Report No. 39737-XK Kosovo Poverty Assessment (In Two Volumes) Volume I: Accelerating Inclusive Growth to Reduce Widespread Poverty October 3, 2007 Poverty Reduction and Economic Management Unit Europe and Central Asia Region Document of the World Bank ACKNOWLEDGEMENTS This report is a joint production of Statistical Office of Kosovo staff in the Household Budget Survey unit comprising Bashkim Bellaqa, Bekim Canoli, and Emina Deliu and World Bank technical team comprising Andrew Dabalen and Anna Gueorguieva, supported by Sasun Tsirunyan and Shpend Ahmeti. The report has benefited from the support of UK's Department for International Development which has generously funded the Trust Fund to support the capacity building and analytic activities of the Western Balkan Programmatic Poverty work. The report would not have been possible without the very close involvement and support of the Social Statistics Department of the Statistical Office of Kosovo. The team graciously acknowledges the analytic work of the IMF (Macro statistics), EAR and Ministry of Agriculture of the PISG, SOK and Vllaznim Bytyqi (Migration). The team has benefited from the comments and guidance of Peter Lanjouw (Peer Reviewer), Pierella Paci (Peer Reviewer), Asad Alam, Ardo Hansson, Elisabeth Huybens, Felix Martin, and Kanthan Shankar. The team is grateful to Julian Lampietti, Ruslan Yemtsov, Kinnon Scott, Gero Carletto, Marcus Goldstein, Gabriel Demombynes, Juan Munoz, and Johan Mistiaen for excellent comments and suggestions. The production of this report benefited enormously from the excellent editing skills of Susana Padilla. VOLUME I TABLE OF CONTENTS EXECUTIVE SUMMARY.......................................................................................................... i CHAPTER 1. MACROECONOMIC AND SOCIAL CONTEXT..................................................... 1 A. GDP Growth has been Poor.................................................................................... 1 B. Agricultural Growth has been Sluggish.................................................................. 3 C. Disappointing Labor Market Performance ............................................................. 4 D. Low Prospects for Poverty Reduction .................................................................... 5 CHAPTER 2. POVERTY TRENDS AND DECOMPOSITION....................................................... 7 A. A Note on Data Quality and Approach................................................................... 7 B. There Was No Change in Poverty .......................................................................... 9 C. Non-Income Dimensions of Poverty .................................................................... 13 CHAPTER 3. POVERTY PROFILE AND POVERTY RISK........................................................ 17 A. Demographic Characteristics of Poor Households ............................................... 17 B. Labor Market Characteristics of the Poor............................................................. 18 C. Educational Attainment and Poverty Incidence.................................................... 19 D. Incidence of Poverty Across Space....................................................................... 20 E. Why Are People Poor and Who is at High Risk of Poverty? ............................... 22 CHAPTER 4. PUBLIC TRANSFERS, REMITTANCES AND POVERTY..................................... 27 A. Composition and Trends of Social Protection Transfers...................................... 27 B. Impact of Social Protection Programs on Poverty................................................ 29 C. Size and Distribution of Remittances ................................................................... 31 D. Impact of Migration on Poverty............................................................................ 33 CHAPTER 5. STRENGTHENING THE FOUNDATION OF POVERTY MONITORING AND EVIDENCE-BASED POLICY MAKING ................................................................................... 37 Annex A. Tables and Figures ...................................................................................... 41 Annex B. Poverty Profile............................................................................................. 45 Annex C. Poverty Decomposition............................................................................... 65 Annex D. Social transfers and Remittances................................................................. 69 References......................................................................................................................... 77 List of Tables Table i: Proposed Policies and Their Cost and Poverty Impact......................................... x Table 1.1: Macroeconomic Trends .................................................................................... 2 Table 1.2 : Alternative Estimates of GDP for Kosovo ...................................................... 3 Table 1.3: Macroeconomic and Agricultural Sector Indicators......................................... 4 Table 1.4: Real Mean Monthly Wages and Number of Observations............................... 5 Table 2.1: Decomposition of Poverty: 2003/04 Compared to 2005/06........................... 11 Table 2.2: Growth Rates in Per Capita Consumption...................................................... 12 Table 2.3: Inequality Indices for 2003/04 and 2005/06................................................... 12 Table 3.1: Absolute and Extreme Poverty Rates Using PA05 Methodology.................. 18 Table 3.2: Poverty and Unemployment ........................................................................... 18 Table 3.3: Employment and Education, 2005/06............................................................. 20 Table 3.4: Poverty Rates and Contribution by Region in Kosovo, 2003-2006 ............... 20 Table 3.5: Impact of Changes in Household Characteristics on Poverty ........................ 24 Table 4.1: Pension and Social Assistance Programs ....................................................... 28 Table 4.2: Social Assistance and Pensions ...................................................................... 30 Table 4.3: Adequacy and Simulation of Social Assistance and Pension Transfer .......... 30 Table 4.4: Migration and Remittances: Summary Statistics for 2005............................. 32 Table 4.5: Propensity Score Matching Results for Secondary Enrollment Rates............ 33 Table 4.6: Remittances from Abroad in 2005 by Recipient Households ........................ 34 Table A.1: Difference in World Bank and IMF Assumptions......................................... 41 Table A.2: Summary Statistics of Main Aggregates by Survey Wave............................ 41 Table A.3: Average Monthly Food Consumption ........................................................... 41 Table A.4: Household Average Monthly Non-food Expenditures.................................. 42 Table A.5: Shares of the Food and Non-food Expenditures over Total Expenditure...... 42 Table A.6: Shares of the Expenditures on Food Categories............................................ 42 Table A.7: Poverty Headcount by Location .................................................................... 43 Table A.8: Kosovo: IMF GDP Estimates at Current Prices, 2004­10. ........................... 43 Table A.9: Poverty Incidence, Gap and Severity............................................................. 44 Table B.1: Poverty Headcount by Location, Region and Ethnic Areas........................... 45 Table B.2: Poverty Contribution by Location.................................................................. 45 Table B.3: Poverty Contribution by Region .................................................................... 45 Table B.4: Poverty Contribution of Ethnic Areas............................................................ 45 Table B.5: Poverty Headcount by Household Size Category.......................................... 46 Table B.6: Poverty Headcount by Household Head Ethnicity ........................................ 46 Table B.7: Poverty Headcount by Household Head Gender ........................................... 46 Table B.8: Fraction of Elderly in the Household and Poverty......................................... 46 Table B.9: Dependency Ratio and Poverty Headcount ................................................... 46 Table B.10: Education of the Household Head and Poverty Headcount......................... 46 Table B.11: Main Activity of the Household Head and Poverty Headcount................... 47 Table B.12: Employment Sector of the Household Head and Poverty Headcount ......... 47 Table B.13: Land Tenure and Poverty Rural areas.......................................................... 47 Table B.14: Ownership of Livestock............................................................................... 47 Table B.15: Ownership of Major Equipment and Poverty in Rural areas....................... 47 Table B.16: Employment and Education, 2002/03.......................................................... 48 Table B.17: Unemployment and Education, 2002/03...................................................... 48 Table B.18: Employment and Education, 2003/04.......................................................... 48 Table B.19: Unemployment and Education, 2003/04...................................................... 49 Table B.20: Employment and Education, 2004/05.......................................................... 49 Table B.21: Unemployment and Education, 2004/05...................................................... 49 Table B.22: Employment and Education, 2005/06.......................................................... 50 Table B.23: Unemployment and Education, 2005/06...................................................... 50 Table B.24: Poverty and Unemployment......................................................................... 50 Table B.25: Gross Enrollment Rates for Primary Schools .............................................. 50 Table B.26: Net Enrollment for Primary Schools............................................................ 51 Table B.27: Net Enrollment Rates for Secondary Schools.............................................. 51 Table B.28: Net Enrollment Rates for Tertiary Education .............................................. 51 Table B.29: Access to Electricity..................................................................................... 52 Table B.30: District Central Heating ............................................................................... 52 Table B.31: Access to Safe Dwelling .............................................................................. 53 Table B.32: Access to Water ........................................................................................... 53 Table B.33: Rural Poverty Headcount Rate and Poverty Contribution........................... 53 Table B.34: Rural Ethnic Divide: Poverty Headcount and Contribution ....................... 53 Table B.35: Rural Area, Educational Attainment............................................................ 54 Table B.36: Rural Access to Electricity........................................................................... 54 Table B.37: Rural Access to Safe Dwelling .................................................................... 54 Table B.38: Rural District Central Heating ..................................................................... 54 Table B.39: Rural Access to Water.................................................................................. 55 Table B.40: Rural Unemployment................................................................................... 55 Table B.41: Rural Gross Enrollment Rates for Primary Schools .................................... 55 Table B.42: Rural Net Enrollment Rates for Primary Schools........................................ 56 Table B.43: Rural Net Enrollment Rates for Secondary Schools.................................... 56 Table B.44: Rural Net Enrollment Rates for Tertiary Education .................................... 56 Table B.45: Vulnerable Group above the Poverty Line .................................................. 57 Table B.46: Vulnerable Group listed below the Poverty Line......................................... 58 Table B.47: Internal Migrants and Their Recipient Location.......................................... 59 Table B.48: Inequality Indices for 2002-2006................................................................. 60 Table B.49: Correlates of Consumption by Year............................................................. 60 Table B.50: Comparison of Enrollment Rates in the Region, 2005 ................................ 62 Table B.51: Contribution to Poverty by Activity of the Household Head....................... 62 Table B.52: Contribution to Poverty by Educational Attainment, 2005/06 .................... 62 Table B.53: Demographic Distribution............................................................................. 63 Table B.54: Demographic Distribution by Type of Settlement and Region, 2005/06. .... 63 Table C.1: Decomposition of Poverty: 2002/03 Compared to 2003/04 ......................... 65 Table C.2: Decomposition of Poverty: 2002/03 Compared to 2004/05 ......................... 65 Table C.3: Decomposition of Poverty: 2002/03 Compared to 2005/06 ......................... 65 Table C.4: Decomposition of Poverty: 2003/04 Compared to 2004/05 ......................... 65 Table C.5: Decomposition of Poverty: 2003/04 Compared to 2005/06 ......................... 65 Table C.6: Urban Poverty Decomposition: 2002/03 Compared to 2003/04................... 65 Table C.7: Urban Poverty Decomposition: 2002/03 Compared to 2003/04................... 66 Table C.8: Urban Decomposition: 2002/03 Compared to 2004/05................................ 66 Table C.9: Urban Poverty Decomposition: 2002/03 Compared to 2005/06................... 66 Table C.10: Urban Poverty Decomposition: 2003/04 Compared to 2005/06................. 66 Table C.11: Urban poverty decomposition: 2003/04 Compared to 2005/06.................. 66 Table C.12: Urban Poverty Decomposition: 2004/05 Compared to 2005/06................. 66 Table C.13: Rural Poverty Decomposition: 2002/03 Compared to 2003/04.................. 67 Table C.14: Rural Poverty Decomposition: 2002/03 Compared to 2004/05.................. 67 Table C.15: Rural Poverty Decomposition: 2002/03 Compared to 2005/06.................. 67 Table C.16: Rural Poverty Decomposition: 2003/04 Compared to 2004/05.................. 67 Table C.17: Rural Poverty Decomposition: 2003/04 Compared to 2005/06.................. 67 Table C.18: Rural Poverty Decomposition: 2204/05 Compared to 2005/06.................. 67 Table D.1: Urban and Rural Households with Remittances and Migrants...................... 70 Table D.2: Regression Results for the 2-stage Estimation of the Effect of Having a Migrant on the Welfare of the Household ........................................................................ 72 Table D.3: Propensity Score Matching Results for Consumption................................... 73 Table D.4: Poverty Rates among Migrant and Non-migrant Households, Propensity Score Matching................................................................................................................. 74 Table D.5: 2-stage IV Regression for the Effect of Having a Migrant in the HH on Consumption, by Urban and Rural ................................................................................... 75 List of Boxes Box 1.1. Has Poverty Increased in Kosovo? NO...............................................................ii Box 2.1: Estimating the Population of Kosovo ................................................................. 7 Box 2.2: Challenges of Using HBS Data for Poverty Analysis......................................... 8 Box 3.1: Three Hypotheses for Deteriorating Conditions among Serb Household......... 19 List of Figures Figure i: Absolute and Extreme Poverty Rates··································································· i Figure ii: Headcount Poverty in the Western Balkans·······················································ii Figure iii: Poverty Rates by Household Size····································································iii Figure iv: Poverty Rate by Labor Force Status·································································iii Figure v: Rural and Urban Poverty Trends······································································· iv Figure vi: Venn Diagram of Non-income and Income Poverty········································· v Figure vii: Relationship between Growth and Poverty Reduction ··································vii Figure 1.1: Employment and Unemployment Rates·························································· 5 Figure 1.2: Registered Unemployment and GDP Growth Rate········································· 5 Figure 1.3: Asset Ownership and Income Levels using HBS Data··································· 6 Figure 2.1: Absolute and Extreme Poverty for 2003/04 and 2005/06····························· 10 Figure 2.2: Distribution of Consumption, 2003/04 and 2005/06····································· 10 Figure 2.3: Growth Incidence Curves for Growth from 2003/4 to 2005/06···················· 11 Figure 2.4: Lorenz Curves for 2003/04 and 2005/06······················································· 13 Figure 2.5: Net Enrollment Rates ···················································································· 14 Figure 2.6: Access to Water····························································································· 14 Figure 2.7: Venn Diagram of Non-income and Income Poverty····································· 15 Figure 3.1: Poverty Rates by Household Size ································································· 17 Figure 3.2: Rural and Urban Poverty Trends··································································· 21 Figure 3.3: Rural Access to Services··············································································· 21 Figure 3.4: Rural Net Enrollment Rates ·········································································· 22 Figure 4.1: Estimated Annual Benefits per Recipient ····················································· 28 Figure 4.2: Social Protection Expenditures, Selected Countries: 2004 ··························· 29 Figure 4.3: Targeting Performance of the Kosovo Social Assistance Program ·············· 31 Figure 4.4: Remittances as a share of GDP in the Western Balkans······························· 32 Figure 4.5: Enrollment Rates for Households with a Migrant and without a Migrant···· 33 Figure C.1: Growth Incidence Curves ············································································· 68 Figure D.1: Undercoverage and Leakage of Social Assistance by Urban and Rural ······ 69 Figure D.2: Top 20 Remittance-receiving Countries as a share of GDP, 2004··············· 69 Figure D.3: Propensity Score Matching for Migration···················································· 74 Figure D.4: Propensity Score and its Frequencies for Treated (households with migrants) and Untreated Groups ······································································································· 75 VOLUME II TABLE OF CONTENTS CHAPTER 1: HOUSEHOLD BUDGET SURVEY (HBS) AND POVERTY MONITORING IN KOSOVO 6 A. There are Problems of Data Comparability 7 (a) Problem # 1: Diary versus Recall 7 (b) Problem #2: Survey Design ­ Redefinition of Consumption Items 7 (c) Problem #3: Survey Design - LSMS versus HBS 8 B. Sample Weights Introduce Additional Uncertainty 8 C. Likely Consequences: Consumption 9 D. Likely Consequences: Poverty Estimates 11 CHAPTER 2: POVERTY ­ ALTERNATIVE ESTIMATES 13 A. Post-Stratification 15 B. Compare Only 2003 and 2005 17 C. Comparable Consumption Aggregate Methodology 18 D. Compare all the Years 20 E. Comparison of Poverty Figures from the LSMS and HBS 21 CHAPTER 3: CONCLUSIONS AND RECOMMENDATIONS 23 A. Recommendations 23 ANNEX A: TABLES AND FIGURES 25 ANNEX B: RESULTS USING DIFFERENT SURVEY YEAR DEFINITION 31 List of Tables Table 1.1: Population Size by Survey Wave and Year 9 Table 1.2: Summary of Survey Constraints and Their Effects on Poverty Estimates 10 Table 1.3: Poverty Headcount by Location and Ethnic areas, using PA05 methodology 12 Table 1.4: Poverty Headcount by Household Head Ethnicity 12 Table 2.1: Overview of the Results of Methodologies for Comparable Poverty Estimates 15 Table 2.2: Summary of Poverty Estimates from the Methodologies Used 15 Table 2.3: Poverty Rates with Current Weights and Reweighted 16 Table 2.4: Sampling procedure for the Bosnia and Herzegovina's Household Budget Survey 17 Table 2.5: Poverty Rates with the PA05 and Comparable CA methodologies 18 Table 2.6: Robust Poverty Lines Based on Consistent Food Items 19 Table 2.7: Poverty Rates using the Abbreviated Consumption Bundle Methodology 19 Table A.1: Comparison of Previous Methodologies 25 Table A.2: Survey Comparison 27 Table A.3: Percent Changes in Main Aggregates from Survey to Survey Comparison 28 Table A.4: Alternative Consumption Aggregate Definitions and Poverty Rates 28 Table A.5: Consistently Asked Questions over the Four Surveys 29 Table A.6: Definition of Consumption Aggregates for the Different Methodologies 30 Table A.7: Poverty Lines in Different Methodologies 30 Table B.1: Introduction of New Questionnaires 31 Table B.2: Poverty Statistics using PA05 Methodology 31 Table B.3: Poverty Rates Using PA05 Methodology 32 Table B.4: Detailed Poverty Diagnostics with Revised Consumption Aggregate 32 Table B.5: Poverty Rates Using Alternative Consumption and Poverty Line Methodologies 33 List of Boxes Box 2.1: Bosnia and Herzegovina HBS: Example of Sampling without a Census 17 Box 2.2: Analysis of Changes 20 List of Figures Figure 1.1: Total Population in Millions and Household Size 8 Figure 1.2: Average Monthly Household Consumption, in Nominal prices 9 Figure 1.3: Poverty Rate Estimates and the Effect of Changes in the Questionnaire 11 Figure 2.1: Cumulative and Density Distribution of Consumption for the Bottom 50 percentile of the Population 16 EXECUTIVE SUMMARY 1. About 45 percent of the population in Kosovo is poor, with another 18 percent vulnerable to poverty. The persistence of poverty levels in the first half of this decade is not surprising within the context of prevailing macro-economic conditions characterized by slow growth, low incomes and tight expenditure constraints. Without the safety net provided through migration and remittances, the welfare of a large fraction of the population would have been even worse. However, the good news is that poverty is shallow in the sense that many people are just above or just below the poverty line. The shallowness of poverty also implies that a small positive change in incomes, through employment generating growth, can pull many people out of poverty. Living standards have stagnated 2. Overall economic stagnation is reflected in the lack of progress in improving living standards. About 15 percent of the population is estimated to be extremely poor, defined as individuals who have difficulty meeting their basic nutritional needs (Figure i). Figure i: Absolute and Extreme Poverty Rates About 45 percent (that is, a little over 2 0. 50 in 5 Kosovars) report a consumption 50 level below the poverty line, which in teulos 45.1 43.5 2002 prices is set at 43 Euros per adult Ab 0. 40 40 equivalent per month. These poverty rates are very high compared to 0. 30 30 neighboring countries (Figure ii) and unlike many countries in the region, have 0. not changed over time. 20 20 e 16.7 metr 13.6 3. Only the top 20 percent of the Ex 0. 10 10 population had a small positive growth 2003/04 2005/06 in consumption, between 2003 and 2005, Confidence Interval while the rest had negative growth. Among the poorest groups, the losses Source: World Bank staff calculations from HBS data. were substantial. The poorest fifth of the population experienced consumption loss of around 10 percent. Examining the changes in consumption separately for urban and rural areas shows that consumption declined for nearly all rural populations, while in urban areas only the bottom fifth of the population reported decreased consumption. 4. In addition to stagnant poverty a large fraction of the population is vulnerable. A shock that reduces incomes by 25 percent could send an additional 18 percent of the population below the poverty line. A similarly positive increase in the incomes of the population can lift as many out of poverty. This reflects the phenomenon that while poverty is widespread, it is shallow in the sense that a large fraction of the population is just around the poverty line. About 40 percent of the vulnerable are estimated to live in Pristina and Prizren. Box 1.1: Has Poverty Increased in Kosovo? NO This report is preceded by two other poverty assessments on Kosovo (World Bank 2001, 2005a). The 2001 report used the Living Standard Measurement Survey (LSMS) of Kosovo and estimated the fraction of the population below the poverty line to be about 50 percent. The 2005 report used the Household Budget Survey conducted in 2002/03 and estimated a poverty rate of 37 percent. In this report, we estimate poverty rate to be about 45 percent. Does this mean that poverty increased? The answer is NO. As explained in volume II of this report, the three data sets are not comparable. In particular, even though the 2005 report and this one use the HBS surveys, changes in how (a) households were asked to remember the period they consumed a reported purchased item (daily, weekly, etc.), and (b) consumption of own-produced items, makes the 2002/03 and subsequent HBS surveys non-comparable. An additional point is that Kosovo appears to be an outlier in the region in terms of the fraction of the population below the poverty line. This should not be surprising given that the GDP per capita is almost half the regional average. Figure ii: Headcount Poverty in the Western Balkans 5 0 4 5 4 0 3 5 3 0 tear 2 5 adcount he 2 0 1 5 1 0 5 0 A lb a n ia 2 0 0 5 B iH 2 0 0 4 F Y R O M 2 0 0 3 K o s o v o 2 0 0 5 M o n te n e g ro 2 0 0 4 S e rb ia 2 0 0 5 C o u n try a n d y e a r o f d a ta H e a d c o u n t p o v e rty Source: Respective World Bank Poverty Assessments. 5. Finally, inequality ­ though low ­ shows signs of being on the increase, especially in rural areas. Overall, inequality in Kosovo is low. The most commonly reported measure of inequality (the Gini index) is about 30 percent in 2005. In 2003, the same index was measured at 27 percent. As expected, urban inequality is higher than that observed in rural areas. However, over time, urban inequality has remained unchanged at 31 percent, while rural inequality has increased from 25 percent to 28 percent. Other measures of inequality also confirm generally low but rising inequality. For instance, the gap between the richest and the poorest deciles widened during the period. The rising inequality in rural areas accounts for the observed increase in overall inequality. But in rural areas, remittances appear to be driving the increased inequality, since the better off households are observed to receive substantially more remittances. ii The poor are mainly concentrated in large families, among the unemployed and the low skilled 6. Larger households are on average poorer. The poverty incidence for households with more than 7 members is at least 7 percentage points higher than households with 1 to 3 members (Figure iii). In addition to household Figure iii: Poverty Rates by Household Size size, the composition of the household introduces additional burden on welfare 2003/04 2005/06 improvement. Households with more 50 51 49 dependents than working adults have 48 48 48 46 higher incidence of poverty compared to 40 40 41 households with more adults than 37 dependents. ecned 30 cini 31 ytr 7. Households with female heads veo 20 have higher poverty incidence. The P poverty incidence is estimated to be 10 higher by 4 percentage points for female headed households compared to male 0 1 to 3 4 to 6 7 to 9 10 to 12 13 + 1 to 3 4 to 6 7 to 9 10 to 12 13 + heads of households. However, over Household size Household size time, this gap has not widened, which Source: World Bank staff calculations from HBS data. suggests that despite the difficult macro- economic situation, female headed households have not lost ground. The estimated poverty incidence for Serb heads of households has increased over time, though data quality issues which appear to be more serious here than general, imply that the magnitude of the deterioration in these trends should be treated with caution. 8. The poverty risk is also higher for the unemployed. Not all the unemployed live in poor households, and not all the employed are free from poverty. The evidence from the surveys suggests that about 70 percent of all the poor are either employed or inactive. However, a comparison of the incidence of poverty between the employed and the Figure iv: Poverty Rates by Labor force status unemployed indicates that the latter have a 20 percentage point higher risk 2003/04 2005/06 50 than the former (Figure iv). In other 51 50 words, while the working poor constitute the largest group among all 40 poor, the likelihood (that is, incidence) 35 of being poor is higher if one is ecned 30 32 cini unemployed than if one has work. The ytr majority of the employed poor are "per veo 20 P diem" workers and employees in the mining sector. 10 0 9. As expected, poverty Employed Unemployed Employed Unemployed incidence declines with higher education of the household head. The Source: World Bank staff calculations from HBS data. poverty incidence for heads of households with tertiary education is 20 percent, but 1 in 2 heads of households with primary education are estimated to live in poverty. The more educated have lower incidence of poverty because they have better employment prospects and better pay. Over iii 70 percent of the people with vocational and tertiary education report being salaried employees, compared to 25 percent of secondary educated individuals. The poor are concentrated in rural areas and in Mitrovica and Ferizaji 10. The majority of the poor live in rural areas. Rural and urban residents faced the same likelihood of poverty, about 42 percent, in 2003. However, by 2005, urban poverty had declined by about 5 percentage points and rural poverty had increased by a similar Figure v: Rural and urban poverty trends magnitude (Figure v), so that more than 50 two-third of all the poor now lived in 49.2 rural areas. Poverty in rural areas is 44.2 40 42.1 highly correlated with lack of land, 37.4 livestock or agricultural equipment. ontialupop 30 11. Poverty incidence varies eht of 20 widely across regions. In 2003, % 18.1 15.6 Mitrovica, Ferizaji, Gjakove and Prizren 14.0 10 12.5 had higher incidence of poverty compared to the rate for all of Kosovo. 0 2003/04 2005/06 2003/04 2005/06 By 2005, only Mitrovica and Ferizaji Rural Urban maintained that distinction. Pristina, Absolute poverty Extreme Poverty together with Gjilani had one of the lowest poverty rates in 2003. However, Source: World Bank staff calculations from HBS data. while this has worsened by 2005 for Pristina, it was still lower than the average for all of Kosovo. Furthermore, while Pristina has lower incidence of poverty than the Kosovo rate, it ranks as the highest contributor to poverty. Indeed, 3 of every 5 poor people live in only three regions ­ Pristina, Prizren and Mitrovica. Non-income dimensions of welfare show better outcomes, but are beset by inequities in access to, and low quality, of key public services 12. Finally, while non-income dimensions show better outcomes, they suffer significant inequities and poor quality. More specifically, there are sizeable differences in access to secondary and tertiary education between the richest and the poorest households. Furthermore, rural families have substantially lower access to central heating and tap water compared to urban families. Also, both quality of water and quality of health services are known to be lower in rural areas. 13. Moreover, a large fraction of the population reports being deprived on multiple dimensions. For instance, about 8 percent are materially poor and have no access to indoor water tap and proper sanitation. Just as many, exactly 9 percent, are poor and have no access to telephone connection or bathroom in the dwelling (Figure vi). These rates are much higher than Romania (or Georgia and Russia) where only 1 percent of the population reported being deprived on multiple dimensions. 14. This brief profile of the poor suggests that a diverse group of households faced hardships during the period under review. Such widespread poverty is not surprising in the context of prevailing macroeconomic conditions. iv Figure vi: Venn Diagram of Non-income and Income Poverty A. Water, sanitation and income B. Telephone, housing and income poverty poverty A: Access to inside water tap A Telephone connection B: Sanitation: flush toilet B Bathroom available C: Poor households C Poor household Source: World Bank staff calculations from HBS data Macro-economic conditions have not enabled massive poverty reduction 15. The current macro-economic conditions provide no prospect for improving living standards. Economic growth surged in the immediate post-conflict period, buoyed by a large inflow of resources for reconstruction. Since then, and especially in the last 4 years, growth has been slow. This is mainly because industrial output has not yet recovered, and agriculture, which contributes a large share of the GDP and where the majority of the population earns their livelihood, remains a low productivity activity. Agriculture is subsistence-based, faces high input costs and poor infrastructure, and operates under poorly defined property rights. Consequently, yields and acreage have not improved, and neither has output. 16. This environment has resulted in poor labor market conditions. About 30 percent of the labor force is estimated to be unemployed, and the conditions are worse for young people. The prevailing poor labor market conditions, no doubt, also partially explain what is suspected to be an increase in the share of informal activity, and possibly the level of under-employment. In addition, unemployment durations are long, and real wages have remained unchanged in the last 4 years (Table 1.4). Social assistance programs are inadequate 17. One of the consequences of the difficult macro-economic conditions is the challenge of balancing huge investment needs and social priorities, especially given a restrictive fiscal rule and conservatism. This has made it difficult to protect many poor people through public support. In their current form, the social assistance programs of Kosovo are characterized by one desirable v feature, which is that the benefits of the program, which is targeted at the poor, reach mostly the poor. Balanced against this good feature are three weaknesses. 18. First, the programs have low coverage, in part because of the tight fiscal space. Specifically, over 75 percent of the poor are not reached by the social assistance program. Second, the value of benefits per recipient household has remained flat between 2002 and 2005. This is understandable given the difficult fiscal conditions. Moreover, given that the overall inflation has also been low or negative, there was probably no erosion in the real value of the benefits, so that this weakness becomes less pressing. Third, also the final weakness, low coverage and low benefits level together have meant that the programs have had little impact on improving the welfare of the population. ...but migration and remittances have been effective mechanisms for reducing poverty 19. At present, about 1 in 5 Kosovars report having at least one household member who is a migrant abroad. Just as many reported having received remittances from abroad. By comparison only 13 percent of the population receives social assistance benefits, which is targeted at the poorest groups. Those with migrants abroad also report higher levels of consumption and are estimated to be less poor. The evidence from the household surveys shows those with a 10 percent higher propensity to migrate abroad report 2 percent more consumption. A comparison of households with migrants abroad and those with similar characteristics but without migrants show that the former have a consumption gain equivalent to 25 percent of the extreme poverty line. The incidence of poverty for the sub-population with migrants is also lower, by 7 percentage points compared to the general population. The higher level of consumption and the lower incidences of poverty for households with migrants are even larger for rural areas. For instance, the incidence of poverty for households with migrants abroad is 20 percentage points lower than similar households in rural areas without migrants. 20. Taken together these findings suggest that without migration poverty incidence would be higher and more concentrated in rural areas. So any efforts that lead to drastic reductions in the current migration patterns has the potential to worsen the well-being of the Kosovo population, to widen the already emerging rural-urban disparities in well-being, and possibly to lead to instability, especially in the rural areas. Looking to the Future 21. Consistent with its European aspirations, Kosovo's development goals aim to create an inclusive society through better living standards. This means achieving progress in a number of areas, such as democratic governance, effective decentralization, and encompassing sector strategies. However, this is not going to be easy because Kosovo continues to face challenges common to all fragile states: huge backlog of investment needs and limited government resources to meet them, fractured societal relations, and weak security. It will be necessary to prioritize actions that will lead to rapid improvements in the welfare of population in order to strengthen social cohesion and ensure lasting peace. While all of this is of course contingent on the resolution of final status negotiations, the analysis in this report points to four areas through which welfare gains can be made. 22. First, generating high and sustained growth is crucial. As noted above, poverty is widespread and many more people are vulnerable to small shocks. When a large fraction of the population is just around the poverty line and inequality is low, we would expect that the sensitivity of poverty to growth (growth elasticity of poverty) will be very high. In the Western vi Balkans, Albania provides the clearest example of how high and sustained growth can reduce poverty on a large scale. In early 2000, about a quarter of Albania's population was poor, and inequality was low. However, with a 6 percent growth rate per year between 2002 and 2005, poverty declined from 25 percent in 2002 to about 18.5 percent in 2005, implying a growth- poverty elasticity of -1.5. In addition to Albania, many countries around the world with living conditions situations similar to Kosovo's ­ widespread but shallow poverty and low inequality ­ had been able to reduce poverty on a mass scale when they have been able to generate high and inclusive growth (Figure vii provides a sample of such countries). To see the promise of growth for Kosovo's poverty reduction prospects, note that a sustained 5 percent growth in GDP for the next 5 years, combined with growth-poverty elasticity like that of Albania, would imply 38 percent poverty reduction, or a decline of absolute poverty from 45 percent to 28 percent. With such a growth rate and growth-poverty elasticity like that estimated for middle-income Commonwealth of Independent States (CIS) countries of -3.1, poverty would decline by 75 percent or from 45 percent to 11 percent in those 5 years. Figure vii: Relationship between Growth and Poverty Reduction EU-8 SEE (w/o Balkans) CIS Middle Income CIS Low Income W Balkans Kosovo +40% +20% yt pover Kosovo ni +0% Change -20% -40% -10% -5% +0% +5% +10% +15% Change in GDP Source:World Bank staff calculations from the World Bank ECAPOV database. 23. While growth is essential, it only makes a big difference in poverty reduction, if it is inclusive. The logical question would be to ask for the sources of such growth. While such analysis is outside the scope of this report, a common experience of many countries which have had successful large scale poverty reductions through growth has been the implementation of strategies that lead to high employment-generating growth. In the case of Kosovo, crucial sectors where employment generation would be expected to be high and conducive to poverty reduction include transport and communications infrastructure, construction, services, and improving the investment climate for small and medium enterprises especially by making vii available affordable and reliable electricity. Such inclusive growth is crucial because simulations in this report show that policies that generate employment are highly correlated with poverty reduction. For instance, a 10 percent reduction in the prevailing unemployment in the population is associated with a 6 percentage point reduction in poverty. Given that majority of the poor are working poor, such growth would provide double dividends in welfare if it can generate employment and improve wages. 24. In addition, it will be critical to improve urban infrastructure and services. Such an action recognizes that urban areas are central to economic growth. In fact the evidence from the flow of internal migrants suggests that the poor may already be moving to urban areas to pursue better opportunities. And urban areas are doing slightly better, despite the much higher flows of remittances to rural areas. Aside from taking pressure off the need for international migration, urban renewal will also go a long way to solving some intractable problems in rural development. In particular, larger rural-urban migration flows could reduce further land fragmentation, improve agricultural labor productivity and improve rural welfare. 25. Sustainable growth eventually must be underpinned by a well-educated, adaptable and healthy work force. The evidence suggests that Kosovo's performance with regard to access to primary and secondary education is reasonably high. Almost 9 in 10 children of primary education age and about 3 in 4 children of secondary education age are enrolled. However, these results mask three potential obstacles to establishing a well-educated population. One problem is the existence of large disparities in enrollment rates, especially in secondary and higher education, between the poorest and richest on one hand and between urban and rural households on the other. The second hurdle is that traditional values that limit girls' education are still prevalent in rural areas. As a result, one observes that secondary enrollment rate for girls is 20 percentage points lower than the rate for boys for secondary age children in rural areas. The final problem is the widespread belief that while enrollment rates may be reasonably high, the quality of schooling is generally poor. As a consequence, learning outcomes for many children are often so low that they drop out early. Simple simulations in this report show that more education is correlated with less poverty. Therefore a general improvement in the quality of education and access to secondary and higher education, especially for the poor and rural girls, will have to be considered as a long term strategy for sustaining high growth. 26. Second, in the near future, maintaining migration flows is essential to protecting livelihoods and stability. The main reason to make this argument is that generating high, sustained and shared growth - that is, growth that generates high-employment - which could enable a large fraction of the population to earn their way out of poverty, and therefore, serve as an effective strategy for mass poverty reduction is unlikely to be attained in the short term. For example, to reduce the current unemployment rate by half (from 40 to 20 percent) in the next 10 years, assuming an annual labor force participation growth rate of 1.9 percent and growth to productive employment elasticity of 1.6, Kosovo would need to grow at about 6 percent per year. Given the current investment climate and backlog of investments in infrastructure needed to jumpstart high growth, such a strategy is likely to be realized only in the medium to long term. 27. By comparison migration remains, by far, one of the most effective mechanisms for reducing poverty in Kosovo. Moreover, unlike growth, the impact is immediate. The need to maintain migration as a mechanism for protecting welfare levels in the medium term may require the Kosovo government taking the bold step of entering into a framework for continued migration flow with key recipient countries. Such a framework can be bilateral or regional (say with several or all EU member states). It could include incentives for migrants to eventually return home through portability of pensions and other entitlements (World Bank, 2006c). viii 28. For as long as migration flows continue, Kosovo can rely on the remittances that are associated with it to protect a sizeable fraction of its population from poverty. But clearly a development strategy that is over-reliant on remittances is neither desirable nor sustainable. The reasons are numerous, but three are worth noting. One is that migrants change. Over time, as migrants settle in their host countries, ties to their "home" country will tend to decline. This deterioration in "attachment" accelerates rapidly in subsequent generations. The other reason is that host country policies change. This is particularly poignant for Kosovo, since a number of countries in the European Union that were major recipients of migrants from Kosovo have either shut their borders or have threatened to do so. The final reason is that remittances, which are essentially private transfers, cannot be effective substitutes for more reliable, domestically generated resources to finance public investments in infrastructure, human capital, social services, and so on. This means that there is a need, eventually, to transition out of high dependence on migration and remittances. 29. Third, improving the targeting of the social assistance program will add to the gains, albeit small. The social assistance programs are modest in size, which is understandable given the fiscal situation. In addition, the social assistance program is well-targeted: about 50 percent of the funds go to the poorest fifth of the population. However, about 25 percent of the non-poor receive benefits. Not all these non-poor are necessarily ineligible given the high vulnerability of the population. However, if these are truly ineligible individuals, this is where marginal gains can be achieved. Over time, however, as fiscal constraints ease, the coverage of the social assistance will also need to be expanded in order to provide meaningful protection. 30. Fiscal constraints will ease with growth, and that is clearly the preferred option to expanding social protection programs. But that is not the only option. Kosovo is currently preparing its medium term expenditure framework, which will be presented for funding at a donor conference later in the year. A logical question to ask is, if the budget envelope expands and there are more resources to spend on social protection, which program should the government expand. To get an idea, we looked at the size of poverty reduction implied by three possible actions: a) increasing the size of pensions (mostly basic pensions) by 10 Euros a month, b) increasing the size of social assistance benefits by 5 Euros a month for current recipients, and c) expanding the social assistance program to all the extreme poor and giving each new recipient the median value of current social assistance benefits. It is to be noted that a similar policy was not attempted for pensions because nearly all the eligible now receive it. The results (Table i) show that expanding the social assistance program to the population of extreme poor will reduce extreme poverty by half (from about 15 to 8 percent). There are two additional points to note regarding the results in Table i. One is that the estimated costs should be considered lower bound since likely additional administrative costs are not included. The other is that the impact of all three actions on absolute poverty is not large. This means that, in addition to expansion, benefit levels may have to rise for a large impact on poverty to be achieved. Conclusion 31. Poverty in Kosovo is widespread and has remained persistent in the first half of this decade. The evidence suggests that poverty is higher among those who live in families that are large, have many unemployed members, and have low education levels. The poor are also geographically concentrated in rural areas and a few regions. The main message of this report is that the slow and volatile growth was doubly disadvantageous. The first disadvantage was that it did not enable a significant fraction of the population to earn their way out of poverty. The second disadvantage was that by constraining the government's revenue base, it made it difficult for many families to receive adequate public protection against shocks. Therefore, to improve ix welfare in the future, the report recommends a focus on generating high and sustainable growth - by improving urban services and infrastructure and addressing inequities in the access to secondary and higher education for the poorest population ­ transitioning out of over-reliance on migration, and improving the targeting and expansion of the social assistance program if the revenue base of the government improves over time. Table i: Proposed Policies and Their Cost and Poverty Impact Cost Absolute Poverty Extreme Poverty Est. number of Marginal Cost Recipients All Recipients All Proposed policies: recipients Social Assistance (Euros/ month) (percentage points) (percentage points) Increase transfer by 5 Euros 43,356 216,780 -0.8 -0.1 -2.3 -0.3 Extend to all extreme poor 37,076 2,345,328 -1.6 -0.4 -34.2 -8.3 Pensions Increase transfer by 10 Euros 127,742 1,277,423 -3.6 -1.1 -2.1 -0.7 Source: World Bank staff calculations from HBS data. Notes: 100 percent propensity to consume out of transfer income assumed. The estimated number of recipients under social assistance (column 2) refers to households. Under pensions it is individuals. x CHAPTER 1: MACROECONOMIC AND SOCIAL CONTEXT The average annual growth rate of real GDP between 2002 and 2006 is estimated at less than one and half percent. Potential sources of growth, especially in mining and energy and agriculture, would benefit enormously from new and modern technology. However, inefficient legacies and uncertain property rights continue to hamper the flow of foreign direct investment. Therefore, the outlook for the economy does not seem bright in the medium term. 1.1 Since the cessation of hostilities in 1999, Kosovo has made progress in improving infrastructure, providing public services and laying the foundations for strengthening state institutions. In 2006, talks began on the resolution of its status. While the status talks continue and await final resolution, the Kosovo authorities had started preparing a Kosovo Development Strategy and Plan (KDSP) to work on the vision and coordination of sector strategies, covering such sectors as agriculture and rural, health, education, infrastructure, gender, and water and sanitation. Since the overall development plan is to improve the welfare of the population, poverty reduction is expected to be an integral part of each of these sector strategies. A focus on poverty reduction in the sector strategies is crucial because, the recent poverty assessment indicated that poverty in Kosovo is widespread and affects disproportionate number of rural residents, children, the elderly, female-headed households, and non-Serb minorities. Furthermore, educational and health outcomes are low, while exposure to health risks is widespread. Moreover, Kosovo continues to face challenges common to all fragile states; huge backlog of investment needs and limited government resources to meet them, fractured societal relations, and weak security. To strengthen social cohesion and ensure lasting peace, public policy making need to focus on developing a strategy to reduce poverty as a matter of urgency. The KDSP can benefit significantly from incorporating the findings of such assessments in its sectoral strategies. 1.2 This note provides estimates of trends in poverty and inequality using the Household Budget Survey data which is implemented by the Statistical Office of Kosovo every year. It updates the information on the size, scope and determinants of poverty which was last done using the data collected in 2002. This chapter provides a short description of the macroeconomic developments. Chapter 2 looks at poverty trends and decomposition. Chapter 3 presents the profile of the poor and factors that determine the risk of being poor. Chapter 4 provides a discussion of the social protection system and its effect on poverty outcomes. It also examines the role played by migration and remittances in shielding household against poverty. Chapter 5 concludes with recommendations for poverty monitoring. A. GDP GROWTH HAS BEEN POOR 1.3 Recent growth has been low, but data quality is an issue. For a brief period following the end to the conflict, growth surged on account of massive reconstruction efforts financed by huge inflows from donors and Kosovar diaspora. GDP growth was estimated at 21 percent in 2000, mostly due to large inflows of foreign assistance for reconstruction activities, and private investments in response to significant trade reforms. But estimated real GDP growth between 2002 and 2005 has been slow and volatile. Real GDP growth was negative in 2002 and 2003. This was followed by a positive upturn in 2004 due to expansionary fiscal stance. The revised estimates place the growth in 2005 at close to zero (at 0.5 percent) and 3 percent growth in 2006. The prediction is for growth to decline to 2.3 percent in 2007 (Table 1.1). Average real GDP growth was around 1.5 percent in the 5 years between 2002 and 2007. This slow expansion is due to a combination of low investment and the ongoing withdrawal of the international community in Kosovo. Table 1.1: Macroeconomic Trends Proj. 2002e 2003e 2004r 2005r 2006r 2007p National Accounts Real GDP growth -0.1 -0.5 2.5 0.5 3.3 2.3 GDP per head (in 2002 Euros) 1,141 1,147 1,156 1,143 1,161 1,168 Investment (in 2002 mil Euros) 634 594 635 627 677 772 External Accounts Current account balance (% GDP) /2 -50.1 -41.6 -39.5 -40.7 -42.2 -40.5 Foreign assistance (% of GDP) /3 42.7 32.4 25.5 22.6 20.7 16.2 Workers' remittances (millions of Euros) /4 35 125 215 262 300 342 Prices CPI Inflation 3.6 1.3 -1.4 -1.4 1.5 2.0 Source: IMF (2007) and World Bank and IMF staff estimates. IMF estimates subject to revision. Notes: e = estimate, r = revision in 2007, p = projection. 2/ Before donor grants. 3/ Total foreign assistance excluding capital transfers 4/ Including pensions from abroad. 1.4 Currently, both the IMF and EAR provide GDP estimates. Numerous revisions of the estimates reflect both inadequacy of data and methodological differences. A key difference is whether to treat UNMIK as a resident contributor to GDP. IMF estimates treat UNMIK as resident, while EAR produces estimates which treat UNMIK as resident and non-resident. As a result, GDP estimate which treat UNMIK as resident was 21 percent and 16 percent higher in 2002 and 2003, respectively, than estimates that treat UNMIK as non-resident. Treating UNMIK as non-resident also explains the negative growth rates in earlier years as UNMIK downsizes. Problems with computation of the CPI add to the difficulties. Table 1.2 shows the scale of the differences. 1.5 Changes in consumption can also be affected by assumptions made about population growth and extra adjustments made to certain consumption items. For instance, in projecting private consumption part of the GDP, IMF assumes a population growth of 1.7 percent per year. It also includes car purchases in consumption, and makes extra adjustments to electricity and food reported from the household surveys. In this report, no adjustments are made to electricity and food expenditures, and car purchases which are infrequent and whose benefits are enjoyed over a long period are not included (Table A.1). There are also differences with previous SOK estimates of consumption: for instance, in SOK (2006b) household expenditure was estimated to have increased by about 5 percent annually because it includes semi-durables and durables in estimating consumption, whereas in this report these goods are excluded as the annual consumption of their benefits cannot be estimated. 2 Table 1.2 : Alternative Estimates of GDP for Kosovo 2002 2003 2004 GPD at current prices (million Euro) Including Donor's sector: EAR 2 589.9 2 505.0 ? IMF estimate 2423.9 2414.7 2434.3 Without Donor's sector 2 139,2 2 157,4 2 008,2 GDP per capita at current prices (Euro) Including Donor's sector 1 363,1 1 296,4 1239,0 IMF estimate 1294 Without Donor's sector 1 125,9 1 116,5 1021,9 GDP real growth rate (%) Including Donor's sector -5.9 +2.8 IMF estimates -0.6 +4.1 Without Donor's sector -1.9 + 7.1 CPI changes (%) -1.6 Source: Kosovo National Accounts, Consultant report. March 2006. B. AGRICULTURAL GROWTH HAS BEEN SLUGGISH 1.6 Poor output is explained in part by sluggish agricultural growth. Agriculture remains the main sector, and largest employer, in Kosovo. In 2004, it accounted for 25 percent of Kosovo's GDP, 16 percent of the value of total exports and between 25 to 35 percent of all employment, mostly in the informal sector (AMP, 2006). While its recorded share of GDP has declined to 19 percent in 2005, this is not believed to be a reflection of productivity gains in the sector. Rather, the evidence appears to point to a recovery in other sectors, especially services. Sluggish agricultural growth is partly a hangover from the war damage such as the destruction of infrastructure, machinery and livestock, and loss of traditional export markets. As a result, estimates of agricultural output show sharp increases in food imports between 2000 and 2003, which have declined only slightly between 2004 and 2005, while exports have increased from a very low base between 2004 and 2005. 1.7 Land devoted to crop production has remained stable. According to the SOK Agricultural Household Survey, agricultural land use remained at the same level of about 360,000 ha in 2004 and 2005. Livestock ownership also does not show an increase. Kosovo's agriculture is dominated by grains (49 percent of crop land is devoted to grains), and characterized by small farm size (about 65 percent of all farms are less than 3 hectares), absence of advisory services, and low productivity all of which constrain its contribution to growth. Acreage under crops peaked in 2004 (also the only year estimated GDP growth was higher than the average for the period) but declined by about 14 percent in 2005 (Table 1.3). 1.8 In addition, yields do not indicate an improvement over the last two years. Agricultural yields declined or remained relatively flat between 2004 and 2005, although it must be noted that two years is too short to draw conclusions, given the importance of weather risk to agriculture. While yields for grains held steady, the biggest declines were especially notable for fruits and vegetables (Table 1.3). When farmers were asked why they left the land fallow, about 30 percent reported low economic profitability, which suggests low productivity agriculture. Not surprisingly, the agricultural production is still predominantly subsistence oriented so that smaller farms reported that 70 percent of output is devoted to households needs in 2005. Even farms in the upper end of the size distribution still designate over 50 percent of their production for 3 domestic use (SOK, 2005a). That said, it is important to keep in mind that there are few distortions in Kosovo agriculture so that what is emerging is built on comparative advantage. Table 1.3: Macroeconomic and Agricultural Sector Indicators 2001 2002 2003 2004 2005 Agricultural output 2 Food imports 283 300 330 193 Food exports 1.5 3.8 7.6 10.3 Total ag surface 3 209,058 196,883 n/a 220,506 190,479 Wheat, ag surface 75,070 70,000 n/a 77,783 80,127 Maize, ag surface 75,038 69,000 n/a 100,970 74,079 Vegetable, ag surface 28,000 28,220 n/a 14,419 14,140 Labor market indicators 1 Registered unemployment 282.3 282.3 302 319.7 Workers' remittances 341 341 345 347 Agricultural Household Surveys 2004 and 2005 4, 5 Household size 7.7 7.8 Agricultural population (million) 1.3 1.3 1.3 Agricultural Land (000s) 291 264.9 260.1 Average yield (000s) 6 Grains 3.1 3.4 Vegetables 16.2 12.6 Fodder Crops 4.5 5.8 Fruits 9.7 5.1 Livestock Numbers Cattle (in thousands) 347 318 351.8 335.2 Poultry (in million) 2.2 3.1 2.6 2.2 Source: 1 WB 2006 Interim Strategy; 2AMP Agricultural Development Plan;3Agricultural Statistical Office- MAFRD (total excludes forage and fruits). SOK Agricultural Household Survey, 2005. Household size 2004 is 4 from the 2004 report, thus not adjusted for changes in the weighting procedure. Agricultural household survey 5 data for the last two years is not comparable to 2001 and 2002 data because of change in methodology. Average 6 yield was calculated by taking the unweighted average of yields reported in Tables 4.1 and 12 in SOK Agricultural Household Survey, 2005. Yields are in kg/hectare, while land is in hectares. C. DISAPPOINTING LABOR MARKET PERFORMANCE 1.9 Sluggish growth has resulted in poor labor market performance in the last two years (Figure 1.1 and Figure 1.2). Overall unemployment rate rose from about 40 percent in 2004 to 42 percent in 2005 (SOK, 2006a). According to the Labor Force Survey conducted by the Statistical Office of Kosovo (SOK, 2006a, 2005b), the unemployment rate among male workers stayed around 30 percent in both years, but 60 percent among female workers. The number of registered unemployed rose from 282,000 in 2002 to 319,000 in 2005 according to official administrative records of registered unemployment (World Bank, 2006). Across age groups, the highest unemployment was among the youth (15-24 year olds) ­ 67 and 65 percent in 02/03 and 04/05 respectively. Finally, unemployment durations are long ­ that is, over 80 percent of the unemployed are in such a status for a year or longer. 1.10 Finally, wages remained unchanged throughout the period. Average real monthly wages reported by wage earners in the household remained flat between 2002 and 2005. The 4 experience is the same whether one looks separately at salaried, professional or manufacturing workers (Table 1.4). Figure 1.1: Employment and Unemployment Rates, Figure 1.2: Registered Unemployment and GDP 2001-2005 growth rate 60 GDP Growth and Unemployment 57.1 55.0 50 3 49.7 3.2 3.0 2.8 2.8 40 39.7 41.4 2 t 2.1 cen 30 27.7 28.5 1 er p 23.8 25.3 20 19.6 0 -0.1 0.3 10 -0.5 -1 0 2002 2003 2004 2005 2001 2002 2003 2004 2005 Registered unemployment Real GDP growth Employment Rate Unemployment rate (in 00, 000) Source: LFS 2001-2005, SOK calculations. World Bank 2006 Interim Strategy. D. LOW PROSPECTS FOR POVERTY REDUCTION 1.11 The prevailing macroeconomic conditions do not provide the platform for significant poverty reduction. The prospect for improved growth is uncertain. According to the IMF, the economy's poor fundamentals and continued donor withdrawal is expected to slow down growth for some time to come. Poor infrastructure and energy are expected to be major bottlenecks. On the positive side, the possible resolution of the status could provide positive signal for clearer property rights and improve the investment climate. Table 1.4: Real Mean Monthly Wages and Number of Observations 2002/03 2003/04 2004/05 2005/06 Total 218.4 214.7 219 209.4 Number of observations (1184) (1590) (1689) (1820) Male 220 219.8 225.7 213.2 Female 211.2 197.2 195.5 193.9 Salaried employee 218.6 214.4 222.6 219.5 Number of observations (1124) (1566) (1577) (1577) Professional 229.6 239.6 249.9 232.3 Manufacturing 204.3 204.6 196.4 210.6 Other 234.6 206.9 204.3 213.1 Per-diem worker 215.5 227.6 168.1 153.5 Number of observations 60 24 112 243 Source: World Bank staff calculations from HBS data. Notes: Cash wages and salaries, net of tax, only for . salaried and per-diem workers, in 2002 prices. included only. Definition of wages changed in 2005 to be restricted to earned in Kosovo and income categories increased from 8 to 13 categories from 2002 to 2005. 1.12 Moreover, a quick look at households' asset holdings does not provide clear evidence of positive welfare changes. The proportion of households owning cars and cell- phones increased during the period (Figure 1.3). The increase is evident in both rural and urban areas, but was faster in urban areas. An additional observation is that relatively high proportion 5 of the households report owning such durable goods as refrigerators, cars, cell phones and washing machines, confirming further that poverty in Kosovo may be widespread but it is shallow. Household income levels appear to have increased between 2002 and 2005 in urban areas and decreased in rural1. This provides mixed and inconclusive evidence for the direction of changes in the poverty rate in Kosovo, since a large fraction of the population lives in rural areas. Figure 1.3: Asset ownership and income levels using HBS data A. Percent of households owning asset B. Individual income levels, 2002 prices Source: World Bank staff calculations from HBS data. Notes: Income is defined as cash wages net of tax, wages in kind, income from per diem work, rent, dividend, interest, social welfare benefits, pensions from Kosovo. 1.13 To conclude, after a boost to growth from reconstruction finance in the initial stages of the post-conflict period, growth has slowed down considerably. Industrial recovery is still uncertain and agriculture, where many earn their living, has become dominated by subsistence orientation. As a result, the labor market prospects for many families have been gloomy. Lack of opportunities to earn one's way out of poverty have led to widespread and stagnant poverty, as the next chapter documents. 1 These figures need to be treated with caution because of concerns about sample weights for the HBS. See Chapter 2. 6 CHAPTER 2: POVERTY TRENDS AND DECOMPOSITION About 45 percent of the population of Kosovo remained in poverty during the period under review. An additional 18 percent were vulnerable to poverty. Furthermore, 15 percent of the population was estimated as extremely poor. While population wide poverty rates remained unchanged, measured urban poverty declined, while rural poverty increased. The evidence suggests that in urban areas, positive growth in consumption was widely shared, while in rural areas, only the top fifth of the population gained. As a result, measured inequality in the population and in rural areas increased. A. A NOTE ON DATA QUALITY AND APPROACH 2.1 This report draws on the Household Budget Survey (HBS) data for much of its content. HBS is a core survey of the Statistical Office of Kosovo (SOK). It provides a sustainable start to monitoring poverty and inequality. However, using the last 4 surveys of the HBS to analyze poverty and inequality presents practical problems. Therefore, it is useful to begin the analysis with a discussion of the challenges and how this analysis has approached them. 2.2 The first major uncertainty concerns representativeness of the samples. Kosovo has not had a reliable census since 1981. Therefore, the current surveys still use the 1981 population frame as the starting point for selecting areas, and therefore households, to include in the sample. However, much has changed since then. Some areas that were highly populated may not be so now and some that had few people may have many more. Without further action, the first introduces large sampling error while the latter introduces bias. For instance, the estimated population from the surveys shows a dramatic decline from 2 million in 2002/03 to 1.5 million in 2005/06, which is not consistent with alternative data sources (Box 2.1). Since there is no clear knowledge of the reference population it is difficult to correct for these problems. Therefore, it is probable that the sample data and statistics obtained from it contain unquantifiable bias. Box 2.1: Estimating the Population of Kosovo There is no single source that provides an accurate count of the population. We rely on two sources to estimate the population total and urban and rural ratios. First, we use the school enrollment data to estimate the overall population. Between 2002/03 and 2005/06, total primary and secondary school enrollment ranged from 410, 000 to 422,000. About 380,000 of these were enrolled in primary ­ ages 6-14. Using the population pyramid of neighboring Albania, and applying the ratio of this age group in the Albanian population, we estimate the population to be about 2 million. We combine this with estimates of the rural population from the Agricultural Household Survey (AHS) conducted by the SOK to estimate the implied rural and urban distribution. This survey first listed all the 1400 villages in Kosovo. Then it took a sample of these villages and listed all the dwellings. They estimated the rural population to be about 1.3 million. Together, this means that rural share is about 65 percent and urban is about 35 percent. 2.3 The second challenge concerns comparability of the surveys. The first survey in the series, conducted in the 2002/03, asked households to record daily expenditures on an open-ended diary. In subsequent years, households were given a list, admittedly encompassing expenditure items reported in the first survey. However, they were also asked to record weekly, not daily expenditures. Moreover, the list has expanded over time, although only by a few items. The second significant change was how the consumption of own-produced goods was reported. In the first survey, it was left to the households to report each item. In subsequent surveys, they were aggregated into a few categories. Rather than report how much wheat was produced for own consumption, households were asked to report grain crops, or fruit and fruit products, and so on. In addition, rather than ask quantities consumed, they were asked to report the value in Euros of the production they have set aside for own use. Furthermore, for own produced consumption, the period of recall changed from daily in the first survey to a month in subsequent surveys. The reason why these changes matter is that moving from a short recall to a longer recall has the tendency to understate consumption and overstate poverty rates (Box 2.2). Box 2.2: Challenges of Using HBS Data for Poverty Analysis The Kosovo Household Budget Survey (HBS) began in June of 2002 and has been administered on a monthly basis by the Statistical Office of Kosovo since then. It has become a core survey in Kosovo's efforts to build a long term monitoring and evaluation system and is fully funded by the government. The survey data, however, poses several practical problems for a reliable estimation of welfare in the first half of 2000s in Kosovo. Two issues warrant a careful use of the results: 1. An outdated sampling frame introduces large sampling errors. As any other micro survey, the HBS relies on a representative sample of the population in order to obtain unbiased estimates of the population statistics. In order to do that, a reliable frame from which to draw the sample is needed. Kosovo's latest reliable census dates back to 1981. The sampling frame (a list of enumeration areas (EAs)) was supplied by SOK and the sampling was done by Statistics Sweden under a project funded by the Swedish International Development Cooperation Agency (SIDA) (Andersson, 2002b). The quality of the list of EAs is questionable: the distinction between urban and rural is purely administrative; the classification of ethnicity does not follow strict rules, and the description of the geographical boundaries of the EAs is outdated (Andersson, 2002a). Finally, survey administration to ensure quality was heavily limited because of lack of resources at SOK and thus misclassified EAs were skipped (Andersson, 2002c), relisting of large EAs was incomplete and field control of enumerators is lacking. There are also issues of undercoverage. Therefore, the HBS demographic statistics each year seem to be from different underlying populations. For instance, the estimate for total population based on the HBS is 2.1 million in 2002- 03 and 1.5 in 2005-06. Such decline is not realistic and contradicts evidence from school enrollment administrative data that population remained relatively stable. 2. Problems of data comparability. Finding how poverty changed in Kosovo using HBS data is not straightforward. There are two changes across surveys that make data non-comparable and require special methodologies to correct comparability, which are discussed in Volume II. The first change is the move from diary to recall method of expenditure collection, starting in 2003. The second change in survey design introduced was the redefinition of consumption items, effectively shortening or increasing the list. One such change, the third change, is the aggregation of consumption of own production from a long list (85 items) to a short one (12) and changing the recall period from daily to a month at the same time. The aggregation of consumption items affects what respondents remember and a shift from a longer list to shorter list is likely to lead to lower reported consumption, and therefore higher poverty. 2.4 To resolve these issues we follow three steps. First, we use the best estimates on the population distribution from other surveys and administrative data to adjust sample weights. We adjust the weights so that the rural to urban ratio remain stable, and call these adjusted weights, post-stratified weights. All our results are therefore weighted using the post-stratified weights. 2.5 Second, we make three decisions to handle the comparability issue. The first decision is to compare only 2003/04 and 2005/06 data for the purposes of examining trends. We dropped 2002/03 because it is the least comparable to the other three surveys. For example, it used an 8 open-ended diary, contained substantially many more items, and had the most expansive definition of own produced goods. We also exclude 2004/05 but the decision is a lot harder. In all aspects, the survey is comparable to 2003/04 and 2005/06. It provided households with the same list of food and non-food items, defined consumption of own production the same way and used the same recall period. That said, we find that the consumption is much higher in 2004/05 compared to the year before or after. Specifically, non-food consumption is about 30 and 18 percent higher compared to 2003/04 and 2005/06, respectively (Table A.1). As a consequence, the poverty rates are much lower and the trends show more volatility. There is no clear explanation for this huge jump in consumption in 2004/05. Therefore, we excluded it in the analysis of the trends. 2.6 The first decision implies establishing the baseline poverty as that obtained for the 2003/04 data. Ideally it would be useful then to obtain the poverty line using the cost of basic needs approach using the same survey. But the surveys after 2002/03 stopped collecting quantity information that is essential for calculating the poverty line. So the second decision was to use the poverty line obtained in 2002/03 and used in the most recent poverty assessment (World Bank 2005), adjusted for price changes. 2.7 The final decision involved undertaking a number of sensitivity analysis. In order to try and use all the surveys, several methods for comparing non-comparable data that are available in the literature were applied to the Kosovo data. In all, four different methods are used. A detailed discussion of these methods and the results obtained from applying them to the Kosovo HBS data is the subject of Volume II of this report. With that as a background, the rest of this report looks at the poverty outcomes using the 2003/04 and 2005/06 data sets. B. THERE WAS NO CHANGE IN POVERTY 2.8 Consistent with sluggish economic activity, the evidence from the household budget survey points to no change in the poverty rate. The poverty rate is around 45 percent in both 2003/4 and 2005/6 (Figure 2.1). The results from several methodologies (see Volume II), which attempt to correct for a number of weaknesses in the data support an unchanging poverty rate. Surprisingly, there is a very large drop in the poverty rate between 2003 and 2004, but it is unlikely that this is capturing a real change in the welfare of the population. More likely, sampling and non-sampling (survey administration) problems contributed to the observed drop (see Box 2-2 for details). Other measures of poverty support stagnating poverty. The poverty gap remained around 12 to 13 percent, and severity of poverty was about 5 percent (Table A.9). 2.9 There was also no change in the fraction of the extremely poor. The extreme poor, defined as individuals who have difficulty meeting basic caloric needs, comprised 13 percent of the population in 2003, but about 16 percent in 2005. The apparent increase is driven almost entirely by a large and improbable increase in the proportion of the extreme poor among household heads who report being Serbian (see Box 3.1). The extreme poverty rate among heads of households who are Albanian rose from about 13 percent to 15 percent during this period, while the rate among Serb heads of households rose from 9 percent to 44 percent in the same period (Table 3.1). 9 2.10 In addition to widespread poverty, vulnerability is high. Although widespread, poverty in Kosovo is shallow. One way to gauge the shallowness of poverty is to look at other measures of poverty that are sensitive to the location of the poor in the distribution of welfare measure. For instance, the poverty Figure 2.1: Absolute and extreme poverty for 2003/04 gap is sensitive to the distance of the and 2005/06 poor to the poverty line. It measures the per capita consumption shortfall 0. 50 50 of the poor, as a fraction of the teulo 45.1 43.5 poverty line, and at 12 to 13 percent Abs 0. 40 in this period it is low. Severity of 40 poverty, which is sensitive to both distance to the poverty line and 0. 30 30 inequality among the poor, is also low, and is estimated at about 5 0. percent. Another way to measure 20 20 e shallowness of poverty is to estimate 16.7 metr 13.6 the fraction of the population that is Ex 0. 10 just around the poverty line. Figure 10 2003/04 2005/06 2.2 (panel A) shows that a shock that Confidence Interval reduces the consumption of those who are now considered non-poor by Source: World Bank staff estimates from HBS data. 25 percent would send an additional 18 percent of the population into poverty. In other words, if the poverty line was 25 percent higher, the poverty rate would be about 63 percent, and not 45 percent. Figure 2.2: Distribution of Consumption, 2003/04 and 2005/06 A. Cumulative distribution for consumption B. Consumption with vulnerability and poverty lines Poverty lines Extreme Absolute Vulnerability Poverty line noital Extreme Vulnerability .6 .8 Poverty Line pu po of .6 %, .4 onit ytisn .4 buirts De di .2 .2 evitalu mu 0 0 C 2 3 4 5 6 0 20 40 60 Log of Monthly Per-capita Consumption, Real total consumption per adult equivalent, monthly in 2002 prices 2003/04 2005/06 2003/04 2005/06 Source: World Bank staff calculations from HBS data. Poverty lines are population-weighted averages for the two survey years. 2.11 The consumption growth was negative for the majority of the population. Changes in consumption for each percentile (groups of households ranked by per capita consumption) of the consumption distribution shows that the bottom four quintiles had negative growth between 2003 and 2005. Moreover, a visual inspection of the gains to the top quintile shows only modest rates, roughly in the order of 5 percent on average. 10 2.12 However, urban populations gained and the growth was more evenly distributed. Figure 2.3, Panel B shows the changes in consumption for each percentile. It indicates that gains for all urban groups, except perhaps the lowest 5 percentile or so were positive. The evidence also suggests that the size of the gains observed for percentiles which gained in the bottom half of the distribution were of roughly the same magnitude as the gains experienced by those in the top half of the distribution. In other words, the gains (for those who gained) were between 5 to 10 percent, regardless of the percentile rank. Figure 2.3: Growth incidence curves for growth from 2003/4 to 2005/06 Total Urban Rural 10 10 10 0 5 0 0 0 0 -1 -1 -5 0 0 -2 -2 0 -1 0 -3 0 5 -3 -1 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 Percentiles Percentiles Percentiles Median spline Growth rate in mean Median spline Growth rate in mean Median spline Growth rate in mean Source: World Bank staff calculations from HBS data. 2.13 But, in rural areas only the top quintile of the population gained. The gains for first three quintiles (up to 60th percentile) were negative, while zero for the fourth quintile (Figure 2.3 Panel C). The consumption losses appear to be quite substantial for the bottom quintile. They report greater than 10 percent loss. We also observe more than 20 percent loss for some population groups in lower percentile ranks. The graph also gives a hint as to why we observe poverty reduction in urban areas and not in rural. In the former, two of the four deciles that are considered poor had positive gains, allowing some of these groups to escape poverty, while in rural areas there was erosion in consumption positions for poor deciles. Since at least 60 percent of the Kosovo population is still rural, it is not surprising that there was overall stagnation in poverty. Table 2.1: Decomposition of Poverty: 2003/04 Compared to 2005/06 Change Growth Redistribution Interaction Total Poverty headcount (P0) 1.6 -0.8 2.5 -0.1 Poverty gap (P1) 1.4 -0.4 1.7 0 Poverty severity (P2) 1.2 -0.2 1.3 0 Urban Poverty headcount (P0) -4.7 -3.3 -0.1 -1.3 Poverty gap (P1) -1.2 -1.5 0.2 0.1 Poverty severity (P2) -0.2 -0.7 0.5 0 Rural Poverty headcount (P0) 5 1.4 3.7 -0.1 Poverty gap (P1) 2.8 0.5 2.3 0 Poverty severity (P2) 1.9 0.2 1.6 0 Source: World Bank Staff calculations from HBS data. 11 2.14 A decomposition of the changes in poverty suggests that redistribution played a larger role than growth in the observed poverty trends. Table 2.1 shows the proportion of the observed trends that is accounted for by growth, changes in the distribution and a residual. The decomposition is done for headcount, poverty gap and poverty gap squared. Note that between 2003 and 2005, estimated headcount increased by about 1.6 percentage point. Table 2.1 suggests that, if the shape of the distribution stayed the same ­ that is, inequality did not change between 2003 and 2005 - then headcount poverty would have declined by 0.8 percentage point on account of the growth in mean consumption. This means that about 16,000 people could have escaped poverty. Instead, we observe an increase in poverty because redistribution within the population leads to a 2.5 percentage point increase in headcount poverty, which offset and exceeded the expected benefits from growth. Put differently, the table suggests that in addition to poor growth, there was a rise in inequality in Kosovo during this period. 2.15 Table 2.2 provides a summary of the changes across the distribution. It shows that between 2003 and 2005, mean of per capita consumption grew by about 1 percent, but that Table 2.2: Growth Rates in Per Capita those at the median had close to no growth at Consumption all. In fact, the mean growth of per capita Growth rate in mean 1.13 consumption for each percentile was negative Growth rate at median -0.50 (Table 2.2, row 3). More importantly, the per Mean percentile growth rate -1.28 capita consumption of the 44 percentile group Poverty line 43 ­ that is, the fraction of the population who Corresponding percentile 43 were classified as poor in 2003 ­ experienced Rate of pro-poor growth -5.78 some of the largest losses, a 5.8 percent Source: World Bank staff calculations from HBS data. reduction. 2.16 In urban areas, growth accounted for a significant reduction in poverty, while in rural areas redistribution accounts for the increase in poverty. About 3.3 percentage point reduction in the observed reduction in urban poverty is accounted for by growth. Moreover, any redistribution that may have taken place appears to have had no adverse effect on poverty. By contrast, almost 4 of the 5 percentage point increase in rural poverty, was due to a change in the shape of the distribution (Table 2.1) Table 2.3: Inequality Indices for 2003/04 and 2005/06 Total Urban Rural 2003/04 2005/06 2003/04 2005/06 2003/04 2005/06 Percentile ratio p90/p10 3.33 3.96 4.05 4.42 3.09 3.68 p75/p25 1.89 1.97 2.14 2.06 1.78 1.9 Generalized Entropy, GE(0) 0.12 0.15 0.16 0.17 0.1 0.14 GE(1) 0.13 0.16 0.17 0.16 0.1 0.14 Gini coefficient 0.27 0.3 0.31 0.31 0.25 0.28 Source: World Bank staff calculations from HBS data. 2.17 In addition to stagnating poverty, there was a slight increase in inequality. As discussed above, there are two avenues for the observed redistribution to lead to more inequality. First, while most of the urban population had positive growth, majority of the rural population had negative growth. Second, in rural areas, where at least two thirds of the population lives, those ranked at the bottom of the consumption distribution had substantial negative losses, while the top quintile had positive growth. The net result should be an increase in inequality, which is what we observe (Table 2.3 and Figure 2.4). The results show that the Gini coefficient rose by about 4 percentage points between 2003 and 2005. Other measures of inequality also show an 12 increase. Moreover, inequality in rural areas rose higher than that in urban areas, even though inequality in urban areas is higher than in rural areas. Specifically, as the visual representation of gains and losses showed (Figure 2.3), the ratio of the richest to the poorest household rose sharply in rural areas (Table 2.3). Figure 2.4: Lorenz Curves for 2003/04 and 2005/06 A. Entire Population B. Urban 1 1 noi noi pt pt .8 .8 um um onsc onsc .6 .6 of of e e arhs .4 arhs .4 evitalu .2 evitalu .2 mu mu C C 0 0 0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1 Cumulative share of population Cumulative share of population 2003/04 2005/06 2003/04 2005/06 Equality Equality C. Rural D. Pristina region 1 1 noi noi pt pt .8 .8 um um onsc onsc .6 .6 of of e e arhs .4 arhs .4 evitalu .2 evitalu .2 mu mu C C 0 0 0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1 Cumulative share of population Cumulative share of population 2003/04 2005/06 2003/04 2005/06 Equality Equality Source: World Bank staff calculations from HBS data. C. NON-INCOME DIMENSIONS OF POVERTY 2.18 Non-income dimensions of welfare show better outcomes. Although the trends in income poverty have been disappointing, there is almost universal enrollment in primary education. There are no significant male-female disparities in primary enrollment, and differences between poor and non-poor or rural and urban are also negligible. Enrollments in secondary education are also within the range of rates estimated for countries in the region. However, the tertiary enrollment rates are slightly lower than the rates estimated for countries in the region, except Albania. For instance, net enrollment rates for the 20-24 year olds was about 17 percent in 2005. By comparison, it was 24 percent in Bosnia. The enrollment for 20-29 year olds is 10 percent for Kosovo, while it is 13 percent in Albania (Table B.50). In addition, a relatively high proportion of the population report living in safe dwellings (with walls of brick, 13 blocks or cement) (Table B.31), dwellings with electricity and indoor water taps (Figure 2.6 and Table B.29). Figure 2.5: Net Enrollment Rates Primary Secondary Tertiary 100 100 30 95 95 25 90 90 20 85 85 2003/04 2003/04 2003/04 80 80 15 2005/06 2005/06 2005/06 75 75 10 70 70 5 65 65 60 60 0 Poorest Quintile 2 Quintile 3 Quintile 4 Richest Poorest Quintile 2 Quintile 3 Quintile 4 Richest Poorest Quintile 2 Quintile 3 Quintile 4 Richest quintile quintile quintile quintile quintile quintile Source: World Bank staff calculations from HBS data. Net enrollment rates = Total enrolled students aged 6- 14/children in that age group. In 2002/03 questionnaire, the question about enrollment was asked for children over 7. In later surveys this was changed to 6 year-olds. The age groups used are: primary 7-14 in 2002/03, primary 6-15 starting 2003/04; secondary 16-18; tertiary 20-24. 2.19 Despite these positive outcomes, there are two major areas of weakness in basic services. First, significant inequities exist in access to basic services. In education, net enrollment in secondary education is on average about 70 percent. However, the enrollment rate for the richest quintile is about 20 percentage points higher than the poorest quintile. The disparity in enrollment is much larger in tertiary, where the net enrollment rates for the richest quintile are twice as high as those of the poorest quintile. In health, childhood immunizations and inadequate nutrition, especially among the minority groups (Roma, Askalja and Egyptians), show poor outcomes (World Bank, 2006a). Similarly huge disparities in access to indoor water tap are observed between the richest and poorest quintiles (Figure 2.6). Figure 2.6: Access to water: Percent of people living in dwellings with indoor water tap, 2003/04 and 2005/06 A. Total, urban and rural B. By quintile 90 100 80 90 70 60 Total 80 50 2003/04 Poor 40 2005/06 70 Rural 30 20 60 10 50 0 Poorest Quintile 2 Quintile 3 Quintile 4 Richest 2003/04 2005/06 quintile quintile Source: World Bank staff calculations from HBS data. The questions about housing are not the same for 2002/03 and later surveys. The number of categories decreases from over 20 to 9. 2.20 Second, near-universal access has not been accompanied with equally good quality of services. Although HBS surveys are unable to capture these issues, it is commonly understood the quality of basic services is poor. In education, there are efforts to modernize the curriculum and to make teaching profession competitive. Yet, overcrowding, especially in urban schools is common, a large fraction of teachers are not professionally qualified and there is no standard or 14 externally referenced assessments (World Bank, 2006a). Therefore, there is no way to know what children learn in schools. On electricity, there are known problems of inadequate supply, especially during peak demand, and water quality is uneven across the wealth gradient and rural/urban divisions. On water, quality differences are especially stark between rural and urban areas. In 2003, 70 percent of water samples from rural areas were found to have bacteriological contamination while only 8 percent of samples from urban areas did. In addition, chemical contamination was observed for 46 percent of rural samples compare to 4 percent of urban (World Bank, 2006a). 2.21 Finally, about 9 percent of the population is deprived on multiple dimensions. About 32 percent of the population is poor and have access to both indoor water tap and proper sanitation. That means the additional 15 percent poor have access to either water tap or proper sanitation but not both. But the most deprived are those who are materially poor and in addition have no access to indoor water and proper sanitation. The evidence from the HBS suggests that about 8 percent are poor and in addition have no access to water tap inside the dwelling or proper sanitation (flush toilet). Similarly, 9 percent of the population is poor and do not have access to telephone connection or bathroom (Figure 2.7). By comparison, a recent study showed that only 1 percent of the populations in Russia, Georgia and neighboring Romania were deprived on multiple dimensions (World Bank, 2005b). Figure 2.7: Venn Diagram of Non-income and Income Poverty A. Water, sanitation and income B. Telephone, housing and income poverty poverty A: Access to inside water tap A Telephone connection B: Sanitation: flush toilet B Bathroom available C: Poor households C Poor household Source: World Bnak staff calculations from HBS data. 2.22 To conclude, the evidence documented in this chapter indicates that income poverty remained widespread and persistent between 2003 and 2005. The decomposition of the poverty trend showed that changes in inequality had a much larger impact on the observed poverty outcomes than growth. Furthermore, these changes in inequality also partially explain the observed differences in urban and rural areas. We also found that while non-income dimensions of poverty show better outcomes, unequal access and poor quality are common. Finally, we found that a large fraction of the population is deprived on multiple dimensions. 15 CHAPTER 3: POVERTY PROFILE AND POVERTY RISK The incidence of poverty is higher for larger households, especially those with many dependents, and families with many unemployed members. There is also evidence that households headed by Serbs and females have higher risk of poverty, although for the former data quality issues are of concern. As in many other situations, higher levels of schooling are associated with lower levels of poverty. Finally, we find that geographically, poverty incidence is higher for rural residents and populations living in Mitrovica and Ferizaji. 3.1 This chapter provides a brief description of the observable characteristics of the poor. It focuses on their human capital endowments, their demographic structure, their location in space, and their performance in the labor market. It pays special attention to the rural poor. First, the discussion will focus on the incidence of poverty. Then, it looks at factors that determine the risk of falling into poverty. A. DEMOGRAPHIC CHARACTERISTICS OF POOR HOUSEHOLDS 3.2 Larger households are, on average, poorer. Households that have over 6 family members have a poverty incidence of nearly 50 percent, compared to 30 percent incidence for households half the size or smaller (1 to 3 members). Over time the rate for larger households has remained unchanged, while that for the smaller households has increased, reflecting poor prospects for the whole population during the period (Figure 3.1 and Table B.5). 3.3 In addition to size, household structure introduces its own drag on welfare improvement. Households with only dependents, that is, those composed of only children and elderly, had 40 percent poverty incidence Figure 3.1: Poverty Rates by Household Size in 2003. By comparison households with no dependents have a poverty incidence of 2003/04 2005/06 29 percent. Over time, the incidence of 50 51 49 48 48 poverty for households with more elderly 48 46 or more dependent has worsened. For 40 40 41 instance, in 2003, the poverty rate for 37 mostly elderly households, defined as ecned 30 cini 31 households with more than half their ytr 20 members being elderly, was about the veo P same as the Kosovo average or the average 10 for households with no elderly. However, in 2005, we observe a significantly higher 0 poverty rate for mostly elderly households. 1 to 3 4 to 6 7 to 9 10 to 12 13 + 1 to 3 4 to 6 7 to 9 10 to 12 13 + Household size Household size Similarly, the difference in the poverty incidence between households with higher Source: World Bank staff calculations from HBS data. and lower dependency ratios has widened over time (Table B.8 and Table B.9). 3.4 Poverty incidence is also higher among female headed households, compared to households headed by males. The incidence of poverty among female headed households was 47 percent in 2003, while that for male headed households was 43. Over time, incidence of poverty has risen in both, by the same order of magnitude, so that the gap has remained unchanged (Table B.7). 3.5 Areas with predominantly large Serb ethnic group and households whose head is identified as Serbian appear to have experienced more hardship, although data quality is of concern. To create a representative sample of households, the Kosovo HBS survey stratifies its sample into Albanian and Serbian areas. The classification of areas into predominantly Serbs or Albanians ethnic population is based on the 1981 Census and is relatively arbitrary. In addition, only about 300 households from the predominantly Serb areas are surveyed each year. Thus the results for ethnicities are more imprecise. With that background the results indicate that the poverty headcount for Serb areas and Serb-headed households seems to have skyrocketed between 2003 and 2005 (see Figure b3.1 and Box 3.1 below). Table 3.1: Absolute and Extreme Poverty Rates by Ethnicity of the Household Head. Absolute Extreme 2003/04 2005/06 2003/04 2005/06 Total 43.5 45.1 13.3 16.6 Albanian 43.6 42.5 13.4 14.7 Serbian 34.7 81.8 8.6 43.5 Other 54.3 51.8 18.5 22.7 Source: World Bank staff calculations from HBS data. B. LABOR MARKET CHARACTERISTICS OF THE POOR 3.6 The unemployed are more likely to be poor and their condition has grown worse over time. The unemployment rate of the poor is about 55 percent in 2005/06 according to the HBS (Table 3.2), and about 17 percent of those who are poor and unemployed are heads of households. In the population, about 16 percent of heads of households are unemployed individuals and this share has remained stable during the period. Therefore, the poverty incidence among the unemployed heads was almost 50 percent, and this has not increased over time. Table 3.2: Poverty and Unemployment 2003/04 2005/06 Poverty rate of the unemployed 50.6 49.6 Poverty rate of the employed 31.9 34.7 Unemployment rate of the poor 59.7 54.5 Unemployment rate of the non-poor 41.8 41.3 Source: World Bank staff calculations from HBS data. Notes: Only ages 15-64 and in the labor force. 3.7 Another group with high incidence of poverty is the per diem workers. About 70 percent of the poor are either employed (salaried or self) or inactive. Among the employed, the per diem workers (probably casual laborers who are paid a wage for specific tasks) have the highest poverty incidence (Table B.11 and Table B.12). This group, together with heads of households who are self-employed and in the mining sector, make up a significant fraction of the working poor (Table B.51). 18 Box 3.1: Three Hypotheses for Deteriorating Conditions among Serb Household The poverty rates among households who are classified as Serbs appear to show a sharp rise in poverty between 2003 and 2005. There are three possible hypotheses for these increases. The first is that Serb areas have become enclaves, isolated economies, which are experiencing gloomier economic prospects within a largely stagnant Kosovo economy. The second is that many better off Serbian households have left for the Republic of Serbia and those left behind are mostly the very poor. Finally, there is the possibility that data quality from mostly Serb statistical areas are poor because the Statistical office does not have much control over the enumerators in the Serbian areas. It is quite possible that all three hypotheses apply. That said, the size of the increase in the poverty rate appears improbably high because neither the conventionally accepted levels of out-migration of Serbs from Kosovo nor their welfare ranking in Kosovo would support these numbers. Furthermore labor force status and wages in predominantly Serb areas do not appear to be very different (see below). Figure b3.1: Unemployment and Labor Table b3.1: Poverty Rates, Labor Force Status and Force Participation Rates by Ethnicity Wages of the Serbian Population 2003/04 2005/06 Serbian population only 2003/04 2005/06 50 eta 47 Poverty rate of the unemployed 43.8 84.9 R Poverty rate of the employed 40 42 dna noi 40 30.0 77.0 38 38 36 Unemployment rate of the poor 40.8 30.3 nte my atpicirt 30 Unemployment rate of the non- 26 Pa plo 24 24 poor 27.4 20.6 e 20 21 emn U rcoFr 20 19 bo 2003/04 2005/06 10 Wages (in 2002 prices) La Albanian 215.2 215.6 0 Albanian Other Albanian Other Serbian 241.8 181.2 Serbian Serbian Other 181.6 182.0 Unemployment Labor Force Participation Source: World Bank staff calculations from HBS Source: World Bank staff calculations from HBS data. Notes: data. Notes: Self-reported unemployment. Self-reported ethnicity and unemployment. 3.8 Self-employed agricultural households face average poverty rates but are the third biggest contributors to poverty. About 40 percent of households headed by self-employed agricultural workers are poor, which makes the poverty incidence in this group slightly below the average for the country. However, this group is quite large as it makes up over 10 percent of all poor (Table B.51). C. EDUCATIONAL ATTAINMENT AND POVERTY INCIDENCE 3.9 Poverty incidence declines rapidly with higher education of the household head. About 43 percent of heads of households have completed secondary school or higher. The poverty incidence is highest among those living in households with heads who have not completed primary school (Table B.10). Most of the poor, however, have either primary or secondary education (Table B.52). The poverty incidence falls sharply for households headed by university educated individuals, down to about 20 percent. 3.10 As expected, the more educated have lower incidence of poverty because they have better employment prospects and better pay. Around 70 percent of people with vocational and tertiary education are salaried employees. By comparison, only 27.7 percent of individuals with secondary education have such jobs, while 41.1 percent report being unemployed. Also a higher 19 proportion of secondary educated individuals report working as "per diem" or "other" workers (Table 3.3). There are also differences in pay. Between 2002 and 2005, real wages of salaried employees have remained steady while those for "per diem" and "other" workers have declined (Table 1.4). Table 3.3: Employment and Education, 2005/06 Uncompleted Primary Secondary Vocational Tertiary Primary Employer 0.3 0.1 1.4 1.1 3.6 Salaried employee 3 6.2 27.7 70.6 69 Subsistence farmer 3.6 7 5.6 2.4 1.1 Per-diem worker 1.8 3.9 4.3 1.6 1.3 Unemployed 16.9 32.1 41.1 15.1 16.6 Housekeeper 55.2 43.8 10.3 2.6 1 Other 19.2 6.8 9.7 6.6 7.3 Total 100 100 100 100 100 Source: World Bank staff calculations from HBS data. Weighted figures for 15-64 year-olds. Students excluded from the calculations. D. INCIDENCE OF POVERTY ACROSS SPACE 3.11 Poverty incidence varies widely across space. In addition to ethnicity, the data is stratified into 7 regions and urban and rural. We use the 7 regions defined in the surveys to obtain the incidence of poverty across space. In 2003, the incidence of poverty was highest in Mitrovica, followed by Ferizaji, Gjakove and Prizreni (Table 3.4). Table 3.4: Poverty Rates and Contribution by Region in Kosovo, 2003-2006 The poverty incidence in all four was Poverty Distribution of the higher than the national headcount. headcount rate poor By 2005, only Mitrovica (with a large 2003-04 2005-06 2003-04 2005-06 Serbian population) and Ferizaji had Gjakova 48.9 45.3 11.5 12.8 poverty incidence higher than the Cjilani 32.5 23.5 7.9 5.6 national headcount. Mitrovica has a Mitrovica 59 69.7 22.7 22.6 larger share of rural population (over Peja 37.8 40.1 9.6 10.2 70 percent ) and larger fraction of Prizreni 48.3 40.5 15.3 15.8 non-Albanian ethnic groups, such as Prishtina 34.3 40.6 22.5 19.8 Serbs (over 10 percent) (Table B.53 Ferizaji 49.8 54.4 10.5 13.3 and Table B.54). Pristina, which had one of the lowest incidences of Total 43.7 45 100 100 poverty, experienced an increase in Source: World Bank staff calculations from HBS data. poverty during the period. This suggests either that a larger proportion of the poor in other urban or rural areas are drawn to the region or that its non-migrant population did not do well during the period. 3.12 The HBS data asks respondents to report members of the household who were born outside their municipality of residence (a proxy for migration). It does not ask how long ago they moved to their current residence or which municipality they moved from. So while the available information does not allow a clear distinction between these processes, it can provide the more likely one. Table B.47 shows the estimated share of the resident population in each municipality, the share of the population that reported being born outside their current municipality and the distribution of the poor with a member born outside the municipality. Mitrovica, Prizren, Peja 20 and Pristina have the highest shares of the population born outside their current municipality. In other words, these four municipalities received about 3 out of every 4 "internal migrant". But when we look at the poverty status of the "migrant" population and where they live, we find that one-third lives in Mitrovica, about a quarter live in Prizren, and an additional 16 percent live in Ferizaj. Only 7 percent of the "migrant" poor live in Pristina. 3.13 The lack of overall progress in poverty reduction masks the divergence between rural and urban areas. In 2003, rural and urban poverty rates were about the same: 44 and 42 percent respectively. By 2005, urban Figure 3.2: Rural and Urban Poverty Trends poverty had declined by 5 percentage points, while rural poverty had increased by 50 49.2 a similar magnitude (Figure 3.2). As a 44.2 40 42.1 result, more than two-thirds of all the poor 37.4 live in rural areas, and this share has onita ul 30 declined only slightly over the period. opp eht of 20 3.14 Individuals without productive % 18.1 15.6 agricultural assets have the highest 14.0 10 12.5 estimated incidence of poverty in the rural population. In 2002 and again in 2005, 0 2003/04 2005/06 2003/04 2005/06 households in rural areas were asked to Rural Urban report whether they owned any land. In Absolute poverty Extreme Poverty between these two years, no such information is available, so it is not possible Source: World Bank staff calculations from HBS data. to track changes in the incidence of poverty between 2002 and 2005 because the consumption data for these two years is not comparable. Instead, we report the poverty incidence for 2005. About 10 percent of households in rural areas reported being landless and we find that nearly 7 of every 10 households in this group are classified as poor in 2005 (Table B.13). In addition, those who reported having no livestock, or agricultural equipment such as tractors, ploughs or trailers, also exhibit higher than average incidence of poverty (Table B.14 and Table B.15). Figure 3.3: Rural Access to Services A. Electricity B. Water C. District central heating 4.5 101 100 100 4 99 80 3.5 98 60 3 97 2003/04 2003/04 96 2.5 2003/04 2005/06 40 2005/06 95 2 2005/06 94 20 1.5 93 92 1 0 t l t l t 0.5 es eli ilet eli es leit onai nt inu 3 hesci onai ilet ilet nt Poor nat qui Q R nat qui oroP naloi nat inuq inu 3 tsehc naloi Q Ri nat inuq 0 Total Rural Rural Poor Source: World Bank staff calculations from HBS data. 3.15 On the non-income dimensions of welfare, rural populations have worse outcomes in access to indoor water tap and central heating. Access to indoor water tap appears to have improved in rural areas, but it is still comparatively lower than urban areas. Figure 3.3 (panel B) shows that access to safe water in rural areas has increased for all quintiles between 2003 and 2005. While this is a positive development, inequality of access between the richest and the poorest quintiles are large. Only 60 percent of the poorest quintile have indoor water tap 21 compared to 80 percent of the richest quintile. Moreover, these rates are still lower than those in urban areas. On central heating, less than 10 percent of the rural population report living in a dwelling with central heating, while in urban areas, there is near universal access. The gap in secondary enrollment rates between rural and urban areas was about 20 percentage points in 2003/04 and fell down to about 10 in 2005/06. For tertiary education, this differential was about 15percent in 2003/04 and 10 in 2005/06. Figure 3.4: Rural net enrollment rates A. Secondary B. Tertiary 90 30 80 25 70 60 20 2003/04 50 2003/04 15 40 2005/06 2005/06 30 10 20 5 10 0 0 ntile le 3 le 4 ntile 2 intile 3 4 orest qui intile 2 inti t quintile Qu Qu Quinti t qui Qu Quintile es Po Riches orest quintileQuintile Po Rich Source: World Bank staff calculations from HBS data. E. WHY ARE PEOPLE POOR AND WHO IS AT HIGH RISK OF POVERTY? 3.16 The preceding discussion looked at poverty trends and profile of the poor between 2003 and 2005. There are two reasons to extend the analysis beyond a look at trends and profiles. First, by definition the poverty profile is a simple correlation between an observable characteristic and poverty status. These correlations do not tell us the independent effect of the observable characteristic that is correlated with poverty status. As an example, a high correlation between poverty and primary education, often does not tell us how much of that correlation is due to the fact that those who have only primary education are also likely to be more unemployed or, even if employed, they are likely to receive lower wages. Therefore, there is a need to understand the link between an observable characteristic and poverty status, when the impact of all the other variables has been "netted" out. 3.17 The second reason to extend the analysis is that a documentation of trends and profiles tells us about what happened in the past, but not what is likely to happen in the future. Past patterns, of course, provide useful information for what to expect in the future. Nonetheless, poverty reducing policies are more interested in what happens in the future. For instance, what happens to the poverty outcomes in the population if certain variables (say fertility, educational attainment, etc.) change from current patterns? Such a thought experiment is all the more interesting, especially if it can be influenced by policy making. The two extensions strengthen the policy content of the analysis of profiles. 3.18 In this section, we extend the preceding analysis in this direction. First we estimate a consumption model in order to understand the magnitude of the consumption shortfall for households with specific characteristics. This first stage highlights the variables that explain the observed differences in consumption. The multivariate nature of the model means that we can infer the size of the shortfall that is attributed to the specific variable of interest. In the second step, we estimate the probability that a household with such observed characteristics will fall into 22 poverty, by taking into consideration their predicted consumption and variance of their consumption (that is, from the unexplained part of consumption). The third step is to select some variables of policy interest, change their values and predict the likely magnitude of the change in the poverty incidence (for a complete discussion of the methods, see El-laithy, Lokshin and Banerji, 2003). The results of the multivariate poverty profile are consistent with the preceding results. Table B.49 presents the results of the consumption model separately for each year. 3.19 First, the key demographic variables appear to be household structure and labor market fortunes of the household members. Households with larger dependents have about 10 percent less consumption, and this gap has been steady during the period, while the households with more unemployed members have 4 to 8 percent less consumption. On the other hand, households with a female head do not appear to have measurably less consumption than male headed households, once we control for other observable characteristics. Finally, households with Serb heads of households appear to report less consumption in the most recent years, suggesting perhaps that their situations might have eroded (but Box 3.1). 3.20 Second, the measured link between education and consumption are in line with the results from the poverty profile. To look at the effect of education on consumption, we use the highest education attained in the household rather than the head of the household head because we find that the former explains the condition of the household better. The model uses secondary education as the comparison group. The results show that all households whose highest education attained is primary or less have at least 12 percent less consumption, while those whose highest education attained is vocational or tertiary have at least 12 percent more consumption. 3.21 Third, the poverty profile indicated that per diem workers have one of the highest incidences of poverty. The results in Table B.49 confirm that households whose main source of income is "per diem" have about 33 percent less consumption than households whose main source of income is remittances. In addition, we find that those whose main source of income is social assistance have one of the highest consumption shortfall compared to those relying on remittances. 3.22 Finally, once we control for location, human capital and demographic characteristics of households, we find that urban households do not have measurably higher consumption than rural households. However, the results confirm that urban households have done better over time as we noted in the evolution of poverty. In 2003, urban households had about 11 percent less consumption than rural households, which has disappeared by 2005. Regionally, Mitrovica, Gjakove and Ferizaji have 18 to 20 percent less consumption than Pristina, and Gjilani has 20 percent more. 3.23 In the second stage of our extended analysis, we use the predicted consumption from the above consumption model together with the unexplained part of consumption (the error term) to obtain the probability of falling into poverty for each household. We can then average these probabilities across groups (say primary educated, rural, etc.) to obtain average probability of being poor for that group. In effect this becomes the average risk of falling into poverty for a group. Then we can use the same model to see how a change in a policy variable changes the risk of falling into poverty. Table 3.5 summarizes the results using the 2005 data. 3.24 The first column of Table 3.5 shows a few policy experiments. For demographic variables, one question to ask is what would happen to poverty risk if female headed households had the same opportunities as male heads of households. Another one attempts to predict the change in poverty risk if the number of dependents were reduced in the households where 23 dependency ratio is higher than the median. Since there is universal primary education, the more interesting education policy questions focus on the changes in poverty risk if secondary education became universal or available to the poorest quintile. The final set of questions looks at how poverty risk might change if rural populations received the same opportunities as urban and all the regions received the opportunities found in Pristina. A discussion of the specific policies that would be required to realize these thought experiments is beyond the scope of this report. It is enough to think about what is the likely impact on poverty risk if these goals were to be achieved. Table 3.5: Impact of Changes in Household Characteristics on Poverty (in percentage points) Absolute Predicted Change in Extreme Predicted Change in poverty probabilit predicted poverty probabilit extreme rate y of being probabilit rate y of being poverty absolute y of being extreme rate poor absolute poor poor Demographic characteristics Female headed households to 49 44.6 -0.9 18.4 20.6 -0.6 male headed households Dependency ratio greater than 49.1 47.4 -4.3 18.9 47.4 -3 median to median Dependency ratio greater than 49.1 47.4 -7 18.9 21.2 -4.8 median reduced by 25 percent Education (max education in household) Change Primary to secondary 64.2 62.2 -10.1 28.5 33.2 -8.4 Change to Secondary for 100 67.9 -2.6 83.5 39.1 -3.3 poorest quintile Change Primary to tertiary 64.2 62.2 -27.4 28.5 33.2 -19.9 Labor market characteristics One less unemployed in HH 47.7 48.3 -1.3 18.7 21.5 -0.9 10% less unemployment 47.7 48.3 -5.9 18.7 21.5 -3.9 Spatial dimension (urban and Pristina are reference) Rural 49.2 47.9 -1.5 18.1 21.4 -1 Gjakova 45.3 46.7 -13.2 13.5 19 -7.4 Gjilani 23.5 24.1 14.1 5.4 8.1 7.3 Mitrovica 69.7 65.3 -16.4 33.5 35.3 -13.7 Peja 40.1 37 -3 18.7 13.6 -1.6 Prizreni 40.5 42.9 -2.2 10.2 16.3 -1.3 Ferizaji 54.4 53.8 -14.7 21.6 25.5 -9.8 Source: World Bank staff calculations from HBS data. Note that the change in extreme poverty rate for households with highest education of "None, cannot read/write" is higher than the observed poverty rate because the change = predicted poverty risk ­ simulated poverty risk, and the predicted in this case is higher than the observed. 3.25 The second column of Table 3.5 is the poverty incidence for the group targeted by the policy change, as computed from the poverty profile. This can be thought of as an empirical probability of poverty for the group. Column 3 is the predicted probability using the consumption model, and as the results show there is close similarity between the two. Column 4 shows what would happen to the predicted probability of being poor if the policy changes in column 1 were achieved, using the absolute poverty line. The last three columns do the same for extreme poverty. 24 3.26 Policies that improve opportunities in lagging regions, improve access to secondary education, and create employment have the largest impact in reducing poverty risk. The risk of poverty would decline by about 15 percentage points if three regions, Gjakove, Mitrovica, and Ferizaji, who together comprise 40 percent of the population, were to close their gap with Pristina. This would be the same as reducing the national level poverty rate in 2005 from 45 to 36 (that is 45-0.4*15). Were all primary educated individuals to receive secondary education, the poverty risk for that group would be reduced by 10 percentage points. The decrease would be even larger if the same group were to receive tertiary education. However, no significant reduction in poverty risk is predicted if the poorest quintile received secondary education. This does not mean that secondary education is not important or relevant for this group. It simply means that under the specific conditions, other interventions may have more impact. A 10 percent reduction in the number of the unemployed in every household would reduce the poverty risk by 6 percentage points. 3.27 In conclusion, this chapter has shown that the poor tend to be concentrated in households that are large, are elderly, have more dependents, more unemployed, and located in rural areas and, regionally, in Mitrovica and Ferizaji. The chapter also looked at determinants of poverty risk, or the factors that increase the likelihood of falling into poverty. It found that labor market success of household members, demographic structure (dependency ratio), education and region of residence are preeminent factors. In particular, the policy experiments indicate that policies that generate employment, improve opportunities in lagging regions and raise the educational attainment of the population have the largest effects on reducing the poverty risk. 25 CHAPTER 4: PUBLIC TRANSFERS, REMITTANCES AND POVERTY The social protection programs in Kosovo have low coverage, but its social assistance program, the explicitly anti-poverty program, is well-targeted. However, the modest size of the programs and the flat benefits per recipient households over time has resulted in only a modest impact on poverty. The results for private transfers are quite different. First, compared to the coverage of social assistance programs, substantially more households have migrants and receive remittances. Second, the impact of migration and remittances on welfare has been large. In rural areas, estimated differences in poverty outcomes between households with migrants and those without is almost 20 percentage points. However, migration might also have contributed to observed inequality in rural areas. 4.1 The discussion thus far has highlighted that about 45 percent of the population in Kosovo lived in poverty between 2003 and 2005, and some 15 percent were extremely poor. A significant fraction remains vulnerable to the slightest of economic downturns. Furthermore, over time, while the same fraction continued to be poor, inequality has widened. In this context, an affordable social protection system that has a wide reach and is well targeted can become an essential instrument to help the poor and vulnerable populations shield themselves from severe hardships. A. COMPOSITION AND TRENDS OF SOCIAL PROTECTION TRANSFERS 4.2 The current social protection system comprises a three-pillar basic pension system, special schemes for war invalids and their next of kin, early retirement for miners to accelerate restructuring of the sector, disability benefits, social assistance benefits, employment assistance programs, and direct and indirect subsidies for narrowly defined vulnerable groups (World Bank, 2006a). 4.3 The basic pension is a flat-rate monthly benefit available to all residents of Kosovo and Kosovar refugees living outside the territory, aged 65 and over, irrespective of prior contributions. The benefit is set to equal the extreme poverty line (the food basket) and it is adjusted annually to reflect changes in costs. Pensions are not targeted to poor families. The social assistance benefit is narrowly targeted on a subset of poor and extremely poor households. It was introduced in 2002 and revised in December 2003. Those eligible fall into two categories: (1) families without resources where no one is capable of work, or expected to make themselves available for work (single mothers, children); and (2) families with at least one child under 5, or caring for an orphan under 15 years. Under the second category, additional members of the household who are capable of working are not eligible as they are required to register as unemployed. Furthermore, eligible families cannot possess income generating assets that exceed 0.5 hectares of land. Also, if eligible families have outside sources of cash that exceed the social assistance benefit then they receive no benefits, but if the outside source is less than social assistance, they receive the difference (World Bank, 2006a). 4.4 The HBS data used in this report does not distinguish the different types of pension or social assistance programs. However, it is important to note that over 75 percent of the social protection system is devoted to basic pension (three-pillars) and social assistance benefits. Although the early retirement program for the mining sector is expected to expand substantially compared to its level in 2003, it is still only a small share of the aggregate social protection system. Therefore, it must be kept in mind that most of the conclusions given here probably apply to the two largest programs. Table 4.1 reports the evolution of the number of recipients and total benefits for the pension and social assistance programs from administrative and survey data. The two data sources tell the same story. Table 4.1: Pension and Social Assistance Programs, Official and HBS Estimated Number of Recipients and Total Value Disbursed 1. Number of recipients: 2002 2003 2004 2005 2006 2007 A: Pension 1 Official figures 116,387 143,045 158,600 169,000 171,400 HBS estimates 68,307 115,871 111,200 127,742 B: Social assistance 2 Official figures 52,329 50,724 46,441 HBS estimates 28,917 36,088 45,420 43,356 2. Benefits: Total disbursement 3 A: Pension Official figures 49,495 61,572 84,711 86,268 89,450 HBS estimates 67,799 68,277 70,052 72,788 B: Social Assistance Official figures 32,293 32,217 32,480 HBS estimates 19,615 26,323 34,880 33,048 Source: MLSW figures from PEIR (2006) and World Bank staff calculations from HBS 2002/2006. Notes: 1 Number of individuals. 2 Number of households. 3 In thousands of Euros. Notes:Official (MLSW) figures for 2002-2004 are actuals and predicted for 2005-2007. The HBS estimated payout is in real 2002 Euro prices. HBS pensions refers to old-age, disability and war invalid pensions. 4.5 First, the social protection system expanded at a fast pace, but the expansion may be slowing down. Both data sources indicate a sharp rise in the number of recipients between 2003 and 2006, and then a slowdown thereafter. The administrative data shows that the number of pension recipients expanded at 6 percent per year between 2004 and 2006, while the HBS estimates a 3 percent annual growth Figure 4.1: Estimated Annual Benefits per Recipient during the same period. By contrast, the 993 number of social assistance recipients is 00 10 estimated to have declined from about 52,000 to about 46,000 when using 0 administrative records. The HBS also 80 768 762 729 estimates about 46,000 recipients of social 678 assistance in the most recent year, which 699 630 0 589 635 may be the result of an earlier expansion. 60 617 570 534 510 522 4.6 Second, the value of benefits per 425 430 0 recipient household has remained flat 40 2002 2003 2004 2005 2006 2007 over time. Figure 4.1 shows trends in the Year estimated benefit per recipient household. Pensions MLSW HBS Again, except perhaps for the 2003 Social Assistance MLSW HBS pension estimate, the administrative and Source: MLSW figures: World Bank (2006a). HBS: World the survey data show similar benefit levels Bank staff calculations from HBS data. and trends. In general, pension benefits 28 per recipient household appear to be lower than social assistance benefits per recipient household. Moreover, both programs show no growth in benefits per recipient household between 2003 and 2006. 4.7 Finally, Kosovo's social protection spending is lower than similar public programs in the region. Figure 4.2 shows the share of social protection programs in per cent of GDP and total government expenditures in some Figure 4.2: Social Protection Expenditures, selected Southeast European countries and high countries: 2004 growth middle income countries from other regions. Whether one chooses to 13.8 Ireland, 2001 look at per cent of GDP or share of 40.9 7.6 Chile, 2003 government spending, Kosovo's social 35.1 16.8 protection programs are the lowest. This Croatia 33.7 11.1 reflects both economic and fiscal FYR Macedonia 30.7 15.9 constraints. The revenue base is limited Bosnia & Herzegovina 28.7 by low growth and borrowing options are 7.5 Albania 25.8 limited. 3.7 Kosovo 12.7 0.6 Thailand, 2001 3.4 4.8 The evolution of the social 0 10 20 30 40 protection spending has to be seen in the % of Government context of the recent history of Kosovo. % of GDP Expenditures The initial expansion was motivated by Source: World Bank, 2006a, and World Bank ECA public the size of the population made vulnerable expenditure database. Figures for Bosnia and Herzegovina and by conflict and economic disruption. On Croatia are based on offical GDP estimates as of March 21, pension alone, there are special programs 2007. for war invalids, disabled (no doubt partly due to conflict), and early retires for mining. There is evidence that recipients list of the disability pension has expanded sharply since 2004 (World Bank, 2006a). However, at present and in the near future, while the need for social protection may be great given the poverty outcomes, expansion will be constrained by a tight budget envelope. This means that existing social protection programs will have to be well targeted and efficient in order to have a larger impact on poverty outcomes. B. IMPACT OF SOCIAL PROTECTION PROGRAMS ON POVERTY 4.9 A key feature of the social protection programs is their low coverage. The top panel of Table 4.2 summarizes the estimated fraction of the population that receives pensions and social assistance. Social assistance covers about 13 percent of the population and this has not changed over time. Furthermore, equal fractions of the urban and rural populations are covered. The program covers about 33 percent of the poorest fifth of the population, and additional 16 percent in the second poorest quintile. The pattern of coverage across quintiles means that only about 23 percent of the poor and about 34 percent of the extreme poor population is reached by the social assistance program. The different types of pensions benefit about 6 percent of the population. The distribution of pension coverage across urban/rural and quintiles is also similar, as it should be, since there are multiple criteria for pension eligibility. 4.10 And very good targeting of the social assistance program. The bottom panel of Table 4.2 shows the distribution of the recipients, that is, the location of the recipient population across space (urban/rural) and welfare ranking (consumption quintiles). First, note that there is no bias towards urban or rural populations. The fractions of the recipients in rural and urban areas are in line with rural and urban population shares. Since over two-thirds of the poor live in rural areas, they are also in line with the distribution of the poor. The leakage of funds used to be only a bit 29 worse in rural areas ­ in 2003/04 about 23 percent of the social assistance recipients were non- poor compared to 11 percent in urban areas--but is currently at par with urban at about 22 percent (Figure D.1). Second, nearly 70 percent of the recipients are poor, and only at most 6 percent are in the richest quintile. Third, given that a large fraction of the population is vulnerable, the results in the table suggest that almost 90 percent of the recipients are either poor or vulnerable, which suggest an excellent targeting effectiveness. Moreover, because the welfare ranking is done on consumption that does not net out the social assistance benefit used for consumption, the targeting effectiveness is probably even better than reported here. Table 4.2: Social Assistance and Pensions: Coverage and Incidence by Urban/Rural and Quintile, percent of individuals Social Assistance Pensions 2003/04 2005/06 2003/04 2005/06 Coverage (% of population) 11.2 13.2 5.8 6.4 Urban 9.7 11.1 5.7 6.3 Rural 11.9 14.4 5.8 6.4 Poorest quintile 28.9 32.6 4.5 7.1 Quintile 2 13.2 16.2 7.1 6 Quintile 3 9.2 9.6 5.8 6.5 Quintile 4 3.3 3.8 6.7 6 Richest quintile 1.3 3.9 5.1 6.4 Beneficiary incidence (distribution of 100 100 100 100 recipients) Urban 30.4 29.3 34.4 34.5 Rural 69.6 70.7 65.6 65.5 Poorest quintile 51.7 49.2 15.7 22.2 Quintile 2 23.6 24.5 23.9 18.9 Quintile 3 16.5 14.5 19.5 20.3 Quintile 4 5.9 5.8 23 18.8 Richest quintile 2.3 6 18 19.9 Source: World Bank staff calculations from HBS data. Notes: Pensions include all types of pensions (basic, disability, war invalid, etc). 4.11 The modest size of the benefit levels and low coverage of the social protection programs suggests that the social protection programs have had low impact on well-being. In the absence of the social assistance, poverty would be higher by about 2 percentage points. It would be even higher in the absence of Table 4.3: Adequacy and Simulations. pensions, by about 4 Social assistance Pensions percentage points (Table Adequacy 2003/04 2005/06 2003/04 2005/06 4.3). Assuming that there is Monthly transfer 60.8 63.5 64.4 63 no overlap between the Poverty rates pension and social Current/post-transfer state assistance recipients, the Absolute poverty rate 43.5 45.1 43.5 45.1 simulations suggest that Extreme poverty rate 13.6 16.7 13.6 16.7 poverty was reduced by Pre-transfer simulation about 6 percentage points Absolute poverty rate 44.9 46.5 47.6 49.6 (or 14 percent). Similarly, Extreme poverty rate 17.2 20 18.6 21.7 in the absence of social Source: World Bank staff calculations from HBS data. Notes: 100% marginal assistance or pensions, propensity to consume from transfer assumed. Real 2002 Euros. Population- extreme poverty would have weighted. 30 been higher by 4 and 5 percentage points, respectively (about 40 percent). Between the two programs, the larger impact on poverty reduction has come from pensions. The simulation of poverty rates with and without social protection programs assumes that the programs have had no influence on the decisions of the participating households, especially with regard to labor supply. Accounting for these behavioral changes would complicate the analysis, and is beyond the scope of this report. It would probably lead to even lower impact of the programs than stated here. However, the quick simulation is useful to gauge the potential size of the impact, even if they may reflect the highest possible impact (the upper bound). Figure 4.3: Targeting Performance of the Kosovo Social Assistance Program B. Percent of Funds going to the Poorest quintile A. Undercoverage and Leakage of Social in Low-Income Countries (LIC) and Middle- Assistance Income Countries (MIC) Undercoverage Kosovo 2003 52 Kosovo 2005 47 80 79.3 77.0 Albania 2002 41 LIC Kyrgyz Rep 37 66.9 65.7 Armenia 30 60 LIC 27 Romania 64 Leakage 40 to the non-poor Lithuania 60 MIC Bulgaria 58 Albania 2005 52 20 21.8 MIC 27 19.5 0 20 40 60 0 % of funds going to the poorest quintile Poor Extreme Poor Non-poor 2003/04 2005/06 Source: World Bank staff calculations from HBS data. Source: World Bank staff calculations from HBS data Notes: Undercoverage is the percent of individuals that are and World Bank (2005). poor but not covered. Leakage is the percent of individual recipients that are not poor. C. SIZE AND DISTRIBUTION OF REMITTANCES 4.12 The formal social protection programs are complemented, and often dwarfed by the private transfers. While these take many forms ­ in-kind and informal transfers for mutual insurance between households, social work services by NGOs, kinship based support networks, and so on ­ remittances are some of the largest and most widespread of these transfers. Kosovo is the 3rd highest remittance recipient in the Western Balkans, when the rank is measured as share of remittances in GDP (Figure 4.4). It is the 11th highest in the world (Figure D.2). Because nearly all of these remittances flow to households, we would expect that they would have significant impact on household welfare. Although HBS does not collect detailed information on migration and remittances on a routine basis, it fielded a module on the extent of external migration and the value of remittances received by households in Kosovo in 2005. This section draws on these data to estimate the size of remittances and their impact on poverty. 4.13 Nearly a quarter of Kosovars have migrants abroad. The number of international migrants is estimated at about 400,000 individuals, and is in line with independent estimates of the size of migrants from Kosovo in OECD countries. Using census and survey data collected from all OECD destination countries, Docquier and Marfouk (2006) estimated the population of 31 migrants from Serbia, Montenegro and Kosovo at about 2 million. Assuming the rates of migration are the same, then apportioning this total according to shares of total population in Serbia, Montenegro and Kosovo suggests 400,000 migrants from Kosovo2. The majority of the migrants come from rural areas. For instance, while 1 in 3 (30 percent) of all households with international migrants live in urban areas, the remaining 70 live in rural areas. Regionally, almost 1 in 5 households with international migrants are residents in Mitrovica, Prizreni and Gjakove (Table 4.4). Figure 4.4: Remittances as a share of GDP in the Western Balkans Bosnia and Herzegovina Serbia and Montenegro Kosovo Albania Croatia Macedonia, FYR 0% 5% 10% 15% 20% 25% Source: Global Economic Prospects 2006: Economic Implications of Remittances and Migration, World Bank. 4.14 The share of the Table 4.4: Migration and Remittances, 2005 population which receives All Migrant Remittances remittances is substantially higher than the fraction Poverty rate 37.2 30.4 29.8 receiving social assistance. As % of population 100 25.9 21.4 noted in Table 4.2, about 13 Urban/rural Distribution percent of the population was Urban 36.2 28.7 27.6 covered by the social assistance Rural 63.8 71.3 72.4 program. By comparison, an Regional Distribution estimated 20 percent received Gjakova 11.5 17.1 15.5 remittances (Table 4.4). The Cjilani 12 10.2 12.4 spatial distribution of recipient Mitrovica 15.1 20.6 20.4 population mimics the pattern Peja 11.2 13.2 11.3 observed for sources of migrants: Prizreni 15.7 18.1 20.1 over 70 percent of the recipients Prishtina 23.3 13.2 12.2 live in rural areas, and Mitrovica Ferizaji 11.2 7.6 8.1 and Prizreni report having the Total 100 100 100 highest fraction of population Source: World Bank staff calculations from HBS 2005 population- weighted data. Notes: 1In 2002 Euros and per adult equivalent. receiving remittance. 2 Population of Serbia, Montenegro and Kosovo is 8 million, 0.7 million and 2 million respectively, for a total of 10.7 million. This means that Kosovo's share is about 19 percent of the total population (=(2/10.7)*100). So its migrant share is 0.19*2 million=380,000. 32 D. IMPACT OF MIGRATION ON POVERTY 4.15 Households with migrants or receiving remittances do not appear to demand more education. Judging from the comparison of enrollment rates, there is no evidence that remittances are used for investment in education. There is a large observed difference in enrollment rates between general population and sub-populations with migrants or receiving remittances for urban areas, but none in rural areas. Net enrollments are about 10 percent lower for sub-populations with migrants or receiving remittances in urban areas (Figure D.1). Separate enrollment rates for male and female children also confirm lower rates for sub-populations with migrants or receiving remittances in urban areas (Figure 4.5). However, by comparing the general urban population to the sub-population of urban residents with migrants, the raw differences may lead to the misleading conclusion that having a migrant reduces enrollment rates. Doing a more careful comparison, that is comparing urban sub-populations with migrants to urban sub-population without migrants but who have the same or nearly the same observable characteristics, shows no difference in enrollment rates (Table 4.5). Figure 4.5: Enrollment Rates for Households with a Migrant and without a Migrant, 2005 Secondary Tertiary 86.7 25 80 23.3 et 71.4 eta 20 Ratn 60 64.2 Rtn 61.2 18.1 llme mellornEt 15 ron 40 13.2 Et 11.9 10 Ne Ne ry 20 5 iarteT 0 Rural Urban 0 Non-migrant HH Migrant HH Rural Urban Non-migrant HH Migrant HH Source: World Bank staff estimates from HBS data. 4.16 Households with migrants have higher consumption. The median value of remittances is 2000 Euros per year, and the mean is higher at 2600 Euros. The average value of remittances to rural areas is about 2800 Euros, compared to about 1500 Euros for urban areas Table 4.6, (panel A). These values are about 3 times higher than the average values from social protection programs to recipient households shown in Figure 4.1. Moreover, self-reported use of remittances shows that the latter are overwhelmingly used for consumption. The survey data indicates that sub-populations with migrants or receiving remittances report 9 percent more consumption per adult equivalent per month, but no higher consumption of food (Table 4.4). Table 4.5: Propensity Score Matching Results for Secondary Enrollment Rates Outcome: Secondary Net Enrollment Rates. Difference T-stat Urban Unmatched -0.09 -3.19 Average Treatment effect for the Treated -0.04 -0.93 (ATT) Rural Unmatched -0.03 -1.07 Average Treatment effect for the Treated -0.01 -0.12 (ATT) Source: World Bank staff calculations from HBS data for 2005. Propensity score method used is single nearest member. 33 4.17 Comparing the mean differences in consumption between the sub-population with migrants and the general population is not necessarily an accurate indication that having a migrant is what brought about the difference. It is quite possible that households with migrants would have had higher consumption anyway, perhaps because there is something about these households that makes them successful, which survey data does not collect. So one way to estimate the true difference in consumption is to compare the sub-population with migrants with a sub-population that does not have migrants but has the same observable characteristics. The estimation has two stages. In the first stage, one estimates the probability of having a migrant abroad on the basis of observable characteristics for that household including education, location, demographic structure, and so on in order to obtain the propensity to migrate (that is, a propensity score). In the second stage, a household with a migrant abroad is matched with another household without a migrant abroad but has the same probability of having a migrant abroad (or very close if there is no exact match). 4.18 Once the matches are done for all households with migrants abroad, the mean difference in consumption or any outcome of interest is calculated. The estimation strategy can be done separately for rural and urban as is done here for Kosovo. Figure D.4 shows the quality of matches for Kosovo, and urban and rural sub-populations. The results show that for migrant households predicted to have a very high propensity to migrate, those with propensity score of 50 percent or higher, there are not very good matches from the sub-population without migrants. For those predicted to have a low propensity to migrate and do migrate, there are very good matches. Table 4.6: Remittances from Abroad in 2005 by Recipient Households, cash and in-kind for the last 12 months A. Mean Remittances and their use B. Mean Remittances by quintiles in urban and rural areas Remittances Mean 2603.2 soru 00 E 40 Median 2000 5002 Urban 2179.4 ni 2977 n 00 Rural 2805.7 30 ea 2666 Poor 1487.1 2140 Extreme Poor 1341.6 nt,muo m 00 1889 A 20 1710 Remittance All or major Minor 1487 Use part part ecnatti 1277 1217 Consumption 82.5 10.6 me 00 978 R 10 Durables 1.7 0.9 1 2 3 4 5 National Income Quintile Home 5.4 3.1 construction Total Urban Rural Savings 2.5 13.6 Other 3.3 7.3 Source: World Bank staff calculations from HBS data. 4.19 The matched results confirm a statistically significant difference in consumption between households with migrants and those without. The results suggests that the difference in consumption between those with migrants and those without for urban population is about 1.2 Euro per adult equivalent per month, which for a household with average size, means an additional 8 Euros per month more consumption, or a gain in consumption that is 25 percent of the extreme poverty line. The differences in rural areas are slightly larger (Table D.4). 4.20 Instead of matching, one can look at the differences in consumption using regression analysis, still accounting for the probability of having a migrant. The process is still two-stage: 34 first the probability of having a migrant abroad is estimated, then the predicted probability is included in a regression of consumption on all observable characteristics. The results from this exercise (IV-estimation) also confirm that households with migrants consume more (Table D.3). The consumption elasticity of having a migrant is 0.17, which means that a household with a 10 percent higher predicted probability of having a migrant abroad will be estimated to have about 2 percent higher consumption. All these suggest that we should expect to see potentially lower poverty rates, which we do. 4.21 The poverty rate for the sub-population with migrants or which receives remittances is 7 percentage points lower than the general population (Table 4.4). The incidence is 13 percentage points lower in rural areas, while in urban it is 6 percentage points higher in households with a migrant (Table D.2). The large difference in rural areas compared to urban areas could be because rural poor receive higher levels of remittances than urban poor. On average the value of remittances to recipient rural households is much higher than the value to recipient urban households. Remittances to rural recipients are much higher than to urban recipients at every quintile, except the poorest (Table 4.6, panel B). Comparing the poverty rates for matched samples shows that the poverty rates are 20 percentage points lower for rural households with a migrant and not statistically different for urban households with a migrant (Table D.5). 4.22 Finally, remittances may explain partly the growing inequality in the population. In general the value of remittances received by poor households is lower. Table 4.6 (panel B) shows that in rural areas, the richer households received substantially larger remittances than the poorest group. In urban areas, there is less divergence, especially among the lowest four quintiles. This pattern may explain the observed sharp increases in inequality in rural areas compared to urban areas. It may also suggest that poverty could have been reduced more sharply if the poor received larger remittances. 4.23 In conclusion, this chapter has highlighted the low coverage of social protection programs, but effective targeting of the social assistance program to the poor. It shows that the modest size of the programs has meant that overall their impact on poverty has been modest. By contrast, the chapter notes that international migration and remittances from migrants have had a wider reach and have much large impacts on welfare, measured as consumption or poverty outcomes. 35 CHAPTER 5: STRENGTHENING THE FOUNDATION OF POVERTY MONITORING AND EVIDENCE-BASED POLICY MAKING The HBS has laid the foundation for poverty monitoring in Kosovo. However as currently designed it has three weaknesses. First, its level of representativeness is uncertain because there is no recent reliable population census to use as reference. Second, it is inadequate to provide a basis for assessing and monitoring all dimensions of welfare, and third, it suffers from inadequate supervisory oversight. To improve the quality and reliability of the surveys, it is recommended that a census be undertaken, but in the immediate term, it is crucial to better manage the updating of the old frame, maintain the tradition of specialized modules, enlarge the sample size and invest in supervision. 5.1 This report looks at the evolution of poverty in Kosovo in the first half of this decade. It documents that growth in this period was low. After a brief discussion of the prevailing macroeconomic environment, it examined the evolution of the welfare of the population of Kosovo. It reached three main conclusions. First, while the non-income dimensions of welfare had generally favorable outcomes, income (or consumption) measures of welfare showed stagnation. In particular, it shows that consumption growth was minimal, and where it was positive, the benefits went mostly to urban and the richest fifth of the rural population. Second, in addition to widespread and persistent poverty, a large fraction of the population is vulnerable. Third, the pattern of gains and losses has led to an increase in inequality, even though the overall inequality in Kosovo is still considered low. 5.2 The report also documented a profile of the poor, and found that the poor are to be found mostly among the larger households, those with more elderly and dependents, with more unemployed members, and households living in rural areas. Finally, the report looked at the two mechanisms for protecting families against economic hardships ­ social protection programs and private transfers in the form of remittances from migrants. It found that the modest size of the social protection programs has meant that they have had only a modest impact on poverty, while the size and reach of migrant remittances has had much larger and positive impact on the welfare of the population, particularly in rural areas. 5.3 In terms of policy, it argues on the basis of the diagnosis regarding the pattern of poverty and the profile of the poor that policies which focus on sustaining high and inclusive growth, raise rural productivity, and improve the skills of the population will have the largest and most lasting impact. 5.4 At the center of this evidence and conclusions, of course, is the HBS data. The HBS provides a solid foundation for monitoring poverty in Kosovo. It has the elements of a sustainable survey: it is a core survey of SOK, it is fully funded by the government, it has dedicated staff (albeit few) supported by technical staff from development partners, and it has been implemented successful every year since it began. In short, HBS has the potential to be the foundations for monitoring and evaluating policies. Such evaluations can be important for evidence-based policy making. However, there is a cloud that hangs over the findings from these surveys and indeed results from all micro-surveys in Kosovo. To be more specific, there are three weaknesses. 5.5 First, there is an underlying uncertainty regarding the representativeness of the samples. Because there has not been a reliable census in almost a quarter of a century ­ the last complete census was done in 1981 ­ there is no known reference population to which the samples can be cross-checked. Over the last five years SOK staff has been updating 400 of the original 3400 enumeration areas drawn up in 1981, so there is information gained every year. However, the uncertainty remains. 5.6 Second, as designed the HBS is inadequate to provide a basis for assessing and monitoring all dimensions of poverty outcomes. The survey collects detailed information on expenditure, but the questions on the non-income dimensions of poverty are very thin and insufficient. It is difficult to monitor a range of education, health, utilities and housing outcomes. In addition to difficulties of monitoring access to services and specific programs, there is difficulty of evaluating the benefits of specific programs because there is no information on beneficiaries and non-beneficiaries. To their credit, SOK staff has introduced modules on specific topics in each round of the survey. One specific example of such module is the migration module in the 2005 survey that was used for Chapter 4 of this report. But the modules are often under-funded, so that their content is sometimes uninformative, and they are not often linked to big policy questions in Kosovo. In addition, while it is good that the samples are stratified by region, urban, and ethnic areas, and therefore, in principle make it possible to obtain statistics at these strata, the 2400 households sampled every year may be too small. For some levels of disaggregation, the sub-samples may be too small to allow outcomes to be monitored accurately. 5.7 Finally, non-sampling errors are not negligible. The main non-sampling errors may occur on the survey implementation and administration side. There are two constraints that introduce non-sampling errors. First, supervision of field work is compromised by inadequate supervision. To see the scale of the problem, note that there are only three core members of the HBS unit, and one supervisor, at the Pristina office. There are about 40 or so enumerators who work on HBS and other surveys. Shortage of personnel, constant attrition, inadequate funding of supervision and no doubt lack of control of field work in Serb areas because of the existing social tensions, will affect the quality of the information collected. Second, lack of experience managing large and complex surveys also compromises quality. Implementation of complex surveys from inception to analysis is relatively new to SOK and began only 5 years ago. Therefore, it is to be expected that during the learning period, there would be lapses on quality. 5.8 To remedy these problems, six recommendations are proposed: · The single most important and beneficial act would be to undertake a census. The availability of the census would establish a credible frame for future surveys and with some effort, would allow remedying whatever biases may be found in existing surveys. But at present, it is not clear when the census will be done. Meanwhile there is still a need to continue to improve the quality of current surveys. While the preparation and final date for the implementation of the census continue, it may be useful to, · Create a master sample from a smaller list of enumeration areas, as was done in Bosnia and Herzegovina (see Volume II). Currently, an update of the master sample is scheduled for early 2007. Using satelite images, about 5, 000 new enumeration 38 areas are created, of which, about 1, 000 will be listed (the number of households will be estimated upon a visit). · Improve the management of the updating of the existing frame. At present there is no clear documentation on how the re-listing of the 400 selected EAs is done each year or how the information is used and at what stage of the process. Establishing an appropriately staffed sampling/methodology unit could help in this effort in a meaningful way. · Continue the tradition of introducing specialized modules, but they should be linked to the policy process. To build ownership and maintain relevance, it is important that the process of selecting the topics and content of the module be consultative. In particular, the line ministries should be drawn in as partners. · Enlarge the sample size. The current sample is too small to monitor welfare outcomes and undertake analysis of policy impact at lower units of disaggregation, such as rural and urban statistics for each region. A new sample size should take into account (a) cost constraints, (b) desired sampling error and confidence intervals, and (c) human resources constraints (a bigger size survey requires excellent supervision and coordination to minimize non-sampling error). Generally, as a rule of thumb, decreasing the sampling error is inversely proportional to the square of the sample size. For instance, to decrease the sampling error by 2 would require increasing the sample size by 4. Experience from similar surveys in the region suggests doubling the current sample size to around 4, 320-4, 800 but a more carefull assessment is needed. · Invest in supervision. This will minimize non-sampling errors, improve quality of surveys, and enhance the credibility of the results from the surveys. It will also strengthen the survey implementation skills of the supervisory staff. · Maintain current expenditure questions to preserve comparability across years. As presented in Volume II of this report, the changes in the questionnaire design constrains comparability over time. Adhering to the most recent questionnaire, coupled with randomized tests of different modules or questions if changes need to be introduced, will help ensure consistent time trends. 39 ANNEX A: TABLES AND FIGURES Table A.1: Difference in World Bank and IMF assumptions and adjustments to the HBS data. IMF assumptions WB assumptions Demographics Population growth 1.7% 0% Population size 2 mil Urban/rural proportions No adjustment 65% rural Economic aggregates Car purchases, electricity, food None Table A.2: Summary Statistics of Main Aggregates by Survey Wave Survey Wave I II III IV Mean sd Mean sd Mean sd Mean sd Household size 6.77 3.53 6.46 3.27 6.09 2.84 6.09 3.08 Adjusted adult equivalent size 6.49 2.41 6.26 2.31 6.00 2.03 5.97 2.16 of HH HH Average Monthly Consumption (Euro) Total Consumption of HH 370.66 282.39 331.04 210.62 362.22 256.08 314.85 221.10 Total Expenditures of HH 317.31 254.80 279.83 194.45 320.01 245.36 277.85 207.58 Consumption of own produced 53.35 86.32 51.21 75.94 42.21 68.73 37.00 70.10 or fetched food HH Average Monthly Expenditures (Euro) Total Expenditures of HH 317.31 254.80 279.83 194.45 320.01 245.36 277.85 207.58 Food expenditures (incl. alcohol 192.75 116.92 167.34 87.56 171.36 102.58 151.59 90.27 and tobacco) Non-Food expenditures 124.57 176.61 112.49 131.72 148.65 172.00 126.26 138.50 Table A.3: Average Monthly Food Consumption, in Current Euro Survey Wave I II III IV Mean sd Mean sd Mean sd Mean sd Food expenditures (incl. 192.75 116.92 167.34 87.56 171.36 102.58 151.59 90.27 alcohol and tobacco) Bread and cereals 36.47 25.88 35.02 25.42 32.65 22.62 27.02 20.84 Meat 34.27 34.73 26.67 20.16 27.51 23.28 24.83 20.27 Fish 1.45 5.66 1.45 3.02 2.13 7.82 1.65 3.26 Milk, cheese and eggs 23.42 24.72 17.33 17.81 18.31 17.57 15.89 15.47 Oil and fats 10.33 8.37 8.41 6.58 8.31 5.41 7.62 5.80 Fruits 9.38 12.04 8.95 8.85 11.04 12.54 9.10 8.97 Vegetables 20.36 21.09 20.12 17.95 17.52 17.88 15.87 14.64 Sweets (sugar, jam, honey, 11.51 9.91 10.52 8.36 11.48 9.76 9.85 8.28 chocolate and confectionery) Other food products 9.00 8.07 9.34 10.30 9.56 9.60 7.64 6.35 Coffee, tea and cocoa 7.84 5.64 7.12 3.99 6.49 4.00 5.78 3.12 Non-alcoholic beverages 9.13 12.91 9.05 9.28 10.06 10.38 8.40 8.24 Alcoholic beverages 1.86 5.98 1.39 5.02 1.36 4.09 1.24 4.20 Tobacco 17.73 21.24 11.96 14.88 14.95 17.33 16.69 21.69 Table A.4: Household Average Monthly Non-food Expenditures (Euro) Survey Wave I II III IV Mean sd Mean sd Mean sd Mean sd Non-Food expenditures 124.57 176.61 112.49 131.72 148.65 172.00 126.26 138.50 Clothing and footwear 25.52 52.58 21.72 38.13 30.09 45.54 23.40 48.34 Housing and related services 43.80 91.62 38.83 53.88 45.86 63.47 41.22 59.41 Health 7.05 17.83 7.72 22.43 10.10 28.22 8.20 17.04 Transport and communication 29.96 51.29 23.69 35.87 30.90 47.19 28.08 31.33 Recreation and culture 4.39 17.88 3.22 22.85 5.65 32.16 3.13 14.24 Education 3.00 16.56 7.87 35.58 9.50 29.81 7.57 28.71 Restaurants and hotels 2.57 13.73 3.72 29.85 4.24 14.45 3.64 17.45 Other services and goods 8.29 23.65 5.71 8.30 12.32 24.45 11.03 23.77 Source: World Bank staff calculations from HBS data. Weighted with original weights. Table A.5: Shares of the Food and Non-food Expenditures over Total Expenditure Survey Wave I II III IV Food expenditures (incl. alcohol and tobacco) 60.75 59.80 53.55 54.56 Clothing and footwear 8.04 7.76 9.40 8.42 Housing and related services 13.80 13.88 14.33 14.84 Health 2.22 2.76 3.16 2.95 Transport and communication 9.44 8.47 9.66 10.11 Recreation and culture 1.38 1.15 1.77 1.13 Education 0.95 2.81 2.97 2.72 Restaurants and hotels 0.81 1.33 1.32 1.31 Other services and goods 2.61 2.04 3.85 3.97 Table A.6: Shares of the Expenditures on Food Categories over Total Food Expenditure Survey Wave I II III IV Bread and cereals 18.92 20.93 19.05 17.82 Meat 17.78 15.94 16.05 16.38 Fish 0.75 0.87 1.24 1.09 Milk, cheese and eggs 12.15 10.36 10.69 10.48 Oil and fats 5.36 5.03 4.85 5.03 Fruits 4.87 5.35 6.44 6.00 Vegetables 10.56 12.02 10.22 10.47 Sweets (sugar, jam, honey, chocolate and 5.97 6.29 6.70 6.50 confectionery) Other food products 4.67 5.58 5.58 5.04 Coffee, tea and cocoa 4.07 4.25 3.79 3.81 Non-alcoholic beverages 4.74 5.41 5.87 5.54 42 Alcoholic beverages 0.96 0.83 0.79 0.82 Tobacco 9.20 7.15 8.72 11.01 Source: World Bank staff calculations from HBS data. Weighted with original weights. Table A.7: Poverty Headcount by Location Survey Wave 2002-03 2003-04 2004-05 2005-06 Original weights Total 37.7 43.7 34.8 45 Rural 34.4 44.2 37.2 49.2 Urban 46.6 42.1 30.3 37.4 Post-stratified weights Total 38.7 43.5 34.8 45.1 Rural 34.4 44.2 37.2 49.2 Urban 46.6 42.1 30.3 37.4 Source: World Bank staff calculations from HBS data. Table A.8: Kosovo: IMF GDP Estimates at Current Prices, 2004­10. In millions of euros, unless otherwise indicated, subject to further revision Prel. Projections 2004 2005 2006 2007 2008 2009 2010 Consumption 2,699 2,735 2,840 2,821 2,811 2,748 2,714 Households 1,921 1,998 2,107 2,185 2,229 2,259 2,291 Public 779 737 733 636 582 489 423 General government 376 337 347 359 401 386 383 Wages 184 195 204 204 204 200 196 Goods and services 192 143 143 155 197 186 187 Donor sector 1/ 403 400 386 276 180 103 40 Wages 333 340 330 232 150 85 33 Expatriates 261 255 248 176 115 65 25 Local employees 72 85 83 55 35 19 7 Goods and services 70 60 56 45 30 18 7 Investment 626 618 681 788 844 909 945 Donor sector 1/ 158 91 78 99 94 74 28 General government 169 151 99 142 204 259 270 Private investment 300 376 504 547 546 576 646 Housing 203 221 241 242 240 229 219 Other 97 155 262 306 306 347 427 Net exports of goods and - - services -1,112 -1,177 -1,283 -1,288 1,235 -1,175 1,092 Exports 212 196 233 255 314 365 411 Exports of goods 79 66 90 113 162 201 230 Exports of services 132 130 143 141 152 165 181 Imports 1,324 1,373 1,516 1,543 1,549 1,540 1,503 Donor imports 216 140 122 120 100 73 29 Other goods and 216 140 122 120 100 73 29 43 services Imports related to the humanitarian assistance 0 0 0 0 0 0 0 Other imports 1,108 1,233 1,394 1,423 1,449 1,467 1,474 Of which: private sector consumer goods 709 785 886 860 819 780 743 Of which: private investment goods 167 241 329 368 403 427 442 GDP 2,214 2,177 2,237 2,320 2,419 2,482 2,567 Workers' remittances (net) 215 262 300 342 346 350 351 Income from abroad (net) 23 30 40 31 4 9 11 GNDI 2,451 2,469 2,577 2,693 2,770 2,841 2,929 Memorandum items: Total foreign assistance 565 491 465 376 275 177 68 Of which: Direct contribution to GNDI 204 198 181 128 93 64 36 Private sector disposable income 2,044 2,050 2,049 2,121 2,242 2,300 2,380 Private sector consump. in percent of disposable income 94 97 103 103 99 98 96 GNDI per capita (in euros) 1,247 1,235 1,268 1,303 1,318 1,329 1,347 Fund staff estimates and projections as of September, 2007. 1/ Donor sector includes UNMIK, KFOR, and other donor spending under the umbrella of the so-called "public investment program", and spending financed by designated donor grants (DDGs). This presentation excludes wages of KFOR personnel as well as consumption of goods imported directly by KFOR. Source: IMF, 2007. Table A.9: Poverty Incidence, Gap and Severity, Corrected for Survey design Estimate Standard 95% Confidence Error Interval Poverty incidence (p0) 2003/04 43.5% 2.0% 39.5% 47.4% 2005/06 45.1% 1.7% 41.6% 48.5% Poverty gap (p1) 2003/04 11.9% 0.7% 10.6% 13.2% 2005/06 13.3% 0.7% 12.0% 14.6% Poverty severity (p2) 2003/04 4.6% 0.3% 4.0% 5.2% 2005/06 5.7% 0.4% 5.0% 6.5% Source: World Bank staff estimates from HBS data. 44 ANNEX B: POVERTY PROFILE Table B.1: Poverty Headcount by Location, Region and Ethnic Areas 2002/03 2003/04 2004/05 2005/06 Total 38.7 43.5 34.8 45.1 Rural 34.4 44.2 37.2 49.2 Urban 46.6 42.1 30.3 37.4 Gjakova 41.1 48.9 34.8 45.3 Cjilani 35.9 32.5 20.7 23.5 Mitrovica 50.2 59 51.2 69.7 Peja 43.8 37.8 31.8 40.1 Prizreni 41.5 48.3 41.6 40.5 Prishtina 26.3 34.3 29 40.6 Ferizaji 55.5 49.8 38.2 54.4 Albanian 38.7 43.7 34.8 43 area Serbian area 34.1 39.4 33.3 80.5 Source: World Bank staff calculations from HBS data. Table B.2: Poverty Contribution by Location 2002/03 2003/04 2004/05 2005/06 Rural 57.8 66.1 69.5 70.9 Urban 42.2 33.9 30.5 29.1 Total 100 100 100 100 Source: World Bank staff calculations from HBS data. Table B.3: Poverty Contribution by Region 2002/03 2003/04 2004/05 2005/06 Gjakova 9.9 11.5 11 12.8 Cjilani 7.8 7.9 7.8 5.6 Mitrovica 20.1 22.7 20.8 22.6 Peja 11.5 9.6 10.2 10.2 Prizreni 14.8 15.3 18.6 15.8 Prishtina 23.1 22.5 21.8 19.8 Ferizaji 12.7 10.5 9.8 13.3 Total 100 100 100 100 Source: World Bank Staff calculations from HBS data. Table B.4: Poverty Contribution of Ethnic Areas 2002/03 2003/04 2004/05 2005/06 Albanian 94.4 96.4 95.9 90.2 Serbian 5.6 3.6 4.1 9.8 Source: World Bank staff calculations from HBS data. Table B.5: Poverty Headcount by Household Size Category 2002/03 2003/04 2004/05 2005/06 1 to 3 31.6 30.7 29.3 40.4 4 to 6 35.4 36.5 32 41.4 7 to 9 41.7 49.5 36 48.5 10 to 12 45.2 51.2 42.6 48.3 13+ 36.2 45.5 34.7 48 Source: World Bank staff calculations from HBS data Table B.6: Poverty Headcount by Household Head Ethnicity 2002/03 2003/04 2004/05 2005/06 Albanian 38.4 43.6 32.1 42.5 Serbian 30 34.7 34.3 81.8 Other 58.7 54.3 67 51.8 Source: World Bank staff calculations from HBS data Table B.7: Poverty Headcount by Household Head Gender 2002/03 2003/04 2004/05 2005/06 Male 37.8 43.3 34.2 44.8 Female 54.2 46.8 44.9 49 Source: World Bank staff calculations from HBS data Table B.8: Fraction of Elderly in the Household and Poverty 2002/03 2003/04 2004/05 2005/06 No elderly 36.4 42.5 35 45.8 1-25% 44.2 46.5 35.1 42.5 26-50% 35.8 41.9 28.1 46.8 51% 59.5 44 42.3 62.3 Source: World Bank staff calculations from HBS data. Table B.9: Dependency Ratio and Poverty Headcount 2002/03 2003/04 2004/05 2005/06 Only dependents 51.9 40.5 27.3 65 Dependency ratio<=1 36.4 44.1 33.9 43.1 Dependency ratio>1 46.4 41.4 37.6 51.5 Source: World Bank staff calculations from HBS data. Table B.10: Education of the Household Head and Poverty Headcount 2002/03 2003/04 2004/05 2005/06 None, can't read/write 43.2 55.5 45.2 52.2 None but can read/write 49.9 55.1 42.1 56.5 Uncompleted primary school 43.3 48.2 45.2 60.8 Primary 43.6 49.1 41 51.4 Secondary 34.9 39.8 33.2 40.5 Vocational 27.4 30 15.3 31.5 University or higher 25.9 25.5 12.7 19.5 Source: World Bank staff calculations from HBS data. 46 Table B.11: Main Activity of the Household Head and Poverty Headcount 2002/03 2003/04 2004/05 2005/06 Employer 37.3 37.2 19.2 18.4 Employed with salary 29.7 38.6 25.5 35.8 Subsistence farmer 39.1 40.9 35.9 42.6 Per-diem worker 48.2 50 48.5 60.8 Other self-employed 36 35.4 18.1 29.6 Retired/disabled 42.8 44.8 35.4 47.5 Unemployed 45 51.4 48.9 58.6 Housekeeper 55.6 22.7 53.3 53.6 Other 20.2 56.8 43.1 58.6 Source: World Bank staff calculations from HBS data. Table B.12: Employment Sector of the Household Head and Poverty Headcount 2002/03 2003/04 2004/05 2005/06 Self-employed, agriculture 39.1 40.9 35.9 42.6 Self-empl, 45.7 45.7 49.2 51.9 mining/construction Self-employed, trade 55.1 43.7 25.2 37.5 Self-employed, other 35.2 42.1 20.5 41.3 Wage earner, professional 21.8 30.1 15.2 27.5 Wage earner, manufacturing 37.6 45.1 34.9 42.3 Wage earner, other 31.4 41.8 30.1 36.6 Unemployed 45 51.4 48.9 58.6 Nonactive 42.8 44.8 37.4 48.5 Source: World Bank staff calculations from HBS data. Table B.13: Land Tenure and Table B.14: Ownership of Livestock Poverty Rural areas and Poverty in Rural areas 2002/03 2005/06 2002/03 2005/06 Landless 43.1 64 No livestock 38.2 55.6 Owns land 33.5 47.8 At least 1 27.9 49.4 More than 1 37.3 43.9 Table B.15: Ownership of Major Equipment (tractor, or trailer) and Poverty in Rural areas 2002/03 2005/06 No equipment 35.5 55.8 At least 1 major equipment 32.6 37 Note: Weighted by individual level weights. Only half of Wave III has observations on land ownership and is thus excluded. Source: World Bank staff calculations from HBS data. 47 Table B.16: Employment and Education, 2002/03. Are Well-educated People more likely to be Employed? Uncompleted Primary Secondary Vocational Tertiary primary Employer 0 0.5 1.6 3.5 3.4 Salaried employee 2.4 8 29.5 65.6 74.5 Subsistence farmer 3.8 6.3 5.2 1.2 0.2 Per-diem worker 1.4 5.5 6.3 1.9 1.5 Unemployed 10.5 28 40.4 13.2 13.8 Housekeeper 59.3 43.8 8.9 3.9 0.8 Other 22.6 8 8.1 10.7 5.8 Total 100 100 100 100 100 Source: World Bank staff calculations from HBS data. Weighted figures for 15-64 year-olds. Other includes self-employed other, unpaid family worker, and retired or disabled. Students are excluded. Table B.17: Unemployment and Education, 2002/03. Are the Unemployed more Likely to be with Lower Educational Attainment? Employed Unemployed Uncompleted primary 2.5 3.9 Primary 26.9 39.4 Secondary 53 53.2 Vocational 7.8 1.6 Tertiary 9.9 1.9 Total 100 100 Source: World Bank staff calculations from HBS data. Weighted figures for 15-64 year-olds. Housekeepers, unpaid family workers and students are excluded. Table B.18: Employment and Education, 2003/04. Are Well-educated People more likely to be Employed? Uncompleted Primary Secondary Vocational Tertiary primary Employer 0.4 0.5 1.4 1.8 0.9 Salaried employee 2.6 6.5 28.8 68.2 72.3 Subsistence farmer 5.4 6.9 5.5 3.4 0 Per-diem worker 0.6 3.2 5.4 0.8 1.1 Unemployed 19.4 35.7 43.5 15.7 18.1 Housekeeper 51.8 41.7 6.5 3.2 0.8 Other 19.8 5.5 8.9 6.9 6.8 Total 100 100 100 100 100 Source: HBS 2003/04. Weighted figures for 15-64 year-olds. 48 Table B.19: Unemployment and Education, 2003/04. Are the Unemployed more Likely to be with Lower Educational Attainment? Employed Unemployed Uncompleted primary 3 5.4 Primary 23.9 42.2 Secondary 55.4 48.8 Vocational 8.1 1.5 Tertiary 9.6 2.1 Total 100 100 Source: World Bank staff calculations from HBS data. Weighted figures for 15-64 year-olds. Housekeepers, unpaid family workers and students are excluded. Table B.20: Employment and Education, 2004/05. Are Well-educated People More Likely to be Employed? Uncompleted Primary Secondary Vocational Tertiary primary Employer 1 0.2 1.6 0.9 1.2 Salaried employee 2.5 5.3 29.1 67.2 75.6 Subsistence farmer 4.3 7.5 6.9 1.5 1.7 Per-diem worker 1.6 5.6 5.1 1.1 1.4 Unemployed 17.8 33 39.9 17.1 13.5 Housekeeper 57.1 42.2 7.7 2.8 0.5 Other 15.7 6.2 9.8 9.3 6 Total 100 100 100 100 100 Source: World Bank staff calculations from HBS data. Weighted figures for 15-64 year-olds. Table B.21: Unemployment and Education, 2004/05. Are the Unemployed more Likely to be with Lower Educational Attainment? Employed Unemployed Uncompleted primary 3.7 6.6 Primary 23.9 41.6 Secondary 52.9 47.7 Vocational 8.4 2.1 Tertiary 11.2 2 Total 100 100 Source: World Bank staff calculations from HBS data. Weighted figures for 15-64 year-olds. Housekeepers, unpaid family workers and students are excluded. 49 Table B.22: Employment and Education, 2005/06. Are Well-educated People More Likely to be Employed? Uncompleted Primary Secondary Vocational Tertiary primary Employer 0.3 0.1 1.4 1.1 3.6 Salaried employee 3 6.2 27.7 70.6 69 Subsistence farmer 3.6 7 5.6 2.4 1.1 Per-diem worker 1.8 3.9 4.3 1.6 1.3 Unemployed 16.9 32.1 41.1 15.1 16.6 Housekeeper 55.2 43.8 10.3 2.6 1 Other 19.2 6.8 9.7 6.6 7.3 Total 100 100 100 100 100 Source: World Bank staff calculations from HBS data. Weighted figures for 15-64 year-olds. Table B.23: Unemployment and Education, 2005/06. Are the Unemployed more Likely to be with Lower Educational Attainment? Employed Unemployed Uncompleted primary 3.3 5.4 Primary 23.8 39.3 Secondary 56.4 52 Vocational 7.1 1.4 Tertiary 9.4 1.9 Total 100 100 Source: World Bank staff calculations from HBS data. Weighted figures for 15-64 year-olds. Housekeepers, unpaid family workers and students are excluded. Table B.24: Poverty and Unemployment 2002/03 2003/04 2004/05 2005/06 Poverty rate of the unemployed 40.8 50.6 40.8 49.5 Poverty rate of the employed 32.2 31.9 25.9 34.7 % of poor unemployed 52.7 63 59 58.6 % of non-poor unemployed 43.5 43.8 42.1 43.3 Source: World Bank staff calculations from HBS data. The questionnaire does not have information on inactivity and thus unemployment rates cannot be calculated. Table B.25: Gross Enrollment Rates for Primary schools 2002/03 2003/04 2004/05 2005/06 Total 100.9 92.4 90.6 93 Poor 100.3 92.8 85.6 92.5 Male 101.1 92.5 90.4 93.1 Females 100.6 92.4 91 92.9 Poorest quintile 98.3 92.3 82.7 91.3 Quintile 2 100.8 92.7 90.8 95.2 Quintile 3 101.5 93.5 93.7 91.5 Quintile 4 103.4 93.7 94.4 94.8 Richest quintile 100.6 89.6 93.2 92.1 50 Table B.26: Net Enrollment for Primary Schools 2002/03 2003/04 2004/05 2005/06 Total 92.2 87.7 85 87.7 Poor 90.1 87.2 78.8 85.9 Male 92.8 88.9 85.6 88.4 Females 91.6 86.6 84.4 87 Poorest quintile 86.4 84.6 75.4 83.3 Quintile 2 92.7 89.5 84.8 89.5 Quintile 3 91.7 87.4 85.3 84.5 Quintile 4 94.1 89 88.9 90.1 Richest quintile 96.7 87.6 92.5 91.5 Source: World Bank staff calculations from HBS data. Gross enrollment rates = Total enrolled students/children in age group. Net enrollment rates = Total enrolled students aged 6-14/children in that age group. In 2002/03 questionnaire, the question about enrollment was asked for children over 7. In later surveys this was changed to 6 year-olds. The age groups used are: primary 7-14 in 2002/03, primary 6-15 starting 2003/04; secondary 16-18; tertiary 20-24. Table B.27: Net Enrollment Rates for Secondary Schools 2002/03 2003/04 2004/05 2005/06 Total 73.1 70.9 66.4 74 Poor 68.3 65.1 58.3 72 Male 75.6 72.6 69.6 81.2 Females 70 69.2 62.9 66.4 Poorest quintile 65.3 62.8 54.6 67.3 Quintile 2 71.4 64.7 65.2 78.1 Quintile 3 81.5 74.5 65.1 71.3 Quintile 4 69.3 77.1 73 72.4 Richest quintile 76.5 80.9 80.5 81 Table B.28: Net Enrollment Rates for Tertiary Education 2002/03 2003/04 2004/05 2005/06 Total 16.7 15.3 14.6 17.7 Poor 12.2 9.3 9.3 13.8 Male 20.4 14.6 17.3 19.9 Females 13 15.9 12.2 15.2 Poorest quintile 8.9 7.6 8.8 12.9 Quintile 2 15.2 9.6 10.8 12.6 Quintile 3 18.5 15.1 14.7 15 Quintile 4 20.6 17 12.7 18.3 Richest quintile 18.2 26.5 24.3 27.4 Source: World Bank staff calculations from HBS data. Approximate net enrollment rates only as we assume that those who reported being a student and are aged 16-19. Approximate net enrollment rates only as we assume that those who reported being a student and are aged 20-24 attend university. Breakdown by type of school (vocational vs. high school) not asked in the survey. The age groups used are: primary 7-14 in 2002/03, primary 6-15 starting 2003/04; secondary 16-18; tertiary 20-24. 51 Table B.29: Access to Electricity: Percent of People Living in Dwellings with Electricity 2002/03 2003/04 2004/05 2005/06 Total 99.2 98.2 98.3 97.9 Poor 98.9 97.8 97.2 97.5 Urban 99.2 99.1 99.5 99.4 Rural 99.1 97.7 97.7 97.1 Poorest quintile 99.1 95.9 95.8 96.8 Quintile 2 98.8 99.6 99 98.1 Quintile 3 99.4 97.4 98.4 98.5 Quintile 4 99.8 99.8 99.1 96.6 Richest quintile 98.7 98.2 99.4 99.4 Table B.30: District Central Heating: Percent of People Living in Dwellings with Central Heating 2002/03 2003/04 2004/05 2005/06 Total 0.6 6.6 4.5 5.6 Poor 0.4 2.7 1.9 4.7 Urban 0.6 16.1 8.8 8.7 Rural 0.6 1.5 2.2 3.9 Poorest quintile 0.4 2.8 2.4 7.1 Quintile 2 0.3 2.3 1.1 2.9 Quintile 3 0.6 6.1 3.5 2.1 Quintile 4 0.7 8 6.1 6.3 Richest quintile 1 13.7 9.2 9.5 Source: World Bank staff calculations from HBS data. The questions about housing are not the same for 2002/03 and later surveys. The number of categories decreases from over 20 to 9. 52 Table B.31: Access to Safe Dwelling: Percent of People Living in Dwellings with Walls of brick, block or cement 2002/03 2003/04 2004/05 2005/06 Total 90.7 95.8 94 95.1 Poor 87.8 95 94.4 94.9 Urban 92.6 95.6 96.9 96.5 Rural 89.7 95.9 92.4 94.3 Poorest quintile 84.5 92.7 93.5 93.2 Quintile 2 91.1 96.9 96.5 96.2 Quintile 3 90.9 96.2 94.9 96.3 Quintile 4 91.8 97.9 90.4 93 Richest quintile 95.2 95.1 94.5 96.6 Table B.32: Access to Water: Percent of People Living in Dwellings with Indoor Water tap 2002/03 2003/04 2004/05 2005/06 Total 63.9 73.6 80.3 83.9 Poor 64.7 63.6 68.3 75.5 Urban 93.7 95.2 96.2 95.2 Rural 47.8 62 71.8 77.8 Poorest quintile 63.7 61.5 64.6 69.7 Quintile 2 63.3 64.7 75.5 77.7 Quintile 3 58 72.9 82.7 87.5 Quintile 4 66.5 82 86.6 88.2 Richest quintile 67.7 86.9 92.1 96.3 Source: World Bank staff calculations from HBS data. The questions about housing are not the same for 2002/03 and later surveys. The number of categories decreases from over 20 to 9. Table B.33: Rural Poverty Headcount Rate and Poverty Contribution Poverty Headcount Rate Poverty Contribution 2002/03 2003/04 2004/05 2005/06 2002/03 2003/04 2004/05 2005/06 Gjakova 43.5 53.7 41.3 46.8 11.5 12.7 11.4 10.7 Cjilani 36.5 35 21.2 27.2 8.9 7.9 8.8 6 Mitrovica 55.2 62 49.7 73.3 23.5 24.8 18.4 24.9 Peja 37.1 34.4 34.9 47.5 11.8 9.5 11 11.4 Prizreni 33.5 44.3 45 40.9 13.6 14.8 18.5 15.1 Prishtina 17.3 36.4 34.6 45.4 17.4 22.1 22.7 19 Ferizaji 54.5 44.8 38.1 59.7 13.2 8.2 9.1 12.8 Table B.34: Rural Ethnic Divide: Poverty Headcount and Contribution Headcount Contribution 2002/03 2003/04 2004/05 2005/06 2002/03 2003/04 2004/05 2005/06 Albanian 34.3 44.2 37.5 46.7 92.6 95 95.3 88.7 Serbian 32 44.6 32.6 83.8 7.4 5 4.7 11.3 53 Table B.35: Rural Area, Educational Attainment 2002/03 2003/04 2004/05 2005/06 Cannot read/write 6.4 6.4 6.3 5.3 No education but can read/write 2.7 3.2 3.9 2.6 Attending primary 20.4 23.1 22.7 23.8 Uncompleted primary 4.6 3.2 4.5 5.3 Primary 36.9 36.6 35.8 34.4 Secondary 25.4 24.5 23.1 25.4 Vocational 1.6 1.8 2.1 1.4 Tertiary 2 1.4 1.6 1.7 Source: World Bank staff calculations from HBS data. Weighted figures for individuals 10 years and older. Table B.36: Rural Access to Electricity: Percent of People Living in dwellings with Electricity Rural 2002/03 2003/04 2004/05 2005/06 Total 99.1 97.7 97.7 97.1 Poor 98.8 97.5 96.4 97.1 Poorest national quintile 98.9 95.3 93.9 95.9 Quintile 2 98.7 99.4 99.1 98.2 Quintile 3 99.4 96.3 97.9 97.9 Quintile 4 99.9 100 98.6 94.9 Richest national quintile 98.6 97.3 99.1 98.9 Table B.37: Rural Access to Safe Dwelling: Percent of People Living in dwellings with Walls of brick, block or cement Rural 2002/03 2003/04 2004/05 2005/06 Total 89.7 95.9 92.4 94.3 Poor 86.5 95.4 93.4 94.1 Poorest national quintile 83.4 94.4 92.2 93 Quintile 2 89.8 96.2 95.8 95 Quintile 3 89.5 95.8 92.9 95.7 Quintile 4 90.4 98.9 87.2 92.6 Richest national quintile 94.4 93.8 93.5 95.2 Table B.38: Rural District Central Heating: Percent of People Living in dwellings with Central Heating Rural 2002/03 2003/04 2004/05 2005/06 Total 0.6 1.5 2.2 3.9 Poor 0.3 0.4 1.7 4.2 Poorest national quintile 0.1 1 2.8 6.9 Quintile 2 0.6 0.2 0.2 2.1 Quintile 3 0.5 1.4 2.5 1.4 Quintile 4 0.6 0.4 2.1 4.3 Richest national quintile 1.2 4.8 3.5 5.2 54 Table B.39: Rural Access to Water: Percent of People Living in Dwellings with Indoor Water tap Rural 2002/03 2003/04 2004/05 2005/06 Total 47.8 62 71.8 77.8 Poor 43.1 49.4 58.1 69.4 Poorest national quintile 44.5 43.9 53.3 61 Quintile 2 37.9 54.1 67.1 73.6 Quintile 3 38.8 61.7 74 83 Quintile 4 58.9 73.5 80.1 82.6 Richest national quintile 54.5 78.4 87.6 93.5 Source: World Bank staff calculations from HBS data. The questions about housing are not the same for 2002/03 and later surveys. The number of categories decreases from over 20 to 9. Table B.40: Rural Unemployment: Percent of Individuals Reporting being Unemployed 2002/03 2003/04 2004/05 2005/06 Total Rural 25.3 31.4 28 29 Rural Poor 27.6 35.3 32.2 31.5 Poorest quintile 28.9 38.6 34.5 35.2 Quintile 2 26.5 33.2 29.1 30.4 Quintile 3 25.2 27.3 30.5 28.4 Quintile 4 25.4 30.6 24.4 26 Richest quintile 21.9 27.3 20.9 23.9 Table B.41: Rural Gross Enrollment Rates for Primary Schools 2002/03 2003/04 2004/05 2005/06 Total 99.7 92.5 90.4 93.1 Poor 97.3 92.9 84.2 92.4 Male 99.6 92.9 89.2 92.9 Females 99.7 92.1 91.7 93.3 Poorest quintile 94.8 91.8 80.9 92.1 Quintile 2 97.5 93.3 89.5 94.6 Quintile 3 101.7 92.6 93.8 89.6 Quintile 4 102.6 93.3 95.4 95.1 Richest quintile 101.4 91.3 95.3 94.6 Source: HBS 2002-05 weighted data for individuals aged 15-64. No information on inactivity thus unemployment rate cannot be calculated. Gross enrollment rates = Total enrolled students/children in age group. In 2002/03 questionnaire, the question about enrollment was asked for children over 7. In later surveys this was changed to 6 year-olds. The age groups used are: primary 7-14 in 2002/03, primary 6-15 starting 2003/04; secondary 16-18; tertiary 20-24. 55 Table B.42: Rural Net Enrollment Rates for Primary Schools 2002/03 2003/04 2004/05 2005/06 Total 92.7 88.3 85.4 88.3 Poor 89.8 88.2 78.3 86.4 Male 93.6 90.1 85.8 88.7 Females 91.7 86.3 84.9 87.8 Poorest quintile 85.7 83.9 74.2 84.3 Quintile 2 92.1 91.3 84.7 89.8 Quintile 3 91.6 87 85 82.5 Quintile 4 97.7 90.8 92.6 93.5 Richest quintile 96.2 87.4 93.5 92.8 Table B.43: Rural Net Enrollment Rates for Secondary Schools 2002/03 2003/04 2004/05 2005/06 Total 69.9 65.3 61.5 70.6 Poor 63.2 58.1 53.4 70.2 Male 75.4 68 66.9 80.5 Females 63.7 62.6 55.4 60.5 Poorest quintile 59.5 54.9 47.3 65.4 Quintile 2 66.6 59.1 62.8 78.6 Quintile 3 77.4 67.7 57.2 60.5 Quintile 4 66.3 72.7 72.4 69.4 Richest quintile 76.8 80 75.4 78.4 Source: World Bank staff calculations from HBS data. Net enrollment rates = Total enrolled students aged 6-14 children in that age group. In 2002/03 questionnaire, the question about enrollment was asked for children over 7. In later surveys this was changed to 6 year-olds. The age groups used are: primary 7-14 in 2002/03, primary 6-15 starting 2003/04; secondary 16-18; tertiary 20-24. Breakdown by type of school (vocational vs. high school) asked in the survey. Table B.44: Rural Net Enrollment Rates for Tertiary Education 2002/03 2003/04 2004/05 2005/06 Total 14.2 11.3 10.8 14.5 Poor 8.1 8 7.8 14 Male 19.4 10.9 14.6 16 Females 9 11.7 7.5 12.7 Poorest quintile 7 4.9 6.6 14.1 Quintile 2 9.1 10.1 8.7 12.9 Quintile 3 16 14.4 9.6 10 Quintile 4 17.3 11.5 7.5 12.3 Richest quintile 16.9 16.1 23.3 24.2 Source: World Bank staff calculations from HBS data. Approximate net enrollment rates only as we assume that those who reported being a student and are aged 20-24 attend university. 56 Table B.45: Vulnerable Group above the Poverty Line and Monthly Consumption below 53.7 Euro in 2002 Prices, in Percent of Individuals 2002/03 2003/04 2004/05 2005/06 Total 18 19.4 18.7 17.5 Distribution by type of settlement Urban 35.5 30.2 36.8 34.3 Rural 64.5 69.8 63.2 65.7 Total 100 100 100 100 Distribution by district Gjakova 11.3 11.5 14.6 14.8 Cjilani 9.2 10.5 6.9 11.2 Mitrovica 15 14.9 10.2 12.1 Peja 4.7 9.8 11.3 9 Prizreni 14.5 15.1 19.8 22 Prishtina 36.5 27.6 29.7 19.7 Ferizaji 8.9 10.6 7.5 11.1 Total 100 100 100 100 Distribution by ethnic area Albanian ethnic area 93.7 98.1 95.1 97 Serbian ethnic area 6.3 1.9 4.9 3 Total 100 100 100 100 Unemployment rate 25.4 28.4 28.2 28.1 Distribution by educational attainment Cannot read/write 7.1 6.8 4.8 5.6 No education but can read/write 2.4 2.9 2.9 1.4 Attending primary 15.2 15.3 15.1 16 Uncompleted primary 3.2 3.7 4.7 4.7 Primary 37.1 36.2 39.1 35.7 Secondary 30.8 30.9 29.3 32.3 Vocational 1.7 2.4 1.9 2.1 Tertiary 2.6 1.9 2.2 2.2 Total 100 100 100 100 Enrollment rates of children Secondary enrollment rate 79.9 74.6 67.1 78.6 Male secondary enrollment rate 84.9 75.4 73.1 89.1 Female secondary enrollment rate 72.1 73.6 60.1 69.2 Tertiary enrollment rate 16.2 16.3 16.8 14.3 Male tertiary enrollment rate 17.8 17.1 16.5 17.7 Female tertiary enrollment rate 14.6 15.6 16.9 10.7 Source: World Bank staff calculations from HBS data. Upper boundary for vulnerability estimated as 25% higher than the absolute poverty line. Ethnic area as defined in survey design. Unemployment rate cannot be calculated because of lack of inactivity data, thus the figures are just percent of population reporting being unemployed. Approximate net enrollment rates only as we assume that those who reported being a student and are aged 16-18 attend secondary school and those aged 20-24 attend university. 57 Table B.46: Vulnerable Group listed below the Poverty Line and Monthly Consumption above 32.3 Euro in 2002 prices 2002/03 2003/04 2004/05 2005/06 Total 17.5 21.4 18.4 21.5 Distribution by type of settlement Urban 44.8 28.4 26.6 27.9 Rural 55.2 71.6 73.4 72.1 Total 100 100 100 100 Distribution by region Gjakova 11.6 15 11.2 14 Cjilani 9.8 7.3 9.4 7.4 Mitrovica 17.1 24.1 16.9 17.4 Peja 8.6 6.8 9.7 9.3 Prizreni 15.4 14.3 20.9 20.2 Prishtina 28.2 23.6 22.2 19.1 Ferizaji 9.2 8.8 9.7 12.5 Total 100 100 100 100 Distribution by ethnic area Albanian ethnic area 95.8 96 93.8 93.5 Serbian ethnic area 4.2 4 6.2 6.5 Total 100 100 100 100 Unemployment rate 26.5 35.2 30.2 28.6 Distribution by educational achievement Cannot read/write 6 6.5 7.2 5.2 No education but can read/write 3.4 4 5.4 2.7 Attending primary 15 17 15.8 18.3 Uncompleted primary 4.4 3 4.9 5.9 Primary 36.6 39 38.2 38.1 Secondary 29.6 27.1 25.7 26.8 Vocational 2.7 2 1.8 1.8 Tertiary 2.2 1.3 1 1.3 Total 100 100 100 100 Enrollment rates Secondary enrollment rate 72.7 64.6 58.6 77.9 Male secondary enrollment rate 77.4 62.1 57.8 91.9 Female secondary enrollment rate 68.2 66.8 60.2 65 Tertiary enrollment rate 16.2 10.5 8.7 16 Male tertiary enrollment rate 20.3 12.6 12.1 17.6 Female tertiary enrollment rate 12.1 8.7 5.8 14.5 Source: World Bank staff calculations from HBS data. Lower boundary for vulnerability estimated as 25% below the absolute poverty line. Ethnic area as defined in survey design. Unemployment rate cannot be calculated because of lack of inactivity data, thus the figures are just percent of population reporting being unemployed. Approximate net enrollment rates only as we assume that those who reported being a student and are aged 16-18 attend secondary school and those aged 20-24 attend university. 58 Table B.47: Internal Migrants and Their Recipient Location (in percent of total individuals) Estimated Share Share of the Distribution of poor of the population population born with a member born outside municipality outside the municipality Gjakova 12.7 1.4 1.5 Decani 1.8 0.4 0 Gjakova 5.3 0.6 1.2 Rahoveci 5.5 0.4 0.3 Gjilani 10.7 11.5 3.9 Gjilani 4.9 6.2 2.4 Kamenica 2.6 1.1 0.3 Viti 3.3 4.2 1.3 Mitrovica 14.6 22.3 33.4 Mitrovica 3.6 8.5 10.7 Leposaviqi 0.1 0 0 Skanderaj 4.7 5.7 9.4 Vushtri 5.9 8 13 Zubin Potok 0.3 0 0 Zvecan 0.1 0.1 0.3 Peja 11.5 17.8 13 Istogu 2.6 1.5 1.3 Klina 3 7.7 2.1 Peja 5.9 8.6 9.6 Prizren 17.5 18.9 25.4 Dragash 0.9 0.3 0.6 Prizren 11.3 16.5 21.5 Suhareka 2.3 1.2 1.7 Malisheva 2.9 0.9 1.6 Prishtina 21.9 17 7.2 Gllogoc 2.8 0 0 Fusha Kosova 1.6 0.5 0.2 Lipjani 3.6 2 0.8 Novo Barda 0.3 0 0 Obiliqi 1.2 0.3 0 Podujeva 5.3 0.1 0 Prishtina 7.1 14.1 6.2 Ferizaj 11 11 15.6 Kacanik 2.2 0.3 0.1 Shtime 2 0 0 Shtarpce 1.1 4.2 9 Ferizaj 5.6 6.5 6.4 Total 100 100 100 Source: World Bank staff calculations from HBS data. Internal migrants are defined as not born in the current municipality. No information on time of migration. 59 Table B.48: Inequality Indices for 2002-2006 2002/03 2003/04 2004/05 2005/06 Percentile ratio p90/p10 3.84 3.33 3.76 3.96 p75/p25 2 1.89 1.97 1.97 Generalized Entropy, GE(-1) 0.17 0.14 0.17 0.18 GE(0) 0.15 0.12 0.15 0.15 GE(1) 0.16 0.13 0.16 0.16 GE(2) 0.19 0.16 0.21 0.2 Gini coefficient 0.3 0.27 0.3 0.3 Atkinson A(0,5) 0.07 0.06 0.07 0.07 A(1) 0.16 0.13 0.16 0.16 A(2) 0.26 0.22 0.25 0.27 Table B.49: Correlates of Consumption by Year 2002/03 2003/04 2004/05 2005/06 Demographic and household head characteristics Age of the household head 0.001 0.001 0.003** 0.001 [0.001] [0.001] [0.001] [0.001] Female HH head -0.01 -0.029 0.037 -0.013 [0.056] [0.063] [0.056] [0.039] Serbian 0.085 0.062 -0.108 -0.560*** [0.062] [0.074] [0.063] [0.053] Other -0.115 -0.127 -0.213** -0.191** [0.071] [0.071] [0.068] [0.062] Dependency ratio -0.123*** -0.091*** -0.098*** -0.104*** [0.023] [0.017] [0.018] [0.020] Number of students in HH -0.036* -0.052*** -0.062*** -0.050** [0.018] [0.014] [0.018] [0.017] Number of unemployed in HH -0.037* -0.063*** -0.088*** -0.082*** [0.017] [0.011] [0.011] [0.012] Highest education attained in the household (secondary omitted) None -0.408*** -0.328* -0.449*** -0.434** [0.122] [0.147] [0.089] [0.165] Primary -0.142** -0.129*** -0.130*** -0.142*** [0.051] [0.031] [0.036] [0.034] Vocational 0.172*** 0.114** 0.172*** 0.107** [0.044] [0.038] [0.043] [0.041] Tertiary 0.158* 0.331*** 0.258*** 0.249*** [0.066] [0.037] [0.037] [0.037] Main source of income for the household (other omitted) Public sector 0.125** 0.123*** 0.076 -0.159*** [0.043] [0.037] [0.042] [0.045] 60 Agriculture 0.046 0.141** -0.022 -0.105 [0.066] [0.053] [0.054] [0.058] Private sector 0.132** 0.251*** 0.082 -0.092 [0.048] [0.036] [0.048] [0.047] Per-diem work 0.011 0.108* -0.06 -0.326*** [0.057] [0.050] [0.057] [0.052] Self-employed/SME 0.096 0.259*** 0.307*** -0.028 [0.086] [0.063] [0.051] [0.055] Pension 0.003 0.119* -0.092 -0.364*** [0.069] [0.060] [0.055] [0.070] Social assistance (dropped) (dropped) -0.426*** -0.633*** [0.072] [0.051] Housing characteristics Brick/cement walls 0.104* 0.180** -0.082 0.024 [0.051] [0.057] [0.051] [0.062] Central district heating 0.232 0.152** 0.263*** 0.121* [0.129] [0.058] [0.070] [0.049] Access to inside water tap 0.042 0.153*** 0.176*** 0.163*** [0.052] [0.033] [0.031] [0.040] Purchase of 5 most-common 0.046*** 0.030** 0.024* 0.017 durables (lighting, textiles) in [0.010] [0.011] [0.010] [0.012] last 1 year (in logs) Regional and area dummy (Pristina omitted) Urban area dummy -0.176*** -0.108** -0.103** 0.02 [0.047] [0.033] [0.031] [0.033] Gjakova -0.09 -0.181** -0.203** -0.176** [0.070] [0.055] [0.061] [0.066] Gjilani 0.046 0.071 0.164*** 0.230*** [0.078] [0.058] [0.049] [0.051] Mitrovica -0.202** -0.120* -0.261*** -0.226*** [0.066] [0.056] [0.049] [0.053] Peja -0.074 -0.014 -0.102* -0.043 [0.114] [0.075] [0.045] [0.064] Prizreni -0.042 -0.143* -0.121* -0.028 [0.077] [0.055] [0.051] [0.056] Ferizaji -0.264*** -0.127* -0.142* -0.203*** [0.071] [0.054] [0.070] [0.054] Constant 3.831*** 3.614*** 4.029*** 4.072*** [0.088] [0.092] [0.106] [0.114] R-squared 0.186 0.251 0.321 0.377 N 2340 2337 2328 2306 * p<0.05, ** p<0.01, *** p<0.001 Source: World Bank staff calculations from HBS data. * p<0.05, ** p<0.01, *** p<0.001. The categories of main source of income increased from 7 to 10 to include social assistance transfers starting in calendar 2005 year. 61 Table B.50: Comparison of Enrollment Rates in the Region and the European Union, 2005 Enrollment rates by age group Tertiary Net Enrollment rates Age in years 5-14 15-19 20-29 20-242 Kosovo 82.3 66.2 9.9 Kosovo 17.7 Albania 87 56 13 Albania 18 EU15 100 82 25 Macedonia 1 27 EU 8 selected 98 85 20 Bosnia and Herzegovina 24.2 Serbia and Montenegro 1 36.3 Sources: Enrollment rates by age group: Kosovo - World Bank staff calculations from HBS data; other countries - Albania PEIR 2006. Tertiary enrollment rates: Kosovo - World Bank staff calculations from HBS data; Albania ­ World Bank staff calculations from LSMS 2005 data; Macedonia ­ Poverty Assessment, 2005; Bosnia and Herzegovina ­ Public Expenditure Review, 2006; Serbia and Montenegro ­ Knowledge for Development Database, World Bank . Notes: /1 Gross enrollment rates reported. /2 Age group 19-23 used in Bosnia and Herzegovina. Table B.51: Contribution to Poverty by Activity of the Household Head. 2003/04 2005/06 Employer 2.9 0.7 Employed with salary 26.2 21.9 Subsistence farmer 9.7 9.9 Per diem worker 7.2 8.4 Other self-employed 3.7 4.2 Retired or disabled 27.4 28.2 Unemployed 22.6 22 Housekeeper 0.1 2.8 Other 0.3 1.8 Total 100 100 Source: World Bank staff calculations from HBS data. Notes: Missing data on employment activity are omitted from the estimation. Weighted by population weights. Table B.52: Contribution to Poverty by Educational Attainment, in percent of individuals, 2005/06 2003/04 2005/06 None and cannot read or write 6.7 5.6 None but can read write 3.3 3.2 Attending primary school 23.6 24.7 Uncompleted primary school and not attending 3.1 5.9 Primary 37.1 34.1 Secondary 23.9 24.3 Vocational 1.3 1.2 University or higher 1 0.9 Total 100 100 Source: World Bank staff calculations from HBS data. Notes: Missing data on education are omitted from the estimation. The question is asked only of those 10 years of age or older. 62 Table B.53: Demographic Distribution by Ethnicity of Household Head and Region, 2005/06. Region Albanian Serbian Other1 Total Gjakovë 95.1 0.2 4.6 100 Gjilani 86.6 12 1.4 100 Mitrovicë 88.4 11.5 0.1 100 Pejë 92.3 1 6.7 100 Prizren 81.4 1 17.7 100 Prishtinë 93.8 5.2 1 100 Ferizaj 90.2 7.9 1.9 100 Total 89.7 5.3 5 100 Source: World Bank staff calculations from HBS data. Notes: In percent of individuals. Population weights used. /1 Turkish, Bosnian/Montenegro, Ashkalian, Roma and others. Table B.54: Demographic Distribution by Type of Settlement and Region, 2005/06. Region Urban Rural Total Gjakovë 42.7 57.3 100 Gjilani 34.1 65.9 100 Mitrovicë 25.6 74.4 100 Pejë 33.1 66.9 100 Prizren 32.8 67.2 100 Prishtinë 38.8 61.2 100 Ferizaj 37.5 62.5 100 Total 35 65 100 Source: World Bank staff calculations from HBS data. Notes: In percent of individuals. Population weights used. 63 ANNEX C: POVERTY DECOMPOSITION Table C.1: Decomposition of Poverty: 2002/03 compared to 2003/04 Change Growth Redistribution Interaction Poverty headcount (P0) -3.9 -5.6 2 -0.2 Poverty gap (P1) -2.5 -2 -0.3 -0.2 Poverty severity (P2) -1.4 -1 -0.4 0 Source: World Bank staff calculations from HBS data. Table C.2: Decomposition of Poverty: 2002/03 compared to 2004/05 Change Growth Redistribution Interaction Poverty headcount (P0) -3.9 -5.6 2 -0.2 Poverty gap (P1) -2.5 -2 -0.3 -0.2 Poverty severity (P2) -1.4 -1 -0.4 0 Source: World Bank staff calculations from HBS data. Table C.3: Decomposition of Poverty: 2002/03 compared to 2005/06 Change Growth Redistribution Interaction Poverty headcount (P0) 6.4 4.2 2 0.2 Poverty gap (P1) 1.6 1.6 -0.1 0.1 Poverty severity (P2) 0.8 0.8 0 0 Source: World Bank staff calculations from HBS data. Table C.4: Decomposition of Poverty: 2003/04 compared to 2004/05 Change Growth Redistribution Interaction Poverty headcount (P0) -8.7 -10.8 2 0.1 Poverty gap (P1) -2.7 -4.2 1.6 -0.1 Poverty severity (P2) -1 -1.8 1 -0.2 Source: World Bank staff calculations from HBS data. Table C.5: Decomposition of Poverty: 2003/04 compared to 2005/06 Change Growth Redistribution Interaction Poverty headcount (P0) 1.6 -0.8 2.5 -0.1 Poverty gap (P1) 1.4 -0.4 1.7 0 Poverty severity (P2) 1.2 -0.2 1.3 0 Source: World Bank staff calculations from HBS data. Table C.6: Urban Poverty Decomposition: 2002/03 compared to 2003/04 Change Growth Redistribution Interaction Poverty headcount (P0) 10.3 10 -0.8 1.1 Poverty gap (P1) 4.1 3.9 0.2 0 Poverty severity (P2) 2.1 1.8 0.4 0 Source: World Bank staff calculations from HBS data. Table C.7: Urban Poverty Decomposition: 2002/03 compared to 2003/04 Change Growth Redistribution Interaction Poverty headcount (P0) -4.5 -9.7 1.9 3.4 Poverty gap (P1) -1 -3.1 2.1 -0.1 Poverty severity (P2) -0.7 -1.4 1 -0.3 Source: World Bank staff calculations from HBS data. Table C.8: Urban Decomposition: 2002/03 compared to 2004/05 Change Growth Redistribution Interaction Poverty headcount (P0) -16.3 -18.9 4.9 -2.2 Poverty gap (P1) -4.6 -6.2 2.1 -0.5 Poverty severity (P2) -1.9 -2.9 1.3 -0.3 Source: World Bank staff calculations from HBS data. Table C.9: Urban Poverty Decomposition: 2002/03 compared to 2005/06 Change Growth Redistribution Interaction Poverty headcount (P0) -9.2 -13.8 3.3 1.4 Poverty gap (P1) -2.2 -4.3 2.4 -0.4 Poverty severity (P2) -0.8 -2 1.5 -0.3 Source: World Bank staff calculations from HBS data Table C.10: Urban Poverty Decomposition: 2003/04 compared to 2005/06 Change Growth Redistribution Interaction Poverty headcount (P0) -11.8 -9.5 -2 -0.3 Poverty gap (P1) -3.6 -3.8 -0.2 0.4 Poverty severity (P2) -1.2 -1.8 0.4 0.1 Source: World Bank staff calculations from HBS data. Table C.11: Urban poverty decomposition: 2003/04 compared to 2005/06 Change Growth Redistribution Interaction Poverty headcount (P0) -4.7 -3.3 -0.1 -1.3 Poverty gap (P1) -1.2 -1.5 0.2 0.1 Poverty severity (P2) -0.2 -0.7 0.5 0 Source: World Bank staff calculations from HBS data. Table C.12: Urban Poverty Decomposition: 2004/05 compared to 2005/06 Change Growth Redistribution Interaction Poverty headcount (P0) 7.1 6.2 1.6 -0.8 Poverty gap (P1) 2.4 2.1 0.2 0.1 Poverty severity (P2) 1.1 1 0.1 0 Source: World Bank staff calculations from HBS data. 66 Table C.13: Rural Poverty Decomposition: 2002/03 compared to 2003/04 Change Growth Redistribution Interaction Poverty headcount (P0) 9.8 12.2 -2.3 -0.1 Poverty gap (P1) 0.9 4.3 -3.7 0.3 Poverty severity (P2) -0.2 2.3 -2.1 -0.4 Source: World Bank staff calculations from HBS data. Table C.14: Rural Poverty Decomposition: 2002/03 compared to 2004/05 Change Growth Redistribution Interaction Poverty headcount (P0) 2.8 0.2 2.2 0.4 Poverty gap (P1) -1.3 0.1 -1.5 0 Poverty severity (P2) -1.1 0.1 -1.1 0 Source: World Bank staff calculations from HBS data. Table C.15: Rural Poverty Decomposition: 2002/03 compared to 2005/06 Change Growth Redistribution Interaction Poverty headcount (P0) 14.8 14.3 -0.1 0.6 Poverty gap (P1) 3.7 4.8 -1.5 0.4 Poverty severity (P2) 1.6 2.5 -0.7 -0.2 Source: World Bank staff calculations from HBS data. Table C.16: Rural Poverty Decomposition: 2003/04 compared to 2004/05 Change Growth Redistribution Interaction Poverty headcount (P0) -7 -11.5 4.4 0 Poverty gap (P1) -2.3 -4.5 2.5 -0.3 Poverty severity (P2) -0.8 -1.8 1.4 -0.4 Source: World Bank staff calculations from HBS data. Table C.17: Rural Poverty Decomposition: 2003/04 compared to 2005/06 Change Growth Redistribution Interaction Poverty headcount (P0) 5 1.4 3.7 -0.1 Poverty gap (P1) 2.8 0.5 2.3 0 Poverty severity (P2) 1.9 0.2 1.6 0 Source: World Bank staff calculations from HBS data. Table C.18: Rural Poverty Decomposition: 2204/05 compared to 2005/06 Change Growth Redistribution Interaction Poverty headcount (P0) 12 12.7 -2.7 2 Poverty gap (P1) 5 5.3 0 -0.2 Poverty severity (P2) 2.7 2.4 0.4 -0.2 Source: World Bank staff calculations from HBS data. 67 Figure C.1: Growth Incidence Curves A. 2002/03 to 2003/04 B. 2002/03 to 2004/05 20 20 10 10 0 0 -1 0 0 -2 0 0 -3 -1 0 20 40 60 80 100 0 20 40 60 80 100 Percentiles Percentiles Median spline Growth rate in mean Median spline Growth rate in mean C. 2002/03 to 2005/06 D. 2003/04 to 2004/05 30 0 -5 20 0 -1 5 10 -1 0 -2 0 5 -2 0 20 40 60 80 100 0 20 40 60 80 100 Percentiles Percentiles Median spline Growth rate in mean Median spline Growth rate in mean E. 2003/04 to 2005/06 F. 2004/05 to 2005/06 10 0 -1 0 5 -1 0 -1 0 -2 0 -2 0 5 -3 -2 0 20 40 60 80 100 0 20 40 60 80 100 Percentiles Percentiles Median spline Growth rate in mean Median spline Growth rate in mean Source: World Bank staff calculations from HBS data. 68 ANNEX D: SOCIAL TRANSFERS AND REMITTANCES Figure D.1: Undercoverage and Leakage of Social Assistance by Urban and Rural Undercoverage 80 79 79 77 77 71 67 60 64 62 Leakage 40 to the non-poor 20 22 23 22 11 0 Urban Rural Urban Rural Urban Rural 1 2 3 2003/04 2005/06 Figure D.2: Top 20 Remittance-receiving Countries as a share of GDP, 2004 Top 20 Remittance-receiving Countries as Share of GDP, 2004 31.06% 27.09% 25.81% 24.78% 22.46% 20.43% 17.41%17.21%16.20%15.49% 13.60%13.46%13.23%12.40%12.38%12.13%11.92%11.71%11.70%11.30% 10.00% a ho iti i nai o a ep. ongaT dov ot dan Ha caia as adorv ovos ci ao na nes m st agua epal R ol m enegr banil N batiri K M Les egovz orJ al Ja ondur Ko epubl Sa A H R Lebanon arci en, er ont Sl ppiil hi kiijaT N H M E P em dn dn anc Y ni Source: Global Economic Prospects 2006: Economic Implications of Remittances and Migration, World Bank. 69 Table D.1: Migration and Remittances: Summary Statistics for 2005, in percent of individuals Entire population Migrant Remittances Poverty rate 37.2 30.4 29.8 Mean Consumption per adult equivalent 1 58 63.3 63.2 Mean Food Consumption per adult equivalent 1 27.5 28.1 27.9 % of population 100 25.9 21.4 Unemployment rate 28.1 32 34.2 Urban/rural Distribution Urban 36.2 28.7 27.6 Rural 63.8 71.3 72.4 Total 100 100 100 Regional Distribution Gjakova 11.5 17.1 15.5 Cjilani 12 10.2 12.4 Mitrovica 15.1 20.6 20.4 Peja 11.2 13.2 11.3 Prizreni 15.7 18.1 20.1 Prishtina 23.3 13.2 12.2 Ferizaji 11.2 7.6 8.1 Total 100 100 100 Ethnic Area Distribution 2 Albanian ethnic area 96 99 99.1 Serbian ethnic area 4 1 0.9 Total 100 100 100 Educational Characteristics Secondary enrollment rate 69.5 60.8 61.5 Male secondary enrollment rate 74 62.4 64.7 Female secondary enrollment rate 64.7 58.5 55.9 Tertiary enrollment rate 14.7 14.6 15 Male tertiary enrollment rate 16 16.1 15.3 Female tertiary enrollment rate 13.3 13.1 14.7 Source: World Bank staff calculations from HBS 2005 population-weighted data. Notes: 1 In 2002 Euros. 2 Ethnic area as defined in survey design. Unemployment rate is only approximate because of lack of inactivity data, thus the figures are just percent of population reporting being unemployed. Net enrollment rates are only approximate as we assume that those who reported being a student and are aged 15-17 attend secondary school and those aged 19-23 attend university. Table D.2: Urban and Rural Households with Remittances and Migrants, in Percent of Individuals Urban Rural Entire HH HH Entire HH HH population with a receiving population with a receiving Migrant Remittances Migrant Remittances Poverty rate 32.4 38 40.6 39.9 27.3 25.7 Mean Cons per AE 64.1 62.7 61.3 54.6 63.6 64 Mean Food Cons per AE 33.6 33.8 33.4 23.9 25.8 25.8 % of population 100 20.5 16.3 100 29 24.2 Distribution across regions Of which: Gjakova 15.2 15.5 12.7 9.4 17.8 16.5 Cjilani 10.9 7.7 9.7 12.6 11.2 13.4 70 Mitrovica 14.7 30.4 31.5 15.4 16.6 16.2 Peja 12.5 16.4 13 10.4 12 10.7 Prizreni 11.6 7 9 18.1 22.6 24.3 Prishtina 23 13.1 12.5 23.5 13.3 12.1 Ferizaji 12.1 10 11.6 10.6 6.6 6.8 Total 100 100 100 100 100 100 Distribution across ethnic areas Albanian ethnic area 96.3 97.6 97 95.9 99.5 99.9 Serbian ethnic area 3.7 2.4 3 4.1 0.5 0.1 Total 100 100 100 100 100 100 Unemployment rate 27 35.3 39.7 28.8 30.6 32 Enrollment rates Secondary enrollment rate 88.5 81.5 72.3 63.6 59.5 61.2 Male secondary enrollment 87.7 84.9 77.8 69.7 63.9 66.8 rate Female secondary 89.5 79.5 68.1 57.3 53.5 51.8 enrollment rate Tertiary enrollment rate 20.7 21.2 22 11.7 13.2 13.5 Male tertiary enrollment 22 19.3 19.7 13.7 15.4 14.4 rate Female tertiary enrollment 19.7 22.8 24.3 9.7 10.8 12.6 rate Source: World Bank staff calculations from HBS data for 2005. 71 Table D.3: Regression Results for the 2-stage Estimation of the Effect of Having a Migrant on the Welfare of the Household Pr(Migrant) Log of Pr(Migrant) Log of Consumption Consumption Predicted probability of a 0.729** 0.694** migrant in HH [0.239] [0.237] HH has member not born in -0.027 this municipality (d) [0.029] Land endowment dummy (d) 0.102** 0.100** [0.033] [0.032] Age of the household head 0.005*** -0.001 0.005*** -0.001 [0.001] [0.002] [0.001] [0.002] Female HH head (d) 0.258*** -0.13 0.260*** -0.122 [0.065] [0.067] [0.065] [0.067] Serbian (d) -0.219*** -0.284** -0.220*** -0.291** [0.017] [0.103] [0.017] [0.103] Other (d) -0.132*** -0.216* -0.134*** -0.221* [0.039] [0.085] [0.039] [0.085] Secondary (d) -0.059 0.209*** -0.06 0.207*** [0.036] [0.039] [0.036] [0.039] Vocational (d) -0.062 0.326*** -0.061 0.323*** [0.047] [0.048] [0.047] [0.048] Tertiary (d) -0.140*** 0.479*** -0.139*** 0.474*** [0.039] [0.066] [0.039] [0.066] Dependency ratio -0.02 -0.077*** -0.02 -0.078*** [0.019] [0.020] [0.020] [0.020] Number of students in HH -0.01 -0.028 -0.01 -0.028 [0.017] [0.019] [0.017] [0.019] Number of unemployed in 0.019* -0.096*** 0.019* -0.095*** HH [0.009] [0.014] [0.009] [0.014] Main source of income: -0.037 0.109*** -0.037 0.108*** public sector (d) [0.031] [0.030] [0.030] [0.030] Main source of income: -0.140*** 0.09 -0.140*** 0.085 agriculture (d) [0.024] [0.056] [0.023] [0.056] Brick/cement walls (d) 0.116** -0.076 0.117** -0.072 [0.042] [0.061] [0.042] [0.061] Central district heating (d) 0.098 0.155** 0.099 0.157** [0.056] [0.051] [0.057] [0.052] Access to inside water tap (d) 0.104*** 0.193*** 0.105*** 0.196*** [0.028] [0.046] [0.028] [0.046] Purchase of 5 most-common 0.019 0.002 0.02 0.003 durables (lighting, textiles) in last 1 year (ln) [0.011] [0.011] [0.011] [0.011] Urban area dummy (d) -0.065* 0.046 -0.063 0.042 [0.033] [0.040] [0.033] [0.040] Gjilani (d) -0.092 0.427*** -0.082 0.424*** 72 [0.050] [0.062] [0.053] [0.061] Mitrovica (d) -0.045 -0.166** -0.033 -0.167** [0.048] [0.057] [0.051] [0.056] Peja (d) -0.012 0.145* -0.002 0.144* [0.053] [0.060] [0.055] [0.060] Prizreni (d) -0.06 0.140* -0.054 0.138* [0.051] [0.055] [0.052] [0.055] Prishtina (d) -0.209*** 0.323*** -0.206*** 0.315*** [0.036] [0.090] [0.036] [0.091] Ferizaji (d) -0.176*** 0.076 -0.172*** 0.069 [0.029] [0.070] [0.029] [0.069] Constant 3.515*** 3.516*** [0.105] [0.105] R-squared 0.318 0.318 N 2312 2312 2312 2312 (d) for discrete change of dummy variable from 0 to 1 * p<0.05, ** p<0.01, *** p<0.001 Source: World Bank staff calculations from HBS data. Table D.4: Propensity Score Matching Results for Consumption Variable Sample Treated Controls Difference S.E. T-stat Log of Consumption Total Unmatched 4.09 3.91 0.19 0.03 6.74 Average Treatment effect for the Treated (ATT) 4.09 3.91 0.19 0.04 4.79 Average Treatment effect for the Untreated(ATU) 3.91 4.11 0.21 Average Treatment Effect (ATE) 0.20 Urban Unmatched 4.09 3.96 0.13 0.04 3.09 Average Treatment effect for the Treated (ATT) 4.09 3.90 0.19 0.06 3.44 Average Treatment effect for the Untreated(ATU) 3.96 4.14 0.18 Average Treatment Effect (ATE) 0.18 Rural Unmatched 4.09 3.82 0.27 0.04 7.44 Average Treatment effect for the Treated (ATT) 4.09 3.83 0.26 0.05 4.91 Average Treatment effect for the Untreated(ATU) 3.82 4.05 0.22 Average Treatment Effect (ATE) 0.23 Source: World Bank staff calculations from HBS data for 2005. Propensity score method used is single nearest member. 73 Figure D.3: Propensity Score Matching for Migration and in (Consumption) tnelav .3 ui 0.3 0.3 eqt ul adrep .2 0.2 0.2 0.2 noi pt 0.1 umsno .1 Clae R Ln 0 Total Urban Rural Average Treatment Effect Unmatched for the Treated Source: World Bank staff calculations from HBS data. Table D.5: Poverty Rates among Migrant and Non-migrant Households, Propensity Score Matching Variable Sample Treated Controls Difference S.E. T-stat Rural Poor Unmatched 25.5% 44.6% -19.1% 0.03 -5.70 ATT 25.5% 45.5% -20.0% 0.05 -3.93 ATU 44.6% 32.9% -11.7% ATE -14.1% Poor Unmatched 28.3% 34.2% -5.9% 0.03 -1.80 ATT 28.3% 36.3% -8.0% 0.05 -1.66 ATU 34.2% 24.2% -10.0% ATE -9.6% Source: World Bank staff calculations from HBS data. Differences in poverty rates with other tables come from the unweighted results. Estimates are at the household level. 74 Figure D.4: Propensity Score and its Frequencies for Treated (households with migrants) and Untreated Groups A. Total B. Urban C. Rural 0 .2 .4 .6 .8 0 .2 .4 .6 .8 1 Propensity Score 0 .2 .4 .6 .8 Propensity Score Propensity Score Untreated Treated Untreated Treated Untreated Treated Table D.6: 2-stage IV Regression for the Effect of Having a Migrant in the HH on Consumption, by Urban and Rural Total Urban Rural Dependent Migrant Log of Migrant Log of Migrant Log of Variable in HH Consumption in HH Consumption in HH) Consumption Predicted 0.675* 0.244 0.597 probability of a migrant in HH [0.261] [0.504] [0.308] HH has member -0.086 -0.021 -0.156* born outside of Kosovo (d) [0.052] [0.055] [0.072] Instruments HH has member 0.003* 0.002 0.007 not born in Kosovo * Years this member is in Kosovo [0.002] [0.001] [0.004] Land endowment 0.087** 0.031 0.098 dummy (d) [0.033] [0.034] [0.055] Other controls, same as in Table D.3, omitted 75 REFERENCES AMP Kosovo. 2006. The Rural Development Context of Kosovo. Agricultural and Rural Development Plan for Kosovo. Downloaded from http://www.ampkosovo.com/ardp1.html Andersson, Claes. 2002a. Household Budget Survey ­ Sample Design. Report from a mission to the Statistical Office of Kosovo. March 11-15,2002. Statistics Sweden. Andersson, Claes. 2002b. Household Budget Survey: Report from a second mission to the Statistical Office of Kosovo, May 20-24, 2002. Statistics Sweden. Andersson, Claes. 2002c. Household Budget Survey: Report from a second mission to the Statistical Office of Kosovo. Oct 7-11, 2002. Statistics Sweden. Bhaumik, S., Ira Gang and Myeong-Su Yun. 2006. "A Note on Poverty in Kosovo." Journal of International Development. 18: 1177-1187. Bhaumik, S., Ira Gang and Myeong-Su Yun. 2005. "Ethnic Conflict and Economic Disparity: Serbians and Albanians in Kosovo." The William Davidson Institute, University of Michigan, Working Paper Number 808. Deaton, A. and Zaidi, S. 2002. Guidelines for Constructing Consumption Aggregates for Welfare Analysis. LSMS Working Paper 135. Deaton, A. and Dreze, J. 2002. Poverty and Inequality in India: A Re-Examination. Economic and Political Weekly September 7, 2002 Deaton, A. 2003. Adjusted Indian Poverty Estimates for 1999-2000. Economic and Political Weekly January 25, 2003 Demery, L. and Groootaert, C. 1993. Correcting for Sampling Bias in the Measurement of Welfare and Poverty in the Cote D'Ivoire Living Standards Survey. The World Bank Economic Review. Vol 77 No 2: 263-292. Docquier, F. and A. Marfouk. 2006. "International migration by educational attainment (1990-2000)", in: Ozden, C. et M. Schi¤ (eds), International migration, remittances and the brain drain, Chapter 5, Palgrave-Macmillan. Elbers, C., Lanjouw, J., Lanjouw, P.,. 2003. Micro-level Estimation of Poverty and Inequality. Econometrica, Vol. 71, No 1 (January, 2003), 255-364. El-laithy, Heba, Michael Lokshin and Arup Banerji. 2003. "Poverty and Economic Growth in Egypt, 1995-2000." Policy Research Working Paper, No. 3068. World Bank, Washington D.C. European Stability Initiative (ESI). 2006 (September). Cutting the lifeline: Migration, Families and the Future of Kosovo. Berlin-Istanbul. Gibson, John. 2005. Statistical Tools and Estimation Methods for Poverty Measures Based on Cross- Sectional Household Surveys. Handbook on Poverty Statistics: Concepts, Methods and Policy Use, 128-205. Downloaded from http://unstats.un.org/unsd/methods/poverty/pdf/Chapter-5.pdf Grimm, M. and Gunther, I. 2004. "How to achieve pro-poor growth in a poor economy: The case of Burkina Faso" in Operationalizing Pro-Poor Growth. Downloaded from www.worldbank.org Howes, S. and Lanjouw, J. 1997. Poverty Comparisons and Household Survey Design. LSMS Working Paper Number 129. IMF, 2007. Aide ­ Mèmoire of the IMF Staff Mission to Kosovo, February 19-27, 2007. Downloaded from http://www.unmikonline.org/press/reports/am_feb_2007.pdf Lanjouw, P., and Kijima, Y. 2003. Poverty in India during the 1990s: A Regional Perspective. World Bank Policy Research Working Paper 3141. Lanjouw, J, and Lanjouw, P. 2001. "How to Compare Apples and Oranges: Poverty Measurement Based on Different Definitions of Consumption" Review of Income and Wealth Series 47, Number 1, March 2001. Lohr, S. 1999. Sampling - Design and Analysis. 1st Edition, Arizona State University, 0534353614 . Ravallion, M. 1994. "A Better way to set poverty lines, " Outreach #15, March 1994. Statistical Office of Kosovo (SOK). 2005a. Agricultural and Household Survey 2005. Series- Agricultural and Environment Statistics, Pristina, Kosovo. Statistical Office of Kosovo (SOK) 2005b. Labour Market Statistics 2004. Downloaded from http://www.ks-gov.net/esk/esk/pdf/english/social/labour_mark_stat_04.pdf Statistical Office of Kosovo (SOK) 2006a. Labour Market Statistics 2005. Downloaded from http://www.ks-gov.net/esk/esk/pdf/english/social/labour_mark_stat_05.pdf Statistical Office of Kosovo (SOK) 2006b. Household Consumption in Kosova 2003-2005. Downloaded from http://www.ks- gov.net/ESK/esk/pdf/english/social/living_cond_03_05/Hous_cons_kos_03_05.pdf Stifell, D. and Christiaensen, L. 2006. Tracking Poverty Over Time in the Absence of Comparable Consumption Data. WPS3810. Tarozzi, A. 2005. Calculating Comparable Statistics from Incomparable Surveys, with an Application to Poverty in India. Downloaded from http://www.econ.duke.edu/~taroz/TarozziComparability.pdf on 9/12/2006) World Bank. 2001. Kosovo: Poverty Assessment (In Two Volumes). Washington D.C. Report No. 23390-KOS World Bank. 2005a. Kosovo: Poverty Assessment. Washington D.C. Report No. 32378-XK World Bank. 2005b. Poverty in Eastern Europe and Central Asia. Washington D. C. 78 World Bank. 2006a. Kosovo: Public Expenditure and Institutional Review. Volume 1. Washington D.C. World Bank. 2006b. Kosovo: Public Expenditure and Institutional Review. Volume II. Washington D.C. World Bank. 2006c. Migration and Remittances: Eastern Europe and the Former Soviet Union. Edited by Ali Mansoor and Bryce Quillin. Washington, D.C. World Bank. 2006d. Global Economic Prospects 2006: Economic Implications of Remittances and Migration. Washington, D.C. 79 Report No. 39737-XK Kosovo Poverty Assessment (In Two Volumes) Volume II: Estimating Trends from Non-comparable Data October 3, 2007 Poverty Reduction and Economic Management Unit Europe and Central Asia Region Document of the World Bank VOLUME I1 TABLE OF CONTENTS CHAPTER 1: HOUSEHOLD BUDGET SURVEY (HBS) AND POVERTYMONITORING IN Kosovo ..................................................................................................................... A. There are Problemsof Data Comparability ............................................................. 7 8 (a) Problem # 1:Diary versus Recall.................................................................... 8 (b) Problem #2: Survey Design-Redefinition of Consumption Items................8 B(c)Sample Weights Introduce Additional Uncertainty................................................. 9 9 C 10 D Likely Consequences:Poverty Estimates.............................................................. ... Problem #3: Survey Design - LSMS versus HBS........................................... Likely Consequences:Consumption..................................................................... 12 CHAPTER 2: A. Post-Stratification .................................................................................................. POVERTY-ALTERNATIVEESTIMATES ...................................................... 15 17 B. Compare Only 2003 and 2005............................................................................... 19 D. Compare all the Years ........................................................................................... C. Comparable Consumption Aggregate Methodology............................................. 20 E. Comparison of Poverty Figures from the LSMS and HBS................................... 22 23 CHAPTER 3: A Recommendations................................................................................................ . CONCLUSIONS AND RECOMMENDATIONS................................................... 25 -25 ANNEXA: TABLES AND FIGURES ..................................................................................... 27 ANNEXB: RESULTS USING DIFFERENT SURVEY YEARDEFINITION................................ 33 3 ListofTables Table 1.1: PopulationSize by Survey Wave andYear ................................................................. 10 Table 1.2: Summary of Survey ConstraintsandTheir Effectson PovertyEstimates..................12 Table 1.3: PovertyHeadcountbyLocationandEthnicareas, usingPA05 methodology............13 Table 2.1: Overviewof the Resultsof Methodologies for ComparablePovertyEstimates.........14 Table 1.4: PovertyHeadcountby HouseholdHeadEthnicity...................................................... Table 2.2: Summary of Poverty Estimatesfromthe MethodologiesUsed................................... 17 17 Table 2.3: PovertyRateswith CurrentWeights andReweighted ................................................ Table 2.4: Samplingprocedurefor the Bosniaand Herzegovina'sHouseholdBudget Survey .... 18 19 Table 2.5: PovertyRateswith the PA05 and ComparableCA methodologies............................ 20 Table 2.6: RobustPovertyLinesBasedon ConsistentFoodItems .............................................. Table 2.7: PovertyRatesusingthe AbbreviatedConsumptionBundleMethodology .................21 21 Table A.1: ComparisonofPreviousMethodologies .................................................................... 27 Table A.2: Survey Comparison Table A.3: PercentChangesinMainAggregates from Survey to Survey Comparison...............29 .................................................................................................... 30 Table A.4: AlternativeConsumptionAggregate DefinitionsandPovertyRates .......................... 30 Table A.6: Definitionof ConsumptionAggregates for the DifferentMethodologies..................-32 Table A.5: Consistently Asked Questionsover the Four Surveys ................................................ 31 Table A.7: Poverty Linesin DifferentMethodologies................................................................. 32 Table B.1: IntroductionofNewQuestionnaires........................................................................... 33 Table B.2: PovertyStatistics usingPA05 Methodology Table B.3: PovertyRatesUsingPA05 Methodology................................................................... .............................................................. 33 Table B.4: DetailedPoverty Diagnostics with RevisedConsumptionAggregate.,...................... 34 Table B.5: PovertyRatesUsingAlternativeConsumptionandPoverty LineMethodologies .....34 35 List of Boxes Box 2.1: BosniaandHerzegovinaHBS: Exampleof Samplingwithout a Census..................... Box 2.2: Analysis ofChanges...................................................................................................... 19 22 ListofFigures Figure 1.1: Total Populationin Millions and HouseholdSize ....................................................... Figure 1.2: Average MonthlyHouseholdConsumption, inNominalprices................................ 9 Figure 1.3: PovertyRateEstimatesandthe EffectofChanges inthe Questionnaire...................10 13 the Population................................................................................................................................ Figure2.1: Cumulativeand DensityDistributionofConsumption for the Bottom50 percentileof 18 4 ACKNOWLEDGEMENTS This report is a joint production of Statistical Office of Kosovo staff in the Household Budget Survey unit comprising Bashkim Bellaqa, Bekim Canoli, and Emina Deliu and World Bank technical team comprising Andrew Dabalen and Anna Gueorguieva, supported by Sasun Tsirunyan and Shpend Ahmeti. The report has benefited from the support of UK's Department for International Development which has generously funded the Trust Fund to support the capacity buildingand analyticactivitiesofthe WesternBalkanProgrammaticPovertywork. The reportwould not have beenpossible without the very close involvementand support ofthe Social Statistics Department ofthe StatisticalOfficeofKosovo. The team graciously acknowledgesthe analyticwork of the IMF (Macro statistics), EAR and Ministry of Agricultureof the PISG, SOK andVllaznim Bytyqi(Migration). The team has benefited from the comments and guidance of Peter Lanjouw (Peer Reviewer), Pierella Paci (Peer Reviewer), Asad Alam, Ardo Hansson, ElisabethHuybens, Felix Martin, Kanthan Shankar, Julian Lampietti, Ruslan Yemtsov, Kinnon Scott, Gero Carletto, Marcus Goldstein, Gabriel Demombynes, Juan Munoz, and Johan Mistiaen for excellent commentsand suggestions. The productionofthis reportbenefited enormouslyfrom the excellenteditingskills of SusanaPadilla. 5 6 CHAPTER 1: HOUSEHOLDBUDGET SURVEY(HBS)AND POVERTYMONITORINGINKOSOVO Since 2002, Kosovo has conducted annual Household Budget Surveys (HBS). At first glance, availability of annual cross-sections of detailed collection of household consumption expenditure data should suggest that one should be able to track poverty and inequality over time. However, examining changes in poverty and inequality over time in Kosovo poses several challenges. The main problem is data comparability because of (9 changes in survey design and (ii) large sampling errors. First, a wide variety of experience in other countries has shown that even small changes in the way expenditure/consumption or income data is collected can have a substantial impact on poverty estimates. These experiences have documented that differences in the poverty estimates over time could be driven by changes in survey design rather than by a real change in household welfare. The survey sampling weights, on the other hand, compound theproblem as they introduce an unquantijable bias or sampling error. The sampling was based on an outdated population frame and with limited survey supervision. In this note, we apply several methods to construct poverty estimates that are consistent over time. First, we make an attempt to construct a comparable consumption by aggregating items that were defined uniformly andfocusing only on the years where the questionnairedid not change. Second, we use an adjustmentprocedure that relies on afew variables whose definition has not changed over time to update the distribution of thepoor over time. The resultsfrom these various methods show that during the period from 2002 to 2006, poverty was high, at around 45 percent, and that there is no evidence of a sustainedimprovementin the welfare of households in Kosovo. The recommendationsfor data collection for poverty monitoring coming from this research are to,first, maintain consistency in the survey questionnaire, second, to conduct a population census, and, third, to emphasize better survey administration and documentation. 1.1 The first poverty assessment for Kosovo was done in 2001 on the basis of a Living Standard Measurement Survey of 2880 householdsconducted between September and December 2000. Although there was no existingcensus, effort was madeto create a representativesample of the populationof Kosovo. Up to date lists of households were created and a sample representative at areas of responsibility (AORs), ruralhrban, and AlbaniadSerbianethnicitywas drawn. 1.2 In June of 2002 Kosovo began to implement the Household Budget Survey (HBS). The HBS is implemented by the Statistical Office of Kosovo with technical assistance from Statistics Sweden, which in turn is financed by SIDA. To date four rounds of HBS have beencompleted(Table 1.1). SOK, together with Statistics Sweden, draw the sampleto be surveyedeach May. The first HBS survey began in June 2002 and 7 ran till May of the following year. The second survey (2003) followed the same cycle. But in 2005, SOK switched to calendar year (January to December of saymeyear) for the introductionof differences in the questionnairebut kept the timing of sampling the same at mid-year. Thus, currently, eachquestionnairespanstwo samples. 1.3 The Household Budget Survey provides a solid foundation for monitoring poverty in Kosovo. The H B S has become a core survey in KOSOVO~S efforts to build a long term monitoring and evaluation system. It has some of the basic tenets of a sustainable survey. It is fully funded by the government and implemented by the SOK staff (with technicalsupport from development partners). The H B S unit of SOK has also introduced innovations to the traditional H B S by including additional modules, most recent of which havebeenmigrationandremittances (2005) andtime use (2006). A. THEREARE PROBLEMS OFDATA COMPARABILITY 1.4 Examining changes in poverty and inequality over time in Kosovo poses several challenges. With a Living Measurements Standards Survey (LSMS)in 2000 and a series of H B S since 2002, it would seem tempting to conclude that tracking welfare changes in the first half of 2000 should be feasible. But there are practicalproblems. A major problem is that data are not comparable. There are three changes across surveys where efforts to compare data present difficultiesto trackingwelfare changes over time. Below we list each of these changes and discuss potential consequences for estimating changes in povertyand inequality. (a) Problem# 1:Diary versus Recall 1.5 The main change between HBS 2002 and subsequent HBS series is how households were asked to recall expenditures of goods and services bought. The first H B S asked households to record expenditures on a daily basis for two weeks. This appliedto food, own-produced consumption and most non-fooditems such as clothing, footwear, and education and health expenditures. A switch from a shorter to a longer recall period (diary to weekly) is likely to make households forget some details of consumption and therefore underreport consumption. The impact is likely to be severe for frequentlypurchaseditems such as food. (b) Problem#2: Survey Design-Redefinitionof ConsumptionItems 1.6 The secondchange which is likely to havean impact onthe comparabilityof data across H B S series is the level of disaggregation of the expenditure items. This took two forms. Inthe 2002 survey, householdsrecorded expenditure items on a blank sheet, but in subsequent years, the list was provided to the households. Between the first and second surveys, the lists did not exhibit substantial differences. It appears that households in the second survey were offeredthe same list that households interviewed in 2002 reported. However, by 2005, the level of disaggregation has increased and the list containedmore items. The more substantialchangewas in how consumption of own- produced items was reported. In this case, the items were aggregated into 12 categories (meat products, poultry, graincrops, and so on). Furthermore, inthe case of consumption of own-produced goods the recall period changed not from daily to weekly as in other items, but from daily to monthly. A shift from a smaller list to a longer list 8 (disaggregation) is likely to lead to higher reported consumption. (c) Problem #3: Survey Design LSMS versus HBS - 1.7 As the first post-conflict household survey, the LSMS estimates and profile o f poverty would be a good starting point to establish the baseline for the monitoring and evaluation system that is now anchored on HBS. However, except for the fact that the LSMS and the H B S are drawn from the same sample frame - that is, the households surveyed in HBS are selected from the same enumeration areas that were drawn for the LSMS - the two surveys differ in a number o f ways (Table A.1). First, distribution o f consumption may differ due to failure to account for seasonality in the LSMS. The latter was conducted for 3 months (September through December o f 2000), while H B S collects information from households (albeit different ones each month) throughout the year. If the three months when LSMS was fielded happen to be a period o f low (high) consumption, then the distribution o f consumption may be lower (higher) than the HBS distribution. Second, the recall period for consumption differs in the two surveys. In the LSMS, food and most frequently purchased non-food items had a recall o f a week or a month, while infrequently purchased goods and services had a 12 month recall. By contrast, the H B S first started with a two-week diary (daily recording) o f food and most non-food expenditures and then switched to a weekly recall. Finally, the LSMS provided households with a much narrower list o f expenditure items (46 food items) compared to HBS list that was over 100. Inpractical terms, a single change is hard to over come, but three makes the problem almost insurmountable. B. SAMPLE WEIGHTSINTRODUCE ADDITIONAL UNCERTAINTY 1.8 Sample weights introduce additional uncertainty. The Kosovo HBS uses the 1981census as the reference population. This is then updated every survey cycle through re-listing o f selected EAs, but it is not clear how the updated information is used in subsequent surveys. In addition to the outdated sampling frame, because o f resource constraints, field supervision o f surveys has been limited. Figure 1.1: Total Population in Millions and Household As a result, there is Size considerable uncertainty Sample Statistics surrounding the H B S *-I demographic statistics each year. shows the implied population count from the sample weights in the HBS. Within each wave o f the survey, it also presents the average household size o f sampled households by year and wave. The. estimated population appears to have 1 2 3 4 declined by about 25 percent between 2002 and 2005. Viewed from the perspective Source: World Bank staff calculations from the HBSdata. 9 o f this period, that is absence o f conflict and or unnatural mortality shocks, there is no clearjustification for this massive change in population estimate. 1.9 Strikingly, sometimes the surveys appear to come from completely different population groups. In particular, while average household size was near 7 in 2002, it drops to around 6 in 2005. Moreover, rural population shares change dramatically. For instance, the rural population decreases from 73 to 65 percent o f the total population. The 2001 LSMS reports rural population as 62.4 percent. The Agricultural Household Survey finds that rural population stayed at round 65 percent in 2004 and 2005. Based on experience in Albania (Carletto et al, 2004), we are expecting the incidence o f internal mobility to remain quite stable over time. One consequence o f this massive change in population estimate is to introduce huge volatility in the estimated count o f fraction o f people below the poverty line. Table 1.1: Population Size by Survey Wave and Year HBS estimates 2002-03 2003-04 2004-05 2005-06 Total population, inmillion 2.1 1.8 1.7 1.5 Number of households, inthousands 306.1 281.6 281 249.4 Householdsize 6.8 6.5 6.1 6.1 Rural as 'YOof total population 72.5 73.4 65.9 64.6 Reference population statistics 2001 - - 2002 2003 - - 2004 2005 Source LSMS LFS andAHS AHS AHS Total Population 1.97 1.9 Rural as % oftotal 62.4 Rural inmillion 1.23 1.3 1.3 1.3 Source: World Bank staff calculations from HBS data and LSMS: World Bank Kosovo Poverty Assessment (2001); Labor Force Survey(LFS) and Agricultural Household Survey (AHS) estimates are from the relevant SOK publications. C. LIKELYCONSEQUENCES:CONSUMPTION 1.10 Experiences around the world have documented the influence and magnitude o f the changes in recall period on consumption. In all cases, longer Figure 1.2: Average Monthly Household recall periods lead to less declared Consumption, in Nominal prices expenditures (Table 1.3). For instance, in India, households who were asked to report weekly food ' Average Monthly Expenditures expenditures had 15 to 20 percent higher per capita consumption than A those asked to report 30 day food expenditures, mainly because p households with shorter recall period reported higher per capita food cI expenditures (Tarozzi, 2002; Deaton, 2001). In another study, Deaton (2003) reports an experiment where reducing recall for food items period Source. World Bank staff calculations from the HBS data. from 30 to 7 days resulted in 30 10 percent higher consumption (1.1 percent per day). Amenuvegbe (1990) shows from Ghana household surveys that for 13 frequently purchased items, reportedexpenditures fell at an average of 2.9 percent for every day added. Lanjouw and Lanjouw (2001) showedthat variationsin food expenditure definitionsthat arise from a disaggregation of the list would lead to significant lower per capita consumption in countries such as Brazil, Ecuador, and El Salvador. For instance, fer capita monthly expenditures in El Salvador were 32 and 15 percent higher at the 10 and 90thpercentiles, respectively,for householdreceivingthe longlist. 1.11 Diagnostic work on Kosovo data indicates that expenditure data has been influenced by changes in recall period. The pattern is consistent with prior expectations as documented above in a number of other countries. It suggests, using Deaton (2003) results and notingthat food accounts for 50 percent of total consumption in Kosovo, that we shouldexpect at least 4 percent lower consumption in 2003 compared to 2002 from changes in recall periodalone (that is, 1.1 percentx 7 x 0.5).Inreality,we find that the mean of total consumption in 2002, which used the diary, was about 10 percent higher than the mean in 2003, where a weekly recall was used. It was15 percent higher than the mean in 2005. The mean of food consumption dropped by 13 percent between2002 and 2003, butby as muchas 21percentbetween2002 and 2005. 1.12 The effect of recall change may have been particularly severe for certain sub-components of consumption. As noted above, recording of own consumption underwent two substantial changes. One is the change in recall from daily to monthly. The other is that, in the second and subsequent surveys, households were given an aggregated list against which to record own consumption. More precisely, the list reported for own consumption changed from 85 in 2002 to 12 in subsequent surveys (Table 1.2). Both changes are likely to lead to underreportingof expenditures. Mean of own consumption fell by 4 percent between 2002 and 2003 and by 30 percent between 2002 and 2005. Giventhat small scale farmers -those with less than 3 hectares of land- report using 70 percent of their production for own consumption (SOK, 2005), the changes introduced in capturing this sub-component of consumption presents serious problems for a credible measure oftotal consumption,and ultimately,poverty inKosovo. 1.13 The possibility of survey design changes driving the changes in consumption (and therefore changes in welfare) cannot be ruled out. Food share fell from 61 to 54 percent between 2002 and 2005. In one view this could be an indication of households getting richer and substitutingaway from food to non-food. However, the evidence for this alternative hypothesis is not strong. First, the macroeconomic data shows a stable inflationregime (possibly even a deflation) and negligible output growth. Second, non- food expenditures remained stable across surveys in sharp contrast to food and its sub- components. Specifically, the share of sub-categories such as bread, meat or eggs and dairy out of total expenditures do not show evidence of substitution away from staples. Taken together, it appears that changes in recall period probably drive much of the observed changes, since as predicted these changes in recall periodare likely to have the biggest impact on frequently purchased items such as food. Simply put, since these changes in consumption (welfare) are observed in the context of several changes to survey design, it is difficult to argue credibly that observed changes are not due to changes in survey design. 11 1.14 Table 1.2: Summary of Survey Constraints and Their Effects on Poverty Estimates Survey and Possibleeffects References Evidence of effect in the Interaction and questionnaire HBS data final effect on design issues poverty Weak sampling Non- Demery and Population estimates: Interactswith all frame representative Grootaert (1994), 2002/03: 2.1 m other survey population. Howes and 2005/06: 1.5 m measurement Household size Lanjouw (1997) Rural proportion: errors. Leadsto and subgroups 2002103: 73% unquantifiable are not stable 2005/06: 65% (Table 1.1) biases. Change from Possibly an Currently no Own production drops by Poverty: open-ended to increase in controlled around30% from 02/03 to Underestimatedin close-ended reported experiments 05/06 05/06 or expenditure consumption (Volume I,Table A.1) overestimatedin questions estimates 02-03 Recall period Decrease in For survey, see Total food expenditure Poverty: change from reported Deatonand drops 21 % from 02/03 to Underestimatedin daily to weekly expenditureof Kozel(2005) 05/06 (Volume I,Table 05/06 or about 4%. A.1) overestimated in 02-03 Change in Decreasein Lanjouw and Own production drops by Interactswith number of reported Lanjouw (2001) around4% after the changes inrecall subcategories of expenditures and many others number of categories period and expenditures changes from over 85 to questiontype. reported 12 Cannot be singled out. Short recall Overstated Gibson (2005) Seasonality inpoverty Overstating period poverty estimates (Table B.4) poverty Source: World Bank staff calculationsfrom HBS data and relevant references. D. LIKELYCONSEQUENCES: POVERTYESTIMATES 1.15 A shift from diary to recall leads to underreporting of consumption, which in turn leads to higher estimated poverty rates. In 2002, the proportion o f people living below the poverty line was estimated at 37 percent. Using a consumption aggregate constructed in the same way and adjusted for inflation, the fraction o f the population below the poverty line increases to 44 percent in 2003, fell to 35 and increased back to 45 percent in 2005. 1.16 Viewed differently, the disaggregation o f consumption items is akin to introducing measurement error into a variable (Table 1.2). If the measurement error i s random, there will be no effect on the estimates o f the mean or the population total if the sample i s large enough. However, such errors will systematically bias poverty estimates. Figure 1.3 shows a situation where an accurate welfare indicator i s compared with an error-ridden indicator. The poverty rate is the area under the welfare function up to the poverty line and it will be affected both by imperfectly measured welfare indicator, or incorrectly specified poverty line. 1.17 12 Figure1.3: PovertyRateEstimatesand the Effectof Changesinthe Questionnaire A. PovertyRatesOver SurveyPeriods, B. TheEffectof RandomMeasurementError Absolute, Extreme on PovertyEstimates Source: World Bank staff calculations from HBS data. Source: Gibson (2005). 1.18 Samplingweights increase the volatility of the estimated poverty. Table 1.3 compares the estimated poverty rates with and without weights. A comparison of the weighted and un-weightedcolumns shows why using weights as currently constructed introduces volatility. The magnitude of changes is further overstated with the weighted statistics. For instance, for urban areas, the weighted poverty rates seem to drop by 5 percentage points whereas the un-weightedby only three. For rural, the value of the supposed increase in poverty is much smaller when the sampling weights are not included. These findings suggest the need for a consistent procedure for calculating samplingweights. Table 1.3: PovertyHeadcountby Locationand Ethnicareas, usingPA05 methodology Weighted Unweighted 2002-03 2003-04 2004-05 2005-06 2002-03 2003-04 2004-05 2005-06 Total 37.7 43.7 34.8 45 45.4 44.5 34.4 44.3 Rural 34.4 44.2 37.2 49.2 42.1 46.2 37.5 49 Urban 46.6 42.1 30.3 37.4 48.1 42.7 31.4 39.7 Albanianarea 37.8 43.8 34.9 43 45.3 45.1 34.5 41.7 Serbianarea 33.5 40.8 33.3 80.4 44.1 38.4 33.8 70.3 Source: World Bank staff calculations from HBS. Notes: Methodology as in the 2005 Poverty Assessment. Weighted refers to individual-level weights, unweighted to householdsize weights. 1.19 These uncertainties persist across several estimates. In additionto nationallevel estimates by wave, poverty rates were estimated for rural and urban residents and Albanian and Serb ethnic groups. For instance, estimates of poverty by ethnicity, whether definedas area occupied mainly by such an ethnic group or ethnicity of headof household, are highly volatile. For instance, the povertyrate for Serbs ranges from 35 to 80 percent. They are especially sensitiveto inclusionof own consumption. For instance, in where we present the poverty rates under different consumption aggregation with the same poverty line, the coefficient of variation (the standard deviation over the mean) of the poverty rate increasedwith the inclusionof own productionfor weighted figures. In 13 all cases, these problems of large changes betweenweighted and unweighted, and within a short time period, are observed. 1.20 The data from 2004-05 (wave 111) seems to be particularly problematic. This survey was done in the same way as waves I1and IV (that is, 2003-04 and 2005-06) so that in theory it should be comparable to these surveys. However, we find that it is particularlysensitive to the inclusionofconsumptionof non-food. The estimatedwelfare swings with and without inclusion of non-foodare (unrealistically) large. This leads to the conclusion that estimated poverty counts are not comparable, especially between 2002 and 2003. Inthe next chapter, we try to resolve this issue in a number of ways and providepreliminaryestimates ofpovertytrends inKosovo. Table 1.4: Poverty Headcount by Household Head Ethnicity Weighted Unweighted 2002-03 2003-04 2002-03 2003-04 2002-03 2003-04 2002-03 2003-04 Albanian 37.4 43.8 32.1 42.5 45.2 45 32.9 41.1 Serbian 30.1 36 34.3 81.7 39.1 35.7 33.8 70.2 Other 57.6 53 67.2 51.7 58.2 59.5 57.6 56 Source: World Bank staffcalculationsfrom HBS. Methodologyas inthe 2005 PovertyAssessment. Weighted refers to individual-level weights, unweightedto householdsize weights. 14 CHAPTER2: POVERTY-ALTERNATIVE ESTIMATES 2.1 The dual problem of (i)possible survey bias inthe data, and (ii)numerous changes in questionnairedesign, make HBS survey estimates merelysuggestive of a trendand shouldbe used only as a guide by policy makers. Numerous changes in survey design do not leadto conclusive comparisons on the levels and trends in poverty between 2002 and 2005. We have shownthat a shift from diary to a weekly or longer recall period,from 2002 to 2003 and thereafter, respectively, is likely to leadto underreportinginconsumption and therefore over- estimationof poverty rates. We have also discussed that aggregation of own consumption items from 85 to 12, in 2002 comparedto 2003 and thereafter, adds to the underreportingof consumption (and by consequence over-estimation of poverty) problems in second and subsequent waves. Finally, the sampling methodology, which indicates a larger population and higher household size in 2002 compared to 2003 and thereafter, is likely to reduce per capita consumption and, for a given poverty line, under-estimatepoverty in 2002 relativeto 2003 and thereafter. While we know the possible direction of impact of these changes in design on consumption and poverty, it is not possible to know with precisionthe magnitude of these changes on consumption or poverty. That is why, the searchfor alternativemethods to establishcomparabilitybecomesnecessary. 2.2 We employ several estimation techniques to correct for some of these problems. In order, we present a brief descriptionof the steps taken to address the (a) sampling issues and(b) non-comparablewelfare measures. (A) Sampling issues: As the HBS data is based on an outdated sample and the survey supervision is very limited, the data suffer from a possible bias. To rectify a part of this problem, we use a post-stratificationprocedure. This method calibrates the weights to make demographic estimates from HBS comparable to external sources. 2.3 Even if sampling issues are addressed, the problem of non-comparability of consumption estimates still persists. Therefore, we apply the following steps to rectify this secondproblem: (B) Comparability of welfare measures: We use two main methods to provide comparableconsumption aggregates. Compare only 2003 and 2005: Since the biggest and the most problematicchanges took place between2002 and2003, one strategy is to ignorethe 2002 survey and start the analysis of poverty from 2003. As a reminder, the 2003 through 2005 data have the same recallperiod. The level of item didaggregationcan also be considered the same, since only minor changes were introduced. For instance, food items declined from 114 to 107 between 2003 and 2004, and similar changes were introduced in non-food items. But overall, the number of the changes in consumption items and their contribution to aggregate consumption were negligible. Our justification for 15 excluding2004 survey is that welfare changes are very sensitive to inclusionof non- food consumption. Therefore, we use three methodologies to compare poverty between2003 and 2005: 0 First, we use the same constructionof consumption aggregate and poverty line as was used for the previous two Poverty Assessments. We refer to this as PA05 methodology(short for PovertyAssessment 2005). Then, we directly compare the povertyrates. 0 Second, a method developed by Lanjouw and Lanjouw (2001) is used to construct a Comparable Consumption Aggregate that includes only consistently recordedexpenditureitems and least volatile items. 0 Third, we constructan Abbreviated ConsumptionBundleconsistingonly of products for which price informationwas collectedby the price unit of the SOK in order to re-calculatethe poverty linefor 2003-04 data 0 Compare all the years: The final optionis to compare all the years. However, as argued above, this cannot be done without additionaladjustment. There are two candidatemethodsfor adjusting povertyratesto arriveat comparability. 0 The first method, calledinverseprobability weighting, aims to matchthe distribution of consumption or any welfare measure between the two surveys. It reweighs the poverty count in 2002 using as weights the probability of an observation belonging to a comparison survey, say year 2003 (Tarozzi, 2005; DiNardo et. al. 1996). Similarly one can compare 2002 to 2004 and2002 to 2005. 0 The second method, which we shall refer to as econometric projection methodology, is to estimate a consumption modelusingthe 2002 data, and then use the estimated parameters from the 2002 model to forecast or predictthe consumptionfor subsequentyears. The final step is to addto the forecast consumption an estimate of unobserved part of consumption (the error term) in order to recover full consumption. The results from this methodology are notyet complete and are notreportedhere. 2.4 All methods suggest that the povertyrate in Kosovo remained in the mid 40s percent from 2002 to 2005. The results of the different estimation methodologies are presented in (Table 2.1). Although the trends are not consistently pointing to the same direction, the pattern that emerges is one of stagnating poverty. The PA05 and abbreviated consumption methods imply a stagnant poverty rate: a change from 44 to 45 percent. The Comparable Consumption Aggregate, suggest a slight decrease from 49 percent in 2003-04 to 46 percent in 2005-06. The Inverse Probability Weighting methodology also confirms that poverty remained very similar from 2002-03 to 2005-06, with only a small increase of about 3 percentage points from 2003 to 2005. These conclusions do not change substantively if survey waves are defined differently. Specificallywhen usingcalendar year 2005 as the last wave, the results show only a small decline in poverty. See Annex B and particularlyTable B.5. 16 Table 2.1: Overview of the Results of Methodologies for Comparable Poverty Estimates ConsumDtion Aggregate (CA) Povertv Line definition definition Poverty Rates CA with PA 05 methodology Poverty line 2002 adjusted with CPI weighted unweighted 2002-03 37.6 45.7 2003-04 43.6 44.5 2004-05 34.8 34.4 2005-06 45 44.3 Comparable CA (Lanjouwand Robust Poverty Line Lanjouw,2001) weighted unweighted 2002-03 2003-04 48.5 48.7 2004-05 41.3 41.3 2005-06 45.6 44.7 CA IUsing Inverse Probability 2002 Poverty Line Weighting (Tarozzi, 2005) weighted unweighted 2002-03 42.7 2003-04 2004-05 2005-06 45.6 Source: World Bank staff estimates from HBS data. 2.5 The results presented in the main part of the report (Volume I) for 2003/04 and are 2005/06 only, re-weightedto match non-HBS based rural and urbanpopulationestimates. In this volume, we present the results from the methodologies discussed above in order to see whether the mainresultofVolume I, of unchangingpovertytrend, is confirmed. Inshort, that inthis volume we undertakea sensitivityanalysis. Table 2.2: Summary of Poverty Estimates from the Methodologies Used Methodology Base year Final year Change Poverty rates A. Samplingissues 2003104 2005106 1. Post-stratification 43.5 45.1 About the same B. Comparability of Welfare Measures Compareonly 2003104 and2005106 2003104 2005106 2. PA05 43.6 45 About the same 3. ComparableConsumptionAggregate 48.5 45.6 Slight decrease 4. AbbreviatedConsumptionBundle 36.3 36.2 The same Compare all surveys2002-2005 2002103 2005106 5. InverseProbabilityWeighting 42.7 45.6 Slight increase Source: World Bank staff estimates from HBS data. A. POST-STRATIFICATION 2.6 Because of an outdated sampling frame and resource-constrained limited survey supervision, the H B S sample is likely to be affected by non-negligible sampling and non- sampling errors. As discussed in the survey sampleseach year appear to come from different populations. This is reflected also in the distribution of the consumption aggregate. As 17 Figure 2.1 shows, the cumulative distributions of consumption from year to year even indicate a stochastic dominance of 2005106over wave 2002103 and 2004105. This figure also shows the similarity of 2003104 and 2005106 and wave 2002103 and 2004105. We suspect that these patterns may be driven by both changes in the questionnaire and the sampling procedure. Figure 2.1: Cumulative and Density Distribution of Consumption for the Bottom 50 percentile of the Population C ionof n Source: World Bank staff estimates from HBS data. 2.7 The sampling process and survey administration is poorly documented. The quality of the list of EAs is poor:the distinctionbetweenurban and rural is purely administrative;the classificationby ethnicity does not follow strict rules, and the descriptionof the geographical boundaries of the EAs is outdated (Andersson, 2002a). In addition, due to Table 2.3: Poverty Rates with Current Weights lackproper supervision misclassified Wave Population Averak and Rewei hted Of Extreme Absolute EAs were skipped (Andersson, 2002c), estimates household poverty poverty relisting of large EAs may be (million) size rate ('YO) rate ('YO) incomplete and field control of Current SamplingWeights enumerators is lacking. Some areas that 2002-03 2.05 6.8 15.0 37.8 were heavily populated in 1981 are 2003-04 1.82 6.5 13.3 43.7 currently not and vise versa This 2004-05 1.71 6.1 10.7 35.5 introduces large sampling errors and 2005-06 1.52 6.1 16.5 45.6 possibly bias to the HI3S estimates. Reweighted There are also issues of undercoverage. 2002-03 1.9 8.5 15.3 38.7 2003-04 1.9 8.0 13.6 43.5 2004-05 1.9 7.4 10.6 34.8 2.8 The reweighting methodology 2005-06 1.9 7.7 16.7 45.1 adjusts the sampling weights attached to each surveyed householdso that the urban and rural populationmatch non-HBS based data Generally survey data and its sampling weights are re-calibrated and post-stratification weights are used to match the distribution to some external data (Lohr, 1999). The adjustment methodology is simple and it uses a scaling factor so that the weighted total population size in all surveys matches that of external sources. Then it also matches the distribution of rural and urban households as compared to that of other surveys (Table 1.1). The resultingweighted populationtotal and household size is muchmore comparable (second half of (Table 2.3). We also match household size distribution in each stratum and obtain very similar results. 18 2.9 The re-weighted poverty rate confirms the time trend of unchanging poverty over time, while the volatility ofthe estimateshas decreased. The poverty rate, when re-weighted, i s again around 45 percent for 2003-04 and 2005-06. At the same time, its decrease in 2002 and 2004 is smaller than when calculated without post-stratification. This procedure, however, seems insufficient in equating the samples. As next steps, the analysis will adjust for other aggregates on which official data is available, as for instance pensioners and students. Box 2.1: Bosnia and Herzegovina HBS: Example of Sampling without a Census Bosnia and Herzegovina's HBS sampling faced similar constraints to those of Kosovo. First, there were no populationregisters or housing registers to be used as sampling frames. Second, there was possibly considerable internal migrationand rapid change amongst the housingstock. Third, the statistical office staff had limited resources and little experience of general population sampling methods (Lynn, 2004). The Bosnia and Herzegovina HBS sampling process follows the steps identified in Table 2.4. The procedure is similar to what currently SOK employs except for several noteworthy differences: census EAs are well delineated and stratified; relisting and questionnaire administrationis better supervised; use ofequal probabilitiesbothat the stage of selectingPSUs and at the stage of selecting households within PSUs. Table 2.4: Sampling procedure for the Bosnia and Herzegovina's Household Budget Survey Stage of Steps Time sampling Implemented only once Pre-sampling Revisedthe census EAs to ensure comprehensivenessand 5 months appropriatemaps Fieldtest A systematic randomsample of 50 EAs to find percent of 1 month unoccupieddwellings. Implementrelisting procedure and follow up visit. Implemented before the survey each year 1st stage Systematic equal-probabilitystratifiedsamplingof 3.65% of EAs. Relisting Semi-intrusiveapproach(observationwhere possible, contact 3 weeks elsewhere). About 1 day visit per EA. 2nd stage Systematic selection of Households from the relist (about 25% of all relisted). Systematic division ofthe sampledhouseholdsinto 12 monthly samples Source: Lynn (2004). B. COMPARE ONLY 2003 AND 2005 PA05 Methodology 2.10 The poverty rate is around 45 percent inboth 2003-04 and 2005-06 with a substantial decline in 2004-05 that is as yet unexplained. We use three methods to compare the poverty rates between 2003 and 2005. The first method uses the same poverty line used for the poverty assessment of 2005 (PA05), adjusted for inflation to estimate the poverty rates. A comparison o f all three years shows that poverty levels remained stagnant between the start and end o f the period. The poverty rate was at 44 percent in 2003 and 45 in 2005. But in 2004105 there i s a large drop in poverty, to 35 percent. While the pattern of change is consistent with the macroeconomic developments - there was a 2.6 percentage point 19 turnaround in GDP growth between 2003-04 and 2004-05 --such a decrease over a short period of time implies unusually high growthelasticityof povertyreduction'. Table 2.5: Poverty Rates with the PA05 and Comparable CA methodologies PA05 Comparable 2.11 Although the last 3 HE3S surveys ' methodology CA appear very similar and seem to be prime Povertyline 2002 PL Robust PL candidates for comparable poverty Povertvrates estimates, changes in the aggregation of 2003-04 43.7 48.5 food items could affect the poverty figures. 2004-05 34.8 41.3 The 2003-2005 HBS surveys usedthe same 2005-06 45 45.6 recall period. Generally, there is a Source: World Bank staffestimatesfrom HBS data. presumption that the groups surveyed are similar: the samples were drawn from three adjacent time periods, between which there had been no expectation of a marked change in poverty. However, they used different levels of aggregation: for instance, there are 107 food items in 2005-06 and 114 in 2003-04 and 2004- 05 surveys. Several additional non-food consumption items were added. Possibly, the changes in survey design produced a (misleading)appearanceof a drop and then an increase in poverty. The 2004-05 is particularlyproblematicand as hasbeenmentionedvery sensitive to inclusionofconsumptionof non-food. C. COMPARABLE CONSUMPTION AGGREGATEMETHODOLOGY 2.12 The second method, which adjusts the poverty line to account for survey-design induced volatility of consumption sub-components shows a slight decline in poverty. We noted that consumption and welfare estimates for 2004/05 survey were noticeably more sensitive to inclusion of consumption of non-food. To address this concern, we use a methodology (Comparable Consumption Aggregate) which constructs the poverty line each year. First, we construct a food poverty line for a reference population using only comparable food consumption items. Then we construct an absolute poverty line each year, non-parametrically(see Box 2.1). 2.13 The differences between the robust and the poverty line from the 2005 Poverty Assessment (Table 2.6) is not only the result of inflationover the period,but also reflects the fact that the 2005 survey embodies a more comprehensive consumption definitionthan 2003 and 2004 surveysas well as the issues arisingfrom biasedsampling and measurementerror in the second half of 2004. On the basis of these robust poverty lines, the incidence of poverty in Kosovo decreased slightly from 48 percent in 2003-04 to 46 percent in 2005-06. This contrasts with the observation that poverty increasedslightly from 2003 to 2005 when only inflation is adjusted for. In addition, the magnitudeof the drop between 2003-04 and 2004- 1 Most likely, the reported higher expenditure by households i s due to survey administration and sampling issues. As shown in the previous sections, the survey methodology could be introducing an unquantifiable bias. Measurement error is also a big concern for the Kosovo HBS as described earlier. Because of limited resources and capacity, survey administration is not at par with international standards: enumerator supervision is compromised while the incentives for respondents changed. This unknown measurementerror poses a special challenge when the focus is on poverty and other distributional statistics, rather than on means and totals. While random measurement error should not affect estimates of the mean or the populationtotal if the sample is large enough, such errors will systematically bias poverty estimates (Gibson, 2005). For poverty rates and other variance-based statistics, the effect of random errors accumulates so errors in measuring household level welfare will be reflectedin inaccurate estimatesof aggregatepovertyrates. 20 05 is much smaller than when consumption o f own-production is included. The trend now shows that poverty declines from 48 percet to 41 percent between 2003 and 2004. Table 2.6: Robust PovertyLinesBasedon Consistent Food Items. Food Poverty Line ExcludedOwn Production 2003-04 2004-05 2005-06 RobustFoodpoverty line 26.2 27.11 22.75 Robustfinal poverty line 41.84 44.2 1 40.39 CPI adjustedPA05 food poverty line 28.35 28.35 28.34 CPI adjustedPA05 poverty line 43 43.01 43 Source: World Bank staff estimates from HBS data. In Euros per adult equivalent, monthly, in June 2002 prices. AbbreviatedConsumption BundleMethodology 2.14 This fourth methodology re-calculates the poverty line for 2003-04 data (second wave) usingnon-HBS price information of 40 items. The calculation of poverty line is based on the household total consumption of certain reference population. Thus, the poverty line calculated for 2002-03 data in the 2005 Poverty Assessment is based on the consumption recorded in 2002-03. As we pointed earlier, consumption in 2002-03 was recorded using a diary method and it is different from later years. Unfortunately, for 2003-04 survey no price information was collected that can allow us to replicate the poverty line for that data Using non-HBS price information we are able to calculate the cost of an abbreviated consumption bundleof 40 items. Table 2.7: PovertyRates usingthe Abbreviated ConsumptionBundleMethodology Survey Adjusted Adult Food line Complete Extreme Complete wave Equivalent Poverty line Poverty Poverty Consumption Rate Rate 2003-04 52.26 22.39 39.01 5.85 36.28 2004-05 59.73 21.84 38.05 4.84 27.23 2005-06 52.12 21.83 38.03 8.89 36.19 units Euro/month Euro/month Euro/month % YO Source: World Bank staffcalculationsfrom HBS data. Wave 2 Poverty line is recalculatedusing40 major food item. The povertylinesfor waves 1,3,4 are deflatedfrom wave 2 povertylines usingCPI. 2.15 Based on these new poverty lines, poverty rates in Kosovo remained stagnant from 2003 to 2005, thus confirming results from other methods. The poverty line is lower than the one calculated for 2002 since it is abbreviated. The poverty line calculated usingH B S price information for the 2002-03 data was 43 Euros per month, while this one is 22 Euros per month. Thus the poverty rate appears to be lower. The lower poverty rates are not driven by any real changes in the welfare but simply by this estimation technique. It i s the poverty trend that is informative. The resulting poverty trend confirms findings from other estimations that poverty rates remained stagnant. 21 Box 2.2: Analysis of Changes Analysis of changes in poverty presentedhere is based on consumption data from the 2003-2004, 2004-05 and 2005-06 Kosovo Household Budget Surveys. The consumptior modules differ over the survey waves: the 2005 HBS included more items than the 2003 an( 2004 surveys. Because the consumption modules differed it was necessary to put together i comparable consumption aggregate (CCA) with each survey. The CCA is a single consumptior value in each survey, constructed such that the sets of components inthe aggregate in the 2003 2004 surveys and the 2005 survey are parallel. Because the CCAs were assembledsolely for thc purpose of maximizing comparability across the two years, the CCA is not identical to the ful consumption aggregate used inthe first part ofthe report, Volume I. Following the methodology developed by Lanjouw and Lanjouw (2001), we define ar abbreviated food poverty line based only on the categories included in the CCA. (Given thc differences between the CCA and the full consumption aggregate, it would not be sensible tc apply the poverty lines based on the full consumption aggregate to the CCA (see Table B.5 Poverty Rates Using Alternative Consumption and Poverty Line Methodologies). The fooc poverty line, z, is defined as the average expenditure on these comparable items by thc populationinthe 30 to 50 percentiles(26.2 Euros for 2003-04). The robust final poverty line, Z derived from this abbreviated food poverty line is 41.8 Euros for 2003-04 surveys and 44.2 an( 40.4 Euros per month for 2004-05 and 2005-06 surveys respectively. Each line is calculate( non-parametricallyby taking average total consumption among sample households with fooc expenditure within 1 percent of z, within 2 percent of z, in increasing bands to within 5 percen of z. The final poverty line, Z, is thenthe average o fthese values. The values are listed in Tablc 2.6. A major assumptionbehindthis methodology is that expenditures on the goods includec inthe CCA have an Engelcurve relation to more comprehensivemeasuresof expenditure. Engel's law postulatesthat the higher the total expenditure, the lower the share of food expenditures. A major assumptionbehindthis methodology is that expenditures on the goods includec in the CCA have an Engel curve relation to more comprehensive measures of expenditure Engel's law postulates that the higher the total expenditure, the lower the share of fooc expenditures. This assumption appears to be met with this data. Other assumptions that needtc be satisfied for this methodology to be robust are stable expenditure patterns and no mis measurement inthe data. The other requirement for the comparisons to be robust is that only thi head count measure of poverty is used. The problemwith higher order poverty measures is tha the relativedistance betweenthe consumption levelof the poor and the poverty line may increasl as the components in the consumptionaggregatebecomemore comprehensive. It should be emphasizedthat the fact that the two surveyswere not identicalmeansthat the CCAs at best are only approximately comparable. As a result, the use ofthe CCAs introduces a levelofunquantifiableerror beyondthe usual sample error. Thus, the apparent changes over time should be interpretedwith caution D. COMPARE ALL THE YEARS 2.16 The procedure employed in this section involve estimating an econometric relationship between welfare and household characteristics with the 2002-03 data, using a set of characteristics common to all surveys. The estimated relationship is then used to update the distribution of the explanatory variables in the later surveys with information on the conditional probability (the estimated relationship) from the 2002 survey (Inverse Probability Weighting (IPW)*. 2The procedure used here is very similar to that of Stifel and Christiansen (2006), drawingheavily on the work of Elbers, Lanjouw, and Lanjouw(2003). 22 InverseProbabilityWeighting 2.17 The IPW consist of two estimation steps that corrects for the difference in the distribution of consumption between two surveys. In the first stage, data for 2002 and a comparison year are combined and a logit or probit model estimated where the dependent variable is 1 if an observation belong to year 2003 and 0 otherwise and the independent variables are a set of variables that have not changedfrom survey to survey. This enables us to obtain the predicted probabilitythat an observation is part of 2003 (propensityscore). In the second stage, the estimated propensity score is used to reweigh the poverty counts in 2002. The reweighted poverty estimate provides a comparable poverty count for 2003. By doingthe same thing for 2004 and 2005, we obtain a series of poverty counts all comparable to 2002 data. 2.18 The estimation results from using IPW methodology also suggest that poverty headcount increased. Our preliminary results using data from 2005-06 and 2002-03 only suggest povertyoutcomes probablyremainedthe same. The results indicatea slight increase, but the magnitudeofthe increasedepends onthe householdcharacteristics specified in step 1. The increase is from 42 to 45 percent between 2002 and 2005 if a large set of household characteristics are used (Table 1.2). If we employ a more limited set of household characteristics,thenthe impliedincreaseof povertyis higher. E. COMPARISON OFPOVERTYFIGURES FROMTHE LSMSAND HBS 2.19 LSMS and HBS data are not directly comparable because of differences in item definitions, disaggregation and recall periods. The 2001 Poverty Assessment (PA) reported the poverty rate at 50 percent usingthe LSMS 2000 data, while in 2005 the PA reported this rate at 37 percent using the HBS 2002 data. While both reports estimated household expenditures and a relevant poverty line, their results are not comparable becausethe LSMS and HBS surveys differ significantly in their representativeness and survey design. The representativeness of the surveys is difficult to gauge because of lack of a recent population census. In addition, the sampling frame of the LSMS was revised for the HBS. There are substantial disparities in the populationestimates for urban and rural areas as well as for the ethnic groups (Poverty Assessment, 2005). 2.20 The questionnaire designs preclude the constructionof comparable poverty indexes. The main differences are in the expenditure item definitions and in the recall period. However, in theory, the inverse probability weighting methods should apply here as well. There are also a few methodologicaldifferences in the constructionof the poverty indicesin these two poverty assessments. The main methodologicaldifferences are the exclusion of durables and the inclusionof health expenses for the consumption aggregate usingthe H B S (Table 1.2). By comparison, the PA using LSMS included durables but excluded health expenditures.Another differencein the estimation of poverty numbers in the two PASstems from the need to account for survey design when calculating point estimates. The H B S survey is stratified at the urbadrural and ethnic group level, but the 2005 poverty estimates did not adjust for this pattern of stratification. It is therefore difficult compare how large a difference the two estimates are from each other. The per-adult equivalent consumption aggregate andthe poverty lines are otherwise constructed inthe same way. 23 CHAPTER3: CONCLUSIONSAND RECOMMENDATIONS 3.1 The HBS demographic estimates suggest that there is an unquantifiable bias due to the outdated sampling frame and limited survey administration. The survey data implies incorrectlya populationreductionfrom about 2 million to 1.6million as well as very volatile urbanandrural dimensions. These estimates are suggestiveof a sampling biasthat cannot be correctedfor without proper sampling frame. The currentlysampling frame for the HBS data (and other SOK surveys such as LFS) dates back to 1981. A sample drawn from the 1981 frame will approximate the population in that frame. The differences between the 1981 populationdistributionand the current are introducingvery large sampling error and possibly bias. Inaddition,the survey administrationhasvery limitedfield supervision. 3.2 Changesin the survey questionnaireadditionallymake povertynon-comparable between years. The H B S questionnaire and how it asks respondents to report their consumption changes dramaticallyfrom 2002/03 to 2004/05 and then additional changes are introducedeach year. As the literatureon the subject prove, changes in the recall periodand the disaggregationofthe categories producedifferent consumption estimates even when there are no consumptiondifferences in reality. 3.3 We employ six methods to address the issues non-comparabilityand sampling. For sampling, we employ post-calibrationto match HBS estimates to other estimates of the population. We also exclude volatile waves, as for instance the consumption patterns from 2004/05 seem to not be comparable to those of 2003/04 or 2004/05. To address non- comparability of how the expenditure data was collected, we use, first, surveys with comparablequestionnaires, and second, all surveysbut correct for differences in consumption patternsby usingeconometric projectionand inverse probabilityweighting. 3.4 Based on this extensive sensitivity analysis, we do not find firm evidence of improvement of householdwelfare in Kosovointhe last three years. The resultthat is robust to specificationand different methodologiesis that there is no significant increaseor decrease inthe povertyrate in Kosovofor the periodfrom 2003/04to 2005/06. A. RECOMMENDATIONS 3.5 Consistencyinthe survey questionnaire should be the goal of next surveys. Even small changes in the way how questions about expenditures are asked can cause differences in reported consumption even if there is no real change in the consumption pattern of the household. Changes in the survey questionnaire should not be introduced without a randomizedexperiment beforehand. A randomizedexperimedmini-surveywill allow to test the effect ofchangingthe question on the reportedexpenditure. 3.6 A census is urgently needed to create a basis for unbiased sampling. It is unfortunatethat four years of data are not reliableenoughto providepolicy guidancebecause of a lack of census. A census will completelyresolve the sampling issues for future surveys, andpossiblycan be usedretroactivelyto adjust earlier samplingweights. 25 3.7 Better survey administrationand documentation of all steps of the process and the data are necessary. Currently, the survey administrators have hardly any supervision. Additional supervision will affect the HBS budget only marginallybut will have highreturns in improveddata quality. Second, documentationof all procedures is very limited. This step is at least costly, but will improve data quality and usefulness. 26 ANNEX A: TABLESAND FIGURES Table A.l: Comparison of Previous Methodologies WB Poverty Assessments 2001~ 20054 Poverty Absolute Extreme Absolute Extreme Rates Total 50.3 11.9 Total 37.0 15.2 Rural 52.0 11.6 Rural 34.1 14.8 Urban 47.5 12.5 Urban Not avail. Pristina 36.4 7.7 Other 47.1 19.1 Data LSMS 2000 HBS2002-2003 Timing: Sept -Dec 2000. Timing: 612002-512003. Sampling frame: Sampling frame: 2, 880 households. 2400 households Rural: basedon the Housing Damage Similar to LSMS with primary sampling Assessment Survey (1999). units revised. Urban OSCE voters' registration. Representativeness:Urbadrural; ethnic Representativeness:Urbadrural. AORs, group. ethnic groups. Consumption Food: Aggregate 1. Purchasedfood inthe last 30 days Food: in39 categories, both quantity and 1. Expenditures on food. value. 2. Own production 2. Stored food inthe last month and 3. Foodout last year, 7 categories. 3. Own production, gifts. 4. Food out. Housing Expenditure and Rent: Housing Expenditure and Rent: excluded. excluded. Non-food: Non-food: personal items, hh services. Personalitems, hhservices Semi-durables Durable goods: Durable goods: rental value. excluded; Education: Education: included, 1-year recall. included, diary method Health: Health: excluded included Price indexes: using unit values (ratio of Price index: values over quantities) after excluding obs > CPI by monthand urbadrural dimension. Information here i s from the Poverty Assessment, 2001 x-1Appendix G of the Kosovo LSMS 2000 Basic Information Document from http:l/www.worldbank.org/lsms 4Kosovo Poverty Assessment (2005) and Tsirunyan, Sasun. 2004. "Poverty and Inequality in Kosovo", backgroundpaper for the Poverty Assessment. 27 2 st.d. Paasche price index. Equivalent EA, = (A, + BC,)'where B= 0.75. EA, = (A, + BC,)'where e= 0.75. Adults Equivalent Adults = (Adults + -75 Equivalent Adults = (Adults + .75 Children),". Children < 15 years old. Children),". Children < 15 years old. Per Adult- Equivalent pEC, = TCI TCI (A, +0 COY X(A0 +6 cole PEC, = Consumption (A, +.ec,)' A, +c, ( A ~ c,)' +e A, +c, where the pivotal householdhas 4 adults where the pivotal householdhas 4 adults and 2 children (Ao=4, C0=2). and 2 children(Ao=4, C0=2). Poverty Food poverty line: Same methodology as for 2001. Food Lines Basedon 2, 100 calories per adult. Caloric basket of 2100 calories is estimatedwith the structure of the 30" to 50" population price information from the HBS. Caloric percentiles. structure of the 30* to 50" population Food line: percentilesfrom the HBS. DM 1A529 per adult per day. Food line: Poverty line: Euro 0.93/day/adult. DM3.498 per adult per day. Usingthe Poverty line: share of non-food items for hh with food Euro 1.41per adult per day. consumptionclose to the FoodLine. Food Same methodology as 2001. Food share= share= 53.97%. 65.9%. Currency Lack of PPP adjustment indexes. PPP not available. conversion Currency conversions use the rates Not indicated. correspondingto the month of the survey. Unofficial exchange rate of 30 to 33 Dinars per DEM. Other Not accounting for stratification. Source: World Bank KosovoPovertyAssessment (2001) and PovertyAssessment (2005). 28 Table A.2: Survey Comparison KOSOVO HBS Wave 1 Wave I1 Wave I11 Wave 1V HBS-2002-2003 HBS-2003-2004 HBS-2004 HBS-2005 Period usedfor analysis 612002-12003 612003-612004 612004-512005 612005-512006 Number of 2400 households 2400 2400 2400 observations ((960 rural, 1440 urban) households households households Survey questionnaire design and its changes Timing of questionnaire 612002 612003 112005 112006 introduction Food consumption expenditure Recall period daily weekly weekly weekly Method diary recall recall recall Questiontype open-ended close-ended close- close ended ended Categories 165 103 103 107 Consumption of own production Quantities no no no In-kind food receivedas Yes Yes Yes gifts, donation Categories 85 12 12 12 Non-food expenditures Education daily diary weekly recall weekly weekly recall recall Categories 14 13 13 13 Health daily diary weekly recall weekly weekly recall recall Categories 6 6 6 11 Other non-food Clothing daily diary weekly recall weekly weekly recall recall Categories 31 10 I O I O Householdtextiles yearly recall weekly recall weekly weekly recall recall Categories 6 6 6 6 Transport daily diary weekly recall weekly weekly recall recall Categories 11 5 5 15 Durables Purchases Yes Yes Yes Ownership quantity of item no no Value no no no When bought no no no Housing consumption Rent no Yes Yes Categories 2 6 6 Estimatedrent ifowned Yes Yes Yes Utilities daily diary weekly recall weekly recall weekly recall Categories - 16 24 24 26 Source: Relevant HBSquestionnaires and datasets. 29 Table A.3: Percent Changt in Main Aggregates from Survey Survey Comparison Changefrom ..... IItoIII IIItoIv IItozv Base is 03/04 04/06 03/04 TotalConsumptionofHH 9% -13% -5% TotalExpendituresof HH 14% -13% -1% Consumptionof ownproduced -18% -12% -28% or fetchedfood Food expenditures (incl. 2% -12% -9% alcohol andtobacco) Non-Foodexpenditures 32% -15% 12% Source: World Bank staff estimates from HBS data. Table A.4: Alternative Consumption Aggregate Definitions and Poverty Rates Poverty Rates Coefficient ConsumDtionAggrePate Weighted of SDecification Variation 2002103 2003104 2004105 2005106 2002-05 Basic food excl ownprod. plusbasic 77.8 81.7 73 79.7 5yo non-foodspendingexcl. utilities Above plus ownproduction 61.3 68.8 58.8 69.6 8Yo Above plus in-kind, food out, alcohol 55.6 63.5 51.4 62.6 10% andtobacco Above plus semi-durablesandutilities 46.4 49.6 38.9 50.1 11% Above plus education 46.4 49.6 38.9 50.1 11% Above plus medical 44.6 47.6 36.8 48.4 12% Source: World Bank staff estimates from HBS data. 30 Table AS: Consistently Asked Questions over the 4 Surveys Variable Wave I 2003-04/III Wave IV Education of head and If7 years or older: What is hidher If6 years or older: What If6 years or older: What max in hh highest level of education is hidher highest level of is hisher highest level of completed? educationcompleted? educationcompleted? 8 categories 8 categories 8 categories Age (of householdhead) How old is heher?Age at last How old is heher? Age at How old is heher?Age at birthday. Note "0" for children last birthday. Note "0" last birthday. Note ``0" under 1year for children under 1year for children under 1year Sex of householdhead What is hisher sex? What is hisher sex? What is hisher sex? StudentAJnemployment What is hisher main activity What is hisher main What is hisher main status duringthe past 12 months? activity during the past 12 activity during the past 12 11categories months? months? 11 categories 11 categories Income source What is the main source of income What is the main source What is the main source for this household? of income for this of income for this 8 categories household? household? 8 categories 10 categories Housing: brick walls What is the main material of the Does your dwelling have Does your dwelling have walls? 4 categories; walls of block, bricks or walls of block, bricks or 2=bricks/cement blocks. cement? cement? Housing: electricity I s this dwelling electrified? Does your dwelling have Does your dwelling have electricity? electricity? Housing:tap water What is the main source of water Does your dwelling have Does your dwelling have for this household? Central indoor water taps? indoor water taps? pipeline, own pipeline, standing water pipe. Purchaseof durables Has anyone in the household Has anyone inthe Has anyone in the duringthe last 12 months householdduringthe last householdduring the last purchased any ...? any...? 12 months purchased 12 monthspurchased 57 categories any...? 57 categories 57 categories Source: World Bank staff estimates from HBS data. 31 Table A.6: Definition of ConsumptionAggregates for the Different Methodologies PA 05 CA Revised* ComparableCA Foodexcluding own production 4 4 4 Alcohol and tobacco 4 4 In-kind(received) 4 4 Own production 4 4 4 Non-foodexcl health and education 4 4 Education 4 4 Health 4 Utilities excl value of housing 4 4 Value of housing Source: World Bank staffestimatesfrom HBS data. Notes: Certainhighvolatility items are excluded (air and sea * transportationexpenses; gamblingand holiday packages;financial andjudicial services). Utilities includedomestic services. Table A.7: PovertyLinesin Different Methodologies(in Euros, per adult equivalent per month) Methodology 2002 PL Robust Endogenousin Endogenous Endogenousin adjusted in Comparable in Econometric with CPI Comparable Surveys Econometric Projection CA Projection (unweighted) (weighted) Wave Povertvline 2002103 43.12 45.6 44.1 2003104 43.10 41.8 37.2 45.6 44.0 2004105 43.34 44.2 38.1 46.8 45.2 2005106 43.10 40.39 38.1 46.8 45.2 Source: World Bank staff estimatesfrom HBSdata. 32 ANNEX B: RESULTSUSINGDIFFERENTSURVEY YEAR DEFINITION 3.8 In this Annex, we present our results using survey waves defined by the introductionof changes inthe questionnaire. 3.9 There is a differencebetween the sampling timing and the timing of the introductionof changes in the questionnaire. The sampling is done for the household being survey in June through May each year. Changes and additions to the survey questionnaire are introduced in January, starting 2005. The selection of which households and EAs are sampled each month is not clear, although 200 households from 25 EAs are consistently surveyed. There is evidence, however, that the surveying consequence is not representative by month or half a year. For instance, much largershare ofthe populationis surveyedeach second half of the year than during the rest of the survey. Partitioningthe sample by calendar year, thus, introduces a bias. Indeed, the resultsusingwaves defined as inthe table belowshow a different trend inpoverty. Table B.l: Introduction of New Questionnaires KOSOVOHBS Wave I Wave I1 Wave111 Wave IV HBS-2002-2003 HBS-2003-2004 HBS-2004 HBS-2005 Period 612002-512003 612003-612004 612004-1212004 112005-1212005 Numberof 2400 households 2400 households 1400 households 2400 households observations (960 rural, 1440 urban) Table B.2: Poverty Statistics using PA05 Methodolopy Absolute Poverty Headcount Extreme Poverty Headcount Weighted Unweighted Weighted Unweighted 612002-512003 37.93 43.56 15.43 18.30 612003-612004 45.14 41.83 13.85 13.64 612004-1212004 35.79 31.61 12.43 10.39 112005-1212005 39.72 39.13 12.68 13.24 Source: World Bank staffestimatesfrom HBS data. Note: Unweightedhere refers to no weights being used andthus these estimatesare at household-levelversus the population-levelestimatesinthe "weighted" column. 33 Table B.3: Poverty Rates Using PA05 Methodology Absolute and extreme poverty rates Urban and Rural Poverty Rates Urban Rural 2002103 46.99 34.49 2003104 42.73 46.01 06-1212004 30.13 38.73 2005 34.95 42.39 ...-. ... _.........................-* _ _ _ _ .. I.. %wo. HBS I, II.111and IV PovmyA-~nant(2005) rnebdolowyuwd Source: World Bank staff calculations from HBS data Table B.4: Detailed Poverty Diagnosticswith Revised Consumption Aggregate Absolute Poverty Headcount Extreme Poverty Headcount By wave weighted unweighted weighted unweighted 612002-512003 40.6 46.0 '17.9 20.1 612003-612004 46.9 42.8 14.5 14.4 612004-1212004 37.3 33.0 12.8 10.7 112005-1212005 42.1 40.7 13.1 13.7 Source: World Bank staffcalculationsfrom HBS data. Unweightedhererefers to no weights beingused and thus these estimatesare at household-levelversus the population-levelestimatesin the "weighted" column. 34 Table B.5: PovertyRatesUsingAlternative Consumptionand Poverty LineMethodologies ConsumptionAggregate (CA) Povertv Rates definition CAL CA withPA05 PL 2002 adjustedwith CPI methodology weighted unweighted By wave 2002-03 37.9 43.6 2003-04 45.1 41.8 2004-05 35.8 31.6 2005-06 39.7 39.1 CA IL ComparableCA (Lanjouw and Lanjouw, 2001) PL 2002 adjustedwith Nonparametric Poverty CPI Line weighted unweighted weighted unweighted By wave 2002-03 40.6 46.0 2003-04 46.9 42.8 39.6 35.9 2004-05 37.3 33.0 39.2 34.8 2005-06 42.1 40.7 38.6 37.5 CA I1in ComparableSurveys PL 2002 adjustedwith EndogenousPoverty Line CPI weighted unweighted weighted unweighted By wave 2002-03 40.6 46.0 2003-04 46.9 42.8 37.8 34.9 2004-05 37.3 33.0 29.9 25.7 2005-06 42.1 40.7 33.2 32.5 CA IUsingEconometricPoverty Projection (Stifel and Christiansen, 2006; Poverty Line Elbers, Lanjouw, and Lanjouw, PL 2002 adjustedwith EndogenouslyDetermined 2003) CPI By wave weighted unweighted weighted unweighted 2002-03 29.7 34.7 38.2 36.2 2003-04 24.7 31.1 30.2 31.1 2004-05 28.8 30.1 38.8 32.3 2005-06 26.8 28.7 35.5 30.7 Source: World Bank staff calculations from HBS data. 35 REFERENCES AMP Kosovo. 2006. The Rural Development Context of Kosovo. Agricultural and Rural Development Planfor Kosovo. (downloaded from http://www.ampkosovo.corn/ardpl .htmlon 9/12/2006). Carletto, C. ,Davis, B. and Stampini, M.,Zezza, A. (2004). A Country on the Move: Internal Mobility and InternationalMigration in Post-Communist Albania. FA0 (mimeo) Deaton, A. and Zaidi, S. 2002. Guidelines for Constructing Consumption Aggregates for Welfare Analysis. LSMS WorkingPaper 135. Deaton, A. and Dreze, J. 2002. Poverty and Inequality inIndia: A Re-Examination. Economic and Political Weekly September 7,2002 Deaton, A. 2003. Adjusted IndianPoverty Estimates for 1999-2000. Economic and Political Weekly January 25,2003 Demery, L. and Groootaert, C. 1993. Correcting for Sampling Bias inthe Measurement of Welfare and Poverty inthe Cote D'Ivoire Living StandardsSurvey. The WorldBank Economic Review. Vol 77No 2: 263-292. Elbers, C., Lanjouw, J., Lanjouw, P.,. 2003. Micro-levelEstimation of Poverty and Inequality. Econometrica, Vol. 71, No 1(January, 2003), 255-364. Gibson, John. 2005. Statistical Tools and Estimation Methods for Poverty Measures Basedon Cross-SectionalHousehold Surveys. Handbook on Poverty Statistics: Concepts,Methods and Policy Use, 128-205 (downloaded from http://unstats.un.or~/unsd/methods/povert~/~df/Chapter-5on 9/12/2006) .pdf Grimm, M. and Gunther, I.2004. How to achieve pro-poor growth ina poor economy: The case of BurkinaFaso. Operationalizing Pro-Poor Growth" (downloaded from " www.wor1dbank.org on 9/12/2006), Howes, S. and Lanjouw, J. 1997. Poverty Comparisons and Household Survey Design. LSMSWorking Paper Number 129. Lanjouw, P., and Kijima, Y. 2003. Poverty inIndiaduring the 1990s:A Regional Perspective. World Bank Policy Research Working Paper 3141. Lanjouw, J, and Lanjouw, P. 2001. "How to CompareApples and Oranges: Poverty MeasurementBasedon Different Definitions of Consumption" Review of Income and Wealth Series 47, Number 1, March2001. 37 Lohr, S. 1999. Sampling - Design and'tlnalysis. 1st Edition, Arizona State University, 0534353614. Lynn, P. 2004. "Development ofa Sampling Methodfor Household SurveysinPost- War Bosnia and Herzegovina" Statistics in Transition June 2004 Vol. 6, No. 6, pp. 953-977 Ravallion, M. 1994. "A Better way to set poverty lines, "Outreach #15, March 1994. Statistical Office of Kosovo (SOK). 2005. Agricultural and Household Survey 2005. Series-Agricultural and Environment Statistics, Pristina, Kosovo. Stifell, D. and Christiaensen, L. 2006. Tracking Poverty Over Time inthe Absence of Comparable Consumption Data WPS3810. Tarozzi, A. 2005. Calculating Comparable Statistics from Incomparable Surveys, with an Application to Poverty inIndia (downloaded from http://www.econ.duke.edu/-taroz/TarozziComparabilitv.pdf on 9/12/2006) 38