65068 The Kyrgyz Republic Profile and Dynamics of Poverty and Inequality, 2009 October 3, 2011 Poverty Reduction and Economic Management Unit Europe and Central Asia Region Document of the World Bank FISCAL YEAR January 1-December 31 Currency Kyrgyz Som Average for 2009 Som 42.9 = US$ 1 Weights and Measures Metric System ABBREVIATION AND ACRONYMS CPI Consumer Price Index ECA Europe and Central Asia FYR Former Yugoslav Republic GDP Gross Domestic Product GNI Gross National Income KIHS Kyrgyz Integrated Household Survey MDG Millennium Development Goals MSB Monthly Social Benefit NSC National Statistical Committee of the Kyrgyz Republic PPP Purchasing Power Parity UMB Unified Monthly Benefit UN United Nations Vice President: Philippe H. Le Houerou Country Director: Motoo Konishi Sector Director Yvonne Tsikata Sector Manager: Benu Bidani Country Manager Alexander Kremer Task Team Leader: Sarosh Sattar TABLE OF CONTENTS PROFILE AND DYNAMICS OF POVERTY AND INEQUALITY....................................... 1 A. Introduction .......................................................................................................................... 1 B. Macroeconomic Background ............................................................................................... 3 C. Development Indicators ....................................................................................................... 3 D. Characteristics of the Poor ................................................................................................... 7 Number and geographic distribution of the poor ....................................................... 7 Inequality ................................................................................................................. 10 Living conditions ..................................................................................................... 12 Demographic characteristics .................................................................................... 13 Gender ...................................................................................................................... 19 Internal migration..................................................................................................... 20 Employment status ................................................................................................... 21 Education ................................................................................................................. 24 Household income and consumption basket ............................................................ 29 Poverty profile using regression model ................................................................... 32 E. The Dynamics of Poverty and Inequality in 2006 – 2009 ................................................. 35 The Trend in Poverty Levels.................................................................................... 35 Changes in Inequality, Consumption, and Income .................................................. 38 LIST OF FIGURES Figure 1: Comparison of GDP per Capita with Selected Regions ............................................. 4 Figure 2: International Poverty Rates for Selected Countries PPP US$2.5 per day, 2008 ........ 4 Figure 3: International Poverty Rates for Selected Countries, PPP US$5.0 per day, 2008 ....... 4 Figure 4: Maternal Mortality Ratio, ........................................................................................... 6 Figure 5: Tuberculosis Prevalence and Incidence Rates, 1990–2008........................................ 6 Figure 6: Primary School Enrollment Rate, 1999–2009............................................................ 6 Figure 7: Absolute and Extreme Poverty in Rural and Urban Areas, 2009 ............................... 7 Figure 8: Absolute and Extreme Poverty Gap and Severity ...................................................... 9 Figure 9: Incidence and Distribution of Poverty, by Oblast ...................................................... 9 Figure 10: Incidence and Poverty Distribution across Mountains and Plains ......................... 10 Figure 11: Gini coefficient by sector and oblast, 2009 ............................................................ 11 Figure 12: Mean Consumption by Quintile, 2009 ................................................................... 11 Figure 13: Average Area of Dwelling per Capita .................................................................... 12 Figure 14: Population’s Access to Housing Amenities ........................................................... 12 Figure 15: Annual Electricity Expenditures per Capita (in som) ............................................ 13 Figure 16: Annual Electricity Expenditures, per capita by Poverty Status (in som) ............... 13 Figure 17: Household Size, by Income and Location .............................................................. 14 Figure 18: Relationship between Household Size and Absolute Poverty Incidence Rate ....... 14 Figure 19: Age Composition of Poor and Nonpoor Households ............................................. 15 Figure 20: Age Composition of Households, by Income Category and Location ................... 15 Figure 21: Distribution of Households by Number of Children and Poverty Status ............... 16 Figure 22: Distribution Households by Number of Children and Poverty Status, for Urban Rural Areas .............................................................................................................................. 16 Figure 23: Average Dependency Ratio, by Poverty Status and Sector.................................... 17 Figure 24: Poverty Incidence and Distribution by Household Head’s Age............................. 17 Figure 25: Gender of the Household Head and Poverty .......................................................... 19 Figure 26: Gender Status of the Household Head and Rural-Urban Poverty .......................... 20 Figure 27: Internal Migration by Household Heads, by Quintiles of Per Capita Consumption .................................................................................................................................................. 21 Figure 28: Employment Status of Household Head and Absolute Poverty Distribution ........ 22 Figure 29: Employment Status of Household Head and Extreme Poverty Distribution ......... 22 Figure 30: Absolute and Extreme Poverty Rates by Employment Status ............................... 23 Figure 31: Poverty Rates and Educational Attainment of Household Head ............................ 25 Figure 32: Education of Household Head and Absolute Poverty Rates, by Location ............. 25 Figure 33: School- Age Children Who Attended School ......................................................... 27 Figure 34: Expenditures on Education..................................................................................... 29 Figure 35: Average Annual Income per Capita ....................................................................... 29 Figure 36: Annual Income per Capita by Poverty Status ........................................................ 30 Figure 37: Annual Per Capita Consumption, by Poverty Status.............................................. 32 Figure 38: Poverty Trends, 2006–09 ....................................................................................... 35 Figure 39: Rural and Urban Poverty Trends, 2006–09 ............................................................ 36 Figure 40: Absolute Poverty and Extreme Poverty Gaps, 2006–08 ........................................ 37 Figure 41: Absolute Poverty and Extreme Poverty Severity, 2006–09 ................................... 38 Figure 42: Gini coefficient of Consumption per Capita, by Urban and Rural, 2006–09......... 39 Figure 43: Annual Per Capita Consumption Growth, by Quintiles, in Real Terms ................ 40 Figure 44: Income Structure of Households, 2007–09 ............................................................ 41 Figure 45: Dynamics of GDP, Poverty, Remittances, and Consumption Growth Rates ......... 42 Figure 46: Growth Incidence Curve, 2008–09 ........................................................................ 43 Figure 47: Extreme Poverty Reduction, 2008.......................................................................... 46 LIST OF TABLES Table 1: Kyrgyz Republic: Key Macroeconomic Indicators, 2006–09 ..................................... 2 Table 2: Selected Social Development Indicators: An International Comparison .................... 5 Table 3: Absolute Poverty Rates, Using Per Capita and Adult Equivalence Scales ............... 18 Table 4: Area of Employment of Household Heads ................................................................ 23 Table 5: Type of Employment of Household Heads ............................................................... 24 Table 6: Education Level of Household Head, by Gender and Consumption Quintiles ......... 26 Table 7: Education Level of Adults Aged 25 and Older, by Poverty Status and Gender ........ 27 Table 8: Children’s School Attendance by Age Group, Poverty Status, and Gender ............. 28 Table 9: Structure of Income by Poverty Status ...................................................................... 30 Table 10: Share of Food Groups in Total Food Consumption, by Poverty Group .................. 33 Table 11: Results of Regression Model: Explaining Per Capita Consumption, 2009 ............. 34 Table 12: Poverty Trends, by Oblasts, 2006–09 ...................................................................... 36 Table 13: Gini coefficient (Per Capita Consumption) by Oblast, 2007–09............................. 39 Table 14: Mean Consumption as Proportion of Poverty Line, by Quintiles, 2007–09 ........... 40 Table 15: Growth and Redistribution Decomposition of Poverty Changes ............................ 42 Table 16: Distribution of Social Protection Benefits and Private Transfers Across Groups, 2008.......................................................................................................................................... 44 Table 17: Benefit Adequacy: Share of Benefits in Total Household Consumption, 2008 ...... 45 BOX 1: Poverty Measures in the Kyrgyz Republic BOX 2: Sensitivity Analysis of Per Capita versus Adult Equivalence Scales Specification BOX 3: To What Extent Does the Existing Safety Net Protect the Poor? ANNEX 1: RE SULT S OF THE STUDY: TO WHAT EXTENT DOES THE EXISTING SAFETY NET PROTECT T HE POOR? ACKNOWLEDGMENTS This report was made possible thanks to the cooperation of the National Statistical Committee (NSC) of the Kyrgyz Republic. Special appreciation goes to Orozmat Abdykalykov (chairman) and Galina Samohleb (head of the Household Survey Department). The report was prepared by the World Bank team led by Sarosh Sattar and included Aibek Baibagysh uulu (economist) and Yeva Gulnazaryan (data analyst). The team would like to acknowledge the helpful comments of peer reviewers Bob Rijkers and Andrew Dabalen. The report was prepared under the guidance of Benu Bidani (sector manager) and Alexander Kremer (country manager). Administrative support was provided by Helena Makarenko and Lilia Saetova. EXECUTIVE S UMMARY The Kyrgyz Republic is a low-income country with high levels of poverty, using either international or national poverty lines. An international comparison of the Kyrgyz Republic’s selected human development indicators shows that it is in a better position than other low- income countries but lags behind other developing countries in the ECA region. It app ears that the Kyrgyz Republic is on track to achieve the majority of the Millennium Development Goals (MDGs), with the exception of those for maternal mortality and tuberculosis. In addition, the data show that primary school enrollment rates gradually declined between 1999 and 2009. The World Bank estimates that the headcount incidence of absolute poverty was 31.7 percent in 2009, based on the national poverty line. 1 This translates to 1.7 million people in the country living below the poverty line. The extreme poverty rate was 3 percent in the same year. The poverty gap, which measures the depth of poverty, reached 6.1 percent and appears to be moderate. Poverty had a strong sectoral and regional dimension. As in other agriculturally based economies, rural poverty dominated. The proportion of the rural population living below the poverty line reached 37.1 percent versus 22.0 percent for the urban population. In regional terms, the oblasts with the highest poverty rates are Issyk-Kul, Naryn, and Osh. In absolute terms, the largest numbers of poor are in Osh, Jalal-Abad, and Issyk-Kul, reflecting the large population base in those oblasts. As expected, the risk of poverty increased with the altitude zone and was reflected in higher poverty rates in high mountainous areas. This report finds that household characteristics matter for poverty status. Poverty rates are high in larger households. Other demographic variables (age, dependency ratio, demographic and family composition, etc.) also play a significant role in defining the poor. By contrast, the correlation between the gender of the household head and poverty is unclear. It appears that for absolute poverty the incidence of male- headed households is higher while for extreme poverty, female- headed households predominate. Internal migration (mainly from rural to urban areas) seems to be one of the strategies households pursue to get out of poverty, many doing so in search of better job opportunities. Yet there are no significant differences between poor and nonpoor households in the distribution of the unemployed. Predictably, however, poverty is indeed higher for unemployed heads of households. The survey data show that lack of education continues to be an important factor explaining poverty: the headcount poverty rate is two times higher for illiterate household heads than for those with technical professional degrees. In general, the poor and the extreme poor stop schooling after 16 years of age. The high cost of education may be part of the reason: the poor spend seven times less than nonpoor on education in absolute terms. The importance of sources of income varies by group. The extreme poor are less reliant on income from employment and more reliant on pensions and social benefits than either the 1 The absolute poverty and extreme poverty lines are terms used by the NSC. They refer to the upper and lower poverty lines. i poor or the nonpoor. This underlines the importance of the social benefit system for reducing poverty in the most destitute group as well as the need to further improve the effectiveness and targeting accuracy of the social safety net. In terms of poverty dynamics, the report shows that the reduction of poverty between 2006 and 2009 by 29.3 percentage points was associated with moderate growth rate in GDP, which averaged 5.7 percent in those years. In fact, the poverty rates to a large extent reflect changes in GDP, total consumption, and remittance growth rates. Thus, in 2009, when the country suffered from the impact of the financial crisis, the poverty rate stagnated, marking the end of the long declining trend in poverty observed in previous years. For the period under analysis, however, the extreme poverty rate continued to decline, albeit at a slower pace, despite stagnation in the absolute poverty trend. The reduction in absolute poverty was experienced more in urban areas, whereas extreme poverty fell more quickly in rural areas. Similarly, the reduction in absolute poverty was experienced in all oblasts, whereas for the declining trend of extreme poverty there is one exception—Bishkek city, where extreme poverty actually increased from 0.6 percent in 2006 to 3 percent in 2009. The reduction in poverty was different for different oblasts; however one region stands out in terms of the significant pace at which poverty has declined. Between 2006 and 2009, the Jalal-Abad oblast halved the absolute poverty rate and showed considerable reduction in the extreme poverty rate. The depth and severity of poverty followed the general poverty trends—that is, rapidly declining in the earlier period of high economic growth but stagnating in the later period for reasons related to the impact of the financial crisis and economic slowdown. The most recent growth decomposition of poverty rates between 2008 and 2009 shows that the dynamics in poverty rates in 2008–09 is to a large extent due to the distribution component rather than the impact of growth in consumption. The trend in inequality was consistent with the trend in poverty rates. The inequality measure stagnated over 2008 and 2009 was after having declined in the earlier period. Overall, the Gini coefficient fell from 31.3 percent in 2006 to 25.5 percent in 2009. That decline was driven mainly by the reduction in rural inequalities; the urban Gini coefficient showed a slower rate of reduction, especially in recent years. ii PROFILE AND DYNAMICS OF POVERTY AND INEQUALITY A. Introduction 1. Poverty reduction is an important goal for governments of many developing countries. This goal is synonymous with economic development and achieving a higher quality of life for all population groups. Thus, monitoring the dynamics of poverty and inequality is implicit in the monitoring of progress in societal development. As the vast literature shows, development progress to a large extent depends on economic and social policies and economic growth. Thus, identifying the relationship between relevant economic variables and poverty and inequality indicators may provide policy guidance on what has furthered the country’s progress. 2. The content of this report is based on data from the Kyrgyz Integrated Household Survey (KIHS) for 2009, which was collected and made available to the World Bank by the National Statistical Committee (NSC) of the Kyrgyz Republic. This report follows a previous assessment of poverty in the Kyrgyz Republic and the most recent for 2008, when the economy had not yet felt the full impact of the financial crisis. It represents the continuation of the efforts to monitor poverty, provide additional insights, and identify the main dimensions of poverty in the Kyrgyz Republic based on new analysis and the most recently available data. 3. The report consists of two main parts. The first part discusses poverty and inequality for 2009 and, thus, from a static perceptive. So, the first section describes and discusses the main features and correlates of the poor. The goal is to provide a brief overview of poverty in the Kyrgyz Republic and describe the characteristics of households and the poor. This is achieved by considering the poverty incidence among households and individuals differentiated by such characteristics as age, household size, employment status, educational attainment, geographic location, gender of the household head, and internal migration status of the household head. This section also incorporates some insights on poverty from a recent study carried out in the Kyrgyz Republic related to the poverty and social impact analysis of selected reforms in the social protection sector. 4. The second section discusses the dynamics of poverty and inequality in the Kyrgyz Republic during 2006–2009. This section relates the trends of macro indicators (gross domestic product [GDP], growth in different sectors of the economy, the consumer price index [CPI], remittances, and social budget expenditures) to changes in micro indicators of interest (poverty and inequality). The objective is to integrate into one coherent picture the recent macro and micro developments. The period of 2006–2009 coincided with the food price volatility and financial crisis of 2008–09. So the distinct feature of this report is that the impact of international instabilities is reflected in the assessment of poverty and inequality. The report uses two types of data: the macroeconomic, national accounts data as regularly reported by the NSC, Ministry of Finance, and National Bank of the Kyrgyz Republic, and microeconomic, primary data derived from the KIHS (also collected by the NSC). These two might not always be consistent with each other, but they do complement one another to provide useful policy insights. 5. The KIHS interviews 5,016 households annually and has collected information on about 19,060 people. The survey is conducted quarterly on a rolling basis. Households remain in the sample for a period of four years, on average. The sample size is sufficiently large to allow for robust estimates of poverty at the national, urban-rural, and oblast (province) levels. The survey collects information on household and individual characteristics, expenditures (food and nonfood), income, assets, living conditions, and labor market activities. 1 Table 1: Kyrgyz Republic: Key Macroeconomic Indicators, 2006–09 2006 2007 2008 2009 Real Growth (in percent) GDP 3.1 8.5 8.4 2.9 GDP, excluding gold 5.7 9.0 6.5 3.4 Agriculture 1.7 1.6 0.9 6.7 Construction 19.1 32.3 10.8 22.1 Industry -10.9 6.3 14.8 -8.1 Services 9.0 12.6 11.0 2.3 Exports of goods and services 7.7 25.8 9.1 -1.1 Imports of goods and services 42.7 11.0 13.6 -19.4 Total consumption, etc. 15.4 2.2 9.7 -11.9 General government consumption -6.0 1.8 -0.8 1.5 Private consumption, etc. 20.0 2.3 11.8 -14.4 Gross domestic investment 29.2 16.5 18.6 -4.9 Net indirect taxes 3.1 8.5 8.4 2.9 US dollars million GDP 2,855 3,823 5,141 4,691 Exports fob 906 1,337 1,874 1,694 Imports cif 1,723 2,417 4,072 3,040 Workers’ remittances (inflow) 731 1,065 1,508 1,094 Current account balance -286 -227 -701 -106 Foreign direct investments (net) 182 208 377 190 Prices Inflation (% beginning to end of period, CPI) 5.1 20.1 20.0 0.0 Wage, average (som per month) 3,270 3,970 5,378 6,161 Wage, average (US$ per month) 81 106 147 144 Exchange rate (som per US$, average of period) 40.2 37.3 36.6 42.9 General government budget (% of GDP) Revenues and grants 27 31 30 32 Grants 1 2 2 5 Tax 18 19 23 22 Social fund contributions 4 4 4 4 Nontax and capital 4 6 5 5 Expenditures 30 31 29 36 Current (other than interest) 24 23 24 28 Transfers and subsidies 9 9 3 3 Wages and salaries 6 7 7 7 Interest payments 1 1 1 1 Capital 4 7 5 5 of which foreign-financed public investment program 3 3 2 3 Net lending 0 0 0 1 Fiscal balance -3 0 0 -4 Miscellaneous GNI per capita, Atlas methodology (US$) 500 620 790 860 Poverty headcount (% of population) 61 55 32 32 Extreme poverty headcount (% of population) 17 12 6 3 Unemployment rate (% of labor force) 8.3 8.2 8.2 8.4 Workers’ remittances (inflow, % to GDP) 26 28 29 23 Current account balance (annualized, % to GDP) -10 -5.9 -13.6 -2.3 Sources: NSC, Nat ional Bank, and Ministry of Finance. 2 B. Macroeconomic Background 6. The growth rate of real GDP in the Kyrgyz Republic over 2006–2009 averaged 5.7 percent but was not stable. In the pre-crisis years, the growth rate of the economy accelerated to above 8 percent, but in 2009 the economy felt the impact of the international crisis and the GDP growth rate slowed to 2.9 percent. If we exclude the direct contribution of the gold mining industry, which accounted for about 5 percent of GDP, the growth rate averaged 6.1 percent. So the slower production in gold mining and generally negative growth in the industrial sector (which fell by 8 percent in 2009) reduced the growth in overall GDP. Despite the recession in some sectors in the crisis year of 2009, two sectors, construction and agriculture, witnessed high growth rates of 22.1 and 6.7 percent, respectively. 7. On the expenditure side of GDP, growth rates in exports, imports, total consumption, and total investments were high and positive before 2009 but turned negative in 2009, reflecting the impact of the financial crisis and reduction in aggregate demand. The high growth rate through 2008 was due to the growth in private consumption and gross fixed investment. The growth in investment, presumably, was driven by the push to complete the construction of the Kambarata 2 hydroelectric station. As the crisis erupted, the Kyrgyz government responded with higher public expenditures—the only positive component of aggregate demand in 2009, though they were also reflected in a higher fiscal deficit for 2009. 8. In the pre-crisis years, high inflation was fueled by high growth rates of 9.7 percent in aggregate consumption as well as rising food prices. In 2009, inflation eventually fell to zero as incomes, including remittances, stagnated and the economy cooled down. The average wage level measured in U.S. dollars fell and the unemployment rate increased marginally. 9. In the past decade, the inflow of remittances played a significant role in the economy as workers abroad sent home funds. As the host countries of the Russian Federation and Kazakhstan fell into recession, the flow of remittances to the Kyrgyz Republic declined in 2009 to 23 percent of GDP. Coupled with a reduction in social transfers and wages, this affected domestic consumption, which eventually reduced imports and resulted in a fall in the current account deficit. C. Development Indicators 10. The Kyrgyz Republic is a low-income, mountainous, and predominantly agrarian country with more than 64 percent of its 5.4 million population residing in rural areas. The country is considered one of the poorest in Central Asia and the Europe and Central Asia region with a GDP per capita of US$ 860 (current US$) in 2009 as seen in figure 1. 11. According to the World Bank’s estimates of poverty based on international poverty lines at a purchasing power parity (PPP) of US$2.50 per day and US$5.00 per day, headcount poverty rates in the Kyrgyz Republic are relatively high compared with other countries in the Europe and Central Asian region (ECA) as seen in figures 2 and 3.2 2 The World Bank’s usage of ―Europe and Central Asia region‖ refers to the countries in Central and Eastern Europe, the Western Balkans, and the Common wealth of Independent States. 3 Figure 1: Comparison of GDP per Capita with Selected Regions current US$ 14,000 13,310 12,000 10,000 8,000 6,596 6,412 6,000 4,245 4,000 2,778 2,000 860 - Kyrgyz Central Asia CIS, Other Western ECA EU10 Republic Balkans Sources: World Development Indicators database (2009), World Ban k. http://data.worldbank.org/data- catalog/world-development-indicators. Notes: Values of country aggregates are averages. CA = Central Asia (Kazakhstan, Kyrgyz Republic, Tajikistan, Turkmen istan, and Uzbekistan); CIS other (Co mmonwealth of Independent States, other) = includes Armenia, Azerbaijan, Belarus, Georgia, Moldova, Russian Federation, and Ukraine. ECA = average for all developing countries of Europe and Central Asia. EU10 = new member states of the European Union (Bu lgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Ro man ia, Slovak Republic, and Slovenia). W B = western Balkans (Albania, Bosnia and Herzegovina, Croatia, the FYR Macedonia, Montenegro, and Serbia). Figure 2: International Poverty Rates for Figure 3: International Poverty Rates for Selected Countries PPP US$2.5 per day, 2008 Selected Countries, PPP US$5.0 per day, 2008 (in percent) (in percent) 60 90 50 80 70 40 60 30 50 20 40 30 10 20 0 10 0 Kazakhstan Romania Azerbaijan Lithuania Macedonia, FYR Russian Federation Moldova Albania Armenia Kyrgyz Republic Serbia Georgia Azerbaijan Serbia Poland Moldova Lithuania Georgia Armenia Albania Montenegro Macedonia, FYR Romania Kazakhstan Russian Federation Kyrgyz Republic Source: World Bank, ECAPOV database. 4 12. In terms of selected human development indicators, the Kyrgyz Republic is in a better position than low-income countries but lagging behind other countries in the Europe and Central Asia (ECA) Region. Yet, the persistence of low social indicators reflects continued economic problems in the country related to the prolonged period of transition and political instability. In addition, given that the country is land-locked and the economy is dependent on energy and trade, it is vulnerable to external conditions, which proved to be very volatile over recent years. This environment is not conducive to sustainable economic growth, which is essential for improving social indicators. Table 2: Selected Social Development Indicators: An International Comparison Kyrgyz Republic ECA Low Income Countries 1990– 2000– 1990– 2000– 1990– 2000– 99 09 99 09 99 09 Prevalence of undernourishment (% of population) 15 14 7 7 38 33 Poverty headcount ratio at PPP $1.25 a day (% of 25 19 4 4 — — population) School enrollment, primary (% net) 90 85 90 91 60 71 Share of women employed outside the agricultural sector (% 49 48 46 47 — — of total nonagricultural employment) M ortality rate, infant (per 1,000 live births) 58 36 42 23 104 81 M alnutrition prevalence, height for age (% of children under 33 18 — — — 46 5) M alnutrition prevalence, weight for age (% of children under 8 3 — — — 30 5) M aternal mortality ratio (modeled estimate, per 100,000 live 88 80 65 38 825 653 births) M ortality rate, under 5 (per 1,000) 68 41 50 27 165 126 Prevalence of HIV, total (% of population ages 15–49) 0.10 0.14 0.14 0.47 3.1 3.0 Incidence of tuberculosis (per 100,000 people) 144 157 90 95 252 308 CO2 emissions (metric tons per capita) 1.4 1 8 6.8 0.5 0.3 Improved sanitation facilities (% of population with access) 93 93 87 89 26 34 Improved water source (% of population with access) 78 86 92 95 55 62 Internet users (per 100 people) 0.1 11.1 0.3 14.4 0.0 1.1 M obile cellular subscriptions (per 100 people) 0 23 1 56 0 8 Telephone lines (per 100 people) 8 8 16 24 0 1 GNI per capita, Atlas method (current US$) 388 482 1946 3779 266 346 Gross capital formation (% of GDP) 18 19 24 22 18 21 Life expectancy at birth, total (years) 67 68 67 68 53 56 Trade (% of GDP) 82 105 57 65 40 55 Source: World Develop ment Indicators (2009 or latest available data), World Bank Notes: The table shows period averages. — = not available. 13. Looking at the country’s development progress through the lens of the Millennium Development Goals (MDGs) shows that the Kyrgyz Republic is partially on track to achieve the declared goals.3 According to the National Progress Report and United Nations data, it appears that goals related to eradicating extreme poverty and hunger, ensuring environmental sustainability, promoting gender parity, reducing child mortality, and developing global partnership for development are well on track and assessed as likely to be achieved. In contrast, some indicators that relate to health and education goals are showing slow progress and require monitoring and policy efforts. 3 For the country table, see http://unstats.un.org/unsd/mdg/. 5 14. Two health indictors are a cause for concern: the maternal mortality ratio and the tuberculosis incidence rate. Both indicators show an upward trend since 2000. The health sector in the country is undergoing sectorwide reforms to address the critical issues that must be tackled in order to improve health outcomes. These are issues of state financing of the sector, low salaries, insufficient payment incentives, ineffective arrangement of service delivery, lack of an integrated approach to sector management, and insufficient resources such as personnel, equipment, and drug management. Figure 4: Maternal Mortality Ratio, Figure 5: Tuberculosis Prevalence and 1990–2008 Incidence Rates, 1990–2008 Tuberculosis prevalence rate per 100,000 population (mid-point) 120 Tuberculosis incidence rate per year per 100,000 maternal deaths per 100 000 live births population (mid-point) 98 100 300 281 252 cases per 100 000 population 81 78 81 236 77 250 228 80 195 200 60 151 158 159 143 143 150 40 100 20 50 0 0 1990 1995 2000 2005 2008 1990 1995 2000 2005 2009 Sources: UN Interagency estimates for maternal mo rtality and UN for tuberculosis. 15. The World Bank data show that the net primary school enrollment rate in the Kyrgyz Republic has been gradually declining in 1999-2009. This indicates that not all boys and girls are entering school and, thus, will not be able to complete the full course of primary schooling or even perhaps achieve basic literacy. Figure 6: Primary School Enrollment Rate, 1999–2009 percent, net 90 88 86 84 82 80 78 76 74 72 70 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Source: World Develop ment Indicators , World Bank 6 D. Characteristics of the Poor 16. This section focuses on trying to answer three questions: who are the poor, where do they live, and why are they poor. The elaboration of these answers provides a profile of the poor which can be useful to policy makers, government officials at the central and local levels, socially oriented civil organizations, and donors. This information can help with the design of programs to not only ameliorate the circumstances of the poor but also raise the living standards of the population Number and geographic distribution of the poor 17. According to World Bank estimates based on the KIHS and using the upper national poverty line, the total headcount ―absolute‖ poverty rate registered at 31.7 percent in 2009 (see Box 1). This means that 1.72 million people in the country were considered to be poor and were unable to meet basic food and nonfood needs. The extreme poverty rate, which measures the incidence of food poverty, leveled at 3 percent in 2009, which implies that 162,000 people were not able to meet their basic food needs. 18. The incidence of rural poverty is significantly higher than that of urban poverty, 37.1 percent versus 22.0 percent, respectively. Similarly, the incidence of extreme poverty in rural and urban areas is 3.1 percent and 2.7 percent, respectively. Because over 64 percent of the population lives in rural areas, the number of the poor in rural areas is three times higher than in urban areas—1,297,000 versus 422,000, respectively. Thus, the majority of the poor, 75.4 percent, reside in rural areas. These numbers point to a continued poverty divide between rural and urban areas. What may be behind this is the greater extent of economic activity and job opportunities in urban areas. Another distinct feature of rural areas is their dependence on agricultural activities, which generally do not provide particularly good income and employment prospects, thus leading to the persistence of poverty in rural areas. Figure 7: Absolute and Extreme Poverty in Rural and Urban Areas, 2009 Poverty Headcount, in percent of population Poverty Distribution, in percent of poor 40.0 37.1 80.0 75.4 67.7 35.0 31.7 70.0 30.0 60.0 25.0 22.0 50.0 20.0 40.0 32.3 15.0 30.0 24.6 10.0 20.0 5.0 3.0 2.7 3.1 10.0 - - National Urban Rural Absolute Extreme Absolute Extreme Urban Rural Source: Staff calculat ions based on KIHS 2009. 7 BOX 1: Poverty Line in the Kyrgyz Republic In this report, the estimation of the absolute poverty line is based on the standard ―cost of basic needs‖ approach. This involves specifying a consumption bundle with food— including home-produced food—and nonfood components. The nonfood expenditures include nondurables and the imputed use value of durables, but they exclude housing costs (both rent as well as the use value of housing). To ensure that the poverty line reflects the consumption patterns of lower-income households, the reference food consumption patterns are derived from households in the third, fourth, and fifth consumption deciles. The extreme or food poverty line is established at the level of expenditures on food needed to consume 2,100 calories per day. The nonfood expenditure component of the poverty line is computed based on those households whose food consumption reaches 2,100 calories. The sum of these two components yields the ―absolute‖ or overall poverty line. This methodology provides reasonable estimates of the minimum food and nonfood expenditures needed in a particular country to achieve adequate nutrition while consuming other nonfood items considered absolutely essential. It should be noted that a poverty line does not reflect what society may think households should consume, nor does it even include all essentials of a dignified life (such as expenditures for school uniforms or health care). The NSC established the extreme and absolute poverty lines in 2003 and subsequently has adjusted them for inflation on an annual basis. In 2008, it updated the poverty lines because of the dramatic relative price changes and ensuing shifts in the consumption patterns of households. Though this change adversely affected the consistency in the measurement of poverty over time, leaving it unchanged would have yielded a biased picture of poverty in the country. However, to discuss the trend in poverty over the last half decade in this report, the poverty rates before and after 2008 were reestimated for comparison purposes. Based on these absolute and extreme poverty lines, poverty can be measured. The three most commonly used poverty indicators are head count index, the poverty gap, and poverty severity. These are all Foster, Greer, and Thorbecke type of poverty measures. An individual is considered poor if his or her per capita consumption is less than the poverty line. The headcount index is the percentage of the population whose per capita consumption is less than the poverty line 19. The poverty gap measures the depth of poverty. In 2009, it was 6.1 percent. On a related note, the measure of poverty severity—the squared poverty gap, which takes into account distribution among the poor—is also quite low at 1.8 percent. Mirroring the fact that poverty in the Kyrgyz Republic is mostly a rural phenomenon, the poverty gap and the severity of poverty are also higher in rural areas than in urban areas, 7.2 percent versus 4.2 percent for the poverty gap and 2.1 percent versus 1.3 percent for the severity of poverty, respectively. 8 Figure 8: Absolute and Extreme Poverty Gap and Severity percent Absolute Poverty Extreme Poverty 8.0 7.2 0.7 0.6 7.0 6.1 0.6 0.5 0.5 6.0 0.5 5.0 4.2 0.4 4.0 0.3 3.0 2.1 0.2 0.2 1.8 0.2 2.0 1.3 0.1 1.0 0.1 - - National Urban Rural National Urban Rural Poverty Gap Poverty Severity Poverty Gap Poverty Severity Source: Staff calculations based on KIHS 2009. 20. The regional distribution of poverty is also related to the rural-urban characterization of the poor. The oblasts that are predominantly agricultural and remote have higher rates of poverty. For example, Issyk-Kul, Naryn, and Osh have the highest rates of headcount poverty—46.1, 44.1, and 38.3 percent, respectively. In contrast, more urbanized and industrialized areas, like Chui oblast and Bishkek city, have the lowest poverty rates—21.2 and 13.2 percent, respectively. Taking into account oblast population levels, the majority of the poor reside in Osh and Jalal-Abad, where the absolute numbers of the poor reach 520,000 and 383,000, respectively. These numbers imply that half of all the poor—52 percent—are located in these two most populous oblasts. Although Bishkek city is the third largest region in terms of population, the absolute number of the poor there is the second lowest (116,000), just above Talas oblast, with an estimated 75,000. These two regions account for just 11 percent of all the poor in the country. So, geographic location is a very important factor explaining the divergences in poverty risks within the country. Figure 9: Incidence and Distribution of Poverty, by Oblast percent a. Share of Oblast’s Population in Poverty b. Distribution of Poor across Oblasts 50 46 44 35 30 45 38 30 40 37 31 33 25 22 35 30 17 18 20 16 16 25 21 14 15 12 12 20 14 10 7 7 8 15 10 10 7 6 3 44 10 3 3 5 5 1 2 2 0 0 Osh Bishkek Issykul Batken Chui Naryn Talas Jalal-Abad Jalal-Abad Bishkek Osh Naryn Talas Batken Chui Issykul Absolute Extreme Absolute Extreme Source: Staff calculations based on KIHS 2009. 9 21. The probability of being poor in the Kyrgyz Republic increases with the altitude at which the household resides. If one considers that the high-altitude zones are not only difficult to access but also remote areas with poor infrastructure and harsh climatic conditions, then it is not surprising to observe that in high mountainous areas the poverty incidence is relatively higher than in the plain zones, 48 percent versus 26 percent, respectively. Similarly, in terms of extreme poverty, the high-altitude zones have extreme poverty rates that are three times higher than in the plains, 6 percent versus 2 percent. Relative numbers, however, mask that in absolute terms the number of the poor in the plain zones is much higher than in the high-altitude zones, because the plain zones have a larger population base. Thus, the overwhelming majority, 62 percent of all the poor and 61 percent of all the extreme poor, reside in plain zones. Figure 10: Incidence and Poverty Distribution across Mountains and Plains percent a. Incidence b. Distribution 60 70 62 61 48 46 60 50 50 40 40 30 26 30 25 20 19 20 20 14 10 6 3 2 10 0 0 High Semi Plain High Semi Plain mountainous mountainous mountainous mountainous Absolute Extreme Absolute Extreme Source: Staff calculations based on KIHS 2009. 22. Location is thus a significant factor defining the poverty status of a household. People living in rural and especially high-altitude remote zones in Issyk-Kul or Osh oblasts are especially vulnerable to poverty in geographic terms. The question is then, what is behind these geographic factors? Given the heterogeneity that one observes in each geographic area, it is difficult to discern the specific factors. Nevertheless, the fact that geography plays an important role points to the notion that one needs to look at the degree of access to markets, transaction costs, and infrastructure to explain the poverty levels in more detail. Inequality 23. In 2009, the Gini coefficient of per capita consumption, a measure of inequality, was 25.5. It was higher for urban areas than for rural areas. Across oblasts, the Gini coefficient varies, with the lowest inequality in Jalal-Abad and the highest in Chui (19.7 and 25.6, respectively). Interestingly, Issyk-Kul and Osh oblasts have higher Gini coefficients than Bishkek, which is the most urbanized city in the country. 10 Figure 11: Gini coefficient by sector and oblast, 2009 percent a: By Rural-Urban Location b: By Oblast 30.0 24.7 25.2 25.6 24.4 23.4 25.1 23.4 Rural 24.1 25.0 19.7 20.0 percent Urban 25.8 15.0 10.0 National 25.5 5.0 0.0 23.0 24.0 25.0 26.0 percent Source: Staff calculations based on KIHS 2009. 24. The estimation of mean consumption as a share of the poverty line and the highest quintile’s mean consumption show the extent of inequality. It appears that households in the first quintile have just 70 percent of the value of the poverty line, whereas those in the fifth quintile enjoy consumption at 2.6 times the value of the poverty line. Similarly, the shares of mean consumption as a proportion of the fifth quintile’s mean consumption show that there is large gap between the poor and the wealthy. The poor’s consumption accounts for only 30 and 40 percent of the mean consumption of the most affluent. Another implication of this statistic is that households in the third quintile seem to be vulnerable to poverty, because their mean consumption is barely above the poverty level. Viewed from this perspective, the third quintile can be considered the near poor and any large negative shocks to the mean consumption of this quintile would induce a large increase in poverty. Figure 12: Mean Consumption by Quintile, 2009 a. As % of Poverty Line b. As % of Top Quintile’s Mean Consumption 3.0 1.2 2.6 1.0 2.5 1.0 2.0 0.8 1.6 0.6 1.5 1.2 0.6 0.5 1.0 0.4 1.0 0.7 0.4 0.3 0.5 0.2 - - 1 2 3 4 5 1 2 3 4 5 Quintile by Consumption per capita Quintile by Consumption per capita Source: Staff co mputations based on KIHS 2009 11 Living conditions 25. Disparities in dwelling conditions are large between the poor and the nonpoor. The poor live in more crowded spaces compared with both the nonpoor and average national levels. Both the total area of dwelling and the living area of dwelling per capita are significantly smaller for the poor. In addition, the poor have problems accessing basic housing amenities such as cold and hot water, gas, and heating. Although even on the national level, the availability of these services is low, the poor have a lower share of access to these utilities, which reflects the gap in living conditions between the poor and the nonpoor. Figure 13: Average Area of Dwelling per Capita square meters 25 20 20 19 15 15 15 14 11 10 5 0 National Poor Non poor Total area of dwelling per capita (m2) Living area of dwelling per capita (m2) Source: Staff calculations based on KIHS 2009. Note: m2 refers to square meters. Figure 14: Population’s Access to Housing Amenities percent 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 Central Individual Water Sewage Hot water Central gas Bath or Telephone heating heating supply system supply supply shower system National Poor Non poor Source: Staff calculations based on KIHS 2009. 12 26. One of the positive legacies of the Soviet era is that power supply coverage is universal (100 percent), both in rural and urban areas and among the poor and nonpoor. Another legacy is that because the power supply is heavily subsidized from the budget, fiscal constraints mean that the infrastructure in the energy sector is gradually deteriorating. Electricity tariffs are quite low in absolute terms at som 0.7 per kilowatt, but despite this there are clear differences in electricity consumption between the poor and the nonpoor and between consumption quintiles. Figure 15: Annual Electricity Expenditures Figure 16: Annual Electricity Expenditures, per Capita (in som) per capita by Poverty Status (in som) 900 798 700 800 591 600 700 588 600 500 472 500 400 359 394 400 335 300 265 300 200 200 100 100 0 0 1 2 3 4 5 Extreme Poor Nonpoor Quintile by Consumption per capita Poor Source: Staff calculations based on KIHS 2009. Demographic characteristics 27. The incidence of poverty is higher in larger households.. The relationship between household size and poverty headcount rate is positive and monotonic: the risk of poverty is more than two times higher for a family of five than for a family of three. Thus, the average size of the poor household is 5.0 persons, whereas in nonpoor households the average is only 3.5 family members, and a national average of 3.8 persons. The positive relationship between household size and extreme poverty rates holds as well: the size of the household for the extreme poor is 5.3 people. A similar relationship holds in urban and rural areas. The average size of a poor rural household is 5.2 people whereas the nonpoor household includes just 3.8 members. Similarly, in urban areas the average poor family consists of 4.5 persons, whereas the size of the average nonpoor family is 3.1 members. 13 Figure 17: Household Size, by Income and Location 6.0 5.2 number of household members 5.0 5.0 4.5 4.2 3.8 3.8 4.0 3.5 3.3 3.1 3.0 2.0 1.0 0.0 Poor Non poor All Poor Non poor All Poor Non poor All National Urban Rural Source: Staff calculations based on KIHS 2009. Figure 18: Relationship between Household Size and Absolute Poverty Incidence Rate percent 60 56.1 50 44.3 40 36.0 Poverty Rate 30 27.1 20 15.1 10 6.1 3.3 0 1 2 3 4 5 6 7 or more household size in number of persons Source: Staff calculations based on KIHS 2009. 28. It appears that behind the positive relationship between household size and poverty status is the demographic composition of the household. On the national level, the numbers of both children and adults are higher in poor households than in nonpoor households (except for the age group of 45– 64 years). Given that the number of elderly in both poor and nonpoor households is relatively equal— 0.20 and 0.23, respectively—it appears that poor households are associated with a higher number of children in the family. This is true regardless of the regional rural-urban divide: poor families are associated with higher numbers of children and working-age adults. 14 Figure 19: Age Composition of Poor and Nonpoor Households Average number of persons Poor Nonpoor All 1.6 1.4 1.3 1.2 1.2 1.0 1.0 1.0 0.9 0.8 0.7 0.8 0.7 0.8 0.7 0.6 0.7 0.6 0.6 0.4 0.4 0.3 0.2 0.2 0.2 0.2 0.0 under 5 5 to 14 15 to 29 30 to 44 45 to 64 65 and older age group Source: Staff calculations based on KIHS 2009. Figure 20: Age Composition of Households, by Income Category and Location Average number of persons a: Urban b: Rural Urban Poor Urban Nonpoor Rural Poor Rural Nonpoor 1.4 1.6 1.5 1.2 1.4 1.2 1.0 1.2 1.0 1.2 1.0 1.0 1.0 0.8 1.0 0.8 0.8 0.6 0.60.7 0.8 0.6 0.8 0.7 0.7 0.7 0.6 0.5 0.6 0.4 0.3 0.2 0.20.2 0.4 0.3 0.2 0.2 0.2 0.0 0.0 under 5 5 to 14 15 to 30 to 45 to 65 and under 5 to 14 15 to 30 to 45 to 65 and 29 44 64 older 5 29 44 64 older age group age group Source: Staff calculations based on KIHS 2009. 29. Further decomposing the households by number of children reveals that poor families are predominate in the categories of households with two to three children and those with more than three children, whereas in the categories of households with no children and with one child, nonpoor families are proportionately more common. In short, adding a child to a family with one child increases the risk of poverty. This correlation is also reflected in rural and urban areas. In both cases, poor households have more children. 15 Figure 21: Distribution of Households by Number of Children and Poverty Status percent 60 Poor 52 50 Nonpoor 42 38 40 All 31 30 25 27 24 18 18 20 11 8 10 4 0 0 1 2-3 4+ number of children in household Source: Staff calculations based on KIHS 2009. Figure 22: Distribution Households by Number of Children and Poverty Status, for Urban Rural Areas percent Urban Rural 60 60 53 49 Poor Poor 50 50 Nonpoor Nonpoor 41 38 40 40 33 29 28 30 26 30 27 22 20 15 20 15 8 10 10 10 4 3 0 0 0 1 2-3 4+ 0 1 2-3 4+ number of children in household number of children in household Source: Staff calculations based on KIHS 2009. 30. Another aspect of the relationship between demographic composition and poverty can be seen by looking at the poverty incidence by dependency ratio.4 It could be assumed that the higher the number of children in the household, the more difficult it is to maintain adequate consumption expenditures above the poverty line. If, however, the larger household size also implies more adults who might work, then one would not see a difference in dependency ratio. The estimated dependency ratio shows that there are indeed large differences in this indicator between poor and nonpoor households as seen in figure 23. At the national level, the dependency ratio for the nonpoor is 0.57, whereas for the poor the ratio is 0.98 (that is, for every adult there is a child to support). This 4 Dependency ratio = (su m of persons over age 64 and under age 15) / (nu mber of persons age 15–64). 16 relationship holds for both rural and urban areas but is more pronounced in rural areas, where the dependency ratio in poor households is significantly higher than in nonpoor households. Figure 23: Average Dependency Ratio, by Poverty Status and Sector 1.2 1.1 Urban 1.0 1.0 Rural 0.8 National 0.8 0.8 0.7 0.6 0.6 0.6 0.6 0.5 0.4 0.2 - Poor Nonpoor All Source: Staff calculations based on KIHS 2009. 31. In terms of age, the incidence of poverty is highest among households whose heads are 30–34 years and 35–39 years of age—46 and 36 percent, respectively. However, given the larger population base of the 65 and above category, this category has the largest share among the poor (16 percent), This observation points to the fact that the risk of falling into poverty varies with the age of the household head; it initially rises, reaching a peak at 30–34 years, and then again rises beginning at the age of 60 years. Figure 24: Poverty Incidence and Distribution by Household Head’s Age percent 50 46 Poverty Rate 45 Distribution of the Poor 40 36 35 32 33 31 29 29 29 30 27 28 25 20 16 14 15 15 13 14 12 10 8 5 5 5 3 0.2 0.2 0 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65+ Household head's age in years Source: Staff calculations based on KIHS 2009. 17 BOX 2: Sensitivity Analysis of Per Capita versus Adult Equivalence Scales Specification The issue that frequently arises in poverty analysis is whether to use per capita or adult equivalence scales in measuring consumption. This is because of the recognition that demographics might be behind the different needs of household members: children, adults, and the elderly. In addition, larger families might benefit from economies of scale in consumption. These might eventually introduce bias in estimates of poverty rates and lead to erroneous conclusions, especially relating to demographics and poverty. Good practice is to test the sensitivity of the results of poverty rates using per capita and adult equivalence scales. We have used the following adult equivalence weights to re- estimate poverty incidence among households with different demographic composition: (Na + 0.67Nc )0.8 , where Na = number of adults, Nc = number of children. The table displays the results of the sensitivity analysis. Table 3: Absolute Poverty Rates, Using Per Capita and Adult Equivalence Scales Number Number of Poverty rate, per Poverty rate, adult Difference of adults children capita, (%) equivalence, (%) (4)-(3) (1) (2) (3) (4) 1 0 3.3 10.8 7.4 1 1 8.8 22.3 13.5 1 2 39.4 39.4 0.0 1 3 28.2 28.2 0.0 1 4 51.3 14.4 -36.9 1 5 100.0 100.0 0.0 1 6 100.0 100.0 0.0 2 0 5.6 17.4 11.9 2 1 14.6 25.6 11.0 2 2 30.2 30.2 0.0 2 3 47.4 37.7 -9.6 2 4 67.7 45.7 -22.0 2 5 71.5 57.0 -14.5 2 6 100.0 97.8 -2.2 3 0 9.4 15.8 6.4 3 1 25.5 32.4 7.0 3 2 30.1 29.6 -0.5 3 3 42.6 33.4 -9.2 3 4 51.0 49.4 -1.5 3 5 71.1 37.6 -33.5 3 6 100.0 100.0 0.0 4 0 24.0 33.4 9.4 4 1 29.6 33.3 3.8 4 2 38.3 37.2 -1.1 4 3 51.0 42.5 -8.5 4 4 85.1 33.6 -51.5 4 5 89.2 89.2 0.0 Total Total 31.7 31.8 0.1 Source: Staff calculations based on KIHS 2009. It appears that, in total, the rates are virtually unchanged when we compare the per capita and adult equivalence poverty rates, 31.7 present versus 31.8 percent. However, household composition has a significant impact on the difference between the two methods. The results show that for smaller household size, using the adult equivalence method renders higher poverty than using the per capita estimation. In contrast but as expected, for larger families with many children, adult equivalence weights show lower poverty relative to the per capita method. The results of this analysis should be kept in mind when interpreting data, but in this report the case for using adult equivalence is not compelling because the choice of parameters (weights) is arbitrary, which introduces subjectivity. In the end, the use of the equivalence scale provides a similar ranking of different households to the ranking obtained when applying the per capita method, which allows us to continue using the per capita method. 18 32. So, it appears that an additional important dimension of poverty is the demographic structure of the family. It is useful to think about this relationship from the perspective of causality: whether poverty creates the larger households or larger households are more prone to poverty. It is quite a complex issue and it is not be possible to resolve it within the framework of this report. Nevertheless, it seems that current policy measures to focus the social safety net toward households with children is warranted and supported by the micro data. Gende r 33. Interestingly, the poverty rate is higher for male-headed households. 5 According to the data, the poverty incidence in male-headed households is 33.3 percent, compared with 28.0 percent in female-headed households. This relationship is more pronounced in rural areas, where male-headed households dominate the poor by 7 percent over female-headed households. In urban areas, it is the opposite: households with female heads have higher rates of poverty. For the incidence of extreme poverty, male-headed households have a rate of 2.9 percent, marginally lower than for female-headed households, at 3.3 percent. Figure 25: Gender of the Household Head and Poverty percent a: Absolute Poverty b: Extreme Poverty Male headed HH Female headed HH Male headed HH Female headed HH 80.0 74.5 80.0 68.2 70.0 70.0 60.0 60.0 50.0 50.0 percent percent 40.0 40.0 33.3 31.8 28.0 30.0 25.5 30.0 20.0 20.0 10.0 10.0 2.9 3.3 0.0 0.0 Absolute Poverty Rate Distribution of the Extreme Poverty Rate Distribution of the Poor Poor Source: Staff calculations based on KIHS 2009. 5 ―Head of household‖ is defined as the person whom all members of the household regard as the head and who is responsible for running the household (regardless of the presence of a partner). 19 Figure 26: Gender Status of the Household Head and Rural-Urban Poverty percent Female headed HH Male headed HH 31.6 Rural poverty rate 38.9 23.2 Urban poverty rate 21.4 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 percent Source: Staff calculations based on KIHS 2009. 34. From the statistics, it is difficult to derive definitive conclusions about the gender and poverty linkages; at best, the relationship is mixed. Whether female or male-headed households are less successful in coping with poverty is a long-researched issue, but one may start thinking about this through the prism of entrepreneurship abilities, which might have a gender dimension during the transition period. Anecdotal evidence might suggest that the transition period opened up new entrepreneurial opportunities for the former Soviet countries and that in the Kyrgyz Republic women were more successful in relative terms in seizing those opportunities. Inte rnal migration 35. Internal migration appears to be important for household welfare. The data show that 26 percent of household heads migrated from elsewhere in the country. As mentioned, the rural-urban divide in poverty hints at the better economic prospects in urban areas. This might be one of the reasons for one-way rural-urban migration. The survey statistics reveal that the urban centers of Bishkek and Chui register the highest number of household heads who were born elsewhere—71 and 60 percent, respectively—whereas rural Jalal-Abad and Batken oblasts have the lowest share of household heads born elsewhere, 1 and 4 percent respectively. The data show that as per capita consumption increases the share of household heads who migrated (predominantly from rural to urban areas, Bishkek city in particular) also increases. This does not directly imply that domestic rural-urban migration is associated with lower poverty rates. However, all else being equal and assuming that the positive relationship between per capita consumption and migration status also holds for the rural poor, migration might be viewed as one of the coping strategies for the poor. 20 Figure 27: Internal Migration by Household Heads, by Quintiles of Per Capita Consumption percent 50% 45% 45% 40% 34% 35% 30% 26% percent 25% 22% 20% 16% 14% 15% 10% 5% 0% All 1 2 3 4 5 Quintiles of per capita consumption Source: Staff calculations based on KIHS 2009. Employme nt status 36. Interestingly, 65 percent of poor household heads are reported as employed and only 35 percent as unemployed. Thus, it appears that unemployment of the household head within the poor category has little bearing on the poverty status of the household. Generally, the shares of employed and unemployed household heads among the poor are not much different to the shares for the whole population of the survey, where the unemployed constitute 33 percent and 67 percent report having employment. It appears that this relationship is driven by rural areas, where the employed dominate the poor category. In contrast, in urban areas in poor households, employed and unemployed heads have relatively equal shares. This fact brings to attention the issue of underemployment, which is pertinent mainly in rural areas. It could be assumed that once the household heads in rural areas could be categorized as underemployed (which is close to unemployed) then one would find the relationship between absolute poverty and employment similar to the one between extreme poverty and employment. For the extreme poor, the employed constitute 45 percent while the unemployed make up 55 percent. This direction of the relationship confirms the theory that the unemployed are more likely to fall into the extreme poverty category. 21 Figure 28: Employment Status of Household Head and Absolute Poverty Distribution percent a. Share of Population Employed by b. Share of Population Employed by Sector Poverty Status and Poverty Status 69 68 Poor Nonpoor 68 68 80 72 70 67 65 67 70 67 60 49 66 50 66 40 65 65 30 65 20 64 10 64 0 All Poor Nonpoor Urban Rural Source: Staff calculations based on KIHS 2009. Figure 29: Employment Status of Household Head and Extreme Poverty Distribution percent HH head employed 45 Extreme Poverty HH head not employed 55 0 10 20 30 40 50 60 Source: Staff calculations based on KIHS 2009. 37. Among the poor, employed household heads are disproportionately represented indicating that many are the working poor who receive low wages and/or are less than fully employed in terms of working hours. Yet, in terms of the poverty headcount rate among all households, families with unemployed heads have a higher poverty incidence (33.7 percent) than those with employed heads (30.8 percent). Similarly for the extreme poor, the poverty rate is higher for families with unemployed heads, at 5 percent, than for families with employed heads, at just 2 percent. This appears to support the conventional wisdom that a positive relationship between unemployment and poverty exists, although the link is rather weak relative to the one that one would expect. 22 Figure 30: Absolute and Extreme Poverty Rates by Employment Status percent 40.0 33.7 35.0 30.8 30.0 25.0 20.0 15.0 10.0 5.0 5.0 2.0 0.0 HH head not HH head employed HH head not HH head employed employed employed Absolute Poverty Extreme Poverty Source: Staff calculations based on KIHS 2009. 38. In terms of principal area of employment for the household head, the survey statistics show that poor households are more represented in the informal, low-paying sectors such as peasant farming and wage work for private individuals. In contrast, nonpoor household heads take up a larger share in the sectors of big enterprise, collective farms, and organizations and institutions that appear to be more formal. This type of relationship holds in both rural and urban areas and suggests that the poor are likely to be working in low paying and less secure jobs. Table 4: Area of Employment of Household Heads percent Nation Urban Rural All Poor Nonpoor All Poor Nonpoor All Poor Nonpoor At an enterprise, organization, 40 34 43 50 43 51 35 32 37 collective farm, agricultural cooperative, institution At a (peasant) farm 18 24 15 1 2 1 27 30 26 On an individual basis 21 19 21 27 26 27 17 18 16 Wage work for pri vate 18 20 17 20 29 19 17 18 16 individuals (individual entrepreneurs) Other 2 2 3 0 1 0 3 2 4 Source: Staff calculations based on KIHS 2009. 39. Analysis of the types of jobs that poor and nonpoor household heads are engaged in reveals that the poor are more likely to be involved in self-employment, whereas the nonpoor are more likely to be in the wage-paying sectors. This might indicate that a lack of wage employment forces the poor to go for self-employment as a job of last resort rather than from choice. Again, this points to the informality and insecurity of the jobs in which the poor find themselves in the labor market. 23 Table 5: Type of Employment of Household Heads percent Nation Urban Rural All Poor Nonpoor All Poor Nonpoor All Poor Nonpoor At own enterprise or own commercial 31 41 26 4 4 4 43 47 40 business As a worker for wage paid in cash or in 67 54 73 96 96 96 54 47 58 kind, or for money allowance As a member of a cooperative, collective 1 4 0 2 4 1 farm, agriculture cooperative Free assistance at an enterprise owned by 1 1 1 1 1 1 relatives Source: Staff calculations based on KIHS 2009. Education 40. As the data show, the relationship between poverty status and the educational attainment of the household head is negative. The statistics confirm that low or no educational attainment by the household head increases the likelihood of household poverty. The incidence of absolute poverty among the illiterate is 63 percent as seen in figure 31. Further, more than 59 percent of household heads with no elementary education are poor. However, with higher educational degrees for the household head, poverty falls only slowly and erratically. Of household heads with a general secondary degree, the poor account for 39 percent; among university degree holders, the incidence of poverty is 15 percent. These findings are in line with observations from other countries, but what is specific to the Kyrgyz Republic is a low level and magnitude of the poverty-reducing premium attached to additional education. Although the highest poverty rate is in the illiterate category, the largest share of the poor (50 percent) is in the general secondary school category, because of the large population share of this category. 41. It appears that the high rates of poverty are driven mainly by poverty in rural areas in all educational categories except for illiteracy and elementary school, in which the urban poverty rates are higher than rural ones as seen in figure 32. So, differentiation of poverty by the educational attainment of the household head highlights the gap in human capital between rural and urban areas. 24 Figure 31: Poverty Rates and Educational Attainment of Household Head percent a: Absolute Poverty b: Extreme Poverty Distribution of the Poor Poverty Headcount Rate Distribution of the Poor Poverty Headcount Rate Illiterate 4 1 63 Illiterate 2 No elementary 1 No elementary 1 59 6 4 Elementary 7 Elementary 4 24 General Secondary 6 General Secondary 7 (incomplete) 36 (incomplete) 4 44 General Secondary General Secondary 50 3 (complete) 39 (complete) Prof. Technical 19 Prof. Technical 14 4 33 Secondary Prof. 6 Secondary Prof. 12 1 26 Incomplete Higher 0 Incomplete Higher 1 0 30 Higher 14 Higher 8 3 15 0 10 20 30 40 50 0 20 40 60 80 percent percent Source: Staff calculations based on KIHS 2009. Figure 32: Education of Household Head and Absolute Poverty Rates, by Location percent Rural Urban Illiterate 62 66 No elementary 63 55 Elementary 21 37 General Secondary (incomplete) 38 31 General Secondary (complete) 43 30 Prof. Technical 38 21 Secondary Prof. 30 20 Incomplete Higher 49 12 Higher 24 8 0 10 20 30 40 50 60 70 percent Source: Staff calculations based on KIHS 2009. 25 42. In addition, decomposing the data by educational attainment, gender of household head, and quintile of per capita consumption shows that regardless of the gender of the household head, those households in which the heads have attained higher education enjoy higher per capita consumption (fourth and fifth quintiles). In contrast, poor households (in the first and second quintiles) dominate in the category of general secondary school. This reinforces the observed relationship that the educational level of a household head plays an important role in defining the poverty status of the household. Table 6: Education Level of Household Head, by Gender and Consumption Quintiles percent Quintiles of Per Capita Consumption 1 2 3 4 5 Education Household Head Household Head Household Head Household Head Household Head Le vel Male Female Male Female Male Female Male Female Male Female Total 100 100 100 100 100 100 100 100 100 100 Higher 7 6 9 15 13 13 22 18 27 29 Incomplete 2 0 0 0 1 1 1 2 3 3 highe r Secondary 9 13 13 17 14 19 17 13 17 22 professional Professional 22 3 11 4 16 8 17 5 14 10 technical General 52 48 58 37 44 33 35 42 30 25 secondary (comple te) General 4 8 5 3 5 7 5 8 5 4 secondary (incomple te) Elementary 4 8 3 12 7 17 3 10 4 6 No 1 2 1 1 0 2 0 1 0 1 elementary Illiterate 0 11 0 11 1 1 0 1 0 0 Source: Staff calculations based on KIHS 2009. 43. Looking beyond the household head, relating poverty status to the age of all adults in the household shows a familiar picture: the nonpoor prevail in the higher and secondary professional education categories and the poor come from general secondary and illiterate categories. 44. Given the role of education in defining the risk of poverty, it is also important to look at the attendance of school-age children by wealth categories. This could provide a hint about the issue of the poverty trap in the context of the country. The data make it possible to estimate the percentage of children who attended school. The picture that emerges from the data is mixed. Comparing the poor to the nonpoor, a higher percentage of the nonpoor attend school, but comparing the extreme poor to the nonpoor shows that the extreme poor have a marginally higher share of attendance. Nevertheless, decomposing school attendance by quintiles of per capita consumption reveals that more wealthy households (fourth and fifth quintiles) have higher rates of school attendance. 26 Table 7: Education Level of Adults Aged 25 and Older, by Poverty Status and Gender percent Extreme Poor Poor Nonpoor All Education level M ale Female M ale Female M ale Female M ale Female Total 100 100 100 100 100 100 100 100 Higher 9 16 7 11 21 23 17 20 Incomplete higher 4 1 1 2 1 1 1 Secondary 5 8 9 12 15 20 13 18 professional Primary professional 24 4 14 6 14 6 14 6 technical (with general secondary) General secondary 49 60 58 57 38 37 44 42 (complete) Primary professional 0 0 1 0 1 1 1 1 technical (without general secondary) General secondary 1 5 5 5 5 5 5 5 (incomplete) Elementary 8 6 4 3 5 6 4 5 No elementary 1 1 1 1 0 1 0 1 Illiterate 0 1 0 3 0 1 0 1 Source: Staff calculations based on KIHS 2009. Figure 33: School-Age Children Who Attended School percent a. By Poverty Status b. By Quintiles of Per Capita Consumption 80 82 81 79 79 81 79 80 78 79 78 78 78 77 78 77 76 77 76 75 77 75 76 76 74 76 73 72 75 1 2 3 4 5 75 Extreme Poor Poor Nonpoor Quintile by consumption per capita Source: Staff calculations based on KIHS 2009. 45. Further decomposing school attendance by age and gender categories shows that the poor and the extreme poor stop schooling after 16 years. The nonpoor are disproportionately represented in the age categories of 14–16, 17–19 and 20–24 years; that is, those age categories correspond to higher or professional education. This seems to be driven by the higher attendance rate in urban areas compared 27 to rural areas. Noteworthy is that at the national level, school attendance by girls is consistently higher than boys across all age categories. 46. Among the extreme poor, there are disparities in school attendance by gender, as shown in the table. For example, girls age 7–10 have lower attendance rates than boys, and this appears to occur in rural areas. A similar fact is observed for the 14–16 age category, where girls’ school attendance is much lower than boys’, but this time the difference seems to occur in urban areas. It is not clear from the data which reasons cause the gender disparities. Table 8: Children’s School Attendance by Age Group, Poverty Status, and Gender percent Age Group Extreme Poor Poor Nonpoor All M ale Female M ale Female M ale Female M ale Female National 7–10 years 98 86 95 98 98 99 97 98 11–13 years 100 99 99 100 100 100 100 100 14–16 years 96 78 90 91 95 97 93 96 17–19 years 9 23 37 60 65 74 57 70 20–24 years 8 33 16 33 26 34 23 33 Urban 7–10 years 99 96 100 99 99 97 99 98 11–13 years 98 100 98 100 100 100 99 100 14–16 years 83 51 89 92 98 97 96 95 17–19 years 20 21 45 45 69 80 64 73 20–24 years 1 40 11 28 41 42 35 39 Rural 7–10 years 97 84 93 97 97 100 95 99 11–13 years 100 99 99 100 100 100 100 100 14–16 years 100 96 90 91 93 98 92 96 17–19 years 4 33 35 66 62 70 53 69 20–24 years 12 20 18 35 17 27 17 30 Source: Staff calculations based on KIHS 2009. 47. Another important question in this context is whether the poor have less education because they cannot afford advanced education or because they do not see the benefit of higher education. The data on household education expenditures (both formal and informal) suggest an answer. It appears that the nonpoor spend a considerable amount on education: seven times as much as the poor. Similarly, more wealthy consumption quintiles (fourth and fifth) spend 10 to 20 times more than lower quintiles. This does not in itself indicate causality between educational expenditure and poverty status, but it hints at the hurdles in affordability of education for the poor. 28 Figure 34: Expenditures on Education som a: By Poverty Status b: By Quintiles of Per Capita Consumption Education Informal payments (KGS), per Education Informal payments (KGS), per capita capita Education expenditures (KGS), per capita Education expenditures (KGS), per capita 5 62 2,079 Quintiles of per capita Nonpoor 38 1,168 4 32 consumption 1,113 14 3 28 Poor 604 159 2 16 344 Extreme Poor 14 86 1 13 101 0 500 1,000 1,500 0 500 1,000 1,500 2,000 2,500 Source: Staff calculations based on KIHS 2009. Household income and consumption basket 48. As expected, the survey data show that levels of income per capita are much higher in urban areas than in rural areas. This reflects the higher-earning opportunities in the urbanized areas. It is also confirmed by looking at the oblast level of income: Bishkek and Chui oblasts have the highest levels of income per capita, som 32,625 and som 26,735, respectively. Interestingly, Osh and Jalal-Abad oblasts are in third and fourth place in terms of income per capita with som 24,227 and som 22,103, respectively. Figure 35: Average Annual Income per Capita som a: By Location b: By Oblast 35,000 32,625 Rural 19,527 30,000 26,735 24,227 25,000 22,103 19,557 20,000 Urban 29,908 14,726 15,000 12,632 11,541 10,000 National 23,203 5,000 0 0 10,000 20,000 30,000 40,000 KGS, som Source: Staff calculations based on KIHS 2009. 29 49. Another interesting statistic is the difference in income levels between poverty groups. The survey results indicate that the extreme poor differ from the poor by a factor of almost two and that the difference between the extreme poor and the nonpoor is a factor of almost four. The income per capita of the poor is one-third of the income of the nonpoor. The difference in income level per capita is also reflected in the structure of income in poverty groups. Thus, the extreme poor are less reliant on income from work but have a higher share of income from pensions, social benefits , and welfare from the local administration, relatives, or friends (including remittances) compared to the poor and the nonpoor. Relative to the nonpoor, the poor have higher shares of pensions and social benefits as income sources but equal shares of income from work. Figure 36: Annual Income per Capita by Poverty Status som 30,000 27,892 25,000 23,203 20,000 15,000 13,121 10,000 7,069 5,000 0 National Extreme Poor Poor Nonpoor Source: Staff calculations based on KIHS 2009. Table 9: Structure of Income by Poverty Status percent Extre me Poor Poor Nonpoor Income from work 66 80 80 Pensions 18 12 11 Social benefits 5 1 0 Material support: 9 5 6 local admin, relatives or friends Other income 2 2 3 Total 100 100 100 Source: Staff calculations based on KIHS 2009. 50. The data show that for the extreme poor, social benefits are less important than pensions and material support from others. For the poor, social benefits are also a less important source of income than other sources. There are two possible reasons for this : one is imperfect targeting and one is the small amount of social benefits. 30 BOX 3: To What Extent Does the Existing Safety Net Protect the Poor In 2010, the World Bank conducted a study using micro data from KIHS 2008 to assess the effectiveness of the Kyrgyz Republic’s social safety net in protecting the poor. The government supports several types of programs that provide a range of benefits and target different groups in society. The study found that the distributional incidence of social transfers varies significantly by program.  Pensions are the most important transfers in terms of poverty reduction, although poverty reduction is not their main objective.  Informal transfers are the second most important transfer in terms of poverty reduction. They reduce the extreme poverty rate by two percentage points.  The Unified Monthly Benefit (UMB) and the Monthly Social Benefit (MSB) target cash transfers to poor and vulnerable households. Their benefits are progressive, with benefits representing a higher share of total household consumption for those in the lower quintiles. More than half of total transfers accrued to the poorest quintile.  Scholarships, utility and housing subsidies, and informal transfers were regressive, with richer households benefiting proportionally more than poorer households. The data show that 34 percent of utility and housing subsidies are allocated to the top 20 percent of households, while the poorest 40 percent receive only 24 percent of the total benefit value. In terms of poverty-reducing impact, the study found that in the absence of any social transfer, extreme poverty rates would be higher. The extreme poverty rate is reduced by four percentage points after accounting for all social transfers. Social transfers reduce the extreme poverty gap from 2.3 percent to 1.2 percent; that is, the extreme poverty gap is almost halved. Noncontributory benefits are far less effective in reducing extreme poverty. Their impact on poverty incidence and the poverty gap is limited. Nonetheless, the UMB remains the most effective targeted transfer. Its impact on poverty reduction could be further strengthened by increasing coverage and benefit levels. For more details on the study, refer to Appendix 3. Source: F. Gassmann. 2011. “To What Extent Does the Existing Safety Net protect the Poor,� PSIA series, World Bank. 51. Differences in the consumption expenditures are directly related to differences in income. Per capita consumption for the extreme poor is nearly one-third of consumption for the poor and more than four times higher than for the nonpoor. Consumption expenditures between the poor and the nonpoor differ by a factor of two, which is what the income statistic showed. As other surveys have shown, there is a significant difference between absolute income and consumption per capita levels. KIHS 2009 shows that households report consumption that is higher than income by more than som 2,000. The difference is notably higher for the nonpoor category. 31 Figure 37: Annual Per Capita Consumption, by Poverty Status som 35,000 33,214 27,621 30,000 25,000 20,000 15,595 15,000 9,828 10,000 5,000 0 National Extreme Poor Poor Nonpoor Source: Staff calculations based on KIHS 2009. 52. It appears that the Engel relationship is also observable for households in the Kyrgyz Republic; that is, with higher levels of income, proxied here by higher consumption (the nonpoor category), the share of food in total consumption declines. The nonpoor spend only 60 percent of their total consumption expenditure on food, whereas the poor and the extreme poor spend 68 percent and 73 percent of total consumption respectively on food as seen table 10. This is additional evidence that the poor are vulnerable to high food inflation. Among food items, the highest shares are devoted to bread, meat, vegetables, and sugar, which together account for 62 percent of consumption expenditure. Comparing poverty groups, the data show that the extreme poor and the poor consume larger shares of bread, potatoes, and vegetable oil and fats than the nonpoor. In contrast, the nonpoor consume more milk, meat, fish, and vegetables and spend more on eating out. Poverty profile using regression model 53. To conclude on the poverty profile section, the standard per capita consumption regression models for rural and urban areas were estimated using KIHS 2009 applying the Ordinary Least Squares method. The model makes it possible to summarize the findings from the poverty profile in a concise and elegant way. The main assumption in applying the regression method is that there are factors that are important in explaining the level of per capita consumption. The explanatory variables are some of the factors that were considered earlier, which are assumed to be exogenous. 54. In general, the model estimates show expected results as seen in table 11. Thus, the geographic coefficients for the urban regression model show negative signs. Since the reference place was the city of Bishkek, the coefficients indicate that residence in urban Issyk-Kul, Jalal-Abad, Naryn, Talas, and Chui is associated with reduced levels of per capita consumption and more poverty. Similarly, for the rural regression, where the reference group was the Jalal-Abad oblast, the rural areas of Batken, Talas, and Chui have positive and significant signs, which indicate higher per capita consumption in those areas. 32 Table 10: Share of Food Groups in Total Food Consumption, by Poverty Group percent National Extreme Poor Nonpoor Poor Share of food consumption in total consumption 61 73 68 60 Food consumption 100 100 100 100 Bread and bakery foods 24 34 31 23 M ilk and dairy produce 9 8 7 10 M eat and meat foods 17 15 15 18 Fish and fish foods 1 0 0 1 Vegetable oil, margarine, and other fats 5 7 6 4 Eggs 2 2 2 2 Potatoes 4 6 5 4 Vegetables, melons, and gourds 10 6 9 10 Fruits and berries 6 4 5 6 Sugar 11 10 11 11 Tea, coffee, cacao 1 2 2 1 Nonalcoholic beverages 1 0 0 1 Other food products 2 2 1 2 Alcoholic beverages 1 0 1 1 Tobacco 1 1 1 1 Eating out 5 2 3 6 Source: Staff calculations based on KIHS 2009. 55. Consistent with our observations, the regression model also shows that household size is a significant variable that explains the level of per capita consumption and that is negative. Thus, with an increase in household size, the level of per capita consumption declines; that leads to more poverty, but that decline is negative—that is, the rate at which consumption declines goes down with household size. This is observed for both rural and urban estimations, but the effect is more significant for urban areas. In terms of household age the linkage to per capita consumption is less obvious. In rural areas, the older the household head, the higher the level of per capita consumption, but in urban areas this relationship is not significant. 56. In contrast, the gender of the household head plays an important role in explaining consumption in the household. The model shows that in both rural and urban areas having a female household head is associated with less per capita consumption. Finally, as expected, the education attainment of a household head is a strong predictor of consumption per capita. Having a lower degree of education is associated with less consumption. The sign are negative and statistically significant, which is supported by the statistical tables on the relationship between poverty and education. 33 Table 11: Results of Regression Model: Explaining Per Capita Consumption, 2009 Urban Rural Coefficient Standard Error Coefficient Standard Error Oblast and type of location Bishkek - Urban (dropped) (dropped) Issykul - Urban -0.130*** 0.02 (dropped) Issykul - Rural (dropped) -0.199*** 0.03 Jalal-Abad - Urban -0.020 0.02 (dropped) Jalal-Abad - Rural (dropped) (dropped) Naryn - Urban -0.067** 0.03 (dropped) Naryn - Rural (dropped) -0.021 0.03 Batken - Urban -0.032 0.03 (dropped) Batken - Rural (dropped) 0.138*** 0.03 Osh - Urban 0.013 0.03 (dropped) Osh - Rural (dropped) 0.043 0.03 Talas - Urban -0.137*** 0.03 (dropped) Talas - Rural (dropped) 0.115*** 0.03 Chui - Urban -0.014 0.03 (dropped) Chui - Rural (dropped) 0.108*** 0.03 Household characteristics Log of household size -0.351*** 0.04 -0.466*** 0.06 Log of household size squared -0.075*** 0.02 -0.038 0.03 Characteristics of household head Log of household head’s age -0.031 0.03 0.235*** 0.04 Gender of household head Male (dropped) (dropped) Female -0.094*** 0.02 -0.075*** 0.02 Education of household head Higher (dropped) (dropped) Incomplete h igher -0.066 0.06 -0.124 0.11 Secondary professional -0.140*** 0.02 -0.133*** 0.03 Professional technical -0.204*** 0.03 -0.133*** 0.04 General secondary (complete) -0.229*** 0.02 -0.199*** 0.03 General secondary (incomplete) -0.265*** 0.04 -0.292*** 0.04 Elementary -0.313*** 0.04 -0.267*** 0.04 No elementary -0.325*** 0.08 -0.428*** 0.08 Illiterate -0.326*** 0.10 -0.339*** 0.07 Constant 5.257*** 0.11 4.206*** 0.15 Nu mber of observations 3,041 1,943 Adjusted R2 0.36 0.41 Source: Staff calculations based on KIHS 2009. Notes: *** p < 0.01, ** p < 0.05, * p < 0.1. The dependent variable is per capita consumption level. Reference categories: rural oblasts, Jalal-Abad; urban oblasts, Bishkek; gender, male; education, household head’s education is higher. 34 E. The Dynamics of Pove rty and Inequality in 2006 – 2009 The Trend in Poverty Levels 57. According to World Bank estimates based on KIHS 2006, 2007, 2008, and 2009, the incidence of poverty in the Kyrgyz Republic fell significantly between 2006 and 2008—by 29.3 percent—but remained unchanged between 2008 and 2009. Despite no change in absolute poverty rates in 2008–09, the extreme poverty rate has been falling steadily, from 17.1 percent in 2006 to 3 percent in 2009. Thus the financial crisis of 2008–09 and subsequent slowing of economic growth appear to have had a mixed impact on poverty trends. While the downward trend in absolute poverty was reversed, the extreme poverty rate continued to decline, which might indicate that the poorest felt little of the international instabilities. Also, it is important to remember that owing to population growth, the absolute number of the poor in the Kyrgyz Republic increased between 2008 and 2009 by an estimated 25,000 people. It is believed that the strong reduction in poverty in the first part of the observed period and the stagnation in the later part is attributable to the high rates of economic growth in 2006–08 and subsequent slowing of the GDP growth rate in 2009. Not least was the role of low inflation in 2009 in reducing extreme poverty rates, because the share of food items in the budget of the extreme poor is about 70 percent. Figure 38: Poverty Trends, 2006–09 percent 70 61 60 54.6 50 percent 40 31.7 31.7 30 20 17.1 11.7 10 6.1 3.0 0 2006 2007 2008 2009 Absolute Poverty Extreme Poverty Source: Staff calculations based on KIHS 2006-09. 58. In regional terms, the impact of the economic downturn was felt more in rural areas than in urban areas, where absolute poverty rates have decreased slightly. However, within the extreme poor group, the dynamics are different. Here the rural extreme poor saw the continued decline throughout the crisis period, but the downward trend for urban extreme poor stopped in 2009. So, the recent estimates show that the rates of extreme poverty are now similar in urban and rural areas, at 3 percent. This theoretically could be attributable to the high food prices in 2008, which negatively affected the urban extreme poor who are net food buyers unlike the rural extreme poor. 35 Figure 39: Rural and Urban Poverty Trends, 2006–09 percent a. Absolute Poverty b. Extreme Poverty 80 25 69 22 70 64 20 60 47 15 50 15 38 37 37 40 10 9 30 23 8 22 20 5 5 3 3 3 10 0 0 2006 2007 2008 2009 2006 2007 2008 2009 Urban Rural Urban Rural Source: Staff calculations based on KIHS 2006–09. 59. The dynamics of poverty over the years are different in different oblasts. Both absolute and extreme poverty rates fell most in the Jalal-Abad, Batken, and Osh oblasts. In urban Bishkek city, the decline in absolute poverty was the slowest, and extreme poverty rates have actually increased from about 1 percent in 2006 to 3 percent in 2009. So the impact of economic development was very different for urbanized centers and rural agricultural areas. Looking at 2008 and 2009, the years of financial crisis, reveals that Naryn, Batken, and Chui oblasts were hit hardest; absolute poverty rates have increased between 2008 and 2009. In terms of extreme poverty rates, the crisis affected Bishkek and Batken oblasts most. Noteworthy is Jalal-Abad oblast, which holds a record for managing to reduce extreme poverty from 30 percent in 2006 to almost 0.5 percent in 2009. This might be due in part to the large construction project to complete the Kambarata 2 hydroelectric station on the Naryn River. Table 12: Poverty Trends, by Oblasts, 2006–09 Absolute Poverty Extreme Poverty 2006 2007 2008 2009 Change 2006–09 2006 2007 2008 2009 Change 2006–09 Bishkek 22 14 15 14 -8 0.6 0.6 2 3 3 Issyk-Kul 64 58 52 46 -18 15 13 17 7 -8 Jalal-Abad 80 78 40 37 -43 30 22 10 0.5 -29 Naryn 75 56 43 44 -30 23 23 12 10 -12 Batken 73 67 21 31 -42 28 14 4 6 -22 Osh 74 72 38 38 -37 20 13 5 2 -19 Talas 61 52 43 33 -27 14 15 5 3 -11 Chui 41 30 16 21 -20 7 2 2 2 -5 Source: Staff calculat ions based on KIHS 2006–09. 36 60. The depth of poverty, as indicated by the poverty gap, has been following the general poverty trends—that is, rapidly declining in the earlier period but stagnating in the years of financial crisis. Between 2008 and 2009, the reduction in the poverty gap appears to have been driven by the decline in rural areas compared with urban areas, where the reduction in the poverty gap was slower. Overall this dynamic indicates that poverty is becoming shallower and that the average per capita consumption of the poor is approaching the level of the poverty line. In addition, the decline in the poverty severity index for all groups indicates that the level of inequality among the poor is also falling—that is, over the period poverty is more uniformly distributed among the poor, reflecting the fact that the poor are quite homogenous. This is evidence that economic growth has been shared among population groups in a manner that benefited the poor. Figure 40: Absolute Poverty and Extreme Poverty Gaps, 2006–08 percent 25 21 20 18 17 15 15 13 10 9 9 8 7 6 5 4 5 4 3 2 2 2 2 1 1 1 1 0.5 1 0 National Urban Rural National Urban Rural Absolute Poverty Gap Extreme Poverty Gap 2006 2007 2008 2009 Source: Staff calculations based on KIHS 2006–09. 37 Figure 41: Absolute Poverty and Extreme Poverty Severity, 2006–09 percent 9 8 8 7 7 6 6 5 5 5 4 3 3 3 3 2 2 2 1 1 0.8 0.9 1 0.4 0.4 0.6 0.4 0.5 0.2 0.2 0.3 0.2 0.1 0 National Urban Rural National Urban Rural Absolute Poverty Severity Index Extreme Poverty Severity Index 2006 2007 2008 2009 Source: Staff calculations based on KIHS 2006-09. Changes in Inequality, Consumption, and Income 61. Inequality, which is measured by the Gini coefficient of per capita consumption, stagnated over 2008 and 2009 but was declining in the earlier period. That decline was driven mainly by a reduction in rural inequalities; the urban Gini coefficient showed a slower rate of reduction, especially in recent years. The story is quite consistent with overall trends in poverty rates. What is different is that urban inequality is consistently higher than rural inequality. In general, the slow fall in inequality levels indicates that economic growth was reaching all population groups and that the gap between the poor and the nonpoor is closing, albeit at a slow rate. 38 Figure 42: Gini coefficient of Consumption per Capita, by Urban and Rural, 2006–09 35.0 31.3 31.2 30.0 29.2 30.0 28.2 25.3 25.9 25.9 25.5 25.8 24.7 24.1 25.0 20.0 percent 15.0 10.0 5.0 0.0 2006 2007 2008 2009 National Urban Rural Source: Staff calculations based on KIHS 2006–09. 62. Although inequality was falling at the national level, the oblasts showed erratic changes in Gini coefficient. Thus, Osh, Jalal-Abad, and Issyk-Kul oblasts witnessed an increase in inequality levels from 2007 to 2009 of 3.5, 0.9, and 1.1 percentage points, respectively. The largest decrease in Gini coefficient was registered in Talas oblast. In a way, this points to the differences in the income sources of the population in the respective oblasts. So, more equitable economic growth of 2007–09 was felt in Talas and Naryn, whereas in Osh growth increased the gap between the poor and the nonpoor. Table 13: Gini coefficient (Per Capita Consumption) by Oblast, 2007–09 Gin i coefficient by Per Capita Consumption 2007 2008 2009 Change 2007–09 Bishkek 24.9 24.8 24.7 -0.2 Issyk-Ku l 24.1 28.4 25.2 1.1 Jalal-Abad 18.8 22 19.7 0.9 Naryn 26.6 24.8 24.4 -2.2 Batken 24.1 20.9 23.4 -0.7 Osh 21.6 23.1 25.1 3.5 Talas 29.5 22.2 23.4 -6.1 Chui 25.8 25.7 25.6 -0.2 Source: Staff calculations based on KIHS 2006–09. 39 63. Looking at the inequality from the perspective of changes in mean consumption among different quintiles shows that over 2007–09, all groups witnessed an increase in per capita consumption as a proportion of poverty line. However, the more affluent groups, like quintiles 3, 4, and 5, have seen the largest increase in their per capita consumption. Table 14: Mean Consumption as Proportion of Poverty Line, by Quintiles, 2007–09 Quintile of Per Capita Consumption 2007 2008 2009 Change Change 2007–09 2008–09 1 0.57 0.67 0.72 0.15 0.05 2 0.76 0.97 0.98 0.22 0.01 3 0.95 1.22 1.23 0.28 0.01 4 1.3 1.56 1.59 0.29 0.03 5 2.21 2.45 2.59 0.38 0.14 Source: Staff calculations based on KIHS 2007–09. 64. By definition, the reduction in poverty was due to growth in per capita consumption, which accelerated in the precrisis years of 2007–08 and slowed down sharplybut remained positive in 2008– 09. In terms of per capita consumption, the crisis year hit the more affluent quintiles 3, 4, and 5 hardest. Although the poor (quintiles 1 and 2) also witnessed declines in the growth rate of per capita consumption, the reduction was less severe than for the more wealthy. This corresponds to the earlier observations that absolute poverty stagnated in the post-crisis years but extreme poverty continued to decline. Another interesting observation is that volatility in the growth rates of consumption expenditure is much higher for the middle quintiles, whereas the extremes (quintiles 1 and 5) had lower magnitudes of variability over the years. This might point to the differential in consumption- smoothing capabilities of the consumption groups. Figure 43: Annual Per Capita Consumption Growth, by Quintiles, in Real Terms 50 44 45 43 40 35 33 growth rate in percent 35 31 30 24 25 20 20 20 18 16 14 14 15 12 11 10 10 8 7 7 5 0 All 1 2 3 4 5 Quintiles of per capita consumption 2006-07 2007-08 2008-09 Source: Staff calculations based on KIHS 2006–09. 40 65. The dynamics of consumption per capita were driven by changes in income, and in this regard it is informative to see how the income structure changed over the observed period. The shares of wages6 and support from friends and relatives (including remittances) in income fell, as figure 44 shows, whereas the share of pensions increased notably. This mirrors the overall macro trends: As international crises led to recession in neighboring countries, migrant workers sent fewer transfers home. As domestic production was slowing, wages declined, but the government responded with fiscal measures that increased pensions and the budget deficit. Figure 44: Income Structure of Households, 2007–09 100% 4 2 3 8 7 6 90% 8 10 11 80% 70% 60% 50% 40% 78 83 80 30% 20% 10% 0% 2007 2008 2009 Salaries Pensions Private transfers Other Source: Staff calculations based on KIHS 2007–09. 66. Overall it appears that the macro and micro indicators of welfare and economic development move closely. The declining poverty trend observed before 2009 stagnated as the growth of economy and income slowed in the crisis years of 2008–09. Private consumption fell but the government increase public expenditures, which increased the budget deficit and public wages and pensions. Together with low inflation, this helped the extreme poor and absolute poverty did not increase. 6 The basis for wages is wage inco me received by indiv idual fro m public and private entities. 41 Figure 45: Dynamics of GDP, Poverty, Remittances, and Consumption Growth Rates 70 GDP real, 60 growth, % 50 40 Absolute 30 Poverty, % percent 20 10 Worker 0 remitances, growth , % -10 2006 2007 2008 2009 -20 Total consumption, -30 growth, % -40 Source: NSC data and staff estimates. 67. To conclude on poverty dynamics, the results of the growth and redistribution decomposition of poverty rates between 2008 and 2009 are briefly presented. As table 15 shows, the slight increase in poverty rates in 2008–09 is in large extent due to the distribution component rather than the impact of growth in consumption. Table 15: Growth and Redistribution Decomposition of Poverty Changes Change in Incidence of Poverty 2008 2009 Actual Growth Redistribution Interaction change Total 31.71 31.75 0.04 -5.93 8.54 -2.57 Urban 22.64 22.04 -0.59 -5.29 4.24 0.45 Rural 36.80 37.06 0.26 -6.27 10.76 -4.22 Source: Staff calculat ions based on KIHS 2008–09. 68. To see how the redistribution affects poverty and the speed at which consumption expenditure changed for different consumption centiles, we present the results of growth-incidence curve estimation between 2008 and 2009. This shows that up to the 30th centile—that is, the very poorest— the growth of consumption expenditure was above the mean growth rate, which is in accord with the observed reduction in extreme poverty. It is also evident that the mean growth of the poor and middle groups (up to the 80th centile) is lower than the mean and growth in consumption expenditures of more affluent groups. So the end effect was that economic growth in 2008–09 mostly affected the extreme poor and bypassed the middle centiles and the poor, whose expenditures rose at much slower rates compared with the mean overall. 42 Figure 46: Growth Incidence Curve, 2008–09 41 Total (years 2008 and 2009) Growth-incidence 95% confidence bounds Growth in mean Mean growth rate 31 21 11 1 -9 1 10 20 30 40 50 60 70 80 90 100 Expenditure percentiles Source: Staff calculations based on KIHS 2008–09. 43 ANNEX 1: To What Extent Does the Existing Safety Net Protect the Poor? The World Bank conducted a study in 2010 using data from KIHS 2008 to assess the effectiveness of the Kyrgyz social safety net in protecting the poor. In general terms, the country’s social system has two pillars: Contributory (for example, old-age pensions) and Non-Contributory (for example, benefits to low-income families with children). The social benefit system has been evolving gradually over the years and, at the moment, is in the midst of an active reform process aimed at increasing target efficiency and optimizing effects for the most needy. Kyrgyz Social Benefit System Contributory (Social Insurance) Non-contributory (Social -Manifestation of event/risk Assistance) -Formal employment record and -Independent of former contribution history contribution history -Old age -Categorical state benefits -Disability -Monthly social benefits -Unemployment -Unified Monthly benefits to poor -Loss of breadwinner with children The study found that the distributional incidence of social transfers varies significantly by program (table 16). Scholarships, utility and housing subsidies, and informal transfers are regressive, with richer households benefiting proportionally more than poorer households. The data show that 34 percent of utility and housing subsidies are allocated to the richest 20 percent of households, while the poorest 40 percent receive only 24 percent of the total benefit value. In terms of targeting accuracy, the UMB and MSB manage to transfer the majority of the funds to the poorest households. More than half of total transfers are accrued by the poorest quintile. Compared with social benefits in other countries in the region, the UMB performs at an average level in terms of targeting accuracy, similarly to social assistance programs in the former Yugoslav Republic of Macedonia and Poland. Table 16: Distribution of Social Protection Benefits and Private Transfers Across Groups, 2008 percent Type of Benefit Quintile I Quintile II Quintile III Quintile IV Quintile V Total Any social transfer 26.9 21.9 14.8 17.3 19.1 100 Pensions 25.7 22 14.7 17.8 19.8 100 Scholarships 16.7 21.1 8.5 16 37.8 100 Monthly social benefit 52.6 5 17.1 19.7 5.7 100 44 Unified monthly benefit 51.9 22.7 19.7 4 1.7 100 Other social insurance benefits 39 24.9 7.8 8.2 20.2 100 Utility and housing subsidies 13.2 11.1 22.5 19.6 33.6 100 Money from relat ives 7.2 14 13.4 19.3 46.1 100 Total consumption 9.8 14 17.6 22.7 35.9 100 Source: Staff calculat ion based on KIHS 2008. Note: Quintiles are based on annual per capita consumption before transfers, assuming a marginal propensity of 25 percent. The ratio of social transfers to the average consumption in each quintile shows the relative importance of transfers or the adequacy benefits. The UMB and MSB, both targeted to poor and vulnerable households, are progressive in relative terms, with benefits representing a higher share of total household consumption for those in the lower quintiles. Still, the magnitude of the transfers remains low because of their low coverage and inadequate benefit values. Table 17: Benefit Adequacy: Share of Benefits in Total Household Consumption, 2008 Quintile I Quintile II Quintile III Quintile VI Quintile V Total social transfer 18.2 15.2 14.9 15.1 14.6 Pensions 21.1 17.9 17.5 16.7 15.5 Monthly social benefit 8.0 2.0 2.9 2.8 2.5 Unified monthly benefit 6.7 2.6 2.6 1.1 0.6 Other social insurance benefits 0.2 0.1 0.0 0.0 0.0 Utility and housing subsidies 0.9 0.8 0.9 1.2 1.1 Money from relat ives 13.8 12.4 13.2 16.7 21.2 Source: Staff calculat ion based on KIHS 2008. Note: Quintiles are based on annual per capita consumption after transfers. In terms of the poverty-reducing impact, the study found that in the absence of any social transfer, extreme poverty rates would be considerably higher. The extreme poverty rate is reduced by four percentage points when accounting for all social transfers. Social transfers reduce the extreme poverty gap from 2.3 percent to 1.2 percent. The extreme poverty gap is almost halved. Pensions are the most important transfers in terms of poverty reduction, although poverty reduction is not their main objective. They are primarily meant to redistribute the income of the life cycle. However, pensions provide significantly higher transfers than other social transfers, because they are often related to previous income. Pensions are the largest social transfer program in terms of allocated resources (government budget and social fund). Informal transfers are the second most important transfer in terms of poverty reduction. They reduce the extreme poverty rate by two percentage points. Noncontributory benefits are far less effective in reducing extreme poverty. Their impact on the poverty incidence and gap are limited. Nonetheless, the UMB remains the most effective 45 targeted transfer. Its impact on poverty reduction could be further increased by increasing coverage and benefit levels. Figure 47: Extreme Poverty Reduction, 2008 a: Incidence b: Gap Any social Any social transfer transfer 12.0 2.5 10.0 Money 2.0 from 8.0 Pensions Money from 1.5 Pensions relatives relatives 6.0 1.0 4.0 0.5 2.0 0.0 0.0 Utility and Monthly Utility and Monthly housing Social housing Social subsidies Benefit subsidies Benefit Other social Unified Other social Unified insurance Monthly Monthly benefits Benefit insurance Benefit benefits Before transfer After transfer Before transfer After transfer Source: Calcu lation based on KIHS 2008. Source: F. Gassmann. 2011. “To What Extent Does the Existing Safety Net Protect the Poor,� Poverty and Social Impact Analysis series, World Bank. 46