meet education demands? 9 1 8 1 M" ..I 15% Public Schools No Private Schools Note: Sample: children aged 7 to 17 enrolled in basic education Source: World Bank using Georgia 2006 HBS data. 133 304. Compared to 2003, school approval rates among the poor have deteriorated, and amongthe non-poor have improved. Estimates in Table 7.9 provide subjective perceptions as to whether school conditions meet education demands for 2003 and 2006. Approval rates increased among households in urban areas from 56 percent to 66 percent, and in the eastern side of the country from 56 percent to 66 percent.On the contrary, approval rates decreased in rural areas from 49 to 40 percent, and in the western side of the country from 48 percent to 36 percent.Approval rates fell among the extreme poor from 60 percent in 2003 to 46 percent in 2006, and increased among individuals in the highest quintile, from 54 percent to 61 percent. These phenomena may be associated with the introductionof per capita financing. A hypothesis is that larger schools with more students (generally locatedin urban areas and in the eastern part of the country) receive more funding than smaller schools (generallylocatedin rural areas and in the western side of the country). Indeed, there is anecdotal evidence suggestingthat some small rural schools (where generally poor children attend) may be or may have become underfundedafter the introductionof the per capita financing scheme (accordingto MoES, the evidence that small schools are underfundedis quite strong). This, as expected, may affect overall education quality. Nevertheless, data at hand do not make it possible to attribute a direct linkage between the two occurrences and, as mentioned, more analysis inthis respect shouldbe conducted. Table 7.9: School Approval Rates in Georgia School Conditions Meet 2003 2006 School Conditions Meet 2003 2006 Educational Demands Educational Demands Strata YOYes ?LOYes Location %Yes %Yes Rural 49.60 40.68 East 56.16 66.15 Urban 56.59 66.49 West 48.06 35.73 Region Quintile Kakheti 38.52 31.70 Poorest Quintile 59.97 46.12 Tbilisi 60.97 79.88 Q2 50.16 53.78 Inner Kartli 52.67 65.37 Q3 55.59 52.93 Lower Kartli 53.48 59.90 Q4 41.75 58.91 Samtskhe-Javakheti 74.78 71.94 Richest Quintile 53.89 60.85 Adjara 60.87 52.37 Ethnicity Guria 35.52 53.14 Georgian 50.64 53.79 Samegrelo 44.25 30.53 Azeri 50.03 51.12 Imereti 39.52 28.39 Armenian 73.32 71.12 Mtskheta-Mtianeti 51.71 43.77 Other 90.84 87.10 Total 52.78 55.48 Total 52.78 55.48 Note: Sample: children aged 7 to 17 enrolled in basic education. Source: World Bank using Georgia 2003 and 2006 HBS data. 305. A large proportion of secondary school students in Georgia rely on "out-of-class" private tutoring to improve their quality of learning and prepare them for tertiary education. Private tutoring has been increasinglyrecognizedto be of major importancein Central and South Eastern Europe, the Former Soviet Union, and Mongolia (OS12006). After the process of reform after the 1990s, private tutoring has been viewed as an effective way for childrenand young adults to adapt to the new reality and cope with system changes. Private tutoring has been perceived as an important "quality" supplement to the mainstream educationalsystem (Bray and Silova 2006; OS12006; Matiashvili and Kutateladze2006; Silova and Bray 2006b). Proliferationof private tutoring in Georgia is relatedto two key factors-highly competitive Higher education entrance examinations and the deteriorating quality of mainstream schooling during the transition years. The admission policy was modified (in 2005) by the government through the introduction of centralized Higher education entrance examinations; demand for private tutoring, however, still remains high. A recent Open Society Institute(09)(2006) study, "Education in a Hidden Marketplace: Monitoring of Private Tutoring," covers eight countries in transition, including 134 Georgia. It indicates that 69 percent of students participating in the survey had received some type of private tutoring at the upper-secondary-educationlevel. The figure is even higher for Georgia, where almost 80 percent of respondents in a sample of higher education students claimedto have receivedsome form of privatetutoring. 306. According to OS1 (2006), the perception of the poor quality of mainstream education emerged as a major factor in demand for private tutoring in Georgia. Students generally use private tutoring as "an enrichment strategy" to perform better in examinationsthat are requiredto access higher education.Dueto data constraints, it is not possibleto know whether privatetutoring is a responseto poor education quality overallor a particular response to poor preparationfor higher educationentry exams (or both). Knowing which motivation would be useful to determine whether the system needs to address broader issues or particular issues related to higher education entry exams. Table 7.10 reveals that 28 percent of all students enrolled in basic educationindicatedthat they receivedsome type of privateout-of- school tutoring in the previous 12 months. Private tutoring is more frequent among children in urban areas (39 percent compared to 17 percent in rural areas) and among children from richer socioeconomic quintiles(50 percent among children in the richest quintile compared to 17 percent among children in the poorest quintile). According to OSI's survey, the most widespread type of private tutoring in Georgia found was "one-to-one" or "small group" tutoring. Among the higher education students who were interviewed in Georgia, the most common subjects for private tutoring were mathematics, foreign languages, native language, and history, which coincided with the subjects required at higher education entrance examinations.One caveat regardingthe OS1 results is that data were collected right at the time when higher education admissions policy was reformed.As such, more time is needed to assess changes in demand for privatetutoring under the new admissions scheme. Table 7.10: One-third of all students in basic education participate in additional out-of-class --- - Did Student Get-Extra-Private- __- - - - . - - - . _ _ _ tutoring.- - _ _ - __-. - Total Urban Rural Poorest Q3 - - - -Richest - Education during Last 12 Quintiles Quintile Months? % yes 28.25 38.92 16.81 16.84 24.62 49.69 % no 71.75 61.08 83.19 83.16 75.38 50.3 1 Why not? % no desire 20.84 17.65 23.33 22.36 21.98 28.33 ?LOlack of money 49.89 64.99 38.07 51.25 47.45 39.01 % lack of time 5.06 4.8 5.27 1.76 5.79 8.32 % no access nearby 16.66 3.3 27.12 18.46 15.39 16.68 % other 7.55 9.26 6.21 6.17 9.39 7.66 Note: Sample: children aged 7 to 17 enrolled in basic education. Source: World Bank using Georgia 2006 HBS data. 307. Affordability and access constraints constitute the main reason why children do not have private tutors, especially among the poor. Privatetutoring may also bring negativeconsequences, such as exacerbating social inequities and decreasing the likelihood of less privileged children to do well in higher education entrance examinations. According to OS1 (2006), the vast majority of students in Georgia admit that "students use private tutoring because the school curricula do not cover everything that is required on the higher education entrance examination" (Box 7.3). Yet, estimates in Table 7.11 indicate that a large share of all children in Georgia do not use private tutoring due to access and/or financial constraints.Financial constraints are the most importantfactor explaining why 65 percent of all children in urban areas and 50 percent of all children in rural areas do not use private tutoring services. Access constraints, such as lack of qualified tutors or tutoring institutions, are other important factors for not using private tutors, especially in rural areas and among children in the poorest quintiles. Not surprisingly, the share of students claiming not to use private tutors due to lack of finances decreasesfor 135 children in upper quintiles. Nevertheless, even at the highest consumption quintile, 4 out of every 10 children do not use private tutors due to lack of financial means. As such, private tutoring is becoming a luxury good that seems to be affecting the likelihood o f children to access tertiary education. According to OS1 (2006), approximately 82.9 percent of the school students receiving private tutoring had parents with higher education compared to 2.3 percent of students with parents having general secondary education. Among the surveyed school students who had some form-of private tutoring, almost all were from families with medium and above medium economic status (OS1 2006; Matiashvili and Kutateladze 2006; Silova and Bray 2006b). It is worth noting that the MoES, in order to make education more accessible, launched a program for social support to students from mountainous regions, conflict zones, representatives of ethnic minorities, and socially vulnerable families. This program offers to beneficiary students preparatory courses for unified national examinations. Table 7.11: Perception of their Socioeconomic Status amongHigher Education Students Who Have UsedPrivateTutors Perception of Family Status Used Private Tutoring (YO) Paid More than 300 GEL ("/o) Good 18.8 21.9 Medium 69 72.2 Bad 12.1 5.9 Total 100 100 Source: Bray and Silova (2006); OS1(2006); Matiashvili and Kutateladze(2006); Silova and Bray (2006b). Box 7.3: Perceptionof Private Tutoring and Higher Education Entrance Examinations in Georgia The connection between higher education entrance examinations and private tutoring fits a larger global pattern of examinations prompting investments in supplementary private tutoring. The subjective perceptions of students also support the finding that tutoring in Georgia is oriented toward preparing students for competitivehigher education entrance examinations. The majority of students in Georgia believed that private tutoring would increase their chances to enter higher education (83.4 percent of the higher education sample and 73 percent of the school sample) and that private tutoring was the only way to pass higher education entrance examinations (58.3 percent of the higher education sample and 53 percent of the school sample). Furthermore, the majority of respondents believed that students who received tutoring were more likely to enter higher educationthan students of equal ability who did not receive tutoring (62.3 percent of the higher education sample and 61 percent of the school sample). Importantly, students believed that private tutoring had practically become a prerequisite for receivingstate funding for higher education (which only 60 percent of students received). Some students mentioned during the interviews that they preferred to invest in private tutoring for a couple of years and get state-fundedplaces than not to invest in private tutoringand pay for higher education for four or more years. ISource: OS1(2006). I 308. Education/ethnicity of the household head, region, and socioeconomic conditions are important determinants of private tutoring usage. Figure 7.11 illustrates the conditional probability (as estimated by a basic probit regression model) of using private tutoring for children enrolled in basic education. The bold circle represents the "zero" effect line. Characteristics associated with a higher probability of using private tutoring are plotted above the "zero" effect line, and those associated with lower probability are plotted below the "zero" effect line. Controlling for other factors, estimates indicate that children from minority households (Azeri and Armenian) are 18 to 19 percent less likely to use private tutoring compared to otherwise similar students of Georgian origin. Also, children living in female-headed households are 4 percent less likely to use private tutoring compared to children living in male-headed households. Geographic location is also an important determinant in explaining the usage of private tutoring. Estimates indicate that children living in Samtskhe-Javakheti and Imereti are 14 to 18 percent more likely to use private tutoring, and those living in Adjara are 7 percent less likely, compared to otherwise similar children living in Tbilisi. As expected, children in urban areas and from richer 136 quintiles display a 4 to 25 percent higher probability of using private tutoring comparedto childrenliving in rural and in poorerhouseholds. Figure7.11: Children living in minority households are less likely to use private tutoring. Head is female 30.0% Head IS Azeri [vs. Head is Georgian] [Vs. Quintile 11 Quintile5 ead is Armenian has some secondary ead has voc sec educalion ead has some tertiary education + [Vs. Headwith Samtsl