Report Report No. 24638-KG No. Kyrgyz Republic 24638-KG Enhancing Pro-poor Growth September 30, 2003 Poverty Reduction and Economic Management Unit Europe and Central Asia Region Kyrg yz Republic Enhancing Pr o-poor Gr owth Document of the World Bank CONTENTS Acknowledgements .......................................................................................................... i Executive Summary ....................................................................................................... ... 111 PartI:Growth-Poverty Linkages ............................................................. 1 1 HasEconomic Growth ReducedPoverty? . ............................................................. 3 A. ThreeEpisodesof Growth. 1996-2001................................................................. 3 Sources of Growth. 1996-2001....................................................................... 7 B. Growth-Poverty Reduction Linkages. 2000-2001 ................................................ 8 Measuring Welfare.......................................................................................... 8 Outcomes ...................................................................................................... 10 Pro-Poor Growth. 2000-2001........................................................................ 13 Inequality...................................................................................................... 15 C. Poverty Comparison Over a Four-Year Period. UsingHBS 1998-2001............ 16 Methodological Issues .................................................................................. 16 Outcomes ...................................................................................................... 19 D. Conclusion: A SeriousPoverty Reduction Challenge Continues...................... 22 Simulation of Impacts of Growth Rates on Poverty Reduction.................... 23 Importance of Broad-Based Growth............................................................. 24 2 Agricultural Growth and RuralPoverty Reduction . ............................................ 27 A. TheSectoral Recovery in Kyrgyz Agriculture ................................................................. 27 Performance.................................................................................... 27 Sectoral Policy .............................................................................................. 31 B. RuralPoverty ...................................................................................................... 33 RuralPoverty Profile .................................................................................... 33 Sources of RuralIncome............................................................................... 33 Assets ............................................................................................................ 35 Determinants of RuralPoverty ..................................................................... 39 C. Conclusions on EnhancingRural Growth andPoverty Reduction..................... 41 LandMarkets. LandUse andLeasing .......................................................... 42 Supporting Development of Agricultural Markets ....................................... 43 Agricultural Support Services....................................................................... 44 Irrigation Policy ............................................................................................ 44 Monitoring RuralDevelopment.................................................................... 45 3 Why is FormalJob Creation So Low? . .................................................................. 46 A. Overall Labor Market Development................................................................... 46 Stable Employment and Low Open Unemployment .................................... 47 Large Hidden Unemployment and Widespread Informal Employment .......50 B. LaborMarket Restructuring andJob Reallocation............................................. 52 Privatization Has Been Faster Than Restructuring....................................... 52 Movement of Labor Towards Agriculture and Trade................................... 54 Positive but Uneven Gains in Labor Productivity andWages...................... 56 Formal Job Creation Remains Very Low ..................................................... 57 C. Barriers and Opportunities for Job Creation....................................................... 59 Public Investment Has Been the Main Source of Demand & Employment Growth .......................................................................................................... 59 Small and MediumEnterprises Still Face a Rather Unfavorable Environment.................................................................................................. 61 Social Security Contributions are High and Contributors Are Prone to Skepticism..................................................................................................... 64 Labor Market Flexibility and Labor Standards............................................. 65 D. The Challengeof MatchingJobs WithPeople ................................................... 66 The Slulls Gap............................................................................................... 66 Regional Mismatch....................................................................................... 67 External and Internal Migration.................................................................... 69 E. Conclusion andRecommendations..................................................................... 71 Part 11: Profile of Poverty ..................................................................... 75 4.Profile of Income Poverty in 2001 ......................................................................... 77 A. Characteristics of Poverty................................................................................... 78 RegionalDimension of Poverty.................................................................... 78 Housing Conditions andPoverty .................................................................. 82 B. Determinants of Poverty..................................................................................... 85 Income Poverty and Labor Market Outcomes .............................................. 84 The Majority o f the Poor are Working Poor................................................. 87 Poverty and Characteristics of the Household Head..................................... 89 Poverty and Household Composition ........................................................... 90 C. Conclusion .......................................................................................................... 91 5 Human and Social Dimensions of Poverty . ............................................................ 93 A. Multidimensional Approach ............................................................................... 93 B. Education andPoverty inthe Kyrgyz Republic.................................................. 96 Education andEnrollments ........................................................................... 95 Household Expenditures on Education......................................................... 99 Differences inEducation Quality................................................................ 102 Education and Earnings .............................................................................. 103 Benefit Incidence Analysis ......................................................................... 104 Conclusion .................................................................................................. 104 C. Health and Poverty in the Kyrgyz Republic ..................................................... 105 Health Outcomes......................................................................................... 105 Public Health............................................................................................... 107 Government Expenditures .......................................................................... 108 HouseholdExpenditures ............................................................................. 112 Health Care Utilization ............................................................................... Achieving the Health Millennium Development Goals (MDGs)...............112 115 Conclusion .................................................................................................. 116 6 Dynamics of Poverty . ............................................................................................. 118 A. What are Chronic andTransient Poverty.......................................................... 118 Why Dynamic Analysis i s Important.......................................................... 119 How Chronic and Transient Poverty Are Measured................................... 119 B. Growth and Changes inWelfare: Determinantsof Transient and Chronic Poverty .............................................................................................................. 121 Economic Mobility and Poverty................................................................. 122 Consumption Volatility............................................................................... 124 Remaining Structural Inequalities............................................................... 125 Regional Inequalities in Consumption........................................................ 126 Assets .......................................................................................................... 128 C. Conclusion and Policy Recommendations ........................................................ 129 Part111: Protecting The Poor ................................................................ 131 7 Social Impactof Reformingthe Electricity Sector . ............................................ 133 A. Reforming the Electricity Sector....................................................................... 134 Impact of Tariff Reform on Households..................................................... 135 B. Impact of Unreliability of Electricity Supply.................................................... 137 Energy Substitution to Dirty Fuels for Heating .......................................... 138 C. Conclusion andPolicy Recommendations ........................................................ 139 Administrative Reform ............................................................................... 140 Social Protection Measures......................................................................... 141 Conclusion .................................................................................................. 143 8 Social RiskMitigation . ........................................................................................... 145 A. Overview of the Social Protection System ....................................................... 145 B. The Impact of Social Protection Programs on IncomeDistribution................. 153 Coverage of Social Protection Programs .................................................... 155 Distribution of Social Protection Benefits Across Groups and Benefit Adequacy .................................................................................................... Dynamic Incidence of Social Protection Programs .................................... Average and Marginal Incidence of Social Protection Programs...............156 157 159 C. TheDuplications Impact of Social Protection Programs on Poverty Reduction ................... 160 and Gaps inCoverage............................................................ 160 Aggregate Impact on Poverty ..................................................................... 161 Cost-Benefit Analysis ................................................................................. 162 Overall Effectiveness of SP Programs inReducing Poverty...................... 164 D. Scaling-Upthe Poverty ReductionImpact of Social ProtectionSpending.......165 E. Conclusion andPolicy Recommendations........................................................ 169 Annexes Annex 1.1: Background Papers Written by Kyrgyz Researchers............................... 173 Annex 2.1: Earlier Poverty Comparisons ................................................................... 174 Annex 3.1: Decomposing the Sources of Growth and Their Impact on Employment............................................................................................. 176 Annex 4.1: Agricultural and Livestock Production.................................................... 177 Annex 4.2: Land and Farm Structures........................................................................ 180 Annex 4.3: RandomEffects Regression of Log Expenditure per Adult Equivalent. Rural Kyrgyz Republic. 1998-2001...................................... 185 Annex 5.1: Determinants of Poverty .......................................................................... 186 Annex 6.1: Enrollment Rates by Age and Selected Household Characteristics. 2001.......................................................................................................... 189 Annex 7.1: Dynamic Analysis.................................................................................... 191 Annex 8.1: Modeling the Behavioral Responseof Householdsto Public Transfers .................................................................................................. 193 Annex 8.2: Calibratinga Proxy-MeansTest for the Kyrgyz Republic ...................... 197 References........................................................................................ 199 Tables Table A PovertyDeclinedin 2001......................................................................... iv Table 1.2 GeneralMacroeconomic Indicators. 1996-2001....................................... 6 Table 1.3 Poverty Measures for 1998.2001. for HBS Data. Basedon the Panel 11 Absolute Poverty Inthe Kyrgyz Republic. 2000-2001........................... andWhole Sample .................................................................................. Table 1.4 11 Table 1.5 Extreme (Food) Poverty Inthe Kyrgyz Republic. 2000-2001................ 13 Table 1.6 Average ConsumptionBy Decile. 2000-2001 (in 2001prices).............. 13 Table 1.7 IncidenceOf Poverty By Location. 2000-2001...................................... 14 Table 1.8 DecompositionOf Poverty Into Growth And Redistribution Components. 2000-2001......................................................................... 15 Table 1.9 Gini Index Of Inequality In2000 And 2001........................................... 16 Table 1.10 Average Monthly Per CapitaConsumptionFor Panel And Whole Sample Households. By Quarter 1998.2001. And By Urban-Rural (In Monthly Dec. 2001 Soms) ................................................................ 18 Table 1.11 DecompositionOf Overall Poverty Change Into Growth And Inequality Change. 1998-2001................................................................ 19 Table 1.12 PovertyIndicatorsFor RuralAnd UrbanSectors. 1998.2001. Panel Households.............................................................................................. 20 Table 1.13 Gini Index Of Inequality 1998.2001. InConsumptionPer Capita. For PanelDataset........................................................................................... 21 Table 2.1 Crop Acreage .......................................................................................... 30 Table 2.2 Rural Poverty Rates (2001)..................................................................... 33 Table 2.3 Rural Market Participation Rates (In Percent) (HBS) ............................ 35 Table 2.4 Distribution Of Cattle By Household. 2001FullRural Sample (9%) ......36 Table 2.5 Distribution Of Small RuminantsBy Household. 2001FullRural Sample (%).............................................................................................. 37 Table 2.6 Distribution Of Agricultural LandArea By Household.2001Full Rural Sample (%).................................................................................... 37 Table 2.7 OwnershipOf Tractors. Trucks And HorsesBy ConsumptionQuintile. 2001 (NumberOf Units)......................................................................... 37 Table 2.8 Access To InfrastructureKJtilities In2000 (Percent Of Population With Access) ........................................................................................... 40 Table 3.1 GDP Growth. Changes InEmployment AndUnemployment. 1995-2000............................................................................................... 47 Table 3.2 Level Of Employment, Unemployment, andLabor Participation Rate By Sex, 1995-2000 (Percent)........................................................................ 49 Table 3.3 Distribution Of Registered Unemployment Spells By Duration ............49 Table 3.4 HiddenUnemployment And Informal Employment, 1996-1998 (Percent) .................................................................................................. 51 Table 3.5 InformalEmployment InThe Kyrgyz Republic (Percent) ..................... 52 Table 3.6 Number O f IndustrialEnterprises, 1989-2000........................................ 53 Table 3.7 Reasons For Being Registeredas Unemployed...................................... 54 Table 3.8 Cumulative Growth InReal GDP AndEmploymentDuring 1996-2000 (Percent) ................................................................................ 54 Table 3.9 Employment By Sector (Thous. O f People And As % Of Total Employment)........................................................................................... 55 Table 3.10 Cumulative Growth InLabor Productivity And Real Wages During 1996-2000 (Percent) ................................................................................ 56 Table 3.11 Gross Job Flow Rates, 1999.................................................................... 58 Table 3.12 Job Creation And Job Destruction: Kyrgyz Republic Versus Selected Job FlowsBy FirmSize And By Ownership Status, 2000 ..................... Countries (%) .......................................................................................... 58 Table 3.13 59 Table 3.14 Decomposition O f Employment Growth (Percent) ................................ 60 Table 3.15 Net Foreign Direct Investment InThe Kyrgyz Republic And Selected Countries (Percent Of GDP) ................................................................... 61 Table 3.16 Contribution Of SMEs To GDP AndEmployment, 1998 ...................... 61 Table 3.17 (Direct) Cost Of Opening A Business InThe Kyrgyz Republic, 2000 ..62 Table 3.18 Social Contributions, 2001 (Percent O f Gross Wages) .......................... Table 3.19 Relaxed L O Unemployment Rate By Level Of Education (Percent) ....65 67 Table 3.20 UnemploymentRates By Region, 2000 (Percent) .................................. 68 Table 3.21 Variability Of Regional Unemployment, 1995-2000 ............................. 68 Table 3.22 Internal Migration And Immigration InThe Kyrgyz Republic ..............70 Table 4.1 Poverty And Inequality By Oblast, 2001................................................ 78 Table 4.2 Percent Of Households With Access To Services, 2001 ........................ 82 Table 4.3 Poverty Rates, By Labor Market Characteristics (Percent) .................... 86 Table 4.4 Headcount Index of Poverty By Education O f Household Head, 2001 .89 Composition Of The Poor By Labor Market Characteristics (Percent) ..87 Table 4.5 Table 4.6 Poverty And Household Composition, 2001 (Percent)........................... 90 Table 5.1 Enrollment-Rate Differentials InBasic Education (Grades 1-9) (InPercent).............................................................................................. 97 Table 5.2 Enrollment-Rate Differentials For Secondary And Higher Education (InPercent).............................................................................................. 98 Table 5.3 Secondary Education Expenditures As A Share Of Household Income, 1996-1998 ............................................................................................. 100 Table 5.4 Total Education Expenditures As A Share Of Household Income, 1997....................................................................................................... 100 Table 5.5 HouseholdExpenditures On Education By Level And Category O f Expenditure. 2001 (Households With Enrolled Members Only. InSoms PerEnrolled HouseholdMember) .......................................... 101 Table 5.6 Headcount Incidence O f Poverty By Education O f HouseholdHead. 2001....................................................................................................... 103 Table 5.7 Benefit Incidence O f Public Expenditures On Education B y Level Of Education And Poverty Status Of Beneficiaries. 2001......................... 103 Table 5.8 Health Outcome Indicators. 1999 ......................................................... Table 5.9 Main Health Status Measures For the Kyrgyz Republic. 2000-2001...105 105 Table 5.10 Infant Mortality Rate. By Region. 2000-2001 ...................................... 106 Table 5.11 Under-5 Mortality Rate. By Region. 2000............................................ 106 Table 5.12 Trends InSelected Communicable Diseases. IncidencePer 100.000 People. 1998-2001................................................................................ 106 Table 5.13 Health Behaviors By Poverty Group. 1998-2001................................. 108 Table 5.14 Public Spending On Health InRelation To GDP. 1995-2001 ..............109 Table 5.15 Average Monthly Salary For Health Sector Employees....................... 111 Table 5.16 Per Capita Health Expenditures (In SOM). By Quintile. Poverty Status. And Area Of Residence ............................................................ 112 Table 5.17 Per Capita Health Expenditures (In SOM). By Poverty Status. And Area Of Residence ................................................................................ 112 Table 5.18 Those With Acute Illness Seeking Medical Assistance Within the Last 30 Days. By Region and Area............................................................... 113 Table 5.19 Those With Acute Illness Seeking Medical Assistance Within the Last Utilization Of Hospital Inpatient Services By Region And Area .........114 30 Days. By Poverty Status .................................................................. Table 5.20 Table 5.21 Utilization Of Hospital Inpatient Services By Poverty Status ..............114 115 Table 5.22 Utilization Of Health Care Services. By Quintile And Type O f Provider............................................................................................ 115 Table 5.23 Utilization Of Health Care Services. By Area O f Residence. Poverty 114 Meeting The Health MDGs .................................................................. Status. And Type Of Provider............................................................... Table 5.24 116 Table 6.1 Transitions In And Out Of Poverty. 1998-2001 ................................... Table 6.2 Mobility InThe Kyrgyz Republic And Other Developing Countries ..118 122 Table 6.3 Changes InWelfare From 1998 to 2001............................................... 123 Table 7.1 Tariff Adjustments. 2000-2002............................................................. 134 Table 7.2 Oil Equivalent Energy Unit Costs. April 2002 ..................................... 136 Table 7.3 Cost Of Life-line Tariff And Cost Of A Direct Income Transfer That Would Provide The Poor With The Same Per Capita Benefit. 2001....142 Table 7.4 Median Monthly Electricity Consumption. kWh. Bishkek. 2000-2002 ............................................................................................. 143 Table 8.1 Social Expenditures Of Consolidated General Government. 1995-2001............................................................................................. 148 Table 8.2 Overview O f The Main Social Protection Programs. Kyrgyz Republic. 2001......................................................................... 150 Table 8.3 Set Of Implicit Poverty Lines Used InThe Implementation O f SP Programs In2001.................................................................................. 150 Table 8.4 Receipt Of SP Benefits: Household-Level Descriptive Statistics. H B S 2001.............................................................................................. 151 Table 8.5 Coverage O f The Social ProtectionPrograms. 2001............................ 156 Table 8.6 Distribution Of Social Protection (SP) Benefits. 2001......................... 156 Table 8.7 Distribution Of Program Beneficiaries By Quintile. Average And Marginal. 2001...................................................................................... 158 Table 8.8 Duplications And Gaps InThe Coverage With Social Protection Programs ............................................................................................... 161 Table 8.9 UMBTargeting Performance ComparedTo A Proxy-Means Test......169 Annex Tables Table A2.1- 1 ReportedPoverty Trends (percent) .......................................... 174 Table A4.1-1 Sown Area by Type of Crop ................................................................. 177 Table A4.1-2 Yield Dynamics of Main Commodities................................................ 177 Table A4.1-3 Yields of Major Commodities .............................................................. 177 Table A4.1-4 Crop Production (000tons)................................................... 178 Table A4.2-1 Land Shares for Permanent Use of the Kyrgyz Republic (as of Jan. 2001) Table A4.2-2 Agricultural Land by FarmStructure (January 2001) (hectares) ..........180 ............................................................................................................... 180 Table A4.2-3 LandRedistribution Fund-Available Landby Oblast (January 2002) (hectares) ............................................................................................... 181 Table A4.2-4 Lease Contracts and Lease Paymentsfor Use o f Land Redistribution FundLand in 2000 (January 1.2002) ........................................................................... 182 Table A5.1-1 Structure of Household Income (% of Total Income). 2001 ................ 186 Table A5.1-2 Estimates of Per Capita Consumption Determinants inthe Kyrgyz Republic. 2001 (total sample) ............................................................... 187 Table A5.1-3 Esiimates of Per Capita Consumption Determinants inthe Kyrgyz Republic. 2001 (Urbanvs.Rural).............................................................. 188 Table A7.1-1 Factors Associated with Changes inper Capita Consumption 2001/1998 ......................................................................................... 191 Table A7.1-2 Determinants of Total. Chronic and Transient Poverty ........................ 192 Table A8.1-1 By How MuchWill Per Capita Consumption Change for a Given Change inPer Capita SP Transfers? .......................................................................................... 193 Table A8.1-2 Static Benefit Incidence Analysis: Changes in the Incidence of the Benefit UnderAlternative Counterfactual Consumption............................................. 194 Table A8.1-3: Marginal Benefit Incidence Analysis: Changes in the Incidence o f the Benefit Under Alternative Counterfactual Consumption.................................... 195 Table A8.1-4: Coverage and Targeting of the Poor: Changes in the Magnitude of Poverty Reduction Impact Under Alternative Counterfactual Consumption........................ 196 Table A8.2-1 Proxy-Means Indicators for Total and Extreme Poverty, by Area o f Residence.......................................................................................... 198 Figures Figure A Poverty Incidence Curves. 1998-2001:Panelof 1108 Households..........v Figure B Average ConsumptionPer Capitain 1998 and 2001. (RegionsRelative to Bishkek)................................................................................................... vi Figure C Access to Public Services i s Greater in UrbanAreas and for Richer Households2001..................................................................................... .. v11 FigureD Simulating the Poverty-ReducingImpact of UniformEconomic Growth. 2001-2006.............................................................................................. ... VIII FigureE Exports and InvestmentSavings.............................................................. ix Figure 1.1 Economic Decline UpTo Mid-1990's And Recovery Since Then. By Sector................................................................................................... 4 Figure 1.2 Real ExchangeRate (Jan 1998=100)........................................................ 5 Figure 1.3 Cumulative Distribution Functions. 2000 And 2001.............................. 12 Figure 1.4 ChangeInPer CapitaConsumptionBy Decile For Urban-Rural Population. 2000-2001(%)..................................................................... 14 Figure 1.5 Lorenz Curves Of ConsumptionInequality. 2000 And 2001 (DashedLine=2000. SolidLine=2001) .................................................. 16 Figure 1.6 FoodPrices RoseFaster Than Those For Non-foodGoods...................20 ConsumptionPer CapitaFor PanelAndWhole SampleHouseholds.... 17 Figure 1.7 Figure 1.8 Poverty Incidence CurvesFor PanelHouseholds. 1998.2001. For RuralAndUrban Sectors ................................................................. 21 Figure 1.9 Simulating The Poverty-ReducingImpact Of Economic Growth. 2002-2006 ............................................................................................... 23 Figure 1.10 Exports And Investment-Savings........................................................... Figure 2.1 Evolution Of Crop AndLivestock Production (1995-2001) .................. 25 28 Figure 2.2 Evolution Of Crop Yields (1995 = 100) ................................................. Figure 2.3 Evolution Of Agricultural Value Added And Labor Productivity..........29 29 Figure 2.4 Relative Prices Of Food And Agricultural Goods.................................. 30 Figure 2.5 ParticipationInCredit Markets By Quintile........................................... 34 Figure 2.6 Credit ReceivedBy Quintile. 2001......................................................... 35 Figure 2.7 36 Land Tenure By Consumption Quintile (2001) ...................................... Evolution Of LandArea Per Household (Panel Data)............................ Figure 2.8 36 Figure 2.9 Gini Coefficients For 2001 (Full Sample) .............................................. 39 Figure 3.1 Migration Flows InThe Kyrgyz Republic. 1989-2000.......................... 69 Figure4.1 Average ConsumptionPer CapitaIn 1998 And 2001. (Regions Relative To Bishkek)............................................................... 79 Figure 4.2 SourcesOf Income For UrbanAnd RuralHouseholds.......................... 81 Figure 4.3 EnvironmentalEIealth Conditions Of Kyrgyz HousingIn2001. 82 Percent Of HouseholdsWith Access To Utilities. 2001......................... (% Of HousesWith Access To Services) ............................................... Figure 4.4 Figure 4.5 Percent Of HouseholdsWith Access To A Telephone Connection .......84 83 Figure 5.1 Average HouseholdExpendituresOn Education And % Poor By Oblast. 2001 ...................................................................................... 99 Figure 5.2 PercentagePoor. Grade 5 Test Pass Rate. And Secondary Examination Scores By Oblast................................................................................... 103 Figure 6.1 Chronic And Transient Poverty By Quintile. 2001 .............................. 120 Figure 6.2 Share Of Individuals With RisingRalling Per Capita Consumption In2001/1998. By 1998 Quintiles.......................................................... 121 Figure 6.3 Chronic/Transient Poverty And HouseholdCharacteristics................. 124 Figure 6.4 Average Change InPer Capita ConsumptionBetween 1998.2001. By Oblast And Area Of Residence....................................................... Figure 6.5 Within The Year. There I s Seasonality InConsumptionPer Capita....125 126 Figure 6.6 Chronic/Transient Poverty And Location............................................. 127 Figure 6.7 Chronic/TransientPoverty And HouseholdCapital............................. 128 Figure 7.1 Share Of Electricity ExpendituresRelative to HouseholdIncome. 2000-2001............................................................................................. 135 Figure 7.2 Average HouseholdElectricity Consumption. kWh. Bishkek. 1999-2002............................................................................................. 136 Figure 7.3 HouseholdAccess To Reliable Electricity By Quintile. 2001 (Interruptions FewerThan Once Per Week)................................................................. 137 Figure 7.4 Primary SourceOf Energy For HeatingBy Quintile. 2002. Ql-Q2 (Percent Of Households)....................................................................... 137 Figure 8.1 Impact Of SP And PT On Poverty And Inequality............................... 162 Figure 8.2 Cost-BenefitRatio Of Social ProtectionPrograms .............................. 164 Figure 8.3 Effectiveness Of Social Protection ProgramsInReducingPoverty.....165 Figure 8.4(a) Share of Individuals Coveredby UMB.by Quintile...................... 166 Figure 8.4(b) Distribution of UMB RecipientsAcross Quintiles......................... 167 Annex Figures Figure4.1.1 Evolution of Crop Production (1995=100) ........................................... 179 Figure4.1-2 Trend of Livestock Inventory (1995=100) ........................................... 179 Figure4.1-3 Trend of Livestock Production (1995=100).......................................... 179 Boxes Box 1.1 Available Data Sources For Poverty Analysis InThe Kyrgyz Republic.................................................................................................... 8 Box 1.2 MeasuringMonetary Welfare ................................................................... 9 Box 1.3 Measures Of Poverty............................................................................... 12 Box 1.4 DecompositionOf A ChangeInPoverty................................................ 15 Box 2.1 The Kyrgyz LandRedistribution Fund................................................... 31 Box 2.2 EmergingAgricultural Markets.............................................................. 32 Box 2.3 PerceivedObstaclesTo Credit Expansion.............................................. 34 Box 2.4 38 Box Tbl. 2.1 Distribution Of Arable LandBy Oblast (2001) ...................................... RegionalDifferences InKyrgyz Agriculture.......................................... 39 Box 2.5 InfrastructureAnd Rural Poverty............................................................ 41 Box 3.1 Defining Labor Market Indicators .......................................................... 48 Box 3.2 Eligibility Conditions For Unemployment Benefits............................... 50 Box 3.3 PerceptionAnd Attitude To Corruption Among ManagersOf Enterprises InThe Kyrgyz Republic....................................................... 63 Box 3.4 Internal Migration And Bishkek's Slums ............................................... 71 Box 5.1 A Broader Picture Of Poverty................................................................. 94 Box 5.2 Education . Access And Opportunity..................................................... 99 Box 5.3 Health Reforms InThe Kyrgyz Republic ............................................ Box 5.4 Informal Payments For Health Services InThe Kyrgyz Republic .......109 111 Box 6.1 Measurement Of Chronic And Transient Poverty ................................ 119 Box 8.1 Kyrgyz PensionsAt A Cross-road: Insurance Or A Coping Instrument? ........................................................................................... 147 Box 8.2 The UnifiedMonthly Benefit(UMB):The Main Poverty Reduction Program................................................................................................. 152 Box 8.3 UMB:Problematic DesignFeatures..................................................... 159 Box 8.4 ImprovingMonitoringAnd Evaluation Capacity InSocial Protection .............................................................................................. 166 ACKNOWLEDGEMENTS This report has been prepared on the basis of extensive and detailed work, carried out jointly by Kyrgyz and World Bank teams. The World Bank team i s led by Radwan Shaban, comprising Dina Abu-Ghaida, Azamat Abdymomunov, Sabine Bemab6, Sarbani Chakraborty, Louise Cord, Dinara Djoldosheva, Oleksiy Ivaschenko, Alexandre Kolev, Julian Lampietti, Mathilde Maurel, Michael Mertaugh, Cem Mete, Gi-Taik Oh, Natalia Pisareva, Pedro Rodriguez, Maria Shkaratan, Verdon Staines, William Sutton, and Emil Tesliuc. The Kyrgyz team i s led by Zarylbek Kudabaev, Chairman of Natstatcom, and comprising Natstatcom staff Shamsia Ibragimova, Mukhtar Minbaev, and Larissa Praslova. Key contributions to the report were also made by eleven Kyrgyz researchers from non-government organizations who were commissioned to write background papers on various aspects of poverty, with support from the Norwegian trust fund. A list of the researchers and their papers i s provided in the Annexes. Stephanie Abdulin edited the report. Erlinda Inglis, James Lynch, Helena Makarenko, Duillo Pighi, and Natalia Turchina provided assistance with the production of the report. Asad Alam, Lilia Burunciuc, Anthony Cholst, Cheryl Gray, Daniela Gressani, James Christopher Lovelace, Mohinder Mudahar, Pradeep Mitra, Samuel Otoo, Dennis de Tray, and Laura Tuck have provided guidance to the report team at various stages. Peer reviewers are Jeni Klugman, Azizur Rahman Khan, Valerie Kozel, Peter Lanjouw, and Kinnon Scott. The peer reviewers offered guidance and advice at various stages of the report analysis and writing as well. Helpful comments and advice have also been provided by Robert Anderson, Gary Fine, Margaret Grosh, Ellen Hamilton, Jane Holt, Mamta Murthi,Michal Rutkowski and Gotz Shreiber. The report benefited from extensive discussions with Kyrgyz authorities and NGOs. Initial poverty diagnostics were jointly presented by Dr. Zarylbek Kudabaev (Natstatcom) and MI-. Radwan Shaban (World Bank) at a workshop at the CDF Secretariat in Bishkek in May 2002, under the guidance of Mr. Kubat Kanimetov, Head of the Economic Policy Department of the Office of the President, and MI-.Leonid Komarover, Head of the Secretariat of the CDF and Deputy Head of the Economic Policy Department. The report was also presented and discussed at the Consultative Group Meeting of donors in Bishkek in October 2002. A major presentation and discussion of the report's findings took place at the CDF Secretariat in December 2002, and was attended by heads of the working groups of the National Poverty Reduction Strategy, NGOs, scholars, and the media. The report utilizes internationally accepted methodology of poverty analysis. i This work was accompanied by a poverty analysis training course in February 2002 in Bishkek, that was provided by the World Bank Institute, with the support of Aline Couduel,GulnaraFebres, andMaria Shkaratan. 11 .. EXECUTIVE SUMMARY 1. The goal of this report has been to assist the Kyrgyz authorities in designing and implementing their National Poverty Reduction Strategy (NPRS). An interim report focusing on the relation between growth and poverty, with detailed analysis of transmission mechanisms in labor market and agricultural sector was presented to the authorities in August 2002. The interim report delivery was driven by the schedule of NPRS preparation. This final poverty assessment report expands the interim report by addressing key additional issues in the links between poverty and key public services such as energy, health, education, and social protection. The process of producing this report has involved close collaboration with Natstatcom, training Kyrgyz researchers on poverty analysis, and commissioning eleven independent researchers to write background papers on various aspects of poverty in the Kyrgyz Republic whose findings have been incorporated selectively inthis report. Analysis inthis report goes through the year 2001, but collaborative work between the Bank and Kyrgyz teams on poverty diagnostics and assessment i s an ongoing process. 2. Sustained and broad-based economicgrowth has to be the central componentin the medium-term poverty reduction strategy. The Kyrgyz economy i s characterized by: 0 very low income levels, with per capita GDP of about US$300 in 2001; 0 moderate inequality, with the Gini index of consumption inequality at the low end of the 0.3-0.4 range; and 0 highlevels of poverty, with about half the population living inabsolute poverty. Yet, the credibility of this forward-looking strategy depends very much on the evidence on recent poverty reduction. Against a legacy o f economic collapse in the first half of the 1990's, the Kyrgyz economy has been growing, though unevenly, since 1996. The key questions are : Has this recent economic growth translated into poverty reduction, and if so, how effectively? How can the country ensure a sustainable path of pro-poor growth over the medium term? A. ECONOMIC GROWTH POVERTYREDUCTION AND Economic Growth 3. The Kyrgyz economy has made major strides in attaining macroeconomic stability and putting the economy on a growth path. The collapse of the Former Soviet Union imposed a huge adjustment burden on the Kyrgyz economy. Output collapsed by 50 percent between 1991 and 1995, and incomes fell substantially due to the loss of transfers from the Union and terms of trade adjustment. Real GDP had a burst of rapid growth in 1996-1997 period, but it slowed down significantly in 1998-1999 in response 111 ... to the regional financial crisis, and then grew again at a rate of 5.4 percent in 2000 and 5.3 percent in 2001. Since 1996, the economy has shown a significant response to stabilization and structural reform efforts. With increased productivity and a shift toward higher-valued products, the agricultural sector has led economic growth since 1996, although gold and trade have also contributed to the recovery. The agricultural sector has particularly responded to price and trade liberalization, equitable land reform, and other supporting policies such as the expansion of micro-credit for farmers. On the other hand, the manufacturing sector has had a dismal performance, and continued to decline through the end o f 2001. Poverty Reduction 4. Here are some key findings on how many people and which groups of people escaped poverty as a result of economic growth during 1998-2001. Economic recoveryfollowing the Russian crisis of 1998 had led to a significant reduction inpoverty. Based on the full nationally representative Household Budget Survey (only comparable since 2000), the number of poorpeople was reduced by an estimated 300 thousand individuals between 2000 and 2001. Table A shows that the headcount index of poverty declined from 62.5 percent to 56.4 percent, using the consumption per capita welfare measure, a 6.1 percent reduction for a population of 4.9 million people. Poverty was reduced significantly in 2001, regardless of the welfare measure used (see Table A). -P TableA: Poverty Declinedin2001 Measure of welfare AbsolutePovertyline, HeadcountIndex of Poverty (%) 2001Soms 2000 2001 Expenditureper capita 7491 52.0 47.6 Consumptionper capita 6975 62.5 56.4 Consumptionper adult equivalent 6975 51.8 45.0 Source: Table 1.4. Analysis of a panel data subset o f the Household Budget Survey which can be used to compare trends over the 1998-2001 period shows a similar trend. Out of every hundred poor people in 1998, 23people are estimated to have escaped poverty by 2001, on average. There was little change in poverty indicators between 1998 and 1999, but poverty declined substantially in 2000 and 2001 (see Figure A). The poorer segments of the population increased their consumption more proportionately than the richer segments. An estimated 80 percent of households in the poorest two quintiles in 1998 increased their consumption levels, while only 40 percent of households in the richest two quintiles increased their consumption levels by 2001. i v Figure A: Poverty Incidence Curves, 1998-2001;panelof 1108 households ____. -2001 ~ 1998 1999 2000 - '1 Poverty Liny t 8 4 c c 1 1 0 460 Consumption PC. Decal PrioeD 568 800 1200 Poverty IncidenceCurves, HBS Panel 0 Poverty reduction was broad-based across all oblasts. However, Naryn, the poorest oblast, benefited very little from poverty reduction. This is a high- mountain oblast with difficult transportation and costly access to the market, that i s predominantly rural, reliant on livestock rather than crop production. In 2001, out of ten Naryn residents, an estimated nine people lived in absolute poverty and six lived in extreme poverty. 0 Rural poverty consistently declined over the 1998-2001period, at an average rate of 8 percent per year. This growth has been largely driven by increasing consumption, which resulted from equitable growth in the agricultural sector. 0 Urban poverty initially increased in 1999 as a result of the regional financial crisis, but then declined over the 1999-2001period as a result of the economic recovery. Growth accompanied by moderate inequality led to a reduction in poverty from 1998-2001. Inequality declined andtherefore growth was pro-poor. Inequality declined slightly in 1999 and 2001, particularly in urban areas, contributing to poverty reduction. The Gini index of consumption inequality declined from 0.303 in 2000 to 0.290 in 2001. The inequality decline ameliorated the poverty-increasing impact of declining urban consumption in 1999, and it augmented the poverty-reducing impact of consumption growth in 2001. B. SELECTEDKEY FINDINGS ONPROFILE OFPOVERTY 5. Selected key findings on the dimensions o f poverty are presented here to inform the design and implementation of policy interventions inthe NPRS. 6. Poverty has a very strong regional dimension. The capital city of Bishkek and nearby Chui oblast are substantially less poor than the rest of the country, while Naryn oblast i s the poorest oblast, and seems to be lagging behind. Talas oblast i s the second V poorest; out of every ten residents in Talas in 2001, seven people lived in absolute poverty and four people lived in extreme poverty in 2001. The northern Issyk-kul oblast and three southern oblasts (Osh, Jalal-abad, Batken) are in the middle interms of poverty. The regional differences in poverty largely reflect regional differences in average consumption levels (see Figure B). A household living in Naryn oblast i s three times as likely to be poor as another household living in Bishkek City or Chui oblast. These differences clearly justify a regional focus in the poverty reduction strategy. This regional approach should take into account the poverty reduction benefits from targeted regional interventions as well as the cost of implementing such interventions. While intervention should mainly focus on creating job opportunities, care should be taken to ensure they reinforce rather than undermine the incipient market economy that has been forming. Also, the types of interventions should include improving access to infrastructure, health services and education, as well as improving social protection. Figure B: Average ConsumptionPer Capitain 1998 and2001, (RegionsRelativeto Bishkek) - I Region Source: Figure 4.1. 7. Despite the recent gains described above, rural poverty continues to be significantly higher than urban poverty. In 2001, two thirds of the Kyrgyz population lived in rural areas, and almost three quarters of the poor lived in rural areas. The rural- urban differential partially reflects the fact that Bishkek city accounts for half the urban population but also has the lowest incidence of poverty. However, per capita consumption of the urban population was greater by 28 percent than that o f the rural population in 2001, even after accounting for regional differences and household characteristics. The rural-urban differential in poverty would clearly have been worse without high agricultural growth and rural poverty reduction. It i s therefore important that both the farm and non-farm rural sectors continue to grow (see Chapter 2). This requires consolidation and operationalization of many of the rural reforms already undertaken, including land reform and efficient allocation of resources, as well as addressing the constraints to agricultural trade and exports in particular. 8. Transientpoverty is high as a result of high consumption volatility. Households have high exposure to shocks and lack the ability to smooth consumption. High consumption volatility makes households even in the richest quintile experience transient poverty at times. Indeed, about 11percent o f households in the richest quintile in 1998 experienced serious shocks to their welfare, falling down two quintiles or more by 2001. Consequently, transient poverty affects households across the welfare spectrum, even vi among the richest quintiles. The low coping capacity and vulnerability to poverty are likely due to the absence of effective saving instruments and imperfections in financial intermediation, insurance, and storage markets. 9. Access to public services such as running water, public sewers, reliable electricity, district heating, or telephone service are very low (See Figure C). Overall, rural households have more limited access to these services than urban households. Moreover, access to these services i s lower for poorer households in both rural and urban areas. An implication of these findings i s that given these services are subsidized, rich households, usually consume more to begin with, and therefore receive most of the subsidies. Moreover, while the lack of infrastructure maybe an important determinant of poverty, increasing access to network services may not be a cost-effective way of reducing poverty with a dispersed population. A strategy on how best to improve access to and reliability of services for poor households needs to be developed. Figure C: Access to Public Services is Greater inUrbanAreas and for Richer Households 2001 Access to Public Serivces, 2001 -Urban -_- I I Eu) 80 0 700 41 a 600 f 500 E0 ! 400 300 p 200 10 0 0 0 Runningwater Public sewer District heating Electricity reliability 90 Access to Public Services, 2001 Rural - .""" 0Poorestquintile 80 0 70 0 94 600 .Richest quintile 50 0 40 0 30 0 20 0 100 00 Runningwater Public sewer District heating Electricityreliability vii C. FUTURE POVERTY REDUCTION Inequality and Poverty Reduction 10. The country faces a huge poverty reduction challenge over the medium term, requiring a strong commitment to reform in order to sustain and accelerate growth. About 2.8 million people, a little over half of the Kyrgyzpopulation, continue to live in absolute poverty, despite significant poverty reduction in recent years. Simulating the impact of uniform consumption growth from its 2001 level while holding inequality constant illustrates the seriousness of the poverty reduction challenge (see Figure D). For every 1 percent in uniform growth in per capita consumption, the headcount index of absolute poverty declines by 0.8 percentage points. A 5 percent annual growth rate in per capita consumption i s expected to: 0 reduce the headcount index of poverty from 56.4 percent in 2001 to 36 percent in 2006; and 0 reduce the number of poor people by 200 thousand per year, with a cumulative reduction of one million people over the 2001-2006 period. A less ambitious annual growth rate of 3 percent inper capita consumption is expected to reduce the headcount index of poverty to merely 45 percent by 2006, at an annual reduction of 116thousand inthe number of poor people per year (see Figure D). FigureD: Simulatingthe Poverty-ReducingImpact of UniformEconomicGrowth,2001-2006 ExpectedReductioninPoverty Headcount with ConsumptionGrowth at 3% and 5% and unchangedinequality 1 0.30 ! 1 I 2001 2002 2003 2004 2005 2006 I Source: Figure 1.9. will be adversely affected if inequality begins to accompany growth. Inequality may 11. Even if desirable and ambitious growth goals are achieved, poverty reduction very well increase if growth i s broadened outside agriculture to industry and modern services, which will have to make more differentiated rewards to factors of production than agriculture. If the Gini index of consumption inequality increases by 0.01 per year from 0.29 in 2001 to 0.34 in 2006, then a 5 percent annual growth rate in consumption i s expected to reduce the headcount index of poverty to merely 41 percent by 2006; instead of to 36 percent, which i s expected if inequality remains unchanged. A 3 percent annual ... V l l l growth rate in consumption with increasing inequality i s expected to reduce the headcount index of poverty to merely 50 percent by 2006. Investment and Trade 12. Future poverty reduction will crucially depend not just on the level of growth but on the forces driving this growth as well. Growth needs to be broad-based. It i s possible that a scenario of 3 percent growth that i s based on trade and private investment could result in greater poverty reduction than a scenario of 5 percent growth that i s primarily supported by fiscal and external imbalances. More specifically, the sources of demand for growth are important for the extent o f poverty reduction. 13. Growth in the most recent 2000-2001 period was more balanced but lacked the necessary exports and private investment dynamism (see Figure E). The authorities began adjusting private consumption and imports since late 1998 when the exchange rate was allowed to float. Further adjustment of public consumption and imports continued during the 2000-01 period, when external adjustment was complemented with fiscal adjustment, and measures to initiate the process that would address the dire strait conditions of the banking and energy sectors. As a result, the trade balance improved substantially, but mainly due to a sharp fall in imports. Exports (other than gold, and those related to the inter-governmental irrigation agreements with Uzbelustan) have shown no dynamism. Similarly, lagging reforms to improve the business environment, as well as in the energy, telecommunications and mining sectors, have lead to an unfavorable investment climate, in turn resulting in very low rates of private investment. Saving rates have improved following strong stabilization efforts, but the ability o f the financial sector to mobilize savings and investments i s still in question. Figure E: Exports and Investment Savings T o t a l Exports, a n d E x p o r t s without power and gold, U S mln 750 650 550 450 350 250 150 1993 1994 1995 1996 1997 1998 1999 2000 2001 Investments and National Savings ( % of GDP) 30 , 25 e. 20 2n 15 10 $ 50 -5 1995 1996 1997 1998 1999 2000 2001 i x 14. Sustaining growth is expected to be much harder over the comingfew years, as the conditionsfor 1996-2001 growth appear unlikely to be replicated over the medium term. Aside from private investments in gold, growth since 1996 has relied substantially on public investments and consumption (including imports). These macroeconomic conditions are difficult to sustain. Current high levels of external debt-at the moment transitorily mitigated by the recent Paris Club flow restructuring operation-pose too many risks to external sustainability. Multi-Sectoral Growth 15. While agriculture has generated a basic level of growth, without which the present levels of poverty would be significantly higher, this sector alone cannot continue to bring about rapid growth. As Chapter 2 shows, success in the agriculture sector can be attributed to a mutually reinforcing set of reforms that the authorities have been implementing consistently over the past six to eight years. These have been supported by several public interventions, some of which can be associated with the subsequent supply response (e.g., rehabilitation of irrigation, market-driven micro-credits schemes). It i s important that agricultural growth i s enhanced through market-driven productivity gains, consolidation of property rights and land reform, and the need to improve the marketing, particularly export orientation, of agricultural products. However, agricultural growth will be significantly conditioned by the extent of improving the export potential, as well as by a rising domestic demand, which in turn requires significant growth in the non-agricultural sectors. Moreover, agriculture may be close to i t s limit in providing a cushion to worsening conditions elsewhere in the economy by absorbing workers from collapsing sectors. Thus, there i s a need for a more balanced growth that relies on multiple sectors growing at the same time, in order to ensure a solid and sustained growth in output and incomes over the medium term. 16. Even after a decade of transition, the manufacturing sector shows no signs of revitalization. A detailed assessment of the current state of manufacturing i s well beyond the scope of this work. Yet, it i s clear that growth in the goldmine sector cannot be the only source of industrial growth for broad-based industrial development. While the goldmine contributes to overall GDP growth, it has limited linkages with the domestic economy and i s therefore unlikely to have significant impact on poverty reduction. The dismal conditions of the manufacturing sector make it clear that the necessary conditions have not been created for stopping and reversingthe rapid decline of this sector. 17. Growth in labor-intensive sectors is more pro-poor, but their progress is undermined by domestic constraints created by an excessively bureaucratic regulatory process, some features of the taxation regime and enforcement, and other factors including corruption. It i s noteworthy that growth in 2001 was driven, in addition to agriculture, by construction, transportation, trade, and catering accounting for about 18 percent of GDP. These sectors are dominated by micro and small-sized enterprises. Only the construction materials sector has a few medium-scale enterprises (a recently privatized cement plant as well as newly created businesses specializing in brick-making and other related activities, mainly producing goods for export to Kazakhstan). Like in agriculture, reforms and liberalization have led to some vibrancy in these sectors, with X notable exceptions in air transport, railways, and part o f the freight truck fleet. As elaborated on in Chapter 3, and as commendable as has been the embracement by the Government of the objective of improving the investment climate for these small firms, significant and tangible gains have yet to materialized. 18. I n short, the analysis of sectoral sources of growth suggests that the significant gains in poverty reduction achieved in recent years are not likely to be sustained over the medium term unless there is a sustained broadening of the sources of growth. A commitment to a strong reform agenda that ensures broad-based growth is therefore crucial. Specific Sectoral Policy Implications 19. Agricultural growth and rural poverty. The key challenge will be to maintain agricultural growth, which has helped to reduce rural poverty. Challenges in the areas o f land markets, land use and leasing include: (i)protecting the interests of small land holders; (ii)supporting the development of land rental markets to improve land distribution and prevent land fragmentation; (iii)addressing the future of land held in Land Redistribution Funds; and (iv) increasing the transparency in leasing arrangements to restructured corporate farms. Challenges in supporting the development of agricultural markets include: (i) phasing-out in-kind payments; and (ii) developing infrastructure to support market access. 20. Formal job creation. The key challenges include: (i) improving incentives for businesses to operate in the formal sector; (ii) improving labor market data to address the problem of inadequate training of workers; (iii) improving collection and dissemination of job opportunities to the unemployed, in order to better manage migration flows; and (iv) improving the business environment to facilitate growth of small enterprises and the creation of productive jobs by reducing registration and licensing procedures and simplifying tax procedures. 21. Education. The key challenges will be: (i)more evenly allocating funding for primary education across regions, to ensure that educational quality i s more equal in all regions; and (ii) ensuring that all primary school students have access to texts, so that the poor receive the same quality education as the non-poor. 22. Health. The key challenge will be to consolidate the health reform and phase-out the system of informal payments for health services, which will improve the access of the poor to health care services. 23. Dynamics of poverty. Key challenges include: (i)the creation o f self-targeted policies to cure transient poverty; (ii)in regions such as Naryn where chronic poverty dominates, policies that emphasize increased access to infrastructure, and improved access to quality education and health services would help alleviate poverty; and (iii) better social risk management policies will help reduce transient poverty due to seasonal factors and imperfections incredit, insurance and storage markets. x i 24. Energy. Key challenges include restructuring of the sector, which should include: (i)service quality improvements; (ii) administrative reform to phase-out the system of informal payments; (iii)social protection measures such as payment smoothing, direct income transfers or establishment of life-line tariffs to increase access of the poor to electricity; and (iv) continued tariff and management reform. 25. Social protection. Key challenges include: (i)extending coverage o f the UMB for the working poor who are living in extreme poverty; (ii)using consistent criteria across all poverty alleviation programs; and (iii) improving monitoring and evaluation of SP programs. 26. This report i s structured in the following manner. It is divided into three sections: the first section comprises Chapters 1through 3, and addresses growth-poverty linkages; the second section comprises Chapters 4 through 6, and presents a profile of both the income and non-income dimensions of poverty, and the third section comprises Chapters 7 and 8, and examines policies of protecting the poor. Chapter 1reveals that economic growth has led to poverty reduction from 1998-2001. Chapter 2 explores the linkages between growth of the agricultural sector and rural poverty reduction over the same period. The linkages between poverty and job creation are examined in Chapter 3. In Chapter 4 i s an analysis of which characteristics are strongly or weakly correlated with income dimensions of poverty. The non-income dimensions of poverty, namely education and health, are explored in Chapter 5. A dynamic analysis of poverty i s undertaken in Chapter 6. Chapter 7 presents an analysis of the impact of electricity sector tariff reforms on the poor. And Chapter 8 analyzes the effectiveness of existing social protection programs in targeting the poor. xii PARTI:GROWTH-POVERTYLINKAGES This first section comprises Chapters 1 through 3, and addresses growth-poverty linkages. Chapter 1 reveals that economic growth has led to poverty reduction from 1998-2001; and shows that economic growth was pro-poor from 2000-2001. Chapter 2 explores the linkages between growth of the agricultural sector and rural poverty reduction, showing that the agricultural sector's average annual growth rate of 6% since 1996 was accompanied by an average decrease in rural poverty by 8% per year from 1998-2001. The linkages between poverty and job creation are examined in Chapter 3, concluding that while the agricultural sector has absorbed workers who lost jobs in other sectors, the problem of a low level of formal job creation partly explains the persistence inwidespread poverty. 1 2 1. HASECONOMICGROWTHREDUCEDPOVERTY? With half the population in absolute poverty, the Kyrgyz Republic faces a serious challenge of poverty reduction. This chapter addresses the fundamental relationship between economic growth and poverty in the Kyrgyz Republic. A key conclusion of the analysis is that economic growth, coupled with moderate and marginally declining inequality, has led to a reduction of poverty. Rural poverty consistently declined over the 1998-2001period; while urban poverty increased in 1999 as a result of the Russian crisis, but declined during 1999-2001. In particular, the year 2001 represents a year of pro-poor growth, in whichpoverty and inequality were reduced while the economy grew. To reach these conclusions, datafrom the Household Budget Survey are used to analyze the linkages between growth, inequality, and poverty in the recent period 2000-2001. Poverty comparisons for earlier years are more challenging, owing to a change in the sample in 2000. The analysis here takes advantage of the panel structure of the Household Budget Survey in order to make consistent poverty comparisons over the whole 1998-2001period. On this basis, we conclude that earlier evidence of a growth- poverty conundrum-i.e., earlier evidence that growth had not resulted in poverty reduction despite non-increasing inequality-was marred by various technical issues of non-comparability of surveys and welfare measures. A. THREEEPISODESGROWTH, OF 1996-2001 1.1 With a per capita GDP of about US$300, the Kyrgyz Republic is the second poorest country in the former Soviet Union, after Tajikistan. Between 1991 and 1995 cumulative decline in GDP amounted 50 percent of the 1990 level (Figure 1.1), due to economic restructuring and the loss of direct transfers from the central union budget, which amounted to 13 percent of GDP in 1991. This loss of income was filled by external borrowing, depletion of assets, and reduction in private consumption and government expenditures. Duringthis period, various reforms were implemented laying the foundation for market-oriented growth: the introduction of the national currency in 1993, liberalization of prices, reforms in commercial legislation, agrarian reform, privatization of assets, and establishment of an open external trade regime. The decline itself and the additional impact of reforms during the early years of transition resulted in hyperinflation, rising unemployment, falling real income, and rising poverty. GDP fell to its lowest level in 1995. However, since 1996 positive GDP growth had resumed (Figure 1.1). 3 Figure 1.1: Economic Decline Up to Mid-1990's and Recovery Since Then, by Sector - Real GDPIndex 1990=100 - GDP (index, 140 1990=100) 120 I -- Agriculture 100 e'' 0 2 m 80 Construction z8 60 a - E - m Manufacturing 40 (incl. Kumtor and power) 20 - _ _ _ . Services Source: World Bank Live Database. 1.2 The pace of economic growth has been uneven with three identifiable episodes of two years each: rapid but concentrated growth in 1996-1997, anemic growth in 1998- 1999, and a more robust and relatively more broad-based economic growth in2000-2001. The initial burst of GDP growth was largely driven by rapid double-digit agricultural growth in 1996 and 1997, and by starting production in the Kumtor gold mine in early 1997. Growth was rapid but concentrated in 1996-1997 due to growth o f the gold mining and agricultural sectors. The value-added of the gold mining sector amounted to 4.2 percent o f GDP in the first year of its production-1997, but its relative contribution stabilized at about 5 percent of GDP in the period since 1998 (see Table 1.1). 1.3 Economicgrowth was anemic during the 1998-1999period, due to thefallout of the Russianfinancial crisis. The regional financial crisis hit the economy in the second half of 1998 and caused a reduction of GDP growth in 1998 to 2.1 percent (0.5 percent on a per capita basis) and, excluding the Kumtor gold mine, only to 0.6 percent (-0.9 percent on a per capita basis). Inthe last two months of 1998, inflation started to increase and the exchange rate depreciated dramatically. The real exchange rate of S o d $ depreciated by 43 percent and SordRussian Ruble appreciated by 23 percent in second half of 1998 (seeFigure 1.2). Exports to CIS countries affected by the Russian crisis suffered and trade impediments in Kazakhstan and Uzbekistan increased; the current account deficit increased from 8 percent in 1997 to 23 percent in 1998. 4 Table 1.1: GDP, SectoralDistribution, and Growth Rates, 1995-2000 1995 1996 1997 1998 1999 2000 2001 GDP constant prices (som m.,1995=100) 16145 17289 19003 19406 20116 21211 22338 Agriculture 6568 7566 8493 8736 9453 9703 10365 Industry 2924 2999 3588 3447 3335 3500 3750 Construction 993 992 823 595 583 757 875 Power sector 547 557 503 466 501 642 631 Kumtor 0 0 800 1010 960 1059 1234 Manufacturing 1384 1449 1462 1375 1291 1042 1010 Services, etc 5469 5456 5528 3783 3759 4246 4262 Transportation 536 602 635 615 635 682 #N/A Trade and catering 1831 1874 1996 2164 2254 2495 #N/A Public administration 1748 1779 1698 1803 1987 1144 #N/A Other and discrepancies 1353 1201 1200 -799 -1117 -75 #N/A Growth rates (% per annum) GDP 7.1 9.9 2.1 3.7 5.4 5.3 Agriculture 15.2 12.3 2.9 8.2 2.6 6.8 Industry 2.6 19.7 -3.9 -3.2 4.9 7.2 Construction 0.0 -17.0 -27.7 -2.0 29.7 15.6 Power sector 1.8 -9.8 -7.3 7.6 28.0 -1.7 Kumtor 0.0 0.0 26.2 -4.9 10.4 16.5 Manufacturing 4.7 0.9 -5.9 -6.2 -19.3 -3.0 Services, etc -0.2 1.3 -31.6 -0.6 13.0 0.4 Transportation 12.4 5.4 -3.1 3.2 7.5 #N/A Trade and catering 2.3 6.5 8.4 4.2 10.7 #N/A Public administration 1.8 -4.6 6.2 10.2 -42.4 #N/A Other and discrepancies -11.3 -0.1 -166.6 39.8 -93.3 #N/A Source: World Bank Live Database. Note: The Kumtor gold mine came on stream in early 1997,thus explaining growth in the "industrial sector" in 1997while other sub-sectorsof industry declined in that year. Figure 1.2. Realexchangerate(Jan. 1998=100) increase correspondstodepreciation 0 180.0 160.0 - 4 & 120.0 140.0 7 100.0 8 80.0 60.0 Source: World Bank Live Database. 5 1.4 Economicgrowth was anemic during the 1998-1999period, due to thefallout of the Russianfinancial crisis. The regional financial crisis hit the economy in the second half of 1998 and caused a reduction of GDP growth in 1998 to 2.1 percent (0.5 percent on a per capita basis) and, excluding the Kumtor gold mine, only to 0.6 percent (-0.9 percent on a per capita basis). Inthe last two months of 1998, inflation started to increase and the exchange rate depreciated dramatically. The real exchange rate o f S o d $ depreciated by 43 percent and Som/Russian Ruble appreciated by 23 percent in second half of 1998 (see Figure 1.2). Exports to CIS countries affected by the Russian crisis suffered and trade impediments in Kazakhstan and Uzbekistan increased; the current account deficit increased from 8 percent in 1997 to 23 percent in 1998. 1.5 The financial sector suffered as well. The rate o f inflation reached 36 percent in 1999, eroding, in real terms, average wages by 15 percent, public sector wages by 19 percent, and pensions by 25 percent. 1.6 GDP growth in 2000-2001 was robust and relatively more broad-based than earlier episodes. Economic recovery increased GDP growth to 5.4 and 5.3 percent in 2000 and 2001, respectively. While the 1996-1997 growth burst was faster and accompanied by a higher rate of investment, the recent growth episode was relatively broader. Over 2000-200 1, sectoral growth occurred in the traditionally growing sectors of agricultural and mining, in addition to healthy growth in construction, power, as well as in the service sub-sectors of transportation and trading and catering. One noted weakness of growth in this two-year period i s the continued deterioration of the manufacturing sector. However, the 2000-2001 growth episode appears more robust and probably less vulnerable to short-run policy reversals. Indeed, this growth episode i s supported by macroeconomic and exchange rate stability, reduced fiscal deficits, tight monetary policy, and single-digit inflation (9.6 percent in 2000 and 3.7 percent in 2001; December to December change) for the first time since independence (see Table 1.2). Table 1.2: General Macroeconomic Indicators, 1996-2001 1996 1997 1998 1999 2000 2001 Population (million) 4.7 4.7 4.8 4.9 4.9 4.9 GDP growth, constant prices (percent) 7.1 9.9 2.1 3.1 5.3 5.3 GDP growth per capita, constant prices (percent) 5.5 8.3 0.5 2.2 4.5 5.4 GDP growth per capita w/o gold, constant prices (percent) 5.5 3.8 -0.9 2.5 3.8 4.3 GDP per capita constant price (1995=100) 105 114 115 117 122 129 M2 (end-of-period, % of GDP) 14.3 13.6 14.4 13.5 11.8 11.1 CPI (percent year-on-year average change) 32.0 23.4 10.5 35.9 18.7 6.9 GDP deflator (percent year-on-year average change) 35.3 19.3 9.1 37.6 27.2 7.4 Exchange Rate (sod$,period average) 13 17 21 39 48 48 Current Account Balance (% of GDP) -24.6 -7.8 -22.9 -15.4 -5.6 -0.7 Fiscal Balance (accrual) (% of GDP) -9.3 -9.0 -11.3 -12.5 -9.8 -5.2 External Debt (% of GDP) 63.0 76.7 89.9 134.0 125.0 111.5 Gross Domestic Investment (% of GDP)-of which 25.2 21.7 15.4 18.0 16.0 15.8 Public 4.3 3.7 6.1 10.1 8.0 5.3 Growth rate of per capita private consumption (%, CPI index) 18.5 -12.0 27.1 -8.9 -5.2 4.7 Source: World Bank Live Database. 6 Sources of Growth, 1996-2001 1.7 The agricultural sector has led overall economic growth since the mid-1990s. While overall real GDP in 2001 was 38 percent higher than its level in 1995, the value- added o f agriculture in 2001 was greater than its 1995 level by 58 percent. As a result, the share of agriculture in GDP increased from 41 percent in 1995 to 45 percent in 1997 and reached 46 percent in 2001. The sustained increase in agriculture has been driven by structural reforms centered around a largely equitable land reform that transferred much of the land to private cultivation by peasant farmers, ended government intervention in price setting and procurement of agricultural commodities, and freed up the agricultural output and input markets. The land reform was conducted in stages, with much o f the land distributed in the 1995-1997 period. A land lease market has since been active, and the temporary moratorium on land sale was lifted in 2001. Improved incentives led to increases in overall crop production through changes in the composition of crops to higher-value crops (wheat, potatoes, sugar beets, and fruits and vegetables), and to recovery of yields (see Chapter 2 on agricultural growth and rural poverty reduction). The value of traditional agricultural outputs (cotton, tobacco, livestock) also increased duringthis periodbut at a much lower rate. 1.8 Incontrast, the manufacturing sector has continuouslydeclinedand is the worst performing sector in the Kyrgyz economy. The relative contribution of manufacturingto GDP declined from 8.6 percent in 1995 to 4.5 percent in 2001. The decline in manufacturing i s attributable partly to economic restructuring, partly to trade liberalization at a time when the machinery and light industry sub-sectors have been unable to compete with imported cheaper and better quality goods. 1.9 Another important reason for limited growth in manufacturing and services i s the unfavorable business environment for small and medium enterprises, which remains characterized by obstacles related to licensing, inspections, tax treatment, and corruption, as well as by trade barriers by neighboring countries. The unfavorable business environment resulted in stagnation in the number of active small and mediumenterprises (at about 9,000 enterprises) and in their contribution to GDP (at 16 percent). As a result, job creation has also been rather low (see Chapter 3 on labor outcomes and poverty). 1.10 Has economic growth since 1996 led to poverty reduction? The economic decline o f the first half o f the 1990's has clearly led to a substantial increase in poverty. However, the implications for poverty reduction of economic growth since 1996 have been less clear. Earlier evidence seems to suggest that poverty continued to increase despite a growing economy (see Annex 2.1). On the basis of the available data (see Box 1.1), the growth-poverty link is addressed below in two stages, to ensure strict comparability o f results. First, the next section demonstrates that economic growth between 2000 and 2001 led to poverty reduction and was pro-poor. Second, using a continuous panel of the same households, it i s demonstrated that overall poverty declined over the 1998-2001 period, as a result of a consistent decline in rural poverty, while urban poverty initially increased in 1999 under the impact of the Russian crisis but continuously declined over the 1999-2001. 7 I Box 1.1: Available Data Sourcesfor Poverty Analysis inthe Kyrgyz Republic The Kyrgyz Poverty Monitoring Survey (KPMS, 1993-1998) i s a multi-purpose individual, household and community survey conducted respectively in October-November 1993 (round I), February- March 1996 (round II), October-November 1996 (round 111), October-November 1997 (round IV), and October-November 1998 (round V). The data are not panel but nationally representative cross-section samples. Rounds 1-111covered approximately 2,000 households each, while subsequent rounds covered 2,700 households to allow analysis at the oblast level. These are essentially living standards measurement surveys, collected with assistance from the World Bank. The household questionnaire contains extensive and detailed questions on food consumption and frequent expenditure items on a two-week recall period, as well as expenditure on other goods and services on a longer recall period. The Household Energy Survey (HES 1999) is a special multi-purpose survey with focus on energy. Informationwas collected from a random sample o f 3000 households inthe fall o f 1999. The HES contains a different and more aggregated expenditure module than the KPMS, making it likely to underestimate the household consumption. The Household Budget Survey (HBS, ongoing since 1952) collects information on household expenditures and income on a monthly basis, using the diary method, and cover a whole year period. The HBS started in 1952 using "quota sampling" o f 1000 enterprise workers, farmers, and pensioners. The sample frame was changed in 1997 to a more population-based sample o f 2000 households, and further expanded to 3000 households in2000. The latest increase inthe sample size was aimed at making the HBS representative at the oblast level, in particular with a better representation o f the southern poorer regions.1 The survey also comes with a set of weights for the whole period 1998-2001 in order to re-calibrates the raw samples.2 The HBS is the only survey for the Kyrgyz Republic that has a panel structure. A household attrition rate o f less than 7 percent i s observed within any given year. Desertions from the sample are systematically replaced in order to maintain a constant sample size. There are 1108 households for whom information exists for the 48 months o f the 1998-2001 period. This panel i s used to estimate welfare changes during the 1998-2001 period. The United Kingdom Department for International Development (DFID) is supporting Natstatcom and i s recommending a replacement of the current H B S with a Kyrgyz Integrated Sample Survey (KISS). This will enhance the quality of data collection for poverty monitoring and policy analysis (in various areas as health, education, energy, etc.) on the basis of a totally new random sample of the population. While introducing this integrated survey may generate a one-time comparability problem, it sets the future poverty monitoring system on a more solid basis. B. GROWTH-POVERTY REDUCTION LINKAGES, 2000-2001 MeasuringWelfare 1.11 The welfare measure utilized throughout the report is per capita consumption. In addition, the sensitivity of key results concerning the poverty trends and profile to the welfare measure are also carried out. Table 1.5 i s based on two welfare measures - The HBS i s a three-stagesurvey. First, there i s a stratificationby area and region. Second, Primary Sampling Unit PSU (streets) are selectedwithin a strata. And third, there is a selectionof households within a PSU. In the second step, the concept of strata i s quite fluid. Within a town or village, streets were randomly selected. In the third step, a number of households were systematically selected from that street. The number of householdto be sampled out of a street was determinedusingquota-sampling. The weights are constructedin order to achieve national representativityby area of residence (urban-rural), regions (8 intotal), andage andgender groups (childrenless than 18years old, female adults of age 18-54; maleadults of age 18- 59; andthe elderly). 8 consumption per capita and consumption per adult equivalent. In addition, consumption expenditure per capita has been usedby Natstatcom for measuring welfare (see Box 1.2). The consumption aggregate used in this report was adjusted using oblast-level price indexes, separately for rural and urban areas, as obtained from the H B S price data. For comparisons across years, consumption was adjusted into 2001 prices using the Consumer Price Index (CPI). While relying primarily on consumption per capita, the report often tests the sensitivity of the results to the welfare measure. For example, Table 1.5 shows growth to be pro-poor, using both consumption per capita and consumption per adult equivalent. Box 1.2: MeasuringMonetaryWelfare As recognised in the 2000/2001 World Development Report (WDR) and the 2000 study of poverty in the World Bank's Making Transition Work for Everyone, poverty has a multi-dimensional nature, encompassing opportunities, security and empowerment. While recognizing the importance of the Droader definition o f poverty, this first report focuses on monetary indicators o f well-being. There are several possible methodologies used to measure household monetary welfare for poverty calculation, which are briefly reviewed below. Natstatcom has been publishing poverty statistics using expenditure per capita as a welfare measure. The other two measures are consumption per capita and consumption per adult equivalent. The expenditure/capita measure includes expenditures on food items, the estimated value of home- produced food items, and purchases of non-food items and services. Inaddition, this measure includes the hullpurchase value ofdurable goods, realestate, livestock, assistanceand gifts to others, as well as taxes and fees. The consumption/capita measure differs from expenditure in including the estimated use value of durables the household owns, with the exception o f housing. The consumption measure does not include the purchase of real estate, livestock, assistance and gifts to others, as well as taxes and fees, as these are not consumed by the household during the year. Consumption is perceived to be a better measure of household welfare than ex enditure because it measures the goods and services directly consumed by the household during the year. s Consumptionper adult equivalent assigns to adult women, children, and pensioners a fraction in order to get a total "adult equivalent" household size. The fractions, or equivalent scales, used here were derived for the Kyrgyz Republic by Barry Popkin (University of North Carolina), using 1993 data. The consumption per adult equivalent measure i s used for identifying whether a particular household is poor or not. The specific scale used is: 1for men 18 to 59 years old; 0.8 for women 18 to 54 years old; 0.78 for pensioners; 0.89 for children 14 to 17 years old; 0.78 for children 7 to 13 years old; 0.64 for children 4 to 6 years old; and 0.49 for children 3 years and younger. Income per capita is not used in this report given that consumption is considered a better welfare measure. Consumption fluctuates less than income and hence measures long-run welfare better. Consumption is also subject to less measurement error, given the large informal economy and desire to avoid taxes. Information on consumer durables are available in the HBS only since 2000. The welfare measure for poverty comparison after 2000 includes an estimate of use value of durables. However, the welfare measure for poverty comparison for the 1998-2001 period excludes the use value of durables in order to ensure comparison consistency across the whole period. 9 1.12 The absolute poverty line was re-constructed, jointly with Natstatcom, using the 2001HBS data and is estimated at 6975 Soms per capita annually. This national poverty line i s very close to the local equivalent o f international poverty line of PPP$2.15 per capita per day, which i s estimated at 6697 Somskapita per year in 2001. Earlier poverty lines were based on the 1996 KPMS data, and were updated using the CPI. It i s important to ground the poverty line in more recent data reflecting current consumption patterns. The estimation of the poverty line followed the same method utilized earlier by the World Bank and Natstatcom for calculating the poverty line using the 1996 KMPS. The absolute poverty line method i s estimated using the cost of basic needs approach, similarly to the established approach in the Kyrgyz Republic and elsewhere. A minimal daily calorie intake of 2100 calories/person/day i s adopted. The cost o f the food basket providing this nutritional level is then estimated, usingconsumption habits of the second- to-fourth deciles o f the population. The cost o f this food basket was 4648 Som per capita annually in 2001. This i s usually adopted in the Kyrgyz Republic as the "food poverty line", and i s treated as a threshold for identifying extreme poverty. Clearly, non-food basic needs should be accounted for as well. The cost of non-food basic needs is estimated on the basis of calculating the food share for households with per capita food consumption in the neighborhood o f the "food poverty line". The calculated food share i s 66.6 percenta4 The overall absolute poverty line i s then estimated at 6975 Soms per capita per year in 2001 prices. Prior to this estimation, Natstatcom updated its poverty line from its 1996 value to be 7491 Somskapita. An update of the World Bank (2000) poverty line would give a poverty line of 8509 Somdcapita in 2001 prices. Outcomes 1.13 The robust economic recovery in 2000-2001 had a significant poverty reduction impact. Growth in GDP per capita was about 5 percent in each o f 2000 and 2001. Consequently, average consumption grew by about 8 percent each in2000 and 2001 from the year earlier. Poverty had a reasonable decline in these two years as well. The headcount index of absolute poverty declined by about 6 percent each in 2000 and 2001 (see Table 1.3). The poverty gap and severity indices had a comparable reasonable reduction in 2000 and 2001.5 One reasonfor suchahighshare i s that the use value of durables computedhere is relatively small at 3 percentof total consumption. The improvementin 2001 of all poverty indicatorsbased on the panel sample i s fully consistent with the results from the whole survey sample discussedin the previous section; the improvementexperienced by the panel households in 2000 is however contrasted by a deteriorationin the whole survey sample--explained, as discussed earlier, by the expansionof the sample that took placein that year. 10 gridSample Whole 1998 1999 2000 2001 Gr.Rate of GDPper capita (% per annum) 0.5 2.2 4.5 5.4 PanelHouseholds Headcount Index (PO) (%) 55.3 54.8 48.2 42.6 Poverty Gap Index (Pl) 0.167 0.163 0.129 0.108 Poverty Severity Index (P2) 0.067 0.064 0.048 0.039 Whole Sample Headcount Index (PO) (%) 50.9 53.8 59.6 52.3 Poverty Gap Index (Pl) 0.156 0.161 0.197 0.154 Poverty Severity Index (P2) 0.063 0.065 0.086 0.062 1.14 Absolute poverty was clearly reduced in 2001 compared to 2000, regardless of thepoverty index or welfare measure. Given the pro-poor growth (Table 1.5), poverty i s expected to decline in 2001 due to both higher average consumption level and reduced inequality. Poverty indeed declined for all welfare measures (consumption per capita, consumption expenditure per capita, and consumption per adult equivalent, see Box 1.2 above) and for the three standard aggregate measures of poverty-Headcount Index (PO), Poverty Gap Index (Pl) and Poverty Severity Index (P2) (see Box 1.3 on Poverty Measures). Using already reported Natstatcom figures, the fraction o f the population below the poverty line declined from 52.0 percent of the population in 2000 to 47.6 percent in 2001 (see Table 1.4).6 Usingthe consumption per capita, the headcount index of poverty declined from 62.5 to 56.4 percent of the p~pulation.~There were fewer people below the poverty line in 2001. In addition, the average consumption level of those remainingbelow the poverty line i s closer to the poverty line, and there are fewer people who remain at the extreme bottom of the consumption distribution, as indicated by the poverty gap andpoverty severity indices. Table 1.4: AbsolutePovertyinthe KyrgyzRepublic,2000-2001 Measure of welfare Absolute HeadcountIndex (PO) PovertyGap (Pl) PovertySeverity poverty (%) (P2) line,2001 (%) soms 2000 2001 2000 2001 2000 2001 Expenditure per 7491 52.0 47.6 16.1 13.9 7 5.6 capita Consumption per 6975 62.5 56.4 21.1 17.2 9.3 7 capita Consumption per 6975 51.8 45.0 15.1 11.7 5.9 4.3 adult equivalent Carrying out tests of significance for the differencesin poverty rates requires additional information on the sample 'designstructure,which are currently unavailable. Using the internationalpoverty line of 6697 Soms per capitaper year (PPP$2.15),the headcount index of poverty in 2001was 53.6percent. The ECAPOV study, World Bank (2000) reports a headcountindex of poverty of 49.1 in 1998, usingthe KPMS. Given these estimatesrely on different survey instruments, the KPMSfor 1998 and HBS for 2001, it would be incorrectto conclude that poverty increasedbetween 1998 and2001. See the next sectionandAnnex 1.1. 11 Box 1.3: Measuresof Poverty This analysis uses the three standardFoster-Greer-Thorbecke(1984) aggregatepoverty measures. The headcount index (PO) i s a measure of the prevalence of poverty. I t denotes the percentage of households who are poor-as defined by the poverty line-as a proportion of total population. This measure was insensitive to the distribution of the poor below the poverty line. The poverty gap index (Pl) is a measure of the depth of poverty, and it denotes the gap between the observed consumption levels of poor householdsand the poverty line. Assuming perfect targeting of resources (transfers), this poverty gap index indicatesthe total amount neededto bring all householdsinpoverty up to the poverty line. The poverty severity index (P2) measures the degree of inequality in distribution below the poverty line, giving greater weight to householdsat the bottom of the consumption distribution. where n= total population, q = number of those with consumption yi less than the poverty line z. 1.15 Indeed,for any poverty line, poverty is lower in 2001 compared to 2000. The cumulative distribution function graph shows the fraction of the population with consumption less than a given level. This graph effectively shows the poverty incidence for different poverty lines. A lower graph means lower poverty. Given that the fraction of the population with consumption less than a given level is lower in 2001 comparedto 2000 (see Figure 1.3), there is clearly less poverty in 2001. This is indeedtrue for any given poverty line and for the urban and rural population, taken separately.8 The implication of this is that social welfare in the Kyrgyz Republic i s definitely higher in 2001 compared with 2000, for the country as a whole and for the urban and rural populationtaken separately. Figure 1.3: CumulativeDistributionFunctions, 2000 and2001 1 - 35 - 5 - z5- Source: HBS data. Note: weight, national sample of 3000 households. ~ The figure also illustrates that rural poverty i s higher than urbanpoverty inboth2000 and 2001. 12 1.16 The reduction in poverty is also evident for extreme poverty. Natstacom has utilized the cost of the food basket that provides minimal nutrition as an extreme poverty line. Adopting this, indicators of the prevalence, depth, and severity o f poverty were lower in 2001 compared to 2000 (see Table 1.5). inthe Kyrgyz Republic, 2000-2001 Measure of welfare Food Headcount Index Poverty Gap (Pl) Poverty Severity (2) poverty line, (PO) (%) (%) 2001s o m (%) 2000 2001 2000 2001 2000 2001 Expenditureper 4510 17.8 13.5 3.9 2.7 1.3 0.8 capita Consumptionper 4648 32.9 24.7 7.7 5.3 2.6 1.8 capita Consumptionper 4648 20.6 14.6 4.1 2.7 1.3 0.8 adult equivalent Pro-Poor Growth, 2000-2001 Economic growth in 2001 was clearly pro-poor. Real GDP grew at a robust rate of 5.4 percent. Using the Household Budget Survey data collected from about three thousand households in both 2000 and 2001 (see Box 1.1), the average consumption per capita increased by 8 percent in 2001 (see Table 1.6). There was a general improvement inwelfare at all levels of the consumption distribution between 2000 and 2001. Growth was pro-poor, since the percentage increase in per capita consumption at the lower deciles exceeded that at the upper deciles. Table 1.6: Average Consumption by Decile, 2000-2001(in 2001 prices) Consumption per capita Consumption per adult equivalent Decile 2000 2001 % change 2000 2001 % change Poorest 1 2658 2945 10.8 3145 3472 10.4 2 3604 4017 11.5 4226 4684 10.8 3 4220 4669 10.6 4969 5468 10.0 4 4847 5395 11.3 5679 6208 9.3 5 5500 6056 10.1 6363 6971 9.6 6 6297 6846 8.7 7246 7825 8.0 7 7262 7892 8.7 8365 8995 7.5 8 8515 9342 9.7 9773 10561 8.1 9 10672 11515 7.9 12081 12890 6.7 Richest 10 17220 17769 3.2 19234 19739 2.6 Total 7084 7650 8.0 8110 8686 7.1 Source: HBS. Note: Weighted data, national sample of 3000 households. 1.17 Growth was morepro-poor in urban areas than rural areas in 2001. All deciles of the population benefited from growth in 2001, regardless of whether they are in urban or rural areas (see Figure 1.4). The average improvement was greater in rural areas, as 13 consumption per capita increased by 8.9 percent in rural areas and 6.5 percent in urban areas in 2001.9 However, growth was clearly more pro-poor in urban areas, with the greater increase in consumption for poorer deciles. The pattern between relative improvement and decile position i s less clear inrural areas. Figure 1.4: Changeinper Capita Consumptionby Decile for Urban- RuralPopulation,2000-2001(%) ~ 180, I - 1- - 2 3 4 5 , . , 6 7 8 9 10 average change in per capita consumption - I change in consumption per capita by decile (urban) -change in consumption per capita by decile (rural) Source: HBS. Note: Weighted data, national sample of 3000 households. 1.18 The poverty reduction in 2001 is also broad-based across the regions of the Kyrgyz Republic. No oblast had an increase in poverty. However, some oblasts had greater reduction in poverty than others. It i s of concern that the oblast with the highest incidence of poverty, Naryn, hardly experienced any poverty reduction (see Table 1.7). This calls for special attention to geographically concentrated pockets of poverty that appear to be caught in a "poverty trap" and unable to benefit from relatively broad-based pro-poor growth. A regional emphasis on poverty reduction in the NSPR is necessary to supplement sustainedpro-poor growth. Table 1.7: Incidence of Poverty by Location, 2000-2001 Headcount index, consumption per capita Absolute Poverty Extreme Poverty 2000 2001 2000 2001 Total 62.5 56.4 31.9 24.7 Urban 53.3 45.4 26.5 17.7 Rural 67.4 62.4 34.7 28.4 Oblast Bishkek 40.0 31.5 19.2 7.8 Issyk-Kul 67.3 63.0 42.9 33.4 Jalal-Abad 72.5 68.0 24.4 24.4 Naryn 89.4 89.1 65.6 61.0 Batken 76.0 54.5 43.6 17.6 Osh 70.8 66.7 38.4 34.1 Talas 83.0 72.3 53.3 39.1 Chui 37.2 33.1 14.5 9.3 Source: HBS.Note: Poverty estimatesare weighted, national sample of 3000 households. Rural consumption levels were still lower than urban consumption level, despite narrowing gap. Consumption per capitainrural areas inrelationto that inurbanareas increased from 82.4 percent in2000 to 84.2 percent in2001. 14 Inequality 1.19 The overall reduction in poverty between 2000 and 2001 is driven by growth, as well as by reduction in inequality. Between 2000 and 2001, consumption per capita grew at a faster rate in rural areas (8.9 percent) than urban areas (6.5 percent). Yet the reduction in absolute poverty was 7.9 percentage points for the urban population, greater than the poverty reduction for the rural population, which was 5 percentage points. The primary reasons for the difference is that rural poverty reduction was basically driven by consumption growth with little change in inequality, while poverty reduction in urban areas was driven by both consumption growth and inequality reduction. To see that, the change in a poverty index of can be decomposed into a "growth component", and a "redistribution component" indicating the impact o f inequality change (see Box 1.4). Practically all rural poverty reduction was practically all driven by growth, about 40 percent o f the urban absolute poverty reduction could be attributed to reduction in inequality among the urban population (see Table 1.8). A change inpoverty between two years can be considered either as a result of changes inaverage standards of living, or a change in the distribution o f consumption. A method developed by Ravallion and Datt (1992) i s used here to decompose the overall change in poverty into a "growth" component", a "redistribution" component, and a residual." The growth component i s the poverty change that would result if the consumption o f all individuals in the second year is increased from its value in the first year equally by the ratio of average consumption between the two years. This would essentially leave the distribution of consumption identical between the two years, thus providing the "growth component" of the poverty change. The redistribution component assumes that the average consumption between the twc years i s identical, thus allowing us to measure the impact of a change in inequality on the poverty index. These "growth" and "redistribution" components do not typically add up to exactly the overall poverty change, and the difference is the "residual" in the equation. Table 1.8: Decompositionof Povertyinto GrowthandRedistributionComponents,2000-2001 Change inthe headcount index. absoluteDovertv Change inthe headcount index. extremeDovertv Actual Growth Redistribution Residual Actual Growth Redistribution Residual All country -6.1 -5.85 -0.40 0.15 -7.2 -5.47 -2.26 0.52 Urban -7.9 -4.37 -3.02 -0.51 -8.8 -3.80 -4.81 -0.19 Rural -5.O -6.53 0.73 0.80 -6.3 -6.22 -0.34 0.26 Source: HBS. Note: Poverty estimatesare weighted, national sample of 3000 households. 1.20 In addition to having higher consumption and lower poverty, the inequality in consumption is clearly reduced between the years 2000 and 2001. The Gini index of inequality i s lower in 2001 compared to a year earlier, for the country as a whole and for the urban and rural sectors separately (see Table 1.9)." This i s indeed expected given lo Datt, G. and M. Ravallion (1992), "Growth and RedistributionComponents of Changes in Poverty Measures: A DecompositionWith Applications to Brazil and Indiain the 1980s," Journal of Development Economics, 38 l1 The Gini inequality indices are low and shouldnot be compared to Gini indices reportedusing a different welfare measure. For example, adding the full value of consumer durables to the welfare measure (and hence using expenditure) is expectedto generate ahigherGini index of inequality. 15 that poorer segments of the population experienced a greater increase in their consumption between 2000 and 2001. Indeed, a Lorenz curve of consumption per capita shows the extent of inequality. The Lorenz curve for the 2001 distribution i s closer to the diagonal, the line o f perfect equality, and hence has less inequality than the 2000 consumption distribution (see Figure 1.5). Given this, any methodologically reasonable measure of inequality will show a decline in inequality in the year 2001. This includes the Gini index, Theil indices, coefficient of variation, Atkinson indices, or any measure that i s consistent with the Lorenz criterion. Table 1.9: GiniIndex of Inequality in2000 and2001 2000 2001 Total 30.3 29.0 Urban 30.7 28.4 Rural 29.2 28.8 Source: HBS data. Note: weighted national sample of 3000 households. Figure 1.5: LorenzCurvesof ConsumptionInequality,2000 and2001 (dashedline=2000,solidline=2001) 0 , , 100 cumulative percent of population 50 C. POVERTYCOMPARISONOVERA FOUR-YEAR PERIOD, USINGHBS 1998-2001 Methodological Issues 1.21 Poverty comparison over 1998-2001 is seriously complicated by the expansion of the HBS sample in 2000.l2About one thousand households were added to increase the sample size to three thousand households, in order to make the sample more representative at the oblast level. However, the added households had a per capita consumption that i s about 70 percent of the consumption level of the original sample households. While households weights were re-computed, these largely accounted for the population's demographic characteristics, and did not fully account for the fact that l2Electronic files of the HBS are not available for years earlier than 1998. Poverty estimates for earlier years largely drew on other data sources, and are reviewed in Annex 1.1. 16 the added households were much poorer on average compared to the original sampled households. Thus, poverty comparison over the four-year period, using the full H B S sample, i s bound to show a worsening of poverty between 1999 and 2000, compared to any real change. Fortunately, the sample has a panel structure tracking the same households over time unless they drop out, when an appropriate household replacement i s found. 1.22 Thus, the four-year poverty comparison is based on a continuous panel of 1108 households who were observed every month during the 1998-2001 period. The panel sample i s the only dataset that can be used for consistent poverty comparisons in the Kyrgyz Republic over the 1998-2001 period. However, some qualifications on using the panel are in order. First, a panel of households does not represent the whole population at any particular moment. The reason for the addition of households in the year 2000 i s to make the sample more representative of regions that are much poorer than average. Thus the poverty levels in the panel sample are likely to be less than the true poverty levels for the population as a whole. Second, households dropping out o f the panel sample are likely to be those with a high value of their time, and these are likely to be poorer households. For poorer households free time is an almost non-existent luxury, and often they do not have time to participate in surveys, engagingin back-breaking and exhausting physical labor from dawn till dusk to make ends meet. Rich households also value time, but unlike the poor they can afford the luxury o f taking time out to answer survey questions. This may make the panel sample biased toward the richer households (thusbiasedinfavor of detecting less poverty than there actually is) or toward households less likely to remain poor (thus biased in favor of detecting a reduction in poverty). These potential biases exist with every panel sample. Yet, there i s a clear advantage of carrying out poverty comparison over time using the panel sample, namely the data are indeed comparable across years, and the analysis will not be adversely influenced by the changing nature o f the whole survey sample. Moreover, while the panel sample may not represent the whole population, it represents a sizable fraction o f the population, for whom a better understanding o f the growth-poverty links will be feasible. Figure1.6: Consumptionper Capitafor PanelandWhole SampleHouseholds 1.23 It is re-assuring to findthat theper capita consumptionfor thepanel andfor the whole sample are fairly close for the years 1998-1999 (see Figure 1.6).13 Indeed, the l3The use value of durables i s not included in the consumption per capita variable that i s used with the panel and for 17 average per capita consumption for the panel households was about 3 percent less in 1998-1999 than that for the whole sample. This supports the view that households in the panel sample are slightly poorer than those dropping out for having a higher opportunity cost of their time. Following the expansion of the whole survey sample in 2000, the average consumption for the panel households i s about 16 percent greater than that for the whole sample in the years 2000-2001. The added households had a consumption level that i s about 70 percent of the consumption of the panel and (pre-2000) sample households. Thus, in comparing welfare change between 1999 and 2000, indicators basedon the whole sample are biased infavor of showing lower consumption and greater poverty compared to indicators derived from the panel households. 1.24 When consumption trends are disaggregatedfor the urbanandrural populations, a similar trend emerges in the relation between panel and whole sample averages. Consumption average for the panel households i s slightly less than that of the whole sample for each of the rural and urban sectors in 1998 and 1999. Yet, consumption for the panel households exceeds that for the whole sample in 2000 and 2001 by about 10 percent for the rural and 13 percent for urban households (see Table 1.10). Thus, the biased nature of expanding the whole sample in 2000 by poorer households seems to apply equally for the urban and rural sectors. Table 1.10: AverageMonthlyPer Capita Consumptionfor Paneland Whole SampleHouseholds, B y 2 0 0 1 , andby Urban-Rural(in monthly Dec. 2001Soms) Quarter/ Year PanelHousehold Whole Samplehouseholds Rural Urban Total Rural Urban Total 41-98 612.0 785.3 666.8 626.2 805.5 690.5 42-98 594.8 733.8 638.8 630.3 757.1 675.8 43-98 672.0 759.3 699.6 681.5 781.7 717.5 44-98 638.6 732.5 668.3 658.4 767.7 697.6 41-99 623.7 751.4 664.2 638.1 782.1 688.6 42-99 617.5 651.9 628.4 621.5 692.7 646.5 43-98 672.8 690.0 678.2 672.1 723.3 690.1 44-99 677.5 731.1 694.5 695.6 751.6 715.3 41-00 676.8 742.6 697.6 633.4 692.8 656.1 42-00 656.3 724.1 677.8 611.9 672.8 635.1 43-00 685.6 747.8 705.3 641.7 674.6 654.2 44-00 791.0 826.3 802.2 756.7 782.9 766.7 41-01 746.4 819.0 769.4 656.7 692.2 670.5 42-01 693.4 760.1 714.5 610.1 637.7 620.8 43-01 769.3 812.0 782.8 652.4 679.1 662.8 44-01 832.0 868.6 843.6 737.1 752.3 743.0 Note: Averages are for sample households only without weights. the whole sample in this section. Information on durables is not available prior to the year 2000. The absolute poverty line used with this consumption variable i s accordingly adjusted to 6785 soms per annum in 2001 prices, in contrast to 6975 soms when the use value of durables is included. The poverty line used for analysis inthis section i s converted to a monthly poverty line o f 568 soms per month in December 2001 prices. This i s one reason that resulting figures for consumption and poverty in this section to differ slightly from those presented in the previous section. Another reasons i s that figures are reported in this section for the panel sample individuals, given that weights differ across years. 18 Outcomes 1.25 Growth was the driving force behind poverty reduction for rural households over the 1998-2001 period. Growth was also an important factor in reducing urban poverty since 1999. In addition, reduction in urban inequality had a poverty-reducing impact in 1999 and 2001. The separate impacts of growth and inequality reduction on poverty duringthe 1998-2001 period can be identified though a decomposition analysis (see Box 1.4). Table 1.11 shows the results o f this decomposition. It clearly shows that the poverty reduction has been mostly driven by growth in average consumption levels. In 1999, the urban households experienced a reduction in average consumption level, increasing the poverty headcount index by 3.9 percent, and a reduction in inequality, reducing poverty by 3.1 percent. The net result was an observed poverty increase in the headcount index of 1.5 percent, taking into account the residual effect. Inequality reduction had a major poverty-reducing impact in 2001 for urban households. Clearly, growth since 1999 has been reducing poverty, unhindered by worsening inequality and in some cases assisted by a reduction ininequality. Table 1.11: Decompositionof OverallPoverty Changeinto Growth& InequalityChange,1998-2001 Actual Changein Growth Inequality HeadcountIndex Component Component Residua1 All households 1999/1998 -0.5 1.5 -1.7 -0.3 2000/1999 -6.6 -5.3 -1.0 -0.3 2001/2000 -5.6 -6.0 -1.8 2.2 Urbanhouseholds 1999/1998 1.5 3.9 -3.1 0.7 2000/1999 -3.9 -4.4 -1.0 1.5 2001/2000 -8.5 -4.8 -5.6 1.9 Ruralhouseholds 1999/1998 -1.3 -0.5 -0.8 0.0 2000/1999 -8.1 -6.0 -0.9 -1.2 2001/2000 -4.2 -5.9 -0.1 1.8 Source: HBS panel dataset. 1.26 The change inpoverty between 1998 and 1999 is more complex, and results from diverging trends in urban and rural poverty. The Russian crisis affected the Kyrgyz economy in the second half of 1998 andfirst half of 1999. The crisis affected the rural and urban sectors differently, thus making the sectoral change in poverty of greater interest than the aggregate change in poverty. For the panel households, rural poverty had a marginal decline while urban poverty increased in 1999. The net result turned out to be very little change in aggregate poverty between 1998 and 1999 (see Table 1.12). Much of the GDP growth (3.7 percent) in 1999 i s accounted for by the growth in agricultural sector (8.2 percent). Indeed, consumption of the urban households declined by about 6 percent in 1999 while it increased by about 3 percent for the rural panel households. 19 1998 1999 2000 2001 Urban HeadcountIndex (PO) (%) 44.6 46.1 42.2 33.7 Poverty Gap Index (P1) 0.127 0.135 0.115 0.091 Poverty Severity Index (P2) 0.054 0.052 0.044 0.034 Rural HeadcountIndex (PO) (%) 60.2 58.9 50.8 46.6 Poverty Gap Index (Pl) 0.185 0.175 0.135 0.116 Poverty Severity Index (P2) 0.073 0.070 0.050 0.041 1.27 Relative food prices increased as a result of the Russian crisis, thus favoring the rural population that i s largely food producers. The inflationary pressure o f the Russian crisis occurred mostly between the middle of 1998 and middle of 1999. During this period, food prices rose much faster than prices of non-food goods (see Figure 1.7).14 Foodproduction i s the major source of income for the rural population, accounting for 53 percent o f their total income in 2001. The comparable ratio for the urban population i s 15 percent, and this is mostly food produced and consumed by the household. As a result, the relative price increase for food products had a favorable impact on the rural population. Figure 1.7: Food Prices Rose Faster Than Those for Non-food Goods _ _ _ _ _Total ............ Non-Food . -.. . -.-Food . Services I 1 I 2.00 1.75 ,- ._e, t 1.25 . I I I 1998 1999 2000 Year 2001 200. CPI Dynamics,Kirghiz Republic, 1998-2001 1.28 There was a small reduction in inequality during the 1998-2001 period (see Table 1.13). Usingthe four-year panel sample, the Gini index shows a small reduction in inequality for urban households in 1999 and in 2001, while rural inequality declined in the year 2000. The Gini index of inequality for the whole panel sample registers an aggregate decline between 1998 and 2001. l5These inequality reductions contributed to poverty reduction inthis period. l4This does not necessarily imply that terms of trade movedin favor of agriculture during the relevantperiod. l5 Given the likely attrition of richer people from a panel sample, due to their higher value of time, inequality indices are likely to be biasedfor apanel where quitting householdsare replaced. The inequality indices computedhere are for a panel wifhouf replacement, so the same set of households are observedall the time. Thus, attrition problems do not 20 Table 1.13: GiniIndex of Inequality1998-2001, inConsumptionper Capita,for PanelDataset Gini Index 1998 1999 2000 2001 Total 0.318 0.306 0.299 0.299 Urban 0.310 0.290 0.294 0.282 Rural 0.315 0.311 0.299 0.304 1.29 The growth-poverty conundrum would emerge if one were to compare poverty during the 1998-2001 using the whole sample. As Table 1.3 illustrates, and using the whole sample, the incidence of poverty would appear to increase slightly in 1999 but more significantly in the year 2000 while GDP was growing at a decent rate. The expansion of the whole sample that added poorer households in the year 2000 i s indeed responsible for this apparent conundrum in poverty comparison between 1999 and 2000. No conundrum emerges when the analysis is carried out in a manner that controls for sample change, inthis case with the panel sample. 1.30 Usingthe four-year panel sample, it is clear that aggregate poverty changed little between 1998 and 1999 and then consistently declined over the 1999-2001 period. Moreover, rural poverty consistently declined over the four-year period. Urban poverty increased in 1999 but consistently declined over the 1999-2001 period. The cumulative distribution functions in Figure 1.8 show the incidence o f poverty for various poverty lines. Given that the poverty incidence curve increased between 1998 and 1999 for urban households, this indicates that urban poverty increased in this period. But the graph also shows that urban poverty consistently declined from its 1999 level. The poverty incidence curves for the rural panel households show a consistent decline over the 1998- 2001 period. (Specific poverty measures are reportedin Tables 1.3 and 1.12). Figure 1.8: Poverty IncidenceCurvesfor PanelHouseholds,1998-2001,for Rural and Urban Sectors _ _ - _ _1998 ............1999 2000 2001 I i- .6 - .4 - .2 - n - 0' 400 568 Consumption PC, Dec-01 Prices 800 1200 Poverty Incidence Curves, Rural, HBS Panel bias the panel inequality indices. The extent to which the panel sample (and therefore its inequality indices) may or may not representthe whole populationwas addressedinparas. 1.20-1.22 above. 21 ............ ..-..-2001 1999 0 0I 400 I Consumption PC, Dec-01 Prices 568 800I 1200 I Poverty Incidence Curves, Urban, H B S Panel D. CONCLUSION: A SERIOUS POVERTY REDUCTION CHALLENGECONTINUES 1.31 This chapter has shown that the Kyrgyz economy experienced robust growth in 2000-2001, recovering from anemic growth during the Russian crisis. The economic recovery resulted in pro-poor growth in 2001, and was accompanied by poverty and inequality reduction. Thus the economy scored well on all three objectives of development in 2001. Moreover, analysis of a continuous panel o f households reveals that there i s no growth-poverty conundrum, and that economic growth, coupled with relatively moderate and somewhat declining inequality, resulted in reducedpoverty for the two consecutive years of economic recovery-2000 and 2001. The 1998-99 poverty change i s more complex, given the increase in relative food prices as a result of the Russian crisis. This relative price increase favored rural households, making rural poverty decline marginally while urban poverty increased in 1999. 1.32 Despite the fact that the Kyrgyz economy experienced poverty-reducing growth for two years in its post-Russian crisis economic recovery, the poverty challenge remains large. One out of every two people lives in absolute poverty, and one out o f every four people lives in extreme poverty. Sustainable economic growth will have to be at the center of the Kyrgyz poverty reduction strategy. For sustainable growth to have a significant impact onpoverty reduction, it is also important to maintain inequality at its moderate level, and prevent it from rising. While growth and poverty reduction have been relatively broad-based, it i s important to recognize that Naryn, the poorest oblast, benefited very little from the recent economic recovery. This calls for an additional regional dimension to the broad poverty reduction strategy, in order to ensure that benefits from growth are more widespread, particularly in the poorest areas. 22 Simulation of Impacts of Growth Rates on Poverty Reduction 1.33 Figure 1.9 presents the expected poverty-reducing impact of economic growth over 2002-2006 period. The graph presents simulation for two growth rates of real per capita consumption: 5 percent and 3 percent per annum. These simulations assume that inequality is held constant at its 2001 level. The starting point for the simulation is the headcount index of absolute poverty of 56.4 percent in 2001, given the full HBS sample. Given the consumption distribution in the Kyrgyz Republic, a 5 percent annual growth rate in per capita consumption i s expectedto reducethe headcountindex of poverty to 36 percent in 2006, a reduction of 4 percentage points per year on average. Out of a population of almost 5 million people, this will effectively reduce the number of people living in absolute poverty from 2.8 million people in 2001 to 1.8 million people in 2006. A 5 percent consumption growth rate translates into 200 thousandescapingpoverty every year, a million peopleover 5 years. Figure 1.9: Simulating the Poverty-Reducing Impact of Economic Growth, 2002-2006 Expected Reduction inPoverty Headcount with GDP Growth at 3% and 5% and unchangedinequality 0.60 * 0.55 3B 0.50 T p 0.45 g2 0.40 0.35 0.30 1 2001 2002 2003 2004 2005 2006 1 1.34 A less ambitious growth rate of per capita consumption of 3 percent per year will only reduce the incidence of poverty to 45 percent by 2006, a 2.4 percentage point reduction per year. This implies that the number of people living in absolute poverty declines by 580,000 people over a five-year period, about 116,000 per year. 1.35 Thesepoverty reduction goals can only be achieved if growth is sustained and inequality does not increase. Inequality may very well increase if growth i s broadened outside agriculture to industry and modern services, which will have to make more differentiated rewards to factors of production than agriculture. If the Gini index of consumption inequality increases by 0.01 per year from 0.29 in 2001 to 0.34 in 2006, then a 5 percent annual growth rate in consumption is expectedto reduce the headcount index of poverty to merely 41 percent by 2006; insteadof the 36 percent that is expected if inequality remains unchanged. A 3 percent annual growth rate in consumption with increasing inequality is expected to reduce the headcount index of poverty to merely 50 percent, rather than 45 percentwith unchanginginequality, by 2006. 23 1.36 Sustaining growth i s expected to be much harder over the coming few years, as the conditions for 1996-2001 growth appear unlikely to be replicated over the medium term. Aside from private investments in gold, growth since 1996 has relied substantially on public investments and consumption (including imports). These macroeconomic conditions are difficult to sustain. Current high levels of external debt-at the moment transitorily mitigated by the recent Paris Club flow restructuring operation-pose too many risks to external sustainability. Importanceof Broad-BasedGrowth 1.37 Future poverty reduction will crucially depend not just on the level of growth but on the forces driving this growth as well. There i s a need for a more balanced growth that relies on multiple sectors growing at the same time, in order to ensure a solid and sustained growth in output and incomes over the medium term. It i s possible that a scenario of 3 percent growth that i s based on trade and private investment could result in greater poverty reduction than a 5 percent growth that i s primarily supported by fiscal and external imbalances. More specifically, the sources of demand for growth are important for the extent of poverty reduction. 1.38 Growth in the most recent 2000-2001 period, was more balanced but lacked the necessary exports and private investment dynamism. The authorities began adjusting private consumption and imports since late 1998 when the exchange rate was allowed to float. Further adjustment o f public consumption and imports continued during the 2000- 01 period, when external adjustment was complemented with fiscal adjustment, and measures to initiate the process that would address the dire strait conditions of the banking and energy sectors. As a result, the trade balance improved substantially, but mainly due to a sharp fall in imports. Exports (other than gold, and those related to the inter-governmental irrigation agreements with Uzbekistan) have shown no dynamism. Similarly, lagging reforms to improve the business environment, as well in the energy, telecommunications and mining sectors, have lead to an unfavorable investment climate, in turn resulting in very low rates of private investment. Saving rates have improved following the strong stabilization efforts, but the ability of the financial sector to mobilize savings and investments i s still in question. 1.39 Growth in labor-intensive sectors i s more pro-poor, but their progress i s undermined by domestic constraints created by an excessively bureaucratic regulatory process, some features o f the taxation regime and enforcement, and other factors. It i s noteworthy that growth in 2001 was driven, in addition to agriculture, by construction, transportation, trade, catering together accounting for about 18 percent in GDP. These sectors are dominated by micro and small-sized enterprises. Only the construction materials sector has a few medium-scale enterprises (a recently privatized cement plant as well as various business start up in brick and other related activities mainly focusing on exporting to neighboring Kazakhstan). Like in agriculture, reforms and liberalization have led to some vibrancy in these sectors, with notable exceptions in air transport, railway, and part of the freight truck fleet. As elaborated in Chapter 3, and as commendable as has the embracement by the government o f the objective of improving 24 the investment climate for these small firms, significant and tangible gains are yet to materialized. 1.40 While agriculture has generateda basic level of growth, without which the present levels of poverty would be significantly higher, this sector alone cannot continue to bring about rapid growth. Moreover, agriculture may be close to its limit in providing a cushion to worsening conditions elsewhere in the economy, by absorbing workers from collapsing sectors. As Chapter 2 shows, success in the agriculture sector can be attributed to a mutually reinforcing set of reforms that the authorities have been implementing consistently over the past 6-8 years. These have been supported by several public interventions, some of which can be associated with the subsequent supply response (e.g., rehabilitation of irrigation, market-driven micro-credits schemes). It i s important that agricultural growth i s enhanced through market-driven productivity gains, consolidation of property rights and landreform, and the need to improve the marketing, particularly export orientation, of agricultural products. However, agricultural growth will be significantly conditioned by the extent of improving the export potential, as well as by a rising domestic demand, which in turn requires significant growth in the non- agricultural sectors (Figure 1.10). Figure 1.10: Exports and Investment-Savings i 750 I 650 ,Total Exports, and Exports without power and gold, U S mln 550 450 350 250 I I 150 I 1993 1994 1995 1996 1997 1998 1999 2000 2001 Investments and National Savings (% o f GDP) 30 zg 20 10 0 8 0 -10 1995 1996 1997 1998 1999 2000 2001 /-Investments 0 0 . w . Public Investments - National Savings 1 1.41 While the gold mining sector contributes to overall GDP growth, it has limited linkages with the domestic economy and i s therefore unlikely to have significant impact 25 on poverty reduction. Even after a decade of transition, the manufacturing sector shows no signs of revitalization.16 The dismal conditions of the manufacturing sector make it clear that the necessary conditions have not been created for stopping and reversing the rapid decline of this sector. 1.42 The analysis of sectoral sources of growth suggests that the significant gains in poverty reduction achieved in recent years are not likely to be sustained over the medium-term unless there i s a commitment to a strong reform agenda that ensures broad- based growth. l6A detailed assessment of the current state of manufacturing is well beyond the scope of this work. 26 2. AGRICULTURALGROWTHAND RURAL POVERTYREDUCTION Rural poverty fell steadily between 1998 and 2001, declining at an average rate of 8percent a year, as the agricultural sector grew at an average rate of 6 percent a year starting in 1996. The agricultural growth rejected an increase in supply and demand, as well as an improved incentive framework and a shift towards higher-valued products. This chapter examines the relationship between agricultural growth and rural poverty. It starts by analyzing the recovery in agriculture that has occurred since 1996 and the key policies that supported it. It then analyzes the linkages between the agriculture sector and rural poverty, before concluding with an assessment of some of the key challenges andpolicy considerationsfor the continued pro-poor growth of the rural sector. A key factor in economic growth and poverty reduction is the increased level of assets (mainly land, livestock, and human assets) among rural households since the mid-1990s. An equitable land reform led to widespread land ownership, with 75 percent of rural households having access to land. Similarly, over half of the rural households own cattle and one third have small ruminants. Liberalizing the agricultural markets also contributed to agricultural growth and rural poverty reduction. Despite these impressive gains in ownership of agricultural assets, the sector remains characterized by traditional production techniques and low levels of market integration. With the high incidence of rural poverty (62.5% in 2001), continued rural growth will be crucial for future reductions in poverty. To capture the full potential for productivity gains, and to ensure an incentive framework that promotes an eficient and transparent allocation of resources in the sector, it is recommended that the Government consolidates reforms undertaken to date, such as: (a) promoting the development of rural land markets in the long-run and improved rental markets in the short run; and (b) supporting agricultural market development to reduce transaction costs and deepen rural markets; (c) enhancing agricultural support services; and (d) completing the institutional framework for water management and continuing investing in irrigation. In addition, data collection and analysis on rural living standards need to be improved. A. THERECOVERYINKYRGYZ AGRICULTURE Sectoral Performance 2.1 Ruralpoverty fell steadily between 1998 and 2001, declining at an average rate of 8 percent a year (Table 1.12), as the agricultural sector reversed a steep decline inthe first 27 half of the 1990s and grew at an average rate of 6 percent a year starting in 1996.' The share of agriculture in overall GDP increased from 41 percent in 1995 to 46 percent in 2001. While overall real GDP in 2001 was 38 percent higher than its level in 1995, the value-added of agriculture in 2001 was greater than its 1995 level by 58 percent (Table 1.1and Figure 1.1). This growth reflected an increase in supply, a recovery in national demand for food products, and an increase in value o f agricultural output, as producers shifted away from fodder towards food crops. 2.2 Reflecting the increased domestic demand for food products, crop production recovered faster than the traditionally more export-oriented livestock sector (Figure 2.1). Crop production almost doubledbetween 1995 and 2001, while livestock production increased by only 20 percent. Within the crop sector, the major output gains were achieved by grains, fruits, vegetables, potatoes and sugar beets, which experienced production increases o f 200 percent or more over this period (Annex 4.1).* In contrast, livestock production increased by only 20 percent during this period, with the largest increases in meat (which almost doubled), followed by eggs and milk. Only wool production declined and by 2001 was less than 80 percent of its 1995 level (Annex 4.1). Under the Soviet system, the Republic had unsustainably high wool production, as the country's ample pasture was supplementedwith fodder production on irrigated land and imported grain. Figure 2.1: Evolutionof Crop and Livestock Production (1995-2001) 250.0 I I 200.0 ---- __- 150.0 2.3 The increase in supply reflected rising yields, a large inflow of labor into agriculture (that was not associated with a decline in labor productivity), and relatively favorable climatic conditions. In general, yields were 60 to 80 percent higher in 2000 than in 1995, with somewhat smaller yield increases for cotton and tobacco (Figure 2.2). Another factor driving the increase in supply was the rise in the share of the workforce in agriculture in the Kyrgyz Republic, which went from 33 percent in 1990 to 52 percent in 1999. Moreover, despite this increase, which normally would lead to a decline in output The analysis of the evolution of poverty between 1998 and 2001 i s based on the panel dataset of 1108 households using constant December 2001 prices, while the analysis of rural poverty in 2001 draws on the full sample of about *3000households. The datafor both are from the Kyrgyz HouseholdBudget Survey. The real value of agricultural and agro-industrial exports dropped by more than half, with their share dipping to 39 percentof total exportsin 2001, comparedto 48 percent in 1995. This decline reflectedadropinwool exports from US$24million in 1995 to US$4-2million in 1998-2001. 28 per worker, value added per worker actually rose between 1995 and 1997 and then remainedconstant between 1998 and 1999 (Figure 2.3). Figure 2.2: Evolution of Crop Yields (1995 = 100) I Yield Dynamicsof Main Crops 170.0 160 0 8 1500 ;; - +Cotton 1400 1300 Sugarbeet 1200 m-%--Tobacco 1100 +Potatoes 100 0 -o- Vegetables 1996 1997 1998 1999 2000 Figure 2.3: Evolution of Agricultural Value Added and Labor Productivity 1 4 0 - V a - 0 1 2 0 a 0 6 0 0 p 0 r 1 0 0 a g w o r k e r 8 0 on d e x 1990.1 0 0 6 0 - - G A P i n a e x 4 0 1 9 9 0 . 1 0 0 2 0 0 I 1 9 9 0 1 9 9 2 1 9 9 4 1 9 9 6 1 9 9 8 2 0 0 0 2.4 The growth in agriculture was also helped by the increase in overall demand, which was an important factor in creating an attractive incentive framework for producers (in theface of the rising domestic supply of food crops). Two factors explain the increase in demand. 2.5 First, the national demand for food products rose steadily in the late 1990s, even during the Russia cri~is.~ Compared to other CIS7 countries, the Kyrgyz Republic was less affectedby the Russia crisis as it was less dependent on agricultural exports than other CIS countries. For example, the demand for Moldovan agriculture was terribly hurt by the Russia crisis, as the value added of agricultural and agro-processed exports in that country i s about equivalent to agricultural GDP; and by 2000 exports were less than half of their 1997 level. Per capita consumption of most food products had recovered, or nearly recovered in the Kyrgyz Republic by 1999, reaching 75-100 percent of 1990 levels. By 2001, the food CPI was 340 percent above its 1995 level, compared to an increaseof only 208 percent for the non-foodCPI (Figure 1.7). 29 2.6 Second, in the Kyrgyz Republic, certain areas o f food-processing started to rebound during the second half of the 1990s, fueled by growing domestic demand, although more export oriented light industry agro-processing activity such as wool and cotton textiles, continued to decline throughout the 1990s. 2.7 Another factor behind the growth of value added in Kyrgyz agriculture was the shift away from fodder crops to support the traditional livestock sector, towardsfood crops, which led to an increase in the value of production. While total land area cultivated remained relatively constant, there was a shift away from fodder towards all other crops reflecting the reduced importance of livestock, with the largest gains experienced by sugar beets, tobacco, potatoes and vegetables (Table 2.1 and Annex 4.1). Table 2.1: Crop Acreage 2000(%) % Change 1996l2000 Grains 56 10 Cotton 3 7 Sugarbeet 3 135 Tobacco 1 67 Potatoes 6 41 Vegetables 4 44 Fodder crop and others 28 -24 2.8 Largely reflecting the net increase in demand, the incentive framework for agriculture improved in the latter half of the 1990s (Figure 2.4), which helped to buoy supply and raise valued added. Real food prices rose steadily throughout the late 1990s andby 2001, the food CPIwas 340 percent above its 1995level, compared to an increase of only 208 percent for the non-food CPI (Figure 1.7).4 Nevertheless, the profitability of agriculture was undermined, by the decliningratio of agricultural producer prices relative to industrial producer prices between 1996 and 2000 (suggesting that input prices and transport costs were increasing). Figure 2.4: RelativePricesof Foodand Agricultural Goods ~ ~ 180 160 4 140 E.. 120 100 -g 80 60 Kyrgyz food 0 +CP C 40 inon- 20 food CPI 0 m m z 5 5 v) IC m z Foodpricesare usedas aproxy for producer prices. Infact, the latter may not haveincreasedas muchas food prices, if marketingcosts are high. 30 Sectoral Policy 2.9 By the late 1990s, the Kyrgyz Republic had establishedrelative macroeconomic stability, which was an important conditionfor the growth of the agriculture sector. Structural reforms in the sector, some of which were put in place before 1995, also began to take hold at this time and played a critical role in supporting the recovery o f the agriculture sector and in transforming it from a command economy structured around 500 state and collective farms to a dynamic private sector with more than 70,000 private farmers. The Government has implemented major reforms inthe area of land policy, the incentive framework, financial services, irrigation and agricultural services. The key reforms and some of the remaining challenges are briefly discussed below. 2.10 The most importantfactor behind the transformation of the rural sector was the land reform program, which began in the early 1990s, the majority of land distributed between 1995 and 2000. By early 2002, the transfer of land previously held in state farms and agricultural cooperatives to peasant farmers was largely complete, with the exception of about 25 percent of arable landheld by the Land Redistribution Fund(LW) and the continued existence of less than 20 state farms (mostly seed and livestock breedingfarms). As of January 2001, the Government haddistributedabout 1.06million ha in the form of permanent land shares to 2.67 million people, and 510,551 families had received landcertificates or titles (Annex 4.2).5 As a result, by this time about 50 percent of total arable land was cultivated by individual farms (762,488 ha), another 20 percent was held individually but managed under a reformed collective or privatized structure,6 and approximately 30 percent was either under a state farm or managed by local governments in the LandRedistribution Fund(Box 2.1). Box 2.1: The KyrgyzLandRedistributionFund An additional feature of the Kyrgyz reform program was to place a large reserve of mainly irrigated land that previously had belonged to state and collective farms, under the control o f local governments (along with pasture and forest lands). A land fund was created to: (1) support the development o f rural settlements; (2) support the establishment of seed farms, livestock breeding stations and experimental peasant farms; (3) support the creation o f special agricultural enterprises, (4) provide land rights to be sold through auction; and (5) allocate land plots to citizens who worked and lived on the farms or came to the farms for permanent residence before July 1, 1996. Initially, about 50 percent of total irrigated arable land, inmost cases the best land, was transferred to a National Land Fund. The land reserve has since been reduced to 25 percent and the National Land Fund was replaced by the Agricultural Land Redistribution Fund (LRF),which as o f 2001 is managed by the local governments. In 2002, the total area of the LRF was 324,048 ha (22 percent o f total arable land). Most of the land i s arable (84 percent, the majority o f which is irrigated) and the remainder is fodder! perennials and fallow. Approximately one third of the land i s in Chui (Annex 4.2). Currently, the LRF land can only be leased. Guidelines allow individuals to lease up to 50 hectares of arable land, but average leases have been much smaller averaging less than 3 ha in 1999.7 All leases of pastureland and LRF land have to be investigated and leases must be drawn by the land management units of GosRegister (which takes a 10percent commission for this). The total amount of land shares distributedto each householdi s done on aper capita basis, and then each household receives one landcertificate for all of its shares 'TheseWorldBank couldeither be ajoint stock company, agro-company, agricultural cooperative or peasant collective. The (2002),Agriculture and Agribusiness: Growth, Opportunities and Obstacles (RevisedDraft) 31 2.11 By the end of the 1990s' the incentiveframework for agriculture was largely liberalized, with the exception of limited interventions in the wheat market, and the continued existence of a small number of government parastatals (e.g., for agricultural machinery and agro-processing). With respect to seed production, the government has been slowly liberalizing the regulatory framework and i s gradually allowing the private sector to produce and market seeds. In some cases, previously state owned enterprises that have been privatized continue to receive preferential treatment. For instance, the government has allowed privatized tractor and chemical input companies to sell subsidized tractors and fertilizer at below market prices. Nonetheless, a recent analysis of agricultural marketing chains suggests that markets are operating reasonably efficiently in the Kyrgyz Republic (Box 2.2). Box 2.2: EmergingAgriculturalMarkets A recent analysis of marketing chains for wheat, fruit and vegetables inthe Kyrgyz Republic suggests that markets operate reasonably efficiently, especially considering the relatively brief period during which they have been functioning. Price differences in these markets largely reflect differences in time, space and quality. Marketing margins are generally low inrelations to cost, and where they are not, there is usually a specific explanation. Risk i s an important factor, primarily because o f the danger of spoilage o f fresh products and unpredicted variation in processing. Unnecessary transaction costs in the form o f bribery are widespread, but are not very high inrelation to total costs (generally amounting to no more than 5 percent of total costs) (Stryker and Livinets, 2002). 2.12 Thegovernment liquidated its state run credit agency and in 1997, introduced a new non-banking institution to provide the rural sector with financial services. The Kyrgyz Agricultural Finance Corporation, a commercially-oriented non-banking agency, was set up with IDA support to temporarily fill in the rural credit gap. It has been very successful in targeting the intended beneficiaries-Le., first-time rural borrowers seeking small loans. KAFC has about 30,000 total beneficiaries, which represent about 4 percent of the rural population.8 2.13 The government has approved a new regulatory and institutionalframework to improve the management and efficiency of both on-farm and off-farm irrigation networks and has invested in rehabilitation of the irrigation network. With respect to on-farm irrigation, the Government has decentralized the management to Water Users Associations (WUAs). Piloted since 1997, the Parliament recently approved (2002) the legal framework for the WUAs and together with donors, including the World Bank, i s providing technical assistance to the emerging associations as well as increasing investment inrehabilitation and maintenance of the sector. 2.14 To support newly privatized farms, the government and several donors including IDA are supporting the development of a farmers' advisory service that In addition to the KAFC, there are two other significant providers of rural financial services, FINCA and the Financial Company on Support and Development of Credit unions, as well as several other smaller providers. 32 would be quasi-independentfrom the Ministry of Agriculture, Water Resources and Food Industry as well as local government officials. It is a flexible service able to provide a variety of assistance to farmers. B. RURAL POVERTY 2.15 This section provides a profile of rural poverty (i.e., sources of income and assets of rural households) and discusses the determinants o f rural poverty based on the results o f a multivariate analysis. RuralPoverty Profile 2.16 Rural poverty fell steadily between 1998and 2001, declining at an averagerate of 8percent a year (Table l.12J9 The decline in rural poverty reflected mainly growth, as the Gini index of expenditure inequality declined only slightly during this period, hovering around 0.3. Poverty remains significantly higher in rural areas, with 62.4 percent o f the rural population beneath the poverty line in 2001, compared to only 45.4 percent o f the urban population (Table 1.7). While the numbers of rural poor are evenly spread between the North and South, the rural poverty rate i s higher in the South, at 68 percent compared to 53 percent in the North (Table 2.2). In both regions, slightly less than one third of the population i s below the extreme poverty line for consumption. - Table 2.2: Rural P o v e 9 R a t e s (2001) 4 Rural Of which: Share of poverty Extreme rural poor rate (%) poor (%) (%I North 53 27 51 South 68 29 49 Total 62 28 100 Source: HBS 2001, weighted sample of about 3000 households. Sources of RuralIncome On average, rural householdsderived 52percent of their incomefrom the sale of home produced goods, 24percent from labor income and the remainder from public and private transfers, including 10percentfrom property sales (Figure 4.2). 2.17 I n 2001, 43percent of all households received credits, the North having somewhat higher access to credit markets (54percent) than the South (36 percent). (Figure 2.5). In general, the average amount of credit provided from 26 and 100 percent ~~ All analysis of trends over time (between1998and 2001) i s done usingthe panel dataset for both years. 33 of total annual consumption and input purchases for households in the first consumption quintile, and 16 to 73 percent of consumption and input purchases for households in the fifth quintile Households in the higher quintiles received greater amounts than households in the lower quintiles (Figure 2.6 and Box 2.3). About 32 percent of households were repaying debts in2001. This i s slightlyless than the share who received credit and the net inflow of funds was positive overall. Figure2.5: ParticipationinCreditMarketsby Quintile 60 50 L5 40 30 .-c 20 l o 0 1 2 3 4 5 consumptionquintile Box2.3: PerceivedObstaclesto CreditExpansion While post 1998 an increasing number o f rural residents have obtained credits, some residents from Uchkun, Naryn oblast spoke o f the bureaucratic and financial barriers to processing the paperwork to obtain credits: "Besides, to obtain a loan, one needs to provide a whole lot o f documents, and to get all these documents one has to go to the town or to the neighboring village where the village council is. Sometimes the right official isn't there, and people go back without the documents they need, and have to pay 15 som for transportation. For many of us, it's not affordable."" Another issue regarding access to credits i s the perception of high risk by bankers, causing them to only offer short-term lending to borrowers. Furthermore there i s the fear of rural residents that they will be unable to abide by the terms o f the loan (e.g., if there i s a crop failure), which deters them from borrowing. One Uchkun resident notes "Many people want to obtain loans, but they're afraid that their crops will be too poor and that they won't be able to repay the loans on time."" lo Narayan,D.andP.Petesch (eds) (2002), Voices ofthe Poor: From Many Lands,New York andWashington, DC, Oxford UniversityPress andThe World Bank Narayan, D. andP. Petesch (eds) (2002), ibid 34 I Figure2.6: CreditReceivedby Quintile,2001 I 3000 2500 2000 1500 North 1000 South 500 0 1 2 3 4 5 consumption quintile 2.18 With the exception of livestock sales, participation in agricultural markets remains relatively low. In 2001, across all quintiles, 60 percent of the households engaged in livestock transactions, while only 25 percent were involved in fertilizer markets, 47 percent in the labor market and 9 percent in land rental markets (renting-in). Ingeneral, participation rates increased duringthe study period, particularly participation inlivestock output markets, fertilizer andlandmarkets (Table 2.3).'* Table 2.3: RuralMarketParticipationRates(inpercent)(HBS) Assets 2.19 An increasingshare of households ownproductive agricultural resources (land, and livestock). According to the panel data, the share of rural households with access to land has increased since 1998 (Figure 2.7). In 1998, more than half of the rural households did not have access to land compared to only 25 percent in 2001. Most land used was privately owned (Figure 2.8). According to the panel data, the share o f land that i s rented has decreased by about 80 percent in both the North and South between 1998 and 2001, while the share of land owned has increased. In2001, households in the fifth quintile rented larger amounts of land (0.1 ha) than households in the bottom l2It appears that while the amount of landrented has decreased, the share of households engaged in renting landhas increased, suggesting that the amount of landrented per transaction has decreased. 35 quintile (0.05 ha), while in 1998 the amount o f landrented was relatively equal across the expenditure quintiles. Figure 2.7: Evolutionof LandArea per Household(panel data) Percent of households 40 20 0 0 ha Oehac2 2<=ha<5 5<=ha<12 12<=ha Area Figure 2.8: Land Tenure by ConsumptionQuintile (2001) I 120% 100% 80% 60% 40% 20% 0% 1 2 3 4 5 (3 life W rent 0 other 2.20 Livestock herd sizes are small, but growing. Table 2.4 shows the household distribution of cattle by herd size for 2001, while Table 2.5 shows the same for small ruminants. The majority of households have one head o f cattle or less and no small ruminants. However, based on panel data, the mean herd size for cattle has increased from 2 in 1998 to 6 in 2001, while the mean small ruminant herd size has increased from 6 to 20. Table 2.4: Distribution of Cattle by Household,2001 FullRural Sample (%) Herd Size Region Ohead 1head 2-6 head 7-11 head >11head Total Total 25 53 17 4 0.9 100 North 33 36 23 6 1 100 South 19 58 18 3 7 100 36 Table2.5: Distributionof SmallRuminantsby Household,2001FullRuralSample (%) HerdSize Region 0 head 1-9head 10-24head 25-59 head >59 head Total Total 65 5 7 13 10 100 North 67 8 6 10 9 100 South 66 6 7 12 9 100 2.21 Despite increases in land and livestock assets, farm size remains quite small and production technologies are fairly rudimentary. The farm sector i s dominated by small farms with less than 2 ha. In 2001, of the 75 percent of the rural households with access to land, the majority had 0.1 to 2 ha (Table 2.6). Since 1998, the share of farms under 2 ha increased from just over 30 percent to more than half of all households (including the landless), while the share of farms with more than 5 ha has grown only marginally. Table 2.6: DistributionofAgriculturalLandArea by Household,2001FullRuralSample(%) LandArea Region 0 ha O