65950 National Statistical Office World Bank MAIN REPORT OF "HOUSEHOLD INCOME AND EXPENDITURE SURVEY/LIVING STANDARDS MEASUREMENT SURVEY", 2002-2003 Ulaanbaatar 2004 ii Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. This report is also available in Mongolian. The opinions expressed here are only those of the authors and do not necessarily reflect those of the institutions involved. For comments, please contact the National Statistical Office at: Government Building III Baga Toiruu 44, Sukhbaatar District, Ulaanbaatar, Mongolia E-mail: nso@magicnet.mn Fax: 976-1-324518 Published by the National Statistical Office Ulaanbaatar, Mongolia, 2004 Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. iii TABLE OF CONTENTS Table of contents iii List of tables v List of figures viii Foreword ix Aknowledgments xi List of abbreviations xii Executive summary 1 Introduction 4 1. Macroeconomic performance and poverty trends 5 1.1. Economic background 6 1.2. Poverty trends 7 1.3. Inequality 8 2. Welfare profile 11 2.1. Consumption patterns 12 2.2. Poverty measures 14 2.3. Sensitivity to the level of the poverty line 14 2.4. Geography 16 2.5. The seasonality of poverty 19 2.6. Household composition 20 2.7. Characteristics of the household head 22 Age and gender 22 Education 23 Employment 25 Migrant status 26 2.8. Assets 27 Livestock 28 Land 30 Financial assets 30 2.9. Housing 31 Dwelling 31 Infrastructure services 31 3. Social sectors, labor market and safety nets 37 3.1. Education 38 Adult educational attainment 38 Public spending 40 iv Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. Net and gross enrollment rates 41 Participation rates 43 Profile of current students 43 School expenditures 45 3.2. Health 47 Morbidity and treatment 47 Spending 48 Knowledge about STD 50 Reproductive health 51 3.3. Labor market 52 Labor force participation 52 Employment 54 Unemployment 55 3.4. Safety nets 56 Extent and importance of transfers 57 Incidence of the transfers received by the household 58 Poverty and transfers received by the household 58 Retirement pensions 59 Poverty and the level of transfers 60 References 61 A. Appendix A: Sample design and data quality 63 A.1. An overview of the HIES-LSMS 64 A.2. The sample design 65 A.3. Data quality 65 B. Appendix B: The construction of the welfare indicator 67 B.1. The choice of the welfare indicator 68 B.2. The construction of the consumption measure 68 Food component 69 Non-food component 69 Durable goods 71 Housing 71 Energy 71 B.3. Price adjustment 73 B.4. Household composition adjustment 74 B.5. The poverty line 75 Food component 76 Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. v Non-food component 76 B.6. Poverty measures 77 C. Appendix C: Sensitivity of poverty estimates to crucial hypotheses 81 C.1. Alternative hypotheses of equivalence scale and economies of size 82 C.2. The inclusion of rent and heating expenses in the consumption aggregate 84 D. Appendix D: Additional statistical tables 89 E. Appendix E: Standard errors and confidence intervals of poverty estimations 119 LIST OF TABLES Table 1.1: National and urban/rural poverty estimates, 2002 8 Table 1.2: Inequality measures 9 Table 2.1: Per capita monthly consumption by main categories 13 Table 2.2: National poverty rates 14 Table 2.3: Poverty and scaling of the poverty line 16 Table 2.4: Poverty and geography 17 Table 2.5: Poverty and analytical domains 17 Table 2.6: The seasonality of poverty 19 Table 2.7: Poverty and household size 20 Table 2.8: Poverty and age of the household head 22 Table 2.9: Poverty and highest level of education completed by the household head 24 Table 2.10: Poverty and labor force participation of the household head 24 Table 2.11: Poverty and sector of occupation of the household head 25 Table 2.12: Poverty and migratory status of the household head 26 Table 2.13: Livestock holdings 28 Table 2.14: Poverty and livestock holdings 29 Table 2.15: Poverty and land access 30 Table 2.16: Poverty and savings 31 Table 2.17: Poverty and type of dwelling 32 Table 2.18: Poverty and infrastructure services 33 Table 2.19: Access to infrastructure services by urban-rural divide 35 Table 3.1: Highest educational attainment of adult population 38 Table 3.2: Highest education level of adult population by poverty and urban-rural divide 39 Table 3.3: Highest education level of adult population by poverty and gender 40 Table 3.4: Net and gross enrollment rates 41 Table 3.5: Enrollment rates by poverty and urban-rural divide 42 Table 3.6: Enrollment rates by poverty and gender 42 vi Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. Table 3.7: Characteristics of current students 44 Table 3.8: One-way distance to school facilities 45 Table 3.9: Spending per pupil in public primary and secondary 46 Table 3.10: Population reporting health complaints 48 Table 3.11: Per capita monthly health spending (Tugrug) 49 Table 3.12: Knowledge about STD 50 Table 3.13: Use of contraceptive methods 51 Table 3.14: Antenatal care 51 Table 3.15: Abortions 52 Table 3.16: Labor force participation rates by poverty status 53 Table 3.17: Unemployment rates by poverty, gender and urban-rural 55 Table 3.18: Safety nets 57 Table 3.19: Poverty and transfers received by the household 59 Table 3.20: Poverty and retirement pensions 59 Table A.1: The HIES-LSMS questionnaire 64 Table A.2: Population by geographical region 66 Table B.1: Maximum monthly fuel consumption during winter 72 Table B.2: Cluster Paasche Index by quarter and analytical domain 74 Table B.3: Food bundle per person per day by main food groups 76 Table B.4: Monthly poverty lines per person 77 Table B.5: Food bundle per person per day 79 Table C.1: Headcount within different groups of households making different assumptions on the extent of economies of scale 83 Table C.2: Lower poverty estimates 86 Table C.3: Upper poverty estimates 87 Table D.1: Inequality measures 90 Table D.2: Decomposition of inequality between and within various population groups (Theil index) 90 Table D.3: Per capita daily caloric intake by main food groups 91 Table D.4: Per capita monthly consumption by poverty status and urban-rural divide 92 Table D.5: Per capita monthly consumption by poverty status and analytical domain 93 Table D.6: Per capita monthly consumption by poverty status and region 94 Table D.7: Per capita monthly consumption by decile 95 Table D.8: Share of total consumption by decile 95 Table D.9: Poverty incidence by characteristics of the household head and urban-rural divide 96 Table D.10: Poverty incidence by characteristics of the household head and analytical domain 97 Table D.11: Poverty incidence by characteristics of the household head and region 98 Table D.12: Poverty incidence by characteristics of the dwelling and urban-rural divide 99 Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. vii Table D.13: Poverty incidence by characteristics of the dwelling and analytical domain 100 Table D.14: Poverty incidence by characteristics of the dwelling and region 101 Table D.15: Characteristics of the adult population by highest level of education attained 102 Table D.16: Enrollment rates comparison, 2002 105 Table D.17: Educational level of current students 106 Table D.18: Characteristics of current students by level of education enrolled 107 Table D.19: Contraceptive methods, all women 15-49 108 Table D.20: Abortions, all women 15 to 49 109 Table D.21: Labor force participation and unemployment rates comparison 110 Table D.22: Participation rates by gender 111 Table D.23: Participation rates by poverty status 112 Table D.24: Population by labor force status 113 Table D.25: Industry, sector and occupation by urban-rural divide and gender 114 Table D.26: Industry, sector and occupation by urban-rural divide and poverty status 115 Table D.27: Unemployment rates by gender 116 Table D.28: Unemployment rates by poverty status 117 Table E.1: Poverty and urban-rural divide 120 Table E.2: Poverty and geography 121 Table E.3: Poverty and analytical domains 122 Table E.4: Poverty and seasonality 123 Table E.5: Poverty and gender of the household head 124 Table E.6: Poverty and highest education level completed by the household head 125 Table E.7: Poverty and type of dwelling 126 Table E.8: Poverty, type of dwelling and urban-rural divide 127 Table E.9: Poverty and livestock holdings 128 Table E.10: Poverty and access to improved water sources 129 Table E.11: Poverty and access to improved sanitation facilities 130 Table E.12: Poverty and access to electricity 131 Table E.13: Poverty and joint access to improved water sources, sanitation facilities and electricity 132 viii Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. LIST OF FIGURES Figure 1.1: Livestock population in Mongolia, 1993-2002 6 Figure 1.2: GDP by sectors, 1998-2002 7 Figure 1.3: Poverty headcount backward projections, 1998-2001 8 Figure 1.4: Lorenz curves for urban and rural areas, 2002/03 HIES/LSMS 10 Figure 1.5: Consumption shares by population quintiles 10 Figure 2.1: Cumulative distribution of per capita consumption 15 Figure 2.2: Density function of per capita consumption 15 Figure 2.3: First order dominance results: Cumulative distribution of per capita consumption 18 Figure 2.4: Poverty and dependency ratio 21 Figure 2.5: Poverty, age and gender of the household head 23 Figure 2.6: Poverty and size of herd 29 Figure 2.7: Access to infrastructure services in urban and rural areas 33 Figure 2.8: Access to infrastructure services by poverty status 34 Figure 3.1: Public spending in primary, secondary and university 40 Figure 3.2: Participation rates 43 Figure 3.3: Spending per pupil in public primary and secondary 45 Figure 3.4: Morbidity rates and probability of seeking treatment 47 Figure 3.5: Labor force participation rates 53 Figure 3.6: Sector of employment by urban-rural divide and gender 54 Figure 3.7: Occupation of the working population by poverty and urban-rural divide 55 Figure 3.8: Characteristics of the unemployed 56 Figure 3.9: Public and private incidence of transfers received by households 58 Figure 3.10: Poverty and net transfers received by the household 60 Figure A.1: Population by age group (Census and HIES-LSMS) 65 Figure A.2: Sex ratio by age group (Census and HIES-LSMS) 66 Figure C.1: Headcount within different groups of households making different assumptions on the extent of economies of scale 83 Figure C.2: Headcount within different groups of households making different assumptions on the extent of economies of scale 84 Figure C.3: Cumulative distribution functions of urban and rural areas (excluding rents and heating costs) 85 Figure C.4: Cumulative distribution functions by region (excluding rent and heating costs) 85 Figure D.1: Public spending in lower and upper secondary 103 Figure D.2: Public spending in primary schools by urban-rural divide 103 Figure D.3: Public spending in secondary schools by urban-rural divide 104 Figure D.4: Public spending in universities by urban-rural divide 104 Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. ix FOREWORD Since the onset of the transition to a market economy of Mongolia our country the need to study changes in people's living standards in relation to household members' demographic situation, their education, health, employment and household engagement in private enterprises has become extremely important. With that pur- pose and with the support of the World Bank and the United Nations Development Programme, the National Statistical Office of Mongolia conducted the Household Income and Expenditure Survey with Living Standards Measurement Survey-like features between 2002 and 2003. Prior to this survey, the first Living Standards Measurement Survey was carried out in 1995 with technical and financial support from the World Bank and the second Living Standards Measurement Survey followed in 1998 with the support from United Nations Development Programme. The integrated Household Income and Expenditure Survey with Living Standards Measurement Survey used new sample design and methodology in accordance with international methodologies, and it combined two dif- ferent types of surveys, namely, the Household Income and Expenditure Survey and the Living Standards Measurement Survey. While doing the survey, we used the principle of using a combination of data. For example, the Household Income and Expenditure Survey collected data based on monthly questionnaires on housing serv- ices, housing, electricity, fuel and similar costs, as well as daily food purchase lists. The Living Standards Measurement Survey collected data on other non-food expenditures through quarterly questionnaires. A total of 11,232 households were surveyed under the Household Income and Expenditure Survey, and a sub-sample of 3,308 was surveyed under the Living Standards Measurement Survey. The integrated processing of data from two different surveys collected at various times at the same survey units provided an opportunity to ensure better link- age between income and expenditures. Moreover, through this experience we have made a contribution to the international practice on these two surveys. The new sample design of the survey was made in such a way as to have national average, by 4 main settlements such as the capital city, aimag centers, soum centers, as well as by urban and rural areas. This enabled to report and analyse the information in accordance with the regions deter- mined by the Government of Mongolia. This survey report has main results on key poverty indicators, used internationally, as they relate to various social sectors. Its annexes contain information regarding the consumption structure, poverty lines along with the methodology used, as well as some statistical indicators. The results of this survey provide the picture of the current situation of poverty in Mongolia in relation to social and economic indicators and will contribute toward implementation and progress on National Millennium Development Goals articulated in the National Millennium Development Report and monitoring of the Economic Growth Support and Poverty Reduction Strategy, as well as toward developing and designing future policies and actions. We are also pleased to note that the survey enriched the national database on poverty and contributed in improving the professional capacity of experts and professionals of the National Statistical Office of Mongolia. We hope that the results of the survey will provide policy makers and decision makers with realistic informa- tion about poverty and will become a resource for experts and researchers who are interested in studying pover- ty as well as social and economic issues of Mongolia. P. BYAMBATSEREN PRATIBHA MEHTA SAHA MEYANATHAN THE CHAIRMAN, RESIDENT REPRESENTATIVE RESIDENT REPRESENTATIVE NATIONAL STATISTICAL UNDP, MONGOLIA WORLD BANK OFFICE OFFICE OF MONGOLIA MONGOLIA x Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. xi AKNOWLEDGMENTS The integrated Household Income and Expenditure Survey and Living Standards Measurement Survey is one of the biggest national surveys carried out in accordance with an international methodology. It is the result of the 3 years cooperation of the staff at all level of World Bank and United Nations Development Programme, the two organizations that gave technical and financial support in undertaking this survey. The staff and experts of National Statistical Office and its local offices participated in conducting the survey. Also, I am pleased to acknowledge the contribution of citizens from more than 11 thousand households of our country who participated in the survey. I would like to express my gratitude and special thanks to Ms. B.Tserenkhand, Director of the Department of Population and Social Statistics of the National Statistical Office, Ms.D.Oyunchimeg, the Deputy Director of the Population and Social Statistics Department, Ms.Yu. Tuul, the Senior Statistician of the Population and Social Statistics Department, Ms.Ts. Amartuvshin, Ms.L. Ganzaya and Ms. B.Enerelt, Statisticians of the Population and Social Statistics Department for the successful organization and conduct of the survey, and Mr. J. Munoz and Ms. V. Evans the World Bank Experts for their cooperation in developing the survey sample design, information pro- cessing program and questionnaire. Also, my deep acknowledgement goes to Mr.L.Carroro and Mr.M.Cumpa, World Bank Experts for their cooperation with the members of the working group in conducting the survey in accordance with an international methodology and technology in writing this report. Finally, I would like to thank all Members of Management Board of the survey and Members of Methodology Working Group and Chairman's Board of NSO for their advice and comments in survey questionnaire and their comments on draft report. I would also like to thank the Aimag, Soum and Bag authorities, and officers of Ulaanbaatar and local offices of National Statictical Office of Mongolia and all the other individuals for conduct- ing the survey and then support all through the process. P. BYAMBATSEREN THE CHAIRMAN, NATIONAL STATISTICAL OFFICE OF MONGOLIA xii 12 Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. LIST OF ABBREVIATIONS AIDS Acquired immunodeficiency syndrome Conf.Interval Confidence interval DPSDD Data Processing and Software Development Department GDP Gross Domestic Product HH Household Hhsize Household size HIES Household Income and Expenditure Survey HIES-LSMS Household Income and Expenditure Survey with Living Standards Measurement Survey IMF International Monetary Fund IUD Intrauterine (contraceptive) device LSMS Living Standards Measurement Survey MEBSD Macroeconomic and Business Statistics Department MECS Ministry of Education, Culture and Science MF Ministry of Finance MH Ministry of Health MSWL Ministry of Social Welfare and Labour NGO Non-government organization NSO National Statistical Office Obs Observation PHC Population and Housing Census PL Poverty line PSSD Population and Social Statistics Department PSU Primary sampling unit Q Quintile STD Sexually transmitted disease Std.Err Standard error UN United Nations UNDP United Nations Development Programme EXECUTIVE SUMMARY 2 Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. This report presents the poverty analysis conduct- Ulaanbaatar displays the lowest level of poverty in the ed using the 2002-2003 HIES-LSMS. Two main objec- country. Five out of nine poor live in rural regions, and tives of this analysis are: 1) the calculation of new the countryside comprises a third of the poor. Poverty poverty estimates for Mongolia, disaggregated at the decreases as one moves eastward, for instance in the regional level (urban/rural areas and geographical West half of its residents are poor, whereas in the East zones); 2) the production of a poverty profile that this figure stands at around one third. describes the main characteristics of the poor in con- Mongolia presents clear seasonality patterns along trast with the non-poor. the year. The incidence of poverty in the second and fourth quarters is five percentage points higher than in The economic background the rest of the year. This seems to be associated main- ly with seasonal livestock activities and weather condi- In the years preceding to the HIES-LSMS survey, tions. economic growth was very modest, a mere 2% in terms of GDP per capita at constant prices between Some characteristics of the household head are 1999 and 2002. However, the overall growth hides a correlated with the level of poverty of the household. very diverse sectoral performance. Agriculture experi- The higher the level of education of the household enced a negative growth as a consequence of extraor- head, the lower the poverty experienced: barely less dinary adverse weather conditions that were responsi- than half of the population living with a head with less ble for a dramatic loss of livestock. On the other hand, than complete secondary is poor, compared to one both industry and services performed very well, grow- ninth if the head has at least a bachelor degree. Being ing respectively by 24 and 44% in real terms. 1999 to employed in agriculture increases the chances of being 2002 the share of agriculture to GDP almost halved poor, while these are the least if working in services. going from 36.5% to 20.1%. Such transformation in Public and state companies seem associated with bet- the GDP composition was both the result of a drastic ter living standards. Migrants show lower levels of absolute decline in agriculture and an opposite positive poverty at the national level than non-migrants, absolute increase of industry and services. although differences are smaller when looking in urban or rural areas. Poverty measures Assets allow households to hedge against eco- nomic insecurity. The main asset owned by the popu- Poverty is a widespread phenomenon in Mongolia lation in Mongolia is livestock. The livestock held by the given that, although using a lower bound poverty line, poor is on average less than half of that of the non- 36.1% of the population is found to be poor. Other poor. Households rearing livestock display lower levels poverty indicators confirm that also depth of poverty of poverty only in rural areas. But regardless of the and inequality among the poor are of substantial mag- region, the more livestock the household holds, the nitude: the poverty gap being 11.0% and the severity less poverty it experiences. The incidence of poverty of poverty 4.7%. Moreover, there is evidence suggest- among households with financial assets is significantly ing that poverty increased in the last five years, but the lower than among households without savings or advance is limited if considering the extreme losses suf- stocks. fered in the agriculture sector. Housing appears to be correlated with poverty Inequality only in urban areas, population living in apartments are the least poor, while the opposite occurs in gers. In Inequality as measured by the Gini coefficient is rural areas, dwellers in houses display a higher inci- 0.33 and there is robust evidence showing that dence of poverty than those living in gers. Access to inequality is higher in urban than in rural areas of the infrastructure services displays a similar pattern, where- country. The richest 20% of the population consumes as in urban areas having access to improved water almost 5.5 times the amount consumed by the poorest sources, improved sanitation facilities or electricity is 20% of the population. associated with less poverty, no clear trend emerges in rural areas. The non-poor and especially urban The main characteristics of the poor dwellers enjoy more access to any of these three serv- ices. Poverty in urban domains is significantly lower than in rural areas, 30% and 43% respectively. Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. 3 Poverty and the education sector ly to have abortions, but if they do, a major reason is the lack of financial means. The educational attainment of the adult popula- tion is very high. A third of the population has either Poverty and the labor market tertiary or vocational studies. The poor display lower attainments than the non-poor, more than half of the The labor force participation rate stands at 65%. poor reach only the 8th grade of secondary compared Urban areas have significantly lower participation rates to one third of the non-poor. Public spending in pri- than rural regions, less than three fifths compared to mary is progressive, largely neutral in secondary and three quarters respectively. The poor display lower regressive in tertiary education. Enrollment rates for rates of participation in the labor market than the non- the poor and non-poor are similar in primary, but in poor. The main sectors of employment are very differ- secondary the non-poor display higher rates. Among ent in urban and rural areas. Livestock activities domi- current students in public institutions, the non-poor nate in rural regions, more than seven out of ten work- spend on average sixty percent more than the poor in ers engage in them, whereas in the capital and aimag both primary and secondary. centers, services account for almost three quarters of the jobs. The likelihood of being a herder or a farmer Poverty and the health sector is higher for the poor, whereas the non-poor are more likely to be managers, professionals and technicians. Morbidity rates are very low, only 6% of the pop- Finally, unemployment is similar in urban and rural ulation reported any health complaint in the month areas but the poor have a rate of unemployment more previous to the survey. The non-poor report more than double that of the non-poor. health complaints than the poor, and the differences grow larger the older the person gets. When they have Poverty and safety nets a health problem, the non-poor are also more likely to seek treatment. Urban dwellers and the non-poor are The extent of safety networks is impressive: four more likely to visit private facilities, but both poor and out of five households either give or receive some sort non-poor have similar chances of being attended by a of transfer. Seventy percent of households are recipi- doctor. The non-poor spend more than three times as ents, while every other family is a donor. Both public much as the poor, and this pattern is even more evi- and private transfers received by the households have dent across quintiles, the richest 20% of the popula- a similar coverage but the former makes up for almost tion spend seven times the amount of the poorest three quarters of the total amount transferred. 20%. Knowledge of sexually transmitted diseases is Nationwide, similar levels of poverty are observed similar among poor and non-poor, although the latter among those living in households getting transfers and are better informed on how to protect themselves. those in households that do not get them. But the net Regarding reproductive health issues, poor women are amount received by the household does matter, the slightly more likely than non-poor women to currently higher the transfer received, the less poverty experi- use contraceptive methods, or if pregnant, to seek and enced. obtain antenatal care. Lastly, poor women are less like- 4 Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. INTRODUCTION dards and its results are deemed to be properly repre- sentative of the country situation. However, its main In July 2003 the Government of Mongolia com- results are not directly comparable with those of previ- pleted the Economic Growth and Poverty Reduction ous LSMS, namely 1995 and 1998, nonetheless the Strategy Paper in which the Government gave high pri- paper also tries to indirectly assess poverty trends in ority to the fight against poverty. As part of that com- the last five years. mitment this paper is a study that intends to monitor The first section of the paper provides information poverty and understand its main causes in order to on the Mongolian economic background, and presents provide policy-makers with useful information to the basic poverty measures that are linked to the eco- improve pro-poor policies. nomic performance to offer an indication of what hap- pened to poverty and inequality in recent years. A sec- The main contributions of this paper are: ond section goes in much more detail in generating 1) new poverty estimates based on the latest avail- and describing the poverty profile, in particular looking able household survey, the 2002-2003 HIES-LSMS; at the geographical distribution of poverty, poverty 2) the implementation of appropriate, and interna- and its correlation with household demographic char- tionally accepted, methodologies in the calculation acteristics, characteristics of the household head, of poverty and its analysis (these methodologies employment, and assets. A final section looks at pover- may constitute a reference for the analysis of ty and social sectors and investigates various aspects of future surveys); education, health and safety nets. The paper contains 3) a 'poverty profile' that describes the main charac- also a number of useful, but more technical appendix- teristics of poverty. es with information about the HIES-LSMS survey (sam- ple design and data quality) (Appendix A), on the The 2002-2003 HIES-LSMS was implemented methodology used to construct the basic welfare indi- using an improved methodology in the selection of the cator, and set the poverty line (Appendix B), some sen- sample using the information of the recent Census, sitivity analysis (Appendix C), and additional statistical instead of administrative data. The sample selection information (Appendix D and E). methodology followed recognized international stan- 1. MACROECONOMIC PERFORMANCE AND POVERTY TRENDS 6 CHAPTER 1. MACROECONOMIC PERFORMANCE AND POVERTY TRENDS 1.1. Economic background was lower than the one of ten years earlier and its composition also changed remarkably with a propor- In the last five years Mongolia's economy has tional increase of goats and decline of camels, cattle undergone very dramatic changes. From 1999 to 2002 and sheep.3 the share of agriculture to GDP almost halved going However, the reduction of the agriculture share of from 36.5% to 20.1%. Such transformation in the GDP was also due to an opposite trend in industry and GDP composition was both the result of a drastic services, which between 1999 and 2002 grew in real absolute decline in agriculture and an opposite positive terms respectively by 24% and 44%4 (Figure 1.2). absolute increase of industry and services. Therefore the collapse of agriculture was counterbal- In Mongolia agriculture consists mainly of livestock anced by the growth of industry and services, and the and only marginally of crops, and throughout the overall per capita GDP growth between 1999 and 1990s livestock population has been growing steadily 2002 was a modest 2%. reaching a peak in 1999. Since 1999 a negative These dramatic changes were accompanied by sequence of extremely cold and harsh winters, known remarkable migration flows and employment shifts as dzuds, and dry summers that lasted until 2002 between economic sectors. Movements from aimag reduced the livestock population by almost 30% (see centers to the countryside, common in the middle of Figure 1.1). Figure 1.1: Livestock population in Mongolia, 1993-2002 40 Camel Horse Cattle Sheep Goat 35 30 25 Millions of heads 20 15 10 5 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Source: Mongolian Statistical Yearbook, 2002 and IMF country report No 99/4, 1999. Animal losses of this magnitude were unprece- the 1990s, were reversed by opposite trends that saw dented, definitely the highest in the last 50 years and an increased urbanization. Such migratory movements much higher than the levels reached at the end of the seem to be well associated with economic opportuni- 1960s, when substantial losses were also recorded.1 ties, and in general with the economic performance of The scale of the disaster was probably augmented by sectors, that have clear urban/rural characteristics. In the uncontrolled growth of herds and their bad man- fact according to administrative data, the population agement,2 but the climatic shock was definitely employed in agriculture reduced both in absolute extraordinary. The overall number of livestock in 2002 terms as well as in terms of share of total employment, 1 Mongolia: Selected Issues and Statistical Appendix, 2002 IMF Country Report No. 02/253. 4 Within industry and services the sectors responsible for growth were manufacturing, 2 See Mongolia Human Development Report, 2003, pages 39-40 for more information on trade, transport and communication, and financial intermediation. And some of their the impact of negdels' dissolution. Negdels were livestock cooperatives with specific tasks growth seems to be well correlated with aid flows by sectors (see "Implementing the of disaster management (grazing land reserves, veterinary support, provision and mainte- Economic Growth Support and Poverty Reduction Strategy", Ministry of Finance and nance of animal shelters and fodder reserves). Economy, page 10). 3 The higher number of goats reflects the new opportunities offered by cashmere trade, but it can also indicate a lower value of the livestock population and its higher vulnerability. In fact, according to a traditional Mongolian way of valuing herd (the bod scale), goats are the least worth livestock, followed by sheep, cattle, horses and camels. CHAPTER 1. MACROECONOMIC PERFORMANCE AND POVERTY TRENDS 7 Figure 1.2: GDP by sectors, 1998-2002 350 Agriculture Industry Services 300 Tugrug (Thousands of millions, 1995 prices) 250 200 150 100 50 0 1998 1999 2000 2001 2002 Source: Mongolian Statistical Yearbook, 2002 and IMF country report No 99/4, 1999. going from 50% in 1998 to 45% in 2002, while in the census, while both the 1995 and 1998 LSMS did not same period employment in services increased from possess recent Census data and adopted a very differ- 34% to 41%.5 ent procedure in the selection of the sample8. According to the elaboration of Census data, net Therefore, problems of comparability cannot be recipients of migratory movements were mainly three resolved, and the welfare indicator used for poverty cities: Ulanbaatar, Erdenet (Bayan-Undur) and analysis as well as the relevant poverty line are very dif- Darkhan. In 2000 about 14% of Ulaanbaatar's popu- ferent. Nonetheless, there is a significant relative dif- lation 5 years and older moved to the capital since ference that should be noted between the current 1995.6 And even higher percentages are recorded for poverty estimates and the previous ones. While previ- Erdenet and Darkhan. It is in these centers that servic- ous surveys found that poverty was higher in urban es and industries grew sensibly. And there are good than rural areas, current findings are reversed and rural reasons to believe that these trends might have only areas are found to be poorer than urban ones. increased in the following years. This basic finding is coherently related to the eco- nomic changes described earlier. Moreover, in order to 1.2. Poverty trends understand what happened to poverty in the last five years it is possible to generate some backward projec- In this macroeconomic scenario what happened to tions based on the available information on GDP com- poverty? Table 1.1 reports poverty estimates obtained position and growth in the three sectors (agriculture, with the analysis of the 2002/03 HIES/LSMS. Estimates industry and services) as well as employment composi- show that 36% of the population is in poverty and in rural areas poverty is sensibly higher than in urban 5 Employment shares in the three sectors estimated with the sample are very similar to those areas (43% against 30%). Similarly the other two of administrative sources: 44.6% in agriculture, 10.7% in industry and 44.8% in services. poverty indexes, the poverty gap and the severity of In addition estimates from the Labour Force Survey also support the accuracy of these val- ues: 46.7% in agriculture, 11.9% in industry and 41.4% in services. poverty7, are higher in rural than urban areas. 6 See "Internal Migration and Urbanization in Mongolia: Analysis based on the 2000 Census", NSO 2003. However, it is important to note that these poverty 7 The poverty gap is an indicator of the depth of poverty, while the severity of poverty takes estimates cannot be directly compared with existing into account also the inequality among the poor, see section 2.2 for more explanations on these indicators. previous estimates, mainly for 1995 and 1998. In fact, 8 Other important differences between the 2002/03 HIES/LSMS and the previous LSMS sur- veys concern the overall sample design: field procedures, interview structure and question- the methodology used to estimate poverty is very dif- naire. Nonetheless, some analysis was undertaken to see the extent of comparability of a ferent and dependent on the dissimilar characteristics modified consumption aggregate, which contained as much as possible similar compo- nents, between the 1998 LSMS and the 2002/03 HIES/LSMS, and between the 1999 HIES of the surveys. In particular, the 2002/03 sample made and the 2002/03 HIES/LSMS. In both cases it emerged that the datasets are not compara- ble, and that the problem does not lie in the theoretical content of the consumption use of an updated sampling frame based on the latest aggregate, but on how (recall period, sampling procedures) and when (during the year) households' information about consumption expenditure was collected. 8 CHAPTER 1. MACROECONOMIC PERFORMANCE AND POVERTY TRENDS Table 1.1: National and urban/rural poverty estimates, 2002 Headcount Poverty Gap Severity National 36.1 (1.4) 11.0 (0.6) 4.7 (0.3) Urban 30.3 (1.7) 9.2 (0.7) 4.0 (0.4) Rural 43.4 (2.4) 13.2 (1.0) 5.6 (0.5) Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. Figure 1.3: Poverty headcount backward projections, 1998-2001 40 35 30 25 Headcount (%) 20 15 10 5 0 1998 1999 2000 2001 2002 Source: Estimation based on the 2002/03 HIES/LSMS and macroecono mic indicators. tion and growth in the same sectors9. Such backward Overall given the tremendous livestock losses, the projections suggest that poverty might have increased, policy of free migration12 seems to have helped reduc- but overall it was a very modest increase10 (Figure 1.3). ing the poverty increase, although especially in the However, these projections are only an indication of capital the government now faces the challenge of one possible scenario of poverty trends assuming that controlling the immigration flow and the consequent different economic growth in the three sectors is the demand of social services and utilities. It is also impor- main driver of poverty changes, while relative inequal- tant to note that aid might have played an important ities within the sectors remain constant11. The hypoth- role in mitigating the effects of the livestock losses. In esis of constant inequality within sectors is not based 9 These projections were performed using the World Bank poverty projections toolkit on any particular information and given the strong designed by Datt and Walker, available at: www.worldbank.org/poverty/psia/tools.htm, growth, especially within services, it is possible that where it is also possible to find more details on the methodology used to make the pro- jections. inequality might have increased within sectors and on 10 A similar result is obtained using the 1998 LSMS as base data and estimating poverty trends up to 2002. the whole. The effect of an increased inequality would 11 Other implicit assumption is that household consumption grew at the same level of GDP, be a higher poverty increase in the last five years. and that the employment of the household head is representative of the main source of household income. Moreover, even though the overall proportion of poor 12 Contrary to the population movement restrictions in place before 1991, which controlled movements especially to Ulaanbaatar, the new Mongolian Constitution approved in 1992 people might not have increased significantly, the geo- declares that every Mongolian citizen has the right to choose where to live in Mongolia. graphical composition of poverty is likely to have Nevertheless, there still exist some formal conditions to get permission to reside in Ulaanbaatar (see "Internal Migration and Urbanization in Mongolia: Analysis based on changed dramatically. the 2000 Census", NSO 2003). CHAPTER 1. MACROECONOMIC PERFORMANCE AND POVERTY TRENDS 9 fact, the Ministry of Foreign Affairs estimated that the er with the Gini index also another inequality measure equivalent of US$ 24 million was received in 2000-01 is reported, namely the Theil index17). From the figures alone (about 2.4% of GDP) for Dzud relief assistance reported in Table 1.2 it emerges that inequality is high- from donor countries, international organizations and er in urban than in rural areas. NGOs13. Moreover, a survey on the nutritional conse- Inequality can also be analyzed using graphical quences of the dzud found no significant differences and more intuitive tools, such as the Lorenz curves. The between dzud affected areas and unaffected areas in Lorenz curve ranks the population of a certain country, general nutrition status and prevalence of micronutri- area or region from the poorest to the richest and ent deficiencies among children and their mothers (see associates population proportions with their fraction of Nutrition Research Centre et al. (2003)). total consumption. Figure 1.4 depicts the Lorenz It is also important to mention that the LSMS cap- curves for urban and rural areas. The further away is tures only a very limited number of migrants. Migrants the Lorenz curve from the line of perfect equality, the in the LSMS are much less than what Census data sug- higher is the level of inequality. The fact that the gest. This could have been the result of an under sam- Lorenz curve for urban areas is always below the one pling of areas with concentration of recent migration14 of rural areas means that inequality is higher in urban or some inaccuracies in the collection of migration areas independently from the specific index used to data. If recent migration was indeed under-represent- measure inequality18, and it is therefore a robust result. ed, there are reasons to believe that this in turn might Finally, a different, but probably more understand- have underestimated the level of poverty. In fact, it is able way to look at inequality is provided in Figure 1.5, likely that recent migrants might be poorer than the which reports the share of national consumption rest of the population. obtained by each population quintile (the population is divided into 5 groups, each containing 20% of the 1.3. Inequality population and ranked from the poorest to the rich- est). It shows that the richest 20% of the population As mentioned earlier, it is more difficult to under- consumes almost 5.5 times more than the poorest stand how the overall level of inequality might have 20%. changed in the last five years, but it is nevertheless important to provide inequality estimates for the latest survey. In 2002-03 the estimated Gini coefficient15 for per capita consumption expenditure, after correcting for price differences, was 0.329. Common values of the index go from 0.2 to 0.5, but comparisons with previous estimates as well as international comparisons should be made with caution. Moreover, they can be very misleading when the index is computed using dif- ferent welfare indicators16. Instead, comparisons are more meaningful across population groups within the country. Table 1.2 reports inequality measures at the national level and within urban and rural areas (togeth- Table 1.2: Inequality measures 13 However, it is not possible to directly assess whether this aid was properly targeted. 14 To support this hypothesis is the fact that listing operations in some primary sampling units Gini coefficient Theil index might have only considered officially registered households (see Appendix A). 15 The Gini coefficient is a measure of inequality that goes from zero to one, where higher values are associated to higher inequality. 16 The most common problem is when inequality measures are based on income values National 0.329 0.183 rather than consumption. In fact, income based measures of inequality tend to be always higher than respective consumption based measures. 17 Also this index can take values from 0 to 1, and higher values indicate higher inequality. Urban 0.331 0.185 The advantage of this index is that, whenever inequality is computed in different popula- Rural 0.313 0.165 tion groups, it is possible to additively decompose the index in two parts: inequality between groups and inequality within groups. This is done for a number of relevant vari- ables and the results are reported in Appendix D (Table D.2). It emerges that inequality Source: 2002/03 HIES/LSMS. within population groups is always the main component, but it is interesting to see that access to infrastructure services (water access, telephone, heating facilities, toilets) are the variables that identify the biggest differences between population groups. 18 As long as the index satisfies the principle of transfers. 10 CHAPTER 1. MACROECONOMIC PERFORMANCE AND POVERTY TRENDS Figure 1.4: Lorenz curves for urban and rural areas, 2002/03 HIES/LSMS 1.0 0.8 Cumulative fraction of consumption 0.6 Line of perfect equality 0.4 Rural areas Urban areas 0.2 0 0 0.2 0.4 0.6 0.8 1.0 Cumulative fraction of population Source: 2002/03 HIES/LSMS. Figure 1.5: Consumption shares by population quintiles 45 40 35 30 Consumption shares 25 20 15 10 5 0 Poorest 2 3 4 Richest Population quintiles Source: 2002/03 HIES/LSMS. 2. WELFARE PROFILE 12 CHAPTER 2. WELFARE PROFILE A welfare profile assesses how living standards average. Among regions, the shares are most stable, vary across different subgroups of the population. This ranging from 46% in the Central region to 52% in the chapter is primarily concerned with the construction of Highland and the East. a poverty profile that will show the characteristics of Among non-food categories, clothing is the most poverty and their correlation with different features of important component and accounts for twelve percent the household and other aspects of welfare. It will sep- of total consumption, with urban and rural areas dis- arate the poor from the non-poor in order to obtain a playing similar figures. The value of housing only rep- better understanding on who the poor are, where they resents 5% of total consumption. In Ulaanbaatar this live, their levels of human capital and wealth, the qual- share rises to 11%, whereas in the rest of the country ity of their housing and the type of work they engage is no larger than 3%. The share of education is 7% in. This may provide useful information for a better and it is stable across regions, only in the countryside design of poverty alleviation efforts. it represents barely 3%. Health expenditures display a steady share across regions of around 5%. Heating 2.1. Consumption patterns consumption stands at 3% of total consumption, rural households having a half the share of their urban The first step to construct a poverty profile is to counterparts. Across regions, families in the West agree on a comparable welfare indicator for the popu- appear to devote more resources to this component of lation. For the purposes of this report, the per capita their consumption. Transportation and communication consumption of the household is used19. It is therefore represents another 5%. Utilities (i.e. electricity and important to show what consumption includes and lighting, water and telephone) account for a similar how is distributed within its components. share. The remaining ten percent of total consumption According to the household survey, the monthly is comprised by entertainment, toiletries, durable per capita consumption in Mongolia during 2002 was goods and alcohol and tobacco. Tugrug 36,750, the equivalent of about US$32 in that year. Table 2.1 displays the average consumption by main expenditure groups and across three different geographical divisions: urban/rural areas, analytical domains (associated also with the degree of urbaniza- tion) and regional areas. Urban areas display consump- tion levels one quarter higher than rural regions. Across analytical domains, the capital ranks first, fol- lowed by aimag centers and on the third place both soum centers and the countryside. Among regions, the West shows the lowest level of consumption, twenty percent lower than the national average, whereas the Central the highest20. The Highland and the East are in between with similar levels. It is worth noticing that whether by domains or by regions, consumption levels in Ulaanbaatar are substantially above the rest of the country. How is the pattern of consumption in the country? The share of food is 44% of the total expenditures, with significant differences between urban and rural areas21. It is expected that urban areas have lower food shares compared to rural ones due to the relative importance of other components of consumption. Indeed, that is the case. In the former, food accounts only for two fifths of total consumption, while in the latter for more than half of it. Across regions, the cap- ital shows a remarkably low food share of around one 19 See Appendix B for a detailed explanation on this and the estimation of the poverty line. 20 Ulaanbaatar is located within the Central region but it is considered as a separate domain third compared to almost three fifths in the country- due to its significance. 21 Unfortunately it is not possible to breakdown this consumption into purchases, home-pro- side. Aimag and soum centers are around the national duction and in-kind transactions due to the way information was collected. Table 2.1: Per capita monthly consumption by main categories (2002 Tugrug, adjusted by regional and temporal price differences) National Urban Rural Analytical domains Geographical regions Ulaanbaatar Aimag Soum Countryside West Highland Central East centers centers a/ Consumption Food 16,350 15,390 17,545 15,477 15,285 14,920 19,043 14,208 17,669 16,913 18,508 Alcohol and tobacco 1,330 1,451 1,178 1,502 1,391 1,284 1,118 1,107 1,346 1,364 1,060 Education 2,519 3,203 1,668 3,480 2,873 2,628 1,120 2,016 1,993 2,453 1,815 Health 1,919 2,204 1,564 2,152 2,266 2,111 1,252 1,599 1,549 2,422 1,642 Durable goods 1/ 410 534 257 601 454 286 240 324 345 312 324 Rent 2/ 1,950 3,083 541 4,583 1,291 542 541 569 803 1,046 789 Heating 3/ 1,199 1,645 644 1,732 1,541 792 559 1,119 791 1,058 966 Utilities 4/ 2,079 2,975 964 3,547 2,292 1,348 745 1,276 1,224 1,782 1,614 Clothing 4,573 4,841 4,239 4,299 5,488 4,310 4,199 4,144 4,839 4,931 4,799 Transportation and communication 1,891 2,236 1,463 2,768 1,599 1,463 1,464 1,427 1,327 1,990 1,146 Others 5/ 2,527 2,785 2,205 2,861 2,694 2,196 2,211 1,934 2,500 2,512 2,622 Total 36,747 40,348 32,269 43,002 37,175 31,881 32,491 29,725 34,386 36,781 35,284 Shares Food 44 38 54 36 41 47 59 48 51 46 52 Alcohol and tobacco 4 4 4 3 4 4 3 4 4 4 3 Education 7 8 5 8 8 8 3 7 6 7 5 Health 5 5 5 5 6 7 4 5 5 7 5 Durable goods 1/ 1 1 1 1 1 1 1 1 1 1 1 Rent 2/ 5 8 2 11 3 2 2 2 2 3 2 Heating 3/ 3 4 2 4 4 2 2 4 2 3 3 Utilities 4/ 6 7 3 8 6 4 2 4 4 5 5 Clothing 12 12 13 10 15 14 13 14 14 13 14 Transportation and communication 5 6 5 6 4 5 5 5 4 5 3 Others 5/ 7 7 7 7 7 7 7 7 7 7 7 Total 100 100 100 100 100 100 100 100 100 100 100 a/ Excludes Ulaanbaatar. 1/ Estimation of the monetary value of the consumption derived from the use of durable goods. 2/ Estimation of the monetary value of the consumption derived from occupying the dwelling. If the household rents its dwelling, the actual rent will be included instead of the imputed rent. 3/ Includes central and local heating, firewood, coal and dung. 4/ Includes electricity and lighting, water and telephone. 5/ Includes recreation, entertaiment, beauty and toilet articles, and household utensils. Source: 2002/03 HIES/LSMS. CHAPTER 2. WELFARE PROFILE 13 14 CHAPTER 2. WELFARE PROFILE 2.2. Poverty measures ed, the severity indicator will indeed rise. The severity measure is 4.7 percent. Unfortunately, there is no easy What are the incidence, depth and severity of or intuitive interpretation of this indicator. However, it poverty in Mongolia? The incidence of poverty in the helps to compare and rank poverty across different country is 36.1% (Table 2.2), which means that groups when similar incidences and gaps are found. Table 2.2: National poverty rates Headcount Poverty Gap Severity 36.1 11.0 4.7 (1.4) (0.6) (0.3) Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. around 900,000 individuals are considered poor22. In 2.3. Sensitivity to the level of the poverty line other words, 36 out of every 100 Mongolians do not have the necessary means to purchase the value of a A natural concern that arises at this stage is to find minimum food and non-food bundle. Although the out how sensitive the poverty measures are with poverty headcount is very easy to understand, it does respect to the level of the poverty line. Yet consider- not provide information on how close or far the poor able effort has been put in deriving a poverty line fol- are from being able to satisfy their basic needs or how lowing a fairly established methodology and trying to consumption is distributed among the poor. This could be as transparent and objective as possible, an be a serious limitation when evaluating alternative pol- unavoidable degree of arbitrariness is involved in the icy options, for example, the implementation of a par- process. Many explicit and implicit assumptions have ticular policy could improve the welfare of the poor been made along the way and not everybody may leaving unchanged the poverty incidence. In order to agree with them. Other poverty lines might be equally obtain a more complete description of the poverty sit- appealing and justified. uation, two other measures are also considered: the poverty gap and the severity of poverty. Stochastic dominance analysis allows us to find the range of poverty lines over which poverty compar- The poverty gap stands at 11% and estimates the isons are robust. It relies on graphical tools and focus- average shortfall in consumption relative to the pover- es on the entire distribution of consumption. Figure ty line. This implies that, on average, the consumption 2.1 shows the cumulative distribution function of per of each person in the country is 11 percent below the capita consumption in Mongolia and provides an poverty line. The indicator has a more practical, example of this sort of techniques26. For a given con- although quite hypothetical, interpretation. If all pover- sumption level on the horizontal axis, the curve indi- ty gaps are added, that amount will be the minimum transfer of income necessary to bring all poor popula- tion out of poverty23. Hence a total annual transfer of 22 Total population for 2002 was 2,475,400 individuals according to the 2000 Census pro- jections. Tugrug 80,848 millions, or US$ 70 millions, would be 23 This estimation assumes both perfect targeting and full consumption of the transfer. Perfect targeting implies that every poor will receive a transfer equal to the difference required to eliminate poverty24. between her consumption and the poverty line, and that no person above the poverty line will receive anything. If the recipient of the transfer also fully consumes it, her consump- The third poverty indicator is the severity of pover- tion will be equal to the poverty line and that person will no longer be considered as poor. Lastly, it will also require no transaction costs. ty. In contrast to the headcount or to the poverty gap, 24 This amount is equivalent to 7% of the 2002 GDP, and was calculated as follows = 0.11 x national poverty line of Tugrug 24,743 x 12 months x 2,475,400 persons. A few caveats this measure is sensitive to the distribution of con- regarding these rather speculative numbers are worth mentioning though. The first is that sumption among the poor25. For instance, if a transfer in practice perfect targeting is impossible, transaction costs will be too high. The second is that even if it were possible, there would be no guarantee that the transfer will be fully occurs from one poor household to a richer household, consumed by the recipients. Finally, it would make little sense to transfer that amount to the poor because strong disincentive effects are likely to appear. the level of poverty should increase. Even though the 25 It weights the shortfall in consumption relative to the poverty line more heavily the poor- poverty incidence and the poverty gap will be unaffect- er the person is. 26 Figures shown cover up to Tugrug 125,000 per person per month, which is a value close to the 99th percentile of the total distribution of per capita consumption. CHAPTER 2. WELFARE PROFILE 15 cates the percent of the population with an equal or dence curve�. It is simple then to assess how much the lesser level of consumption on the vertical axis. If one headcount will change when the poverty line is shifted thinks of the chosen consumption level as the poverty upward or downwards. At a poverty line of Tugrug line, the curve will show the associated poverty head- 24,743 per person per month, around 36% of the count, and hence it can be seen as a “poverty inci- population are poor. Nonetheless, given that the slope Figure 2.1: Cumulative distribution of per capita consumption 1.00 Cumulative fraction of population 0.75 0.50 0.25 Poverty line 0 25 50 75 100 Per capita real consumption (Thousands of Tugrug per month) Source: 2002/03 HIES/LSMS. Figure 2.2: Density function of per capita consumption Estimate of density Poverty line 25 50 75 100 Per capita real consumption (Thousands of Tugrug per month) Source: 2002/03 HIES/LSMS. 16 CHAPTER 2. WELFARE PROFILE of the distribution is relatively steep around that level, and pose particular challenges for economic develop- it is likely that small changes in the poverty line will ment. What is then the link between poverty and have a larger impact on the poverty incidence. geography? Table 2.3: Poverty and scaling of the poverty line Scaling of Headcount Poverty Line National Urban Rural 200 78.3 73.2 84.7 150 62.9 56.0 71.5 125 50.6 44.0 58.8 110 41.5 35.5 49.0 100 36.1 30.3 43.4 90 29.9 24.7 36.4 75 20.2 16.9 24.2 50 6.5 5.7 7.4 Source: 2002/03 HIES/LSMS. The concentration of households around the There are substantial disparities in poverty across poverty line can be illustrated with a related concept, regions. Table 2.4 displays poverty measures consider- the density function27. Figure 2.2 depicts the kernel ing a division of the country based on geographical density estimate of the per capita consumption. It areas. Mongolia can be divided in 4 main regions: shows two important characteristics of the distribution West, Highland, Central and East. Ulaanbaatar is locat- around the poverty line. First, a significant clustering ed within the Central region but is considered as a sep- occurs close to that point. Second, there is more prob- arate one due to its significance. Poverty decreases as ability mass below the poverty line than above it. The one moves eastward. The poverty incidence in the implication of both features is that the poverty meas- West reaches more than half of its population, almost ures are less sensitive to scaling up the poverty line two fifths in the Highland and around one third in both than to scaling it down. Table 2.3 confirms this by esti- Central and East. Ulaanbaatar has the lowest incidence mating the headcount when the poverty line is scaled of poverty, slightly more than one quarter of the capi- up and down. On the one hand, it reveals that 12 per- tal residents is poor. The West comprises one sixth of cent of the population lies within plus or minus 10 per- the population but one quarter of the poor. By con- cent of the poverty line and almost one third within trast, the capital accounts for one third of the popula- plus or minus 25 percent. On the other hand, when tion and one fifth of the poor. Another quarter of the the poverty line is doubled, the incidence of poverty poor live in the Highland, a fifth in the Central area increases in less (from 36 to 78%), but when the and the remaining tenth in the East. poverty line is halved, the headcount decreases much Urbanization is another factor to take into more (from 36 to 7%). account. For instance, the West and the Highland, the two poorest regions, are the less urbanized ones. 2.4. Geography Generally rural areas are less developed than urban ones and hence show lower levels of living standards. Mongolia presents a very diverse geography. It is Table 2.5 shows a division of the country based on not only a landlocked country but also displays a high altitude level. Its territory encompasses deserts, 27 The notion of the density function is very similar to that of histograms. Traditional his- steppes, forests, lakes and high mountains, each one tograms divide a range of the variable of interest into certain number of intervals of equal width and draw a vertical bar for each interval with height proportional to the relative fre- with its own particular features in terms of climate, quency of observations within each interval. A kernel density function can be thought of as a "smoothed" histogram. It estimates the density, or relative frequency, at every point soil, flora and fauna. These characteristics are impor- rather than at every interval. Hence, say in the case of consumption, the area between two consumption levels is the proportion of the population with consumption within that tant to determine living standards across the country range (it follows that the total area under the curve is 1 or 100 percent of the population). CHAPTER 2. WELFARE PROFILE 17 Table 2.4: Poverty and geography National West Highland Central East Ulaanbaatar Headcount 36.1 51.1 38.7 34.4 34.5 27.3 (1.4) (3.5) (2.9) (3.0) (4.4) (2.6) Poverty Gap 11.0 14.6 12.3 10.1 12.4 8.1 (0.6) (1.3) (1.3) (1.4) (2.3) (1.0) Severity 4.7 5.7 5.2 4.3 6.6 3.3 (0.3) (0.7) (0.7) (0.8) (1.6) (0.5) Memorandum items: Share below PL (%) 100.0 24.0 25.8 18.6 8.9 22.8 Number below PL ('000) 894.0 214.4 230.5 166.3 79.1 203.8 Population share (%) 100.0 17.0 24.1 19.5 9.3 30.2 Population ('000) 2,475.4 419.8 596.1 483.4 229.0 747.3 Household size 4.3 4.6 4.1 4.1 4.4 4.4 Dependency ratio (%) 43.3 46.7 43.0 44.3 45.2 40.6 Children (% household size) 31.2 37.1 31.2 29.7 36.0 27.5 Age of household head 44.5 42.5 43.1 44.7 41.9 47.3 Male household head (%) 82.5 91.6 85.2 79.4 85.7 76.6 Urbanization (%) 55.4 34.8 31.3 40.6 42.0 100.0 Note: Total population for 2002 is based on the projections from the 2000 Census. Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. Table 2.5: Poverty and analytical domains National Urban Rural Total Ulaanbaatar Aimag Total Soum Country centers centers side Headcount 36.1 30.3 27.3 33.9 43.4 44.5 42.7 (1.4) (1.7) (2.6) (2.2) (2.4) (3.0) (3.3) Poverty Gap 11.0 9.2 8.1 10.5 13.2 14.4 12.6 (0.6) (0.7) (1.0) (1.0) (1.0) (1.5) (1.3) Severity 4.7 4.0 3.3 4.7 5.6 6.4 5.1 (0.3) (0.4) (0.5) (0.7) (0.5) (0.9) (0.7) Memorandum items: Share below PL (%) 100.0 46.5 22.8 23.7 53.5 20.0 33.6 Number below PL ('000) 894.0 415.3 203.8 211.5 478.7 178.5 300.2 Population share (%) 100.0 55.4 30.2 25.2 44.6 16.2 28.4 Population ('000) 2,475.4 1,372.1 747.3 624.8 1,103.3 400.8 702.5 Household size 4.3 4.4 4.4 4.4 4.2 4.4 4.1 Dependency ratio (%) 43.3 41.8 40.6 43.3 45.2 42.4 46.6 Children (% household size) 31.2 29.7 27.5 32.2 33.0 33.2 32.9 Age of household head 44.5 46.2 47.3 45.0 42.4 43.7 41.7 Male household head (%) 82.5 79.6 76.6 83.1 86.1 85.3 86.5 Note: Total population for 2002 is based on the projections from the 2000 Census. Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. 18 CHAPTER 2. WELFARE PROFILE Figure 2.3: First order dominance results: Cumulative distribution of per capita consumption 1.00 1.00 Aimag Rural centers 0.75 0.75 Urban Ulaanbaatar 0.50 0.50 0.25 0.25 Poverty line Poverty line 0 0 25 50 75 100 25 50 75 100 Cumulative fraction of population 1.00 1.00 Soum Countryside centers Countryside 0.75 0.75 Ulaanbaatar Soum 0.50 0.50 centers Aimag centers 0.25 0.25 Poverty line Poverty line 0 0 25 50 75 100 25 50 75 100 1.00 Highland 0.75 Ulaanbaatar West Central 0.50 East 0.25 Poverty line 0 25 50 75 100 Per capita real consumption (Thousands of Tugrug per month) Source: 2002/03 HIES/LSMS. urban and rural areas, and on the four analytical rural areas are the opposite. One third of the poor lives domains considered for the survey design. Poverty in in the countryside, one quarter in aimag centers and urban domains is significantly lower than in rural areas, one fifth in the soum centers. 30% and 43% respectively. Among urban domains, What is the sensitivity of these findings to the level Ulaanbaatar is less poor than aimag centers. However, of the poverty line? Again, stochastic analysis allows us the incidence of poverty in soum centers and the coun- to evaluate the robustness of the results. At the region- tryside, both rural areas, is very much alike, with soum al level, the West is the poorest region and centers being slightly worse-off. Fifty five percent of Ulaanbaatar is the least poor (Figure 2.3). Nothing con- the population lives in urban areas, but only around clusive can be said regarding the other three regions forty five percent of the poor, whereas the figures in because their curves intersect each other, which means CHAPTER 2. WELFARE PROFILE 19 that their ranking will be affected depending on the is clear that these estimates are not related to different chosen poverty line28. Regarding the urban-rural divide, household characteristics in the four quarters: urban- the three previous points stand. First, urban areas are ization and demographic features do not show signifi- always better-off than rural areas. Second, cant variations. This supports the argument that pover- Ulaanbaatar is less poor than the aimag centers. Third, ty fluctuations are the result of the seasonality charac- although the ranking between soum centers and teristic of the Mongolian economic cycle. countryside is quite sensitive to the chosen poverty Both urban and rural areas are affected by season- line, the poverty incidence is almost the same in both ality fluctuations, but in rather differing ways (this is domains. Overall then, the capital is the least poor, fol- not shown in the table). Urban areas enjoy a consump- lowed by aimag centers and then by rural areas. tion surge in the first quarter, but they are not affect- ed by any seasonality effect in the summer period. On 2.5. The seasonality of poverty the contrary, in rural areas the third quarter emerges as the period with the highest consumption. One impor- A relevant feature of poverty in Mongolia is its tant message associated to these results is that house- seasonality. In particular livestock activities, but also holds, especially in rural areas, are unable to smooth other factors determine remarkable fluctuations in consumption and this requires both improved market consumption levels along the year29. Typically summer integration as well as policies that can strengthen the time (the third quarter) is a period of relative abun- role of credit markets. Table 2.6: The seasonality of poverty National Quarter I Quarter II Quarter III Quarter IV Headcount 36.1 29.1 40.3 33.5 41.2 (1.4) (3.0) (2.7) (2.8) (2.9) Poverty Gap 11.0 8.0 11.7 10.3 13.7 (0.6) (1.0) (1.1) (1.2) (1.4) Severity 4.7 3.1 4.9 4.4 6.1 (0.3) (0.5) (0.6) (0.6) (0.8) Memorandum items: Share below PL (%) 100.0 19.6 25.9 23.6 30.9 Population share (%) 100.0 24.3 23.2 25.4 27.0 Household size 4.3 4.3 4.3 4.3 4.4 Dependency ratio (%) 43.3 42.8 42.6 44.4 43.3 Children (% household size) 31.2 31.7 30.9 30.6 31.6 Age of household head 44.5 43.5 44.7 45.4 44.3 Male household head (%) 82.5 83.9 84.3 80.6 81.6 Urbanization (%) 55.4 55.6 58.5 54.9 53.1 Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. dance while the long winters are associated with lower consumption, interrupted only by the increased spend- 28 By plotting two or more per capita consumption cumulative functions in the same graph, it is possible to infer first-order stochastic dominance. Distribution A first-order stochasti- ing associated to the festivity period of the new lunar cally dominates distribution B if for any given level of per capita consumption, the share of the population with a lesser or equal level of consumption will always be lower in dis- year, which generally falls in January or February. From tribution B. In other words, if curve A always lies above curve B, distribution B will have a Table 2.6 it is evident how poverty measures fluctuate higher level of welfare and hence lower poverty. However, if the curves intersect each other, the criteria does not apply and it is not possible to infer which distribution has a during the year, with the poverty headcount higher by higher level of welfare. 29 It is important to mention that, as explained in appendix B, the consumption aggregate 5 percentage points in the second and fourth quarters. has been adequately corrected for seasonal price differences, and some of the consump- tion components (rent and utilities) are also adjusted by seasonal consumption because From the memorandum items reported in the table it are derived from annual consumption before being expressed in monthly terms. However, food consumption as well as non-food consumption was collected on a quarterly basis. 20 Table 2.7: Poverty and household size CHAPTER 2. WELFARE PROFILE National Household size 2.6. Household composition 1 2 3 4 5 6 7 8 plus Headcount 36.1 1.2 7.4 15.5 23.5 34.4 48.5 57.4 69.4 (1.4) (0.9) (1.8) (1.7) (2.0) (2.2) (3.0) (4.0) (3.7) Poverty Gap 11.0 0.4 1.8 3.6 6.2 9.1 14.9 19.0 26.1 (0.6) (0.4) (0.4) (0.5) (0.6) (0.8) (1.2) (1.7) (2.3) Severity 4.7 0.2 0.5 1.2 2.3 3.5 6.1 8.1 13.3 tion, some are comprised by nuclear or by extended families, others have a high proportion of children, and Households differ in their demographic composi- (0.3) (0.2) (0.1) (0.2) (0.3) (0.4) (0.7) (0.9) (1.6) Memorandum items: Poor share (%) 100.0 0.0 0.9 5.6 15.0 21.2 21.2 15.7 20.5 Population share (%) 100.0 1.3 4.2 13.0 23.0 22.3 15.8 9.9 10.7 Dependency ratio (%) 43.3 61.4 46.8 37.6 42.2 43.4 44.0 43.2 42.4 Children (% household size) 31.2 0.0 11.5 26.2 36.7 37.9 39.9 39.4 37.3 Age of household head 44.5 56.3 50.1 42.0 40.5 43.5 44.5 47.1 49.5 Male household head (%) 82.5 43.7 67.0 79.9 89.1 87.5 91.4 89.8 83.2 Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. size of the household. The incidence of poverty tion? Table 2.7 shows first how poverty varies with the correlation between poverty and household composi- some are comprised only by elderly people. Is there any increases monotonically with household size. This is CHAPTER 2. WELFARE PROFILE 21 hardly surprising given that our welfare indicator is per poverty incidence and the dependency ratio for urban capita consumption, which implicitly assumes that and rural areas. The higher the dependency ratio, the there are neither different needs among members nor higher the poverty experienced by the household. economies of size within the household30. The likeli- Usually a higher share of children and elderly people hood of being poor if one lives in households of up to relative to the total number of members in the family three members is barely more than 10 percent. One of means that “earners� have to support more people, every five Mongolians lives in those households but hence there is less income and consumption available they make up for less than one tenth of the poor. The to each household member and therefore more pover- poverty incidence in households of four and five mem- ty. This relationship holds up to values of 75%, above bers, the typical household size in the country, is 24 these levels poverty declines, which is likely to reflect and 34 percent respectively. These households com- the fact that in households where the share of depen- prise just less than half of the population and two out dants is really high, these households are mainly com- of every five poor. By contrast, poverty reaches at least prised by elderly people still working or receiving some 50 percent among households of more than five mem- steady income, like a pension or remittances from a bers, which represent a third of the population but relative, that defends them against poverty32. more than half of the poor. The level of poverty is par- Figure 2.4: Poverty and dependency ratio 80 Headcount (%) 60 Rural 40 Urban 20 0 25 50 75 100 Dependency ratio Source: 2002/03 HIES/LSMS. ticularly dramatic among those households with at least eight members, where seven out of every ten people is below the poverty line and they represent a fifth of the poor. A second way to analyze the demographic com- position of the households is through the dependency ratio. This is a common indicator to capture the demo- 30 The sensitivity of these two assumptions to eight different family compositions is exam- graphic composition of the families. It will be defined ined in more detail in Appendix C.1. 31 Alternatively, it can be also defined as the ratio between the non-working-age population as the ratio between the non-working age population and the working-age population, typically those less than 15 or more than 64 to those 15 to 64 years old. Thus it represents the number of "dependants" for each "earner" in the and the number of members in the household31. Thus household. However, in Mongolia a different age-cut is used to define working-age pop- it represents the share of “dependants� in the house- ulation: men aged 16 to 59 and women aged 16 to 55. 32 Indeed, 80% of the households with dependency ratios higher than 75 are comprised of hold. Figure 2.4 displays the relationship between the one or two elderly members. 22 CHAPTER 2. WELFARE PROFILE 2.7. Characteristics of the household head first group, increases with the second and finally falls, although remains higher than at young ages. More A common practice when doing poverty compar- than three out of five poor live in households with mid- isons is to classify households according to the charac- dle-aged heads, a quarter have an older head and one teristics of the household head33. Although not with- tenth a younger one. Differences in the composition of out limitations, it does provide a simple and useful way the households across these three groups may explain to make comparisons across households34. Often living much of the observed poverty levels. For instance, chil- standards and household demographic composition dren account for forty percent of the family size are linked with the characteristics of the head, who is among households with middle-aged heads but likely to be the main source of economic support with- decreases to less than that among those with older in the household. For instance, a head with tertiary heads, which also are more likely to be headed by a education is likely to live in urban areas and have a woman. smaller than average number of children. In this sec- Is the pattern the same when comparing female tion, the connection between poverty and age, gen- against male-headed households? Available evidence der, education, employment and migratory status of suggests that female-headed families are better off at the household head is examined. younger ages, but after the head reaches around 30 years old they are consistently worse off (Figure 2.5). Age and gender These results must be taken with caution because the comparison is assessing families with very dissimilar What is the link between the age of the household structures. More than four of every five female heads head and poverty? Table 2.8 displays the poverty Table 2.8: Poverty and age of the household head National 15 - 29 30 - 49 50 plus Headcount 36.1 27.0 40.2 31.6 (1.4) (3.0) (1.8) (2.1) Poverty Gap 11.0 7.5 12.2 10.0 (0.6) (1.0) (0.7) (0.8) Severity 4.7 2.9 5.2 4.3 (0.3) (0.5) (0.4) (0.5) Memorandum items: Household size 4.3 3.4 4.7 4.1 Dependency ratio (%) 43.3 35.9 42.2 48.5 Children (% household size) 31.2 32.4 40.8 14.6 Age of household head 44.5 25.6 39.3 61.6 Male household head (%) 82.5 93.7 85.6 72.5 Share below PL (%) 100.0 8.4 64.9 26.7 Population share 100.0 11.3 58.2 30.5 Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. 33 The LSMS applies a precise definition to identify the head of the household. It is the per- measures according to three age groups of the house- son who is acknowledged as the head by the other members, plays the main role in organ- izing others, bears full responsibility for household problems, and takes most of the house- hold head, 15 to 29 years old, 30 to 49 and 50 and hold financial decisions. 34 For instance, sometimes the eldest person is considered as the head as a sign of respect, more. The incidence of poverty is lowest among the although he or she does not fulfill the given definition. Another example is when female widows, who may be in practice the heads of the household, refer to their eldest son as the head of the family. CHAPTER 2. WELFARE PROFILE 23 Figure 2.5: Poverty, age and gender of the household head 60 Female 40 Headcount (%) Male 20 0 20 35 50 65 Age of the household head Source: 2002/03 HIES/LSMS. are widows, divorced or separated, while more than households where the head has finished at least the nine out of ten male heads are married. Female heads 8th grade of secondary and one quarter of Mongolians are older and more likely to live in rural areas. Finally, has a household head with tertiary education. By con- nationwide, female-headed households comprise trast, less than one fifth lives in households where the around fifteen percent of the total households and a head has no education or only primary school. As similar share of the poor. expected, the higher the level of instruction complet- ed, the less the poverty experienced. The returns to Education education seem to increase considerably if the head has finished complete secondary, for levels lower than A fundamental indicator of human capital is edu- that, the incidence of poverty is around 45 percent but cation. It is widely recognized as one of the main fac- for higher educational attainments only 25 percent. tors to increase the living standards of the population. This hides differences within each of these two broad People with none or little education are likely to be groups. Poverty levels are similar for heads with no employed in labor-intensive industries, which generally education, only primary or up to 8th grade of second- exhibit less productivity and hence lower salaries, have ary. But completion of secondary reduces the head- a small degree of labor mobility and are more vulnera- count measure to almost one third, having a diploma ble to adverse shocks. Education enlarges not only job to one quarter and receiving at least a bachelor degree opportunities but also helps people to realize the sig- to almost one tenth. Vocational education appears to nificance of other aspects of welfare, like the impor- be the exception among higher levels of instruction. tance of a better health or to participate more actively Urban and rural disaggregation introduces two minor in society. changes. In soum centers and the countryside only a Table 2.9 displays information on poverty meas- diploma or a university degree are found to reduce the ures by the highest level of education obtained by the level of poverty, completing secondary or vocational household head. Before commenting on the relation- education does not seem to be enough. The counter- ship between education and poverty, it is important to part of this finding is that in the capital and aimag cen- note that education levels of household heads are very ters, these two levels do decrease the chances of being high, more than 80 percent of the population lives in poor. 24 CHAPTER 2. WELFARE PROFILE Table 2.9: Poverty and highest level of education completed by the household head National None Primary Secondary Complete Vocational Diploma University 8th grade Secondary Headcount 36.1 45.8 45.6 45.5 34.9 40.7 23.4 11.6 (1.4) (4.9) (3.6) (2.3) (2.3) (3.4) (2.5) (2.1) Poverty Gap 11.0 12.8 16.4 13.8 9.3 13.1 6.7 2.9 (0.6) (1.7) (1.7) (0.9) (0.9) (1.5) (0.9) (0.7) Severity 4.7 4.8 7.9 5.7 3.6 6.0 2.7 1.1 (0.3) (0.9) (1.1) (0.5) (0.4) (0.9) (0.5) (0.3) Memorandum items: Household size 4.3 3.2 4.2 4.6 4.4 4.5 4.3 4.1 Dependency ratio (%) 43.3 59.3 49.7 42.1 41.6 43.4 38.4 39.1 Children (% household size) 31.2 19.3 24.7 35.0 36.1 38.8 28.8 25.7 Age of household head 44.5 55.9 52.3 41.0 38.8 40.7 46.3 46.7 Male household head (%) 82.5 60.7 73.1 89.2 88.1 83.7 80.5 82.7 Share below PL (%) 100.0 5.4 17.9 34.6 18.2 11.5 8.8 3.7 Population share 100.0 4.2 14.2 27.5 18.8 10.2 13.6 11.5 Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. Table 2.10: Poverty and labor force participation of the household head National Employed Unemployed Out of Total Agriculture Industry Services Labor Force Headcount 36.1 33.6 41.0 33.2 26.9 48.7 41.6 (1.4) (1.7) (3.0) (3.4) (1.9) (5.4) (2.2) Poverty Gap 11.0 9.7 12.0 9.1 7.7 16.7 14.0 (0.6) (0.6) (1.2) (1.3) (0.7) (2.4) (1.1) Severity 4.7 3.9 4.8 3.6 3.1 7.4 6.6 (0.3) (0.3) (0.6) (0.7) (0.3) (1.3) (0.7) Memorandum items: Household size 4.3 4.4 4.2 4.4 4.5 4.7 4.1 Dependency ratio (%) 43.3 40.7 44.8 35.8 38.1 42.2 50.4 Children (% household size) 31.2 33.9 33.4 33.2 34.6 40.9 23.0 Age of household head 44.5 41.0 41.0 39.7 41.4 37.7 54.3 Male household head (%) 82.5 86.7 88.3 89.0 84.4 86.3 71.1 Share below PL (%) 100.0 66.5 34.2 8.1 24.3 4.0 29.4 Population share 100.0 71.5 30.2 8.8 32.6 3.0 25.5 Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. CHAPTER 2. WELFARE PROFILE 25 Employment almost a third in families whose head is not actively participating in the labor market. The distribution of One of the most evident determinants of house- the population follows a very similar pattern, except hold welfare is whether or not their members can par- that agriculture decreases its share and the contrary ticipate in the labor market and particularly, if occurs to services. employed, the type of job that they can engage in. In The relationship between poverty and employ- Mongolia, this issue received some attention since the ment can be further explored by looking at the sector transition to a market economy started. Initially, the of employment. Table 2.11 separates employed house- shrinkage of manufacturing and the public administra- hold heads in herders, working in the private sector, in tion pushed many people back to agriculture. the public sector and in state companies36. An addi- However, in recent years the combination of natural tional second breakdown is done among those out of disasters and the surge of the services sector have the labor force into pensioners and the rest. A few turned that trend. Table 2.11: Poverty and sector of occupation of the household head National Employed Unemployed Out of Labor Force Herders Private Public State Pensioners Others Headcount 36.1 39.2 34.7 25.9 21.6 48.7 35.7 51.4 (1.4) (3.2) (2.2) (2.5) (5.7) (5.4) (2.7) (3.1) Poverty Gap 11.0 11.4 9.9 7.5 4.9 16.7 10.9 19.2 (0.6) (1.2) (0.8) (0.9) (1.8) (2.4) (1.1) (2.0) Severity 4.7 4.5 4.1 3.0 1.7 7.4 4.7 9.6 (0.3) (0.7) (0.4) (0.4) (0.8) (1.3) (0.6) (1.4) Memorandum items: Household size 4.3 4.2 4.4 4.5 4.5 4.7 3.9 4.6 Dependency ratio (%) 43.3 45.0 39.4 37.2 32.2 42.2 57.0 37.5 Children (% household size) 31.2 32.6 36.0 33.5 31.6 40.9 17.1 34.4 Age of household head 44.5 41.1 40.0 42.3 41.4 37.7 61.9 39.6 Male household head (%) 82.5 88.6 87.3 82.0 91.6 86.3 63.1 86.8 Share below PL (%) 100.0 28.8 23.2 12.8 1.8 4.0 15.6 13.8 Population share 100.0 26.5 24.1 17.9 3.0 3.0 15.8 9.7 Note: Pensioners refer to household heads receiving any pension or benefit from the state. Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. Table 2.10 combines information on participation findings are worth emphasizing. First, the population on the labor force, main sector of employment and in households whose head is involved in livestock activ- poverty35. Population living in households where the ities experiences higher poverty than those whose head is currently working has higher living standards head is employed anywhere else. Second, public and than those whose head is either unemployed or out of especially state jobs seem to offer better living stan- the labor force. Among the employed, poverty levels dards to the twenty percent of Mongolians living in are lower in families whose head works in services compared to those in industries and significantly lower 35 A person participates in the labor force if she worked during last week, did not work but than those in agriculture. More than a third of the poor had a job or did not work, did not have a job but looked for work. Otherwise, she is con- sidered out of the labor force. No age considerations were taken into account for the esti- lives in households whose head engages in agriculture, mation of Table 2.10. 36 After transition, state companies lost their major role in the economy. Nowadays they are a quarter in services, less than a tenth in industry and limited to a few sectors in the economy, mainly utilities, transportation and textiles. 26 CHAPTER 2. WELFARE PROFILE those households. Third, poverty levels in households opportunities. A lot of them went back to rural areas with heads employed in the private sector are some- to pursue herding during the beginning of 1990s. where in between, although much closer to those rear- Others, especially recently, have returned or migrated ing livestock than to heads with public posts. to the cities and aimag centers. For instance, according to the household survey almost ten percent of the pop- Fourth, families with an unemployed head experi- ulation can be considered as migrants37. Half of them ence a fifty percent chance of being poor. However, migrated in the last ten years and a quarter since 1998. they comprise less than five percent of the poor. Fifth, Four out of ten migrants reported that they moved there are two very different groups among heads that because of work or to live close to the market. What is are not participating in the labor market: pensioners the observed connection between poverty and migra- and non-pensioners. The probability of being poor in tion? households where the head is a pensioner is signifi- cantly lower than in families where the head is not, Twelve percent of Mongolians live in a household almost one third compared to one half. Each one of whose head is an immigrant. They experience less these two groups comprise around fifteen percent of poverty than those the rest of the population, 31% the poor. Sixth, demographic indicators provide some and 37% respectively (Table 2.12). Although this find- useful information. For instance, those employed in ing is significant at the national level, it is not when the public and state jobs tend to be older than those in the comparison is done within urban areas. In both private sector. Pensioners are the eldest, but heads out domains, families with a head that migrated are bet- of the labor force that are not pensioners have similar ter-off than those with a head born in the same soum, ages than those working. Finally, the population living but the differences are lower. Immigrants are concen- with a head that is a pensioner has the highest chance trated in urban areas, almost four fifths of the popula- of having also a female head. tion with an immigrant head are in the capital and in aimag centers. A tenth of the poor lives in households Migrant status headed by an immigrant, and seventy percent of them Table 2.12: Poverty and migratory status of the household head National Urban Rural Migrant Non-migrant Migrant Non-migrant Migrant Non-migrant Headcount 31.2 36.8 29.0 30.5 38.7 43.7 (2.9) (1.5) (3.2) (1.9) (5.9) (2.4) Poverty Gap 9.6 11.2 8.4 9.4 13.9 13.2 (1.2) (0.6) (1.3) (0.8) (2.8) (1.0) Severity 4.4 4.7 3.8 4.0 6.6 5.5 (0.7) (0.4) (0.8) (0.4) (1.5) (0.6) Memorandum items: Household size 4.2 4.3 4.3 4.4 4.1 4.3 Dependency ratio (%) 43.8 43.3 42.9 41.6 47.1 45.0 Children (% household size) 29.9 31.4 29.4 29.7 31.8 33.1 Age of household head 47.2 44.1 47.6 45.9 45.9 42.1 Male household head (%) 79.7 82.9 78.8 79.7 82.6 86.3 Share below PL (%) 10.6 89.4 16.5 83.5 5.5 94.5 Population share 12.3 87.7 9.6 45.9 2.7 41.8 Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. As it was pointed out in the previous section, changes in the structure of the economy during the 37 The definition considers population born in a different soum in which they are currently living and people that originally emigrated from their soum of birth but returned to live in last decade saw many people looking for other job there. Using a similar definition, but with aimags as the space of reference instead of soums, the 2000 Census estimated a considerably higher figure of about 25%. CHAPTER 2. WELFARE PROFILE 27 are in urban areas. Finally, no major distinctions are domains, the more rural is the area, the higher are the found when looking at demographic indicators, except average holdings. Across regions, it is the East the one that rural immigrant heads are older and more likely to that consistently has a higher livestock per capita for be female. almost all species (the exception being camels). The fact that most of its territory consists of vast steppes 2.8. Assets and grasslands, a critical element for herding, favors these activities in that region. On the other hand, the Ownership of assets is an essential factor to deter- West is a domain where ownership is well spread, mine the living standards of the population. It allows ranks second after the Highland, but livestock per households to hedge against economic insecurity or herder is the lowest. Finally, more poor people are seasonal patterns in agriculture. If the main breadwin- involved in rearing animals but their average livestock ner is suddenly unemployed or if a natural disaster held is less than half that of the non-poor. This pattern occurs, such as heavy snowstorms, droughts or floods, is similar for all species of livestock. the household can use its assets to smooth their con- What is the connection between livestock hold- sumption. For instance, livestock can be slaughtered or ings and living standards? Table 2.14 compares pover- money taken out from savings. Assets are generally ty measures by urban-rural divide and by whether or crucial to access credit markets. Hence this wealth indi- not the household keeps livestock. The evidence seems cator works as insurance to avoid vulnerability. Three to suggest that the impact of rearing livestock is very types of household assets will be examined: livestock, different in those two domains. In urban areas it is land and financial assets. linked with a higher level of poverty, probably reflect- ing the fact that in cities reliance on agriculture activi- Livestock ties is not enough, households must diversify in order to improve their livelihood. However, in rural areas, Livestock is the main factor of production in agri- owning livestock does increase the welfare level of the culture in Mongolia. Almost half of the labor force population, the incidence of poverty is significantly engages in agriculture, mainly herding and related lower for the population that engages in livestock activities. Livestock rearing involves mainly five types of activities and their gap and severity of poverty indexes animals in the country, each one reflecting different are even proportionally smaller when compared to opportunities for the household, having goats implies population without livestock. Across regions, it is in the been involved in the cashmere business, owning sheep East and Central where herders enjoy higher living or camels is related to the wool commerce, and cattle standards than non-herders, but only in the East the and horses are associated with meat, milk and dairy level of poverty is considerably lower among the pop- production. ulation involved in herding. In the West the incidence Table 2.13 shows livestock holdings for the main of poverty appears to be lower among non-herders, five species and by various geographical divisions. whereas in the Highland is about the same across both Almost four out of ten people hold animals. Cattle, groups. horses, goats and sheep are held by around one fourth This result implies that, at least in rural areas, there to one third of the population, whereas camels are is a negative link between poverty and livestock hold- only brought up by less than one tenth. Patterns vary ings. Does the number of livestock held matter? Figure by region, less than 10% of urban dwellers owns ani- 2.6 displays the incidence of poverty relative to the mals compared with almost three quarters in rural level of per capita livestock among herders. It is found areas. Ulaanbaatar is the domain where ownership of that indeed poverty declines with a higher number of animals is lowest, not even four percent. By contrast, per capita livestock in both urban and rural domains. in the countryside close to ninety percent of the popu- Although in urban areas, the share of population own- lation holds some type of animals. A more even pat- ing livestock is worse-off compared to those that do tern is observed when looking at the west-east divide, not, among owners, the more livestock they hold, the with the Highland as the region where holdings are less poverty they experience. The relationship is clearer higher, especially for sheep and goats. in rural areas, yet for holdings greater than twenty The average livestock per capita among herders is 7 bods, or an equivalent of 7 horses38 (see also Table 38 The purpose of the bod scale is to calculate the size of the herd by transforming all live- 2.13). Not surprisingly, rural areas have more than stock held into equivalent horses. One horse is assumed to be the same as one cattle (cow or yak), 0.67 camels, six sheep or eight goats. double the levels of urban domains. Among analytical 28 Table 2.13: Livestock holdings Cattle Horses Camels Sheep Goats Bods Holders Average Holders Average Holders Average Holders Average Holders Average Holders Average (%) among (%) among (%) among (%) among (%) among (%) among CHAPTER 2. WELFARE PROFILE holders holders holders holders holders holders Urban 7.2 1.9 3.3 2.4 0.3 0.7 4.1 8.2 4.7 4.5 9.1 3.3 Rural 54.5 2.3 57.5 2.9 17.4 1.1 61.7 12.4 64.0 12.5 72.3 7.6 Ulaanbaatar 3.2 1.2 1.1 4.0 0.1 1.0 1.1 5.8 1.0 3.4 3.8 2.6 Aimag centers 11.9 2.1 5.9 2.0 0.7 0.6 7.7 8.6 9.2 4.6 15.4 3.5 Soum centers 39.0 1.7 29.3 2.3 4.4 0.5 34.0 7.9 36.4 7.1 46.9 4.5 Countryside 63.4 2.6 73.6 3.0 24.7 1.1 77.5 13.5 79.8 13.9 86.7 8.5 West 40.2 1.6 41.9 1.4 15.3 0.5 43.5 8.7 49.9 12.4 54.4 5.0 Highland 42.5 2.6 45.2 2.9 8.1 1.4 52.1 10.7 50.4 11.7 58.3 7.3 Central a/ 28.5 1.5 24.1 3.3 9.7 1.8 30.7 14.2 31.2 11.3 40.1 6.6 East 50.8 3.7 47.5 4.5 15.9 0.6 38.2 20.9 45.2 12.5 53.8 11.5 Non-poor 28.3 2.7 27.1 3.7 7.8 1.4 29.1 15.6 30.0 13.6 35.7 9.0 Poor 28.3 1.4 27.9 1.4 8.1 0.5 30.9 6.1 33.3 9.0 39.9 3.9 National 28.3 2.3 27.4 2.9 7.9 1.0 29.8 12.1 31.2 11.8 37.2 7.0 a/ Excludes Ulaanbaatar. Note: The bod scale was used to estimate the size of the herd. These factors transform cattle, camels, sheep and goats into equivalent horses. One horse is assumed to have the same value as one cattle, 0.67 camels, six sheep or eight goats. Cattle includes cows and yaks. Source: 2002/03 HIES/LSMS. CHAPTER 2. WELFARE PROFILE 29 Table 2.14: Poverty and livestock holdings National Urban Rural Non-herders Herders Non-herders Herders Non-herders Herders Headcount 34.6 38.7 29.9 33.7 53.5 39.5 (1.6) (2.6) (1.8) (5.1) (3.2) (2.9) Poverty Gap 10.9 11.2 9.2 8.9 17.5 11.6 (0.7) (1.0) (0.7) (2.3) (1.7) (1.1) Severity 4.8 4.5 4.0 3.5 8.0 4.6 (0.4) (0.5) (0.4) (1.2) (1.1) (0.6) Memorandum items: Household size 4.3 4.3 4.3 4.7 4.2 4.3 Dependency ratio (%) 42.6 44.6 41.6 44.5 46.5 44.6 Children (% household size) 30.8 31.8 29.7 29.4 35.4 32.1 Age of household head 45.3 43.0 46.0 48.8 42.8 42.2 Male household head (%) 79.4 87.8 78.9 87.0 81.4 87.9 Share below PL (%) 60.1 39.9 41.8 4.7 18.3 35.2 Population share 62.8 37.2 50.4 5.0 12.4 32.2 Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. Figure 2.6: Poverty and size of herd 60 Headcount (%) Rural 40 20 Urban 0 5 15 25 Per capita livestock (bods) Source: 2002/03 HIES/LSMS. bods per capita, poverty appears to be stable. A possi- relied only in one particular activity. The fact that 75% ble explanation is that the more animals the household of herders owns at least three of the main five types of own, the more productive activities it can engage, so, animals provides support to this hypothesis39. by diversifying, the household minimizes its exposure to negative shocks that may hit them harder if they 30 CHAPTER 2. WELFARE PROFILE Land Financial assets Land is typically recognized as one of the most A significant component of household wealth is important assets of households, particularly in agricul- generally made of financial assets. If income exceeds tural economies. However in Mongolia farming is lim- expenditure, people can accumulate savings, but if they ited and it is of limited relevance when compared with are more concerned with daily survival, this is unlikely to herding activities. According to the household survey, happen. In Mongolia, only one tenth of the population only 13% of the population uses land for growing lives in households that have financial assets in the form crops, with no major differences in urban or rural of either bank accounts or stocks in companies40. In areas. Furthermore, being engaged in farming appears urban areas the share is 15% compared to barely 7% to reduce the chances of having higher living standards in rural domains. This may reflect however the low in both domains. The poor are more likely to be degree of financial intermediation in the country and it involved in agriculture than the non-poor, 17% and could be argued that people save by holding cash, 11% respectively. A few factors may help to explain something that is not captured in the survey. Yet more why agriculture is not developed in the country. First, than 90% of non-savers responded that they did not exposure to weather conditions makes farming diffi- save because they do not have enough money to do so. Table 2.15: Poverty and land access National Urban Rural Non-farmers Farmers Non-farmers Farmers Non-farmers Farmers Headcount 34.5 47.2 29.2 39.0 41.4 54.3 (1.5) (3.2) (1.8) (4.4) (2.5) (4.4) Poverty Gap 10.3 15.4 8.8 12.6 12.4 17.7 (0.6) (1.5) (0.7) (1.8) (1.0) (2.3) Severity 4.4 6.8 3.8 5.5 5.1 7.9 (0.3) (0.9) (0.4) (1.1) (0.5) (1.5) Memorandum items: Household size 4.2 4.8 4.3 4.9 4.2 4.8 Dependency ratio (%) 43.4 43.1 41.6 43.7 45.6 42.6 Children (% household size) 30.8 33.9 29.5 30.8 32.5 36.5 Age of household head 44.4 45.2 45.9 49.0 42.4 41.9 Male household head (%) 81.9 87.8 78.9 86.0 85.6 89.2 Share below PL (%) 83.2 16.8 40.1 6.4 43.1 10.4 Population share 87.2 12.8 49.5 5.9 37.6 6.9 Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. cult, production can be easily lost due to weather haz- Moreover, it is evident that having financial assets is ards. Second, productivity is affected by the quality of strongly correlated with low poverty levels, particularly the soil and the low share of irrigated land. Third, more in Ulaanbaatar and aimag centers, where the poverty investment may be required for farming than, say, for incidence among savers is one third that among non- herding, both in terms of labor and capital. Fourth, it is savers (Table 2.16). In soum centers and the country- not a traditional activity performed by households, just side, the poverty headcount among savers is forty per- until a few years ago the state used to run farms in the country. Fifth, farming is harder to reconcile with the 39 The other case would be if households focus in only one or two livestock activities, which movements involved in the long-established way of may allow them to specialize and reach some economies of scale in the production process. breeding livestock. 40 When state owned companies were privatized, shares were given away or sold to the pop- ulation. CHAPTER 2. WELFARE PROFILE 31 Table 2.16: Poverty and savings National Urban Rural Non-savers Savers Non-savers Savers Non-savers Savers Headcount 38.8 15.5 33.8 11.1 44.6 27.5 (1.5) (2.4) (1.8) (2.4) (2.4) (5.9) Poverty Gap 12.0 3.1 10.5 2.0 13.7 6.1 (0.6) (0.6) (0.8) (0.5) (1.0) (1.5) Severity 5.2 0.9 4.6 0.5 5.8 2.0 (0.4) (0.2) (0.5) (0.2) (0.6) (0.7) Memorandum items: Household size 4.3 4.3 4.4 4.3 4.2 4.4 Dependency ratio (%) 43.5 42.1 42.1 40.1 45.0 47.6 Children (% household size) 31.1 32.1 29.5 30.3 32.7 37.1 Age of household head 44.5 44.5 46.5 44.7 42.2 44.2 Male household head (%) 82.0 86.5 78.6 85.1 85.8 90.4 Share below PL (%) 95.0 5.0 43.8 2.6 51.2 2.4 Population share 88.3 11.7 46.9 8.6 41.5 3.1 Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. cent less than among non-savers. The pattern is even es, a third in apartments and only a fifth in gers, where- more clear-cut when comparing the other two poverty as in rural domains three quarters of the people live in measures. Lastly, 5% of the poor own financial assets gers and the remaining mainly in houses. Table 2.17 compared to 15% of the non-poor. displays the relationship between poverty and type of dwellings. The incidence of poverty is higher in gers, 2.9. Housing lower in houses and the least in apartments. The same trend is observed in urban areas, the chances of being Another key determinant of living standards for the poor living in an apartment are less than half of those population is the type of housing they occupy and the living in houses and a third of those occupying gers. But access to basic infrastructure services. Households can in rural domains another pattern emerges: the level of quickly improve their welfare if they are provided with a poverty is higher in houses than in gers. The poor are better dwelling or with services that make them less vul- more likely to live in a ger, more than half of them do, nerable and expand their options and opportunities. A a third in houses and barely one tenth in apartments. In proper infrastructure will lift some of the constraints they Ulaanbaatar and aimag centers though half of the poor face to increase their productivity, for example, it could lives in houses, a third in gers and a sixth in apartments. make a big difference if instead of fetching water from a In rural domains the distribution of the poor follows the place half an hour away from the dwelling, household distribution of the population, three out of four live in members could obtain water from an improved source, gers and the remaining in houses. say a public standpipe, located closer to the dwelling, or even better, if they could be connected to the water net- Infrastructure services work. Two aspects of housing will be examined, type of dwellings and access to basic services. Living standards are increased by adequate infra- structure services such as access to an improved source Dwelling of water, proper sanitation facilities or electricity41. Lack of safe water or basic sanitation affects the health Gers are the most common type of housing in Mongolia, 45% of dwellers live there, a third in houses 41 Access to an improved water source refers to the percentage of the population with household connection, public standpipe or protected well or spring. Unimproved sources and a fifth in apartments. This varies by regions, in include vendors, tanker trucks and unprotected wells and springs. Sanitation refers to the urban areas almost half of the population lives in hous- percentage of the population with access to improved sanitation facilities such as ade- quate excreta disposal facilities (private or shared but not public). They can range from simple but protected pit latrines to flush toilets with a sewerage connection. 32 Table 2.17: Poverty and type of dwelling National Urban Rural CHAPTER 2. WELFARE PROFILE Ger House Apartment Others Ger House Apartment Others Ger House Apartment Others Headcount 43.4 38.2 16.6 30.0 47.5 33.9 14.3 31.2 41.9 48.5 41.8 20.0 (2.2) (1.9) (2.3) (6.7) (3.2) (2.2) (2.1) (7.1) (2.7) (3.7) (10.4) (18.4) Poverty Gap 13.5 11.3 5.0 9.1 14.7 10.5 3.9 9.6 13.0 13.3 16.8 4.6 (0.9) (0.9) (1.1) (2.5) (1.4) (1.0) (0.7) (2.7) (1.1) (1.5) (8.0) (4.2) Severity 5.7 4.8 2.3 3.2 6.2 4.7 1.6 3.5 5.5 5.2 9.2 1.0 (0.5) (0.5) (0.7) (1.3) (0.8) (0.6) (0.4) (1.5) (0.6) (0.8) (5.3) (1.0) Memorandum items: Household size 4.2 4.6 4.0 4.5 4.5 4.7 4.0 4.6 4.1 4.6 4.7 4.2 Dependency ratio (%) 45.7 42.1 40.3 40.4 45.7 41.1 40.5 40.7 45.8 44.3 37.2 38.3 Children (% household size) 31.9 32.3 28.0 30.5 31.0 30.8 27.5 29.5 32.2 35.7 34.1 38.3 Age of household head 43.5 44.6 46.1 48.4 47.3 45.4 46.5 49.7 42.3 42.8 40.8 38.6 Male household head (%) 84.0 82.4 79.6 82.4 77.6 81.1 78.8 82.0 86.0 85.5 91.0 85.9 Share below PL (%) 53.1 37.1 9.2 0.7 15.3 23.3 7.3 0.7 37.8 13.8 1.9 0.1 Population share 44.2 35.1 19.9 0.9 11.6 24.8 18.3 0.8 32.6 10.3 1.7 0.1 Note: Others include public and students dormitories, and other public apartments. Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. CHAPTER 2. WELFARE PROFILE 33 Figure 2.7: Access to infrastructure services in urban and rural areas 100 Urban Rural 80 60 Population (%) 40 20 0 Improved water sources Sanitation Electricity All three Source: 2002/03 HIES/LSMS. Table 2.18: Poverty and infrastructure services Improved water sources a/ Sanitation b/ Electricity All three Yes No Yes No Yes No Yes No Headcount 33.0 40.9 30.2 42.5 34.0 41.8 26.9 42.8 (1.6) (2.4) (1.7) (2.1) (1.5) (3.2) (1.8) (1.9) Poverty Gap 9.9 12.7 9.0 13.1 10.3 12.8 7.9 13.2 (0.7) (1.0) (0.7) (0.9) (0.6) (1.3) (0.7) (0.8) Severity 4.2 5.4 3.8 5.6 4.4 5.5 3.3 5.7 (0.4) (0.6) (0.4) (0.5) (0.3) (0.7) (0.4) (0.5) Memorandum items: Household size 4.3 4.3 4.3 4.3 4.4 4.1 4.3 4.3 Dependency ratio (%) 42.6 44.4 41.6 45.1 42.4 45.8 41.3 44.8 Children (% household size) 30.8 31.7 29.6 32.8 30.8 32.1 29.0 32.8 Age of household head 45.3 43.2 45.8 43.0 45.4 42.0 46.2 43.2 Male household head (%) 80.7 85.4 81.6 83.6 81.3 85.7 81.5 83.3 Share below PL (%) 55.8 44.2 43.5 56.5 68.9 31.1 31.3 68.7 Population share 61.0 39.0 52.0 48.0 73.1 26.9 42.0 58.0 a/ It refers to the percentage of the population with access to an improved water source such as household connection, public standpipe or protected well or spring. Unimproved sources include vendors, tanker trucks and unprotected wells and springs. b/ Sanitation refers to the percentage of the population with access to improved sanitation facilities such as adequate excreta disposal facilities (private or shared but not public). They can range from simple but protected pit latrines to flush toilets with sewerage connection. Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. 34 CHAPTER 2. WELFARE PROFILE of the population by increasing the chances of illness- water, sanitation or electricity is poorer than those es that are quickly transmitted in those environments. with access to them. The contrast is more evident Lack of electricity has a direct effect on education and when comparing access to all of the three basic servic- investment prospects. How does Mongolia fare in es, only one quarter of the population receiving them these dimensions of welfare? is poor compared with more than two fifths among those who do not. Table 2.19 provides the poverty The household survey indicates that three fifths of measures by an urban-rural divide. The picture varies the country have access to improved sources of water, substantially depending on what area one is looking half to improved sanitation facilities, three quarters to at. In urban areas the incidence of poverty is consider- electricity, and four out of ten individuals to all of ably lower among those receiving any service or all of them. However, there is a considerable urban bias. them than among urban dwellers lacking access to Figure 2.7 shows that the availability of these services infrastructure services. By contrast, in rural regions in urban areas is far more established than in rural findings are a bit puzzling. The incidence of poverty is regions. At least three quarters of urban dwellers have higher among those obtaining any of the services, access to each one of them compared to a quarter of although the joint access to the three of them does the rural population. Even more significant is the com- seem correlated to higher living standards yet the dif- parison among those receiving all of the three basic ference is not large enough to be regarded as statisti- services, 63 percent in urban areas and only 16 percent cally significant. in rural regions. Another factor -not fully captured in the survey- is the quality of the services. Urban areas Figure 2.8 shows the availability of infrastructure generally have access to better services than rural services by poverty status of the population. The non- areas. For instance, tap water may be regarded as of poor have better access to improved water sources, better quality than water coming from a well, which, sanitation facilities and electricity than the poor, and even when is protected, could be more exposed to the gap is substantial when considering joint access. contamination. Again, the national picture disguises regional patterns. Whereas in urban areas a larger share of the non-poor Table 2.18 displays the association between the receives these services, in rural domains access is simi- level of poverty and access to basic infrastructure serv- lar for both groups. ices. Nationwide, population lacking appropriate Figure 2.8: Access to infrastructure services by poverty status 80 Non-poor Poor 70 60 50 Population (%) 40 30 20 10 0 Improved water sources Sanitation Electricity All three Source: 2002/03 HIES/LSMS. Table 2.19: Access to infrastructure services by urban-rural divid Improved water sources a/ Sanitation b/ Electricity All three Urban Rural Urban Rural Urban Rural Urban Rural Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Headcount 28.3 38.8 46.3 41.7 26.0 41.9 45.0 42.8 29.6 63.5 46.7 41.0 24.0 41.0 41.4 43.8 (1.9) (3.3) (3.1) (3.0) (1.9) (2.8) (3.3) (2.7) (1.7) (8.8) (2.7) (3.3) (2.0) (2.4) (3.8) (2.7) Poverty Gap 8.6 11.9 13.7 13.0 7.7 13.4 13.9 13.0 8.8 29.9 14.7 12.1 7.0 13.0 12.3 13.4 (0.7) (1.5) (1.3) (1.3) (0.7) (1.3) (1.6) (1.1) (0.7) (7.3) (1.4) (1.3) (0.7) (1.2) (1.7) (1.1) Severity 3.7 5.2 5.8 5.4 3.2 6.1 6.2 5.3 3.7 18.8 6.3 5.0 2.9 5.8 5.3 5.6 (0.4) (0.9) (0.7) (0.7) (0.4) (0.8) (0.9) (0.6) (0.3) (6.0) (0.8) (0.7) (0.4) (0.7) (0.9) (0.6) Memorandum items: Household size 4.3 4.6 4.3 4.2 4.3 4.6 4.5 4.2 4.4 4.5 4.4 4.1 4.2 4.6 4.4 4.2 Dependency ratio (%) 41.6 42.9 45.6 44.9 41.4 43.1 42.6 46.0 41.6 53.6 44.7 45.5 41.2 43.0 42.1 45.7 Children (% household size) 29.4 30.8 34.7 32.1 28.6 32.7 33.4 32.9 29.6 32.2 34.3 32.1 28.2 32.4 33.0 33.0 Age of household head 46.3 45.7 42.4 42.3 46.7 44.7 42.5 42.3 46.2 47.2 43.2 41.8 46.9 44.9 42.8 42.3 Male household head (%) 79.2 81.2 84.8 86.8 80.3 77.6 86.4 86.0 79.9 62.1 85.5 86.5 80.5 77.9 86.8 86.0 Share below PL (%) 35.1 11.3 20.7 32.9 29.3 17.2 14.2 39.3 44.7 1.8 24.2 29.4 23.3 23.2 8.0 45.5 Population share 44.9 10.5 16.1 28.4 40.6 14.8 11.4 33.2 54.4 1.0 18.7 25.9 35.0 20.4 7.0 37.6 a/ It refers to the percentage of the population with access to an improved water source such as household connection, public standpipe or protected well or spring. Unimproved sources include vendors, tanker trucks and unprotected wells and springs. b/ Sanitation refers to the percentage of the population with access to improved sanitation facilities such as adequate excreta disposal facilities (private or shared but not public). They can range from simple but protected pit latrines to flush toilets with a sewerage connection. Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. CHAPTER 2. WELFARE PROFILE 35 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NETS 38 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET A major constraint that the poor face to escape ment rates. Lastly, it analyzes the extent and impor- poverty is their low levels of human capital. Investing in tance of safety nets, both formal and informal, and education and health is a significant step towards their effect on the living standards of the population. improving the living standards of the poor. International experience has shown that it boosts the 3.1. Education productivity of labor, which is typically the main and sole asset they own. It also provides the means to the This section reviews the evidence for the educa- poor and their children to lead better lives. A second tion sector in the country. It examines the extent and limitation the poor confront is the scarce assistance the degree of inequality in terms of access to schools they obtained from public and private sources to help and education endowments. It starts by looking at the them cope with economic insecurity. Safety networks educational attainment of the adult population and often play an important role in mitigating adverse then focusing on those currently attending school. shocks the household face and in alleviating poverty. Enrollment is examined through net and gross enroll- This chapter focuses first on the education and health ment rates as well as participation rates, and later a sectors, it examines if the provision of these services is students’ profile is developed. equitable and the differences in endowments between Adult educational attainment the poor and the non-poor. Then it evaluates the main features of the labor market such as participation The adult population in Mongolia has reached a rates, characteristics of employment and unemploy- very high educational attainment42. According to the Table 3.1: Highest educational attainment of adult population None Primary Secondary Complete Vocational Higher University Total 8th grade Secondary diploma Location Urban 2.2 6.8 18.6 31.3 9.4 16.3 15.4 100.0 Rural 9.0 19.1 32.9 20.8 7.7 7.3 3.2 100.0 Ulaanbaatar 2.0 6.4 17.2 31.3 8.5 16.7 17.9 100.0 Aimag centers 2.5 7.2 20.4 31.4 10.6 15.7 12.2 100.0 Soum centers 3.0 9.0 26.6 29.3 12.3 13.4 6.5 100.0 Countryside 12.3 24.8 36.5 16.0 5.2 3.9 1.3 100.0 West 7.7 15.6 26.3 23.7 10.0 10.8 6.0 100.0 Highland 8.1 14.4 30.9 24.2 5.7 10.7 6.1 100.0 Central a/ 3.4 14.5 25.9 26.9 11.5 9.6 8.2 100.0 East 7.6 14.0 30.2 23.3 9.0 11.0 4.9 100.0 Gender Men 4.9 11.8 29.2 24.9 8.8 10.4 9.9 100.0 Women 5.2 12.1 20.6 28.6 8.7 14.3 10.6 100.0 Quintile Poorest 6.8 15.5 36.5 23.4 9.4 6.6 1.9 100.0 Q2 6.0 13.0 30.3 27.8 9.1 9.8 4.1 100.0 Q3 5.0 12.5 24.7 29.2 8.7 11.9 7.9 100.0 Q4 4.4 10.5 19.8 28.8 9.2 14.1 13.2 100.0 Richest 3.8 9.7 16.2 24.9 7.6 17.7 20.1 100.0 National 5.1 12.0 24.7 26.9 8.7 12.5 10.2 100.0 a/ Excludes Ulaanbaatar. Source: 2002/03 HIES/LSMS. 42 Adult population refers to the population 18 years old and more. Less than 10% of them are still attending educational institutions. CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET 39 household survey, more than 80% of the population cation is that they are overwhelmingly rural dwellers. 18 years and older has at least finished the 8th grade Around seven out of ten are rural dwellers and more of secondary, almost a tenth vocational educational than 80 percent of them are from the countryside. and more than a fifth tertiary education (see Table 3.1, A division of the population based on consump- and also Table D.15 in Appendix D). Only one of every tion quintiles illustrates an evident pattern43. The better twenty adults has not completed primary school and -off the individual in terms of consumption, the higher around one tenth has only completed primary. Urban its educational attainments. Almost one out of five adults have higher levels than rural residents. For adults from the poorest quintile completes no more instance, four out of ten urban dwellers have finished than primary school compared to half that share studies beyond secondary school, i.e. either vocational among the wealthiest. On the other hand, almost half or tertiary education, compared to less than one fifth of the richest adults have more than a secondary rural residents. By contrast, less than one tenth of degree but less than a fifth of the poorest have urban adults have none or primary education, whereas achieved the same. Within each educational level, the this share is three times higher in rural areas. distribution by quintile, up to vocational degrees, is rel- Across analytical domains, the capital presents the atively uniform, with each quintile contributing around highest attainments, followed by aimag centers, soum one fifth. However, for tertiary education that is no centers and lagging further behind the countryside. In longer the case. More than a third of those with a the four domains, around half of the population has higher diploma come from the richest quintile com- finished either 8th grade of secondary or completed pared to less than one tenth from the lowest. Among secondary. However, differences are clearer both at those with university degrees, the gap is even wider, the low and top end of education levels. Between almost fifty percent come from the richest group and three and four tenths of adults have attained educa- less than 5% from the poorest quintile. tion levels beyond secondary school in urban areas and What is the link between poverty status and edu- soum centers, while in the countryside this percentage cation levels? The poor display lower educational plummets to one tenth. The opposite finding is found Table 3.2: Highest education level of adult population by poverty and urban-rural divide Urban Rural National Non-poor Poor Non-poor Poor Non-poor Poor None 1.7 3.5 8.9 9.2 4.5 6.4 Primary 5.6 10.0 19.0 19.3 10.7 14.7 Secondary 8th grade 13.7 32.2 31.6 35.1 20.6 33.7 Complete secondary 31.8 30.0 20.9 20.7 27.6 25.3 Vocational 9.1 10.5 7.2 8.7 8.3 9.6 Higher diploma 18.7 9.7 8.3 5.6 14.7 7.6 University 19.5 4.1 4.2 1.5 13.6 2.8 Total 100.0 100.0 100.0 100.0 100.0 100.0 Source: 2002/03 HIES/LSMS. attainments than the non-poor. More than half of the at lower levels: almost 40 percent of adults in the poor only reach the 8th grade of secondary compared countryside have no more than primary education but to one third of the non-poor. Around 10% of the poor in the rest of the country this share stands at around 10 percent. This accounts for the fact that the only 43 Quintiles are defined in terms of per capita consumption, at the national level and on a unambiguous feature of those with low levels of edu- population basis. Thus each quintile comprises twenty percent of the population. 40 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET Table 3.3: Highest education level of adult population by poverty and gender Men Women Non-poor Poor Non-poor Poor None 4.0 6.9 4.9 5.9 Primary 10.6 14.5 10.8 14.9 Secondary 8th grade 24.8 39.0 16.8 28.9 Complete secondary 26.9 20.6 28.2 29.5 Vocational 8.2 10.1 8.5 9.1 Higher diploma 12.5 5.9 16.7 9.2 University 13.0 3.0 14.2 2.5 Total 100.0 100.0 100.0 100.0 Source: 2002/03 HIES/LSMS. complete tertiary education, while almost three out of institutions than in universities. Three out of five adults ten non-poor achieve the same feat. These patterns with a diploma are female compared to five out of nine are similar in urban areas but they are more even in women in universities. Table 3.3 introduces the pover- rural regions (see Table 3.2). For example, three out of ty element to this comparison. Still non-poor women five rural dwellers complete only up to the 8th grade have better educational levels than non-poor men. of secondary, regardless of their poverty status, and Among the poor, both men and women display similar the share of non-poor with tertiary education is less levels, although women are more likely to finish sec- than double that of the poor. ondary and tertiary education. The gender dimension shows that women have Public spending higher education levels than men. The disparity starts to build up at early stages i.e. whereas three out of ten Mongolia devotes around 9% of its Gross men stop at the 8th grade of secondary, only one fifth Domestic Product to the education sector. What is the of women do so. Women are more likely to finish a pattern of this spending across different levels? Figure tertiary degree and this result is partially driven by a 3.1 plots the cumulative percentage of beneficiaries slightly higher female completion in higher education from public education against the cumulative share of Figure 3.1: Public spending in primary, secondary and university 100 80 Secondary Cum. share of benefits/beneficiaries 60 Primary 40 University 20 0 0 20 40 60 80 100 Cum. percentage of population (rank by per capita consumption) Source: 2002/03 HIES/LSMS. CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET 41 the population ranked by per capita consumption. This Net and gross enrollment rates analysis requires information on unit costs and fre- quency of service for each level of education. Unit A standard approach to measure the access and costs are assumed to be constant within each level, efficiency of the educational system is with enrollment thus the distribution of beneficiaries is identical to the rates47,48. Table 3.4 shows both rates along a number of distribution of public spending in the respective level44. students’ characteristics. In primary school, the net Table 3.4: Net and gross enrollment rates Net enrollment rates Gross enrollment rates Primary Secondary Primary Secondary (8-11) (12-17) (8-11) (12-17) Location Urban 89 83 110 91 Rural 88 64 109 68 Ulaanbaatar 87 82 111 90 Aimag centers 91 85 109 93 Soum centers 93 76 116 81 Countryside 84 54 105 58 Gender Men 89 72 108 79 Women 88 78 111 84 Quintile Poorest 89 65 108 70 Q2 87 70 117 78 Q3 92 78 113 84 Q4 86 82 100 89 Richest 88 85 106 93 National 89 75 109 82 Source: 2002/03 HIES/LSMS. The incidence of public spending is very different in enrollment rate is almost 90%. No major differences each level of schooling, in primary spending is progres- are found across urban and rural areas, gender or con- sive, in secondary is largely neutral and in tertiary edu- sumption quintiles49. Across regions, only the country- cation is highly regressive. This pattern is a reflection of side appears with a relatively low rate. Overall then, the lower attendance of the poor to higher education attendance to primary at the right age does not seem levels. While the poor are more likely to benefit from to be a concern. But the gross enrollment rate stands primary education than the non-poor, their chances become even at the secondary and significantly lower 44 The overall pattern of spending in education is not plotted because of the lack of disag- gregated information on expenditures for primary, secondary and university. at tertiary levels45. The assessment along urban and 45 A further breakdown of secondary into lower (covering the first 4 years) and upper (the rural areas favors the latter. Primary and secondary in last 2) reveals no major differences across these two levels. 46 Figures showing these findings can be found in Appendix D. rural regions is highly pro-poor whereas in urban areas 47 Net enrollment rate is defined as the ratio of the number of children of official school age who are enrolled in school to the population of the corresponding official school age. is neutral. Tertiary education is regressive in rural areas Gross enrollment ratio is the ratio of total enrollment, regardless of age, to the population but less than in urban ones46. of the age group that officially corresponds to the level of education shown. Two levels of education are considered: primary (ages 8 to 11) and secondary (ages 12 to 17). 48 Table D.16 in Appendix D shows a comparison between the enrollment rates calculated from the household survey and the official figures. 49 There might be differences in the quality of education, but it is not possible to perform such analysis with the available data. 42 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET at 109%. This signals that a significant share of stu- Enrollment rates by poverty status and urban-rural dents attending primary are over-aged, which is likely divide are shown in Table 3.5. Net rates for primary to reflect mainly a late entrance to school. In general, school in urban areas are the same regardless of living the further apart are these two rates, the more serious standards but in rural areas they slightly favor the non- is the problem of over-aged students50. poor. Gross rates vary especially in urban areas, where the difference is significant across the poverty levels. Enrollment rates in secondary school reveal anoth- Hence, although poor and non-poor have similar er situation. First, they are much lower than in primary, access to primary, the poor are less likely to attend this suggesting that attendance to secondary school at any level at the right age. In secondary school, the non- age is not as common as in primary. Second, gross and poor have higher net and gross rates, and their differ- net rates are less far apart than in primary. This indi- ences are larger. This points to the fact that a larger cates that a smaller proportion of over-aged students share of the non-poor attends secondary, even though attend secondary and that only children that started they may be over-aged, compared to the poor. In other primary at the correct age continue for further educa- words, attendance of the poor to school drops more tion. Third, both rates differ a lot across different char- than the non-poor after primary. Table 3.5: Enrollment rates by poverty and urban-rural divide Urban Rural National Non-poor Poor Non-poor Poor Non-poor Poor Net enrollment rates Primary (8-11) 89 89 90 86 90 87 Secondary (12-17) 88 75 69 59 81 67 Gross enrollment rates Primary (8-11) 107 115 110 108 108 111 Secondary (12-17) 96 82 73 64 88 73 Source: 2002/03 HIES/LSMS. acteristics of the students. For instance, enrollment is Information by poverty status and gender is dis- significantly higher in urban areas and among children played in Table 3.6. Again very similar net rates are from the richest quintile compared to rural regions and found for primary levels. Gross rates are reasonably sim- children from the poorest quintile respectively. ilar, except among poor women where the problem of Table 3.6: Enrollment rates by poverty and gender Men Women Non-poor Poor Non-poor Poor Net enrollment rates Primary (8-11) 90 88 90 87 Secondary (12-17) 78 63 83 71 Gross enrollment rates Primary (8-11) 110 106 107 116 Secondary (12-17) 86 69 89 77 Source: 2002/03 HIES/LSMS. 50 It can also reflect high repetition rates. CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET 43 misalignment of grade by age is extremely acute. In the instance, among 15 years old, only one out of twenty case of secondary, the non-poor exhibit better net and in urban areas do not attend school compared to one gross rates of enrollment, reflecting that a higher share out of four in rural regions. The gap grows wider when of both male and female non-poor attend secondary, the inspection is done across quintiles. By the time stu- whether it is at the right or at a later age. dents are supposed to finish secondary, the share of those attending school among the poorest quintile is Participation rates almost half than that of the well-off, 48% and 91% respectively. Another way of looking at enrollment is through participation rates51. Figure 3.2 shows these indicators Profile of current students Figure 3.2: Participation rates By gender By urban and rural areas 100 100 WOMEN 80 80 URBAN MEN 60 60 RURAL 40 40 20 20 PERCENTAGE 0 0 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 By poverty status By quintile 100 100 80 80 NON-POOR RICHEST 60 60 POOR POOREST 40 40 20 20 0 0 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 AGE Source: 2002/03 HIES/LSMS. by age and several variables of interest such as gender, What are the characteristics of those currently urban-rural divide, poverty status and consumption attending school? Table 3.7 shows the level of educa- quintile. Overall, women, urban residents, the non- tion in which students are enrolled, the proportion of poor and individuals from the richest quintile are more female students and the share of those attending pub- likely to attend school. A couple of findings hold for lic institutions, by an urban-rural divide and poverty the four comparisons. First, participation rates for pri- status52. Disparities in attendance to education levels mary school (ages 8 to 11) are almost universal, more are patent. On average the non-poor attend higher than 90% on average. Second, by the second or third levels than the poor. For instance, almost half of the year of secondary school differences start to appear, poor attend primary school compared to less than one although remain less than ten points. But by the time third of the non-poor. Only one of twenty poor is students are supposed to be enrolled at the 8th grade 51 Participation rate is defined as the percentage of the population currently attending any of school (or 8th grade of secondary as it is called in educational institution. It excludes pre-school attendance. Mongolia), disparities are quite significant. For 52 Additional information on the characteristics of current students can be found in Tables D.17 and D.18 in Appendix D. 44 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET Table 3.7: Characteristics of current students Urban Rural National Non-poor Poor Total Non-poor Poor Total Non-poor Poor Total Level of education (%) Primary 26.4 38.2 29.9 36.1 51.0 43.1 29.5 44.5 34.9 Secondary 50.0 55.0 51.5 47.7 44.8 46.4 49.3 50.0 49.6 University 22.1 6.1 17.3 14.6 2.4 8.9 19.7 4.3 14.1 Vocational, other 1.5 0.7 1.2 1.6 1.8 1.7 1.5 1.2 1.4 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Female students (%) Primary 52.2 51.2 51.8 44.1 49.4 47.1 49.1 50.2 49.6 Secondary 50.9 54.4 52.1 55.9 54.8 55.4 52.5 54.6 53.2 University 56.6 70.3 58.1 66.8 55.4 65.3 59.0 66.2 59.8 Vocational, other 47.4 18.9 42.7 56.7 41.0 48.9 50.5 34.7 45.5 Total 52.5 53.9 52.9 53.2 51.9 52.6 52.7 52.9 52.8 In public schools (%) Primary 95.3 99.4 96.9 99.0 100.0 99.6 96.8 99.8 98.2 Secondary 96.7 99.7 97.6 99.9 99.6 99.7 97.7 99.6 98.4 University 74.2 67.6 73.5 77.8 81.0 78.2 75.1 71.3 74.6 Vocational, other 91.0 85.5 90.1 100.0 100.0 100.0 94.0 95.9 94.6 Total 91.3 97.5 93.2 96.3 99.3 97.7 92.9 98.4 94.9 Source: 2002/03 HIES/LSMS. going to vocational or tertiary education, while one tions, which points out to the increase of private uni- fifth of the non-poor does so. Attendance to second- versities, mainly in urban areas. Moreover, among ary school is similar, around half of both groups is cur- those attending, the non-poor have more chances to rently attending that level. The same overall trend is benefit from public education, particularly if they live in observed in both urban and rural areas, although the soum centers or in the countryside. former display a significantly higher enrollment in ter- Another aspect that influences school attendance tiary education. is given by the facilities to access the school. Table 3.8 Nationwide, female students account for barely displays the average one-way distance and time to get more than half of the students. The higher the level, to the school from the dwelling of the students. the larger the proportion of women. The exception is Primary and secondary schools are on average less vocational and other education but these levels com- than 2 kilometers away from home. The poor are clos- prise less than 2% of all current students. That pattern er to schools than the non-poor but in terms of time is more evident among the poor, perhaps reflecting spent to get there, both groups spend similar amounts, the fact that poor men sometimes prefer to enter the around 15 minutes. This is explained by the fact that a labor market rather than to continue their studies. larger proportion of the non-poor go to school by car These findings hold generally for urban and rural areas. rather than walking. Public schools are widespread in the country, par- ticularly for primary and secondary. Less than 2% of students attending those levels go to private school. However, the evidence suggests that in urban regions the non-poor are more likely to attend private institu- tions. No differences are found in rural areas. Once students go to tertiary education, two things change. Only a quarter of these students go to public institu- CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET 45 Table 3.8: One-way distance to school facilities Urban Rural National Non-poor Poor Total Non-poor Poor Total Non-poor Poor Total Distance (kms) Primary 1.7 1.2 1.5 1.9 1.4 1.6 1.8 1.3 1.6 Secondary 1.8 1.8 1.8 1.3 1.6 1.4 1.6 1.7 1.7 University 6.3 7.7 6.4 5.2 4.0 5.0 6.0 6.6 6.1 Vocational, other 3.9 1.6 3.6 2.8 1.1 1.9 3.6 1.2 2.8 Total 2.8 1.9 2.5 2.1 1.6 1.8 2.6 1.7 2.3 Distance (minutes) Primary 14 14 14 12 13 13 13 14 13 Secondary 14 15 14 11 13 12 13 14 14 University 25 30 26 23 15 22 25 26 25 Vocational, other 26 15 24 13 24 19 22 22 22 Total 17 16 16 13 13 13 16 15 15 Source: 2002/03 HIES/LSMS. School expenditures spending per student in urban areas is higher than in rural regions. But this hides differences along the What are the levels of household spending in pub- poverty dimension. In primary schools, the urban non- lic education in the country? Figure 3.3 provides infor- poor spend more than their rural counterparts, but the mation on these expenditures per pupil by poverty sta- opposite occurs among the poor. In secondary schools tus and urban-rural divide53. First, non-poor students similar levels are observed. spend more than the poor, on average sixty percent Figure 3.3: Spending per pupil in public primary and secondary Primary National Poor Non-poor Rural Poor Non-poor Urban Poor Non-poor 1,000 2,000 3,000 4,000 5,000 6,000 Tugrug per pupil per month more in both primary and secondary. This holds within urban and rural regions, although the extra spending 53 Only public education was considered because the proportion of private students in pri- mary and secondary is negligible. University was not included because the breakdown into in rural primary schools is only a quarter more. Second, urban and rural areas, and poor and non-poor reduces drastically the number of cases. 46 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET Secondary National Poor Non-poor Rural Poor Non-poor Urban Poor Non-poor 1,000 2,000 3,000 4,000 5,000 6,000 Tugrug per pupil per month Source: 2002/03 HIES/LSMS. Table 3.9 shows the distribution and levels of around one tenth for the richest while they are school expenditure per pupil in public primary and sec- insignificant for the rest. The main component of ondary schools by quintiles. The cost of schooling rises spending is books, around 45%, although for the with the position of the households in the consump- poorest it rises close to 60%. Uniforms and food and tion distribution. Students from the richest quintile other expenses while away from home account for spend more than double than the poorest both in pri- another quarter of total expenditures. mary and secondary. Expenditures in tuition represent Table 3.9: Spending per pupil in public primary and secondary Poorest Q2 Q3 Q4 Richest Total Primary (%) Room rent 0.0 0.1 0.0 0.4 1.0 0.3 Food to pay for room 1.9 6.7 1.0 5.9 2.2 3.5 Transport 1.1 3.3 3.6 6.6 3.4 3.6 Tuition 0.2 2.5 2.3 2.0 9.1 3.3 Books 59.1 45.7 46.0 42.0 42.5 46.9 Uniforms 15.7 13.7 16.4 13.0 12.6 14.3 Expenses away from home 4.8 8.7 12.2 13.6 14.3 10.8 Other 17.2 19.3 18.4 16.6 15.0 17.3 Total (Tugrug/person/month) 2,239 3,052 3,707 4,050 4,790 3,348 Secondary (%) Room rent 0.3 1.0 0.0 0.4 0.4 0.4 Food to pay for room 1.1 2.9 1.4 3.8 2.2 2.4 Transport 3.9 6.2 7.3 6.8 5.8 6.2 Tuition 1.7 0.6 3.3 3.5 10.9 4.5 Books 57.8 46.4 43.1 41.8 35.4 43.4 Uniforms 10.9 11.6 10.1 8.4 7.6 9.5 Expenses away from home 7.6 10.9 18.3 15.9 17.6 14.9 Other 16.7 20.4 16.4 19.5 20.0 18.7 Total (Tugrug/person/month) 2,670 3,607 4,390 4,778 6,004 4,233 Source: 2002/03 HIES/LSMS. CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET 47 3.2. Health older the person gets. This is a result found also in other countries and could be a reflection of many fac- This section examines some features of the health tors such as education or interaction with health sector in Mongolia. It looks first at intermediate indica- providers, which are elements where the non-poor tors such as morbidity rates and utilization of health have usually an advantage over the poor, and make care facilities. Later, other outcomes are analyzed such them more aware of their health conditions. Hence the as spending on health and knowledge about sexually non-poor tend to report more accurately their health transmitted diseases. Finally, reproductive health is problems, while the poor tend to ignore them. Fourth, evaluated by assessing the use of contraceptive meth- seven out of ten people reporting health complaints ods, antenatal care and delivery assistance, and the sought treatment. Although there are differences in incidence of abortion. the likelihood of seeking treatment across age groups, differences as not as large as for reporting health com- Morbidity and treatment plaints, and there is no specific emerging trend. The non-poor sought treatment more often than the poor, One indicator of the health status of the popula- on average three out of four non-poor looked for tion is the morbidity rate. Although not without limita- Figure 3.4: Morbidity rates and probability of seeking treatment 25 90 NON-POOR POOR 80 O -PO R NN O O PO R 20 70 60 15 ith plaints () ) 50 Population (% A ongthose w com 40 10 30 m 5 20 10 0 Less than 10 10-19 20-29 30-39 40-49 50-59 60plus Total 0 Less than10 10-19 20-29 30-39 40-49 50-59 60plus Total Source: 2002/03 HIES/LSMS. treatment compared to three out of five poor people. tions, it can provide useful information54. Figure 3.4 displays these rates along with the probability of seek- Table 3.10 provides information among the popu- ing treatment, conditional on having reported a health lation with health complaints. The most usual types of complaint, by poverty status and age groups. A few complaints are heart, circulatory and respiratory prob- findings are worth highlighting. First, the self-reported lems. The first two are more common among the non- morbidity rate in the month previous to the survey is poor, whereas the latter is more frequent among the very low, only 6% of the population had any health poor. Similar patterns are found across urban and rural complaints. Second, with the exception of the popula- areas as well as by gender. The share of population tion less than 10 years, the older the person, the high- with health complaints that saw their daily activities er the chances of reporting a health complaint. For disrupted is slightly higher than fifty percent. The non- instance, one out of seven individuals in their fifties poor reported more disrupted days in the last month had some health complaint compared to a quarter of that share among those in their twenties. 54 The morbidity rate from the household survey is based on the perception of the respon- dents on their health status during the last month. But people perceiving themselves as healthy or ill will probably vary according to their own and particular circumstances. For Third, the non-poor report more health complaints instance, someone who has been ill for some time might report no health complaints, than the poor, and the differences grow larger the when in fact what has happened is that he has already adapted to his illness. Or it could be the case that the person is not answering by himself, so the respondent might not know whether or not the other household member had a health complaint. 48 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET Table 3.10: Population reporting health complaints National Urban Rural Men Women Non-poor Poor Complaints (% population) 6.4 6.5 6.4 5.2 7.6 7.4 4.6 Among those with complaints (%), Type of health complaint Heart, circulatory 30 33 25 28 31 33 20 Respiratory 28 27 30 28 28 26 36 Digestive 14 14 15 15 14 15 12 Mental 10 10 10 11 9 9 12 Urinary/sexual 14 15 13 11 16 14 13 Other 19 17 21 20 18 19 18 Disrupted daily activities (%) 53 51 55 55 52 54 50 Days in the last month (days) 16 17 14 16 16 17 14 Sought treatment? (%) 71 74 68 70 72 74 63 Visited public facilities 94 92 97 95 94 92 100 Among them, place of treatment was Central hospital, specialized clinic 24 33 12 24 24 25 20 District (aimag) clinic 26 32 18 23 28 28 20 Family clinic 31 14 53 33 30 28 41 Home 18 19 16 20 17 18 19 Attended by a doctor 97 100 94 96 98 97 97 Not sought treatment (%) 29 26 32 30 28 26 37 Reasons for not seeking Not serious enough 58 64 53 60 57 49 76 Treated myself 26 24 28 25 27 34 9 Other 16 13 19 16 16 17 15 Source: 2002/03 HIES/LSMS. than the poor, perhaps reflecting the fact that they can treatment if the person reported a health complaint? afford doing so. Three out of five regarded the complaint as not serious enough and a quarter treated the complaint by them- The extensive health system in the country is selves. This pattern varies by poverty status, three out reflected in the fact, that among the population that of four poor did not take seriously the complaint com- sough treatment, 94% visited public facilities. Urban pared to half of the non-poor. Self-treatment is more dwellers and the non-poor are more likely to visit pri- usual among the non-poor than among the poor, 34% vate providers. Treatment for three out of ten of those and 9% respectively. visiting public facilities was provided in a family clinic, one quarter went to district or aimag clinics, and Spending another quarter to central hospitals or specialized clin- ics. The poor, and especially rural residents, are more Health spending represents 5% in the total con- likely to benefit from family clinics. No differences are sumption of the household. Table 3.11 displays the found by gender. Lastly, almost all the people who levels and patterns of per capita monthly health expen- looked for treatment were attended by a doctor, simi- diture across urban and rural areas, poverty status and lar figures are observed across poverty status and gen- consumption quintiles. The first finding is that the vari- der, yet rural residents are less likely to have been ation in the level of spending is much larger than the attended by a doctor than their urban counterparts. differences on the share of health in consumption. For What are the main reasons for not looking for instance, whereas in both urban and rural areas expen- diture shares are similar, urban spending is forty per- Table 3.11: Per capita monthly health spending (Tugrug) National Urban Rural Non-poor Poor Poorest Q2 Q3 Q4 Richest Outpatient visits 819 1,016 572 1,196 152 107 244 528 625 2,591 Service (%) 80 84 70 80 75 80 74 80 79 81 Transportation (%) 17 12 28 17 22 17 22 18 19 16 Gifts (%) 3 4 2 3 3 3 4 2 2 3 Medicines 904 971 822 1,107 546 503 597 837 1,009 1,577 Public hospital stays 137 132 144 178 65 50 86 170 161 221 Service (%) 72 75 67 72 66 68 70 82 66 69 Transportation (%) 22 16 28 21 28 24 26 13 26 23 Gifts (%) 7 9 4 7 6 8 5 5 7 8 Private hospital stays 52 75 24 74 13 8 16 24 71 141 Service (%) 67 65 75 69 46 50 44 83 70 67 Transportation (%) 31 32 25 28 54 50 56 17 30 29 Gifts (%) 2 3 0 3 0 0 0 0 0 4 Reproductive health a/ 5 7 3 6 3 4 4 3 4 12 Total health spending 1,919 2,204 1,564 2,561 782 672 947 1,564 1,871 4,542 Share in total consumption 5.2 5.5 4.8 5.4 4.5 4.9 4.2 5.1 4.4 6.1 a/ Refers to expenditures related to pregnancies in the last year. Includes the cost of pre-natal consultations and delivery expenditures. Note: Gifts given or bribes paid during the outpatient visits or stays in the hospital. Source: 2002/03 HIES/LSMS. CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET 49 50 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET cent higher than rural expenditure. Second, the non- er gaps in the levels of spending are observed when poor spend more than three times as much as the excluding medicines across poverty status, the non- poor. This pattern is even more evident when looking poor spend more than five times the amount of the across quintiles, the richest 20% of the population poor in both the cost of the service and transportation. have an expenditure almost seven times higher than Knowledge about STD the bottom 20%. Third, spending on self-prescribed medicines represents almost half the total spending on Sexually transmitted diseases (STD) are a major health, and this rises to two thirds among the poor. health concern worldwide and can impose significant The better-off the person in the consumption distribu- burdens to the population, especially to the poor and tion, the less the significance of medicines in health the less educated. The household survey collected spending: among the bottom 20% this figure stands information about knowledge of STD only from people at three quarters of total expenditure, whereas among 15 years and older who were available to answer indi- the top 20% only at one third. vidually such questions. Overall information is available Fourth, excluding spending on medicines, total for 43% of all people 15 years and older (see Table health expenditure can be divided into the cost of the 3.12). Rural residents, women and the non-poor are service per se, transportation and gifts given to the more likely to provide answers in this section of the health providers. The service per se accounts for almost questionnaire. Among the respondents, more than four fifths of total spending, transportation to the nine out of ten have heard about STD, which is an health facility makes up for one fifth, and the remain- extremely high percentage. Knowledge is more com- ing consists of gifts given to the health provider. No mon in urban areas than in rural regions but no differ- major differences in the share patterns are found ences are found by gender or poverty status. What are across poverty status or quintiles. However, in rural the diseases that the population has heard about? regions transportation becomes more important, not Almost nine out of ten with awareness of STD knew only its share increases to almost one third but also the about AIDS, seven out of ten about syphilis and two level of spending is twenty percent higher than in thirds about gonorrhea. Patterns are similar across urban areas. By contrast, in urban areas the service per gender and poverty status. However urban dwellers se is more significant, its share is higher and the are consistently more familiar with any STD than rural amount spent is double that in rural regions. Even larg- residents. Lastly, having only one sex partner or using Table 3.12: Knowledge about STD National Urban Rural Men Women Non-poor Poor Answering by themselves (%) 43 40 47 38 48 45 40 Among those answering by themselves, Heard about STD? (%) 92 95 88 91 92 92 91 Among those that heard (%), Diseases Syphilis 72 74 69 71 72 73 69 Gonorrhea 66 73 59 65 67 69 61 AIDS 88 91 84 90 87 88 87 Others a/ 28 33 20 18 34 29 24 Don't know 4 2 6 4 4 4 5 What do to? (%) One partner 59 64 53 58 60 61 55 Use of condoms 62 68 54 60 63 63 58 Others b/ 44 49 38 38 49 46 39 Don't know 11 7 17 12 10 9 15 a/ Genital warts, condylomata, and others. b/ Abstinence, avoid sex with prostitutes, seek medical treatment, and others. Source: 2002/03 HIES/LSMS. CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET 51 condoms were mentioned by three fifths of the of education (see Table 3.13). The highest the level of respondents as the most known ways to protect them- education attained, the higher the chances of had selves against these diseases. The non-poor, women used contraceptive methods. Among women that have and urban residents are generally better informed ever used contraceptive methods, the share of women about these two methods of protection. currently using them is very high, more than nine out of ten are doing so. Poor women are more likely to be Reproductive health currently using contraceptive methods but no clear pattern emerges by education level. Which methods According to the household survey 63% of all cur- are the most prevalent? Almost half of women use rently married women between 15 and 49 years had IUD, followed by pills and calendar. IUD and injections used contraceptive methods55. Although no major dif- are most frequent among the poor and rural, whereas Table 3.13: Use of contraceptive methods National Urban Rural Non-poor Poor Poorest Q2 Q3 Q4 Richest Ever used contraceptive methods (%) 63 63 62 63 62 64 59 64 61 65 None, primary 45 54 43 35 56 47 61 39 25 42 Sec. 8th grade 55 55 55 52 60 65 48 59 49 49 Complete secondary 66 63 69 67 63 63 64 63 68 72 Vocational, tertiary 68 66 74 69 67 73 62 75 66 68 Among women that had used, Current use of contraceptive methods (%) 93 93 93 92 95 96 94 91 94 91 None, primary 90 94 89 93 88 83 93 100 100 70 Sec. 8th grade 92 94 92 88 96 99 93 82 95 87 Complete secondary 95 94 97 94 97 98 97 94 95 93 Vocational, tertiary 92 92 92 91 94 95 91 91 92 91 Which method? (%) IUD 49 46 52 46 54 55 48 51 45 46 Pill, drugs 17 18 15 18 14 13 20 16 20 16 Calendar 14 17 9 17 7 5 10 17 16 18 Injection 10 6 14 8 13 15 11 9 8 7 Condom 8 10 6 8 7 8 6 6 8 11 Others a/ 3 2 4 3 4 3 5 3 3 2 a/ Includes abstinence, withdrawal, patch, male or female sterilization, diaphragm, and spermicide. Source: 2002/03 HIES/LSMS. Table 3.14: Antenatal care National Urban Rural Non-poor Poor Poorest Q2 Q3 Q4 Richest Pre-natal consultations (%) 98 100 96 97 99 99 97 95 96 100 Number of consults 9 10 8 9 8 9 8 9 9 9 Paid consults (%) 7 9 6 7 8 7 12 4 5 10 Delivery in a hospital (%) 98 100 97 98 98 100 95 97 98 99 Paid delivery (%) 20 25 16 20 19 21 19 17 23 18 Gifts given for the delivery (%) 23 28 19 25 20 22 18 21 26 30 Source: 2002/03 HIES/LSMS. ferences are found by urban-rural divide, poverty sta- 55 It also includes unmarried women living with a partner. The household survey collected infor- tus or quintiles, some distinctions are observed by level mation on all women 15 years and older but the analysis will focus on those between 15 to 49 years. See Table D.19 in Appendix D for information on all women 15 to 49 years old. 52 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET pills and calendar are preferred among non-poor and frequent motive among the non-poor and women in urban women. the richest quintile than for the rest of women. By con- trast, lack of money becomes more important among Antenatal care has reached almost universal levels the poor and particularly among women in the poor- in Mongolia, almost all women who had children in est quintile. the two years previous to the survey consulted a health care professional during their pregnancies (see Table Table 3.15: Abortions National Urban Rural Non-poor Poor Poorest Q2 Q3 Q4 Richest Ever had abortions? (%) 19 25 12 21 14 15 13 20 21 25 None, primary 10 23 8 10 10 12 6 13 11 9 Sec. 8th grade 12 18 9 11 13 17 7 10 10 16 Complete secondary 17 23 9 19 13 12 13 19 25 16 Vocational, tertiary 26 29 21 28 20 16 21 29 24 32 Reasons for abortion (%) Due to health 30 25 42 30 28 27 27 27 28 35 Do not want a child 21 22 18 24 11 12 13 18 22 31 Too soon to give birth again 22 23 19 20 26 20 31 26 21 16 Lack of money 19 21 16 16 29 35 22 23 17 9 Others a/ 8 10 5 9 6 5 8 6 11 9 a/ Attending school, not married, others. Source: 2002/03 HIES/LSMS. 3.14). Urban women are more likely to seek pre-natal 3.3. Labor market treatment than rural women but no differences are observed by poverty status. Nine consultations is the This section briefly reviews some characteristics of average number of pre-natal check-ups among the labor market and employment in the country. It women who sought medical advice, and most of these starts by looking at labor participation rates. Then the consultations are free. Virtually all deliveries are done main sectors of employment and occupations of the in a hospital and some payment was made in a fifth of working population are examined. Finally, unemploy- them. Delivery expenditures and gifts given to the ment rates are analyzed. health provider are more common in urban areas and among the non-poor. Labor force participation A final subject regarding reproductive health is The standard approach to measure labor force that of abortions. One fifth of currently married participation for the economically active population is women between 15 to 49 years reported having had defined by the International Labor Organization56. In an abortion during their life (see Table 3.15). Clearer Mongolia, the labor force participation rate stands at trends appear when analyzing this topic. Urban, non- 65%57. Urban areas have significantly lower participa- poor and more educated women are more likely to tion rates than rural regions, less than three fifths com- have had abortions. For instance, one quarter of pared to three quarters respectively. But this finding is women with vocational or tertiary education reported an abortion compared to one tenth of those with less 56 The labor force is comprised by all people employed or unemployed i.e. those that worked than complete secondary. Three out of ten women in the last week, those that did not work in the last week but had a permanent job, and those that did not work in the last week, did not have a permanent job but looked for said that the main reasons for the abortion were health work. The rest of the population is considered out of the labor force. 57 The labor force participation rate is the ratio between the labor force and the population considerations, and this is particularly important in in the relevant age group. Typically labor force statistics are based on the population rural areas where this share increases to more than between 15 and 64 years old. However, in Mongolia, a different age-cut is used, 16 to 59 for men and 16 to 54 for women. Table D.21 in Appendix D compares the labor force par- four tenths. Not wanting the child is relatively a more ticipation rates according to both definitions. The Mongolian approach shows participa- tion rates higher than the international approach. The table also displays figures from two other sources: the 2003 Labor Force Survey and administrative offices. CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET 53 Figure 3.5: Labor force participation rates 90 80 70 60 50 Percentage 40 30 20 10 0 National Urban Rural Ulaanbaatar Aimag Soum Countryside Poorest Q2 Q3 Q4 Richest centers centers Source: 2002/03 HIES/LSMS. driven by a very high participation rate in the country- ondary, especially among people with tertiary educa- side, whereas in the rest of the country results are tion. quite similar (see Figure 3.5). Across regions, the high- Participation rates by poverty status are shown in est participation is found in the Highlands, where three Table 3.16, which also displays results along the gen- out of four residents participate in the labor force, and der dimension and urban-rural divide. A few results are the lowest in the Central region, where just three out worth noticing. First, the poor are less likely to partici- of five do so58. The analysis by quintile reveals no major pate in the labor market compared to the non-poor, variation among participation rates, except perhaps particularly in urban areas and among women. when comparing the poorest with the richest. Second, men have consistently higher participation Education levels display a more mixed picture, rates are rates than women, which is a result that holds also Table 3.16: Labor force participation rates by poverty status Men Women National Non-poor Poor Total Non-poor Poor Total Non-poor Poor Total Urban 60.8 51.7 58.2 58.7 49.4 56.1 59.7 50.5 57.1 Rural 81.9 76.0 79.5 74.0 70.2 72.5 78.0 73.1 76.0 Total 69.1 64.6 67.6 64.4 59.9 62.9 66.7 62.2 65.2 Source: 2002/03 HIES/LSMS. lowest among those with complete secondary and highest among those with degrees higher than sec- 58 See Tables D.22 and D.23 in Appendix D for more information on labor force participa- tion rates by gender and poverty status. 54 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET across quintiles and education levels. Third, urban Appendix D). The reverse finding is found among the dwellers participate less in the labor force, especially non-poor. Urban and rural areas display the same among men. The gap is substantial for those of national pattern but in the former it is more pro- younger age (less than 25) and for those with less than nounced. A closer look within services in urban areas complete secondary. reveals that a fifth of these jobs is in trade, almost one out of seven is in the public administration and a quar- Employment ter in the education and health sectors. A similar com- position is observed among the poor and the non- Services is the main sector of employment in poor. Mongolia and agriculture ranks second in a very close position, 46% and 43% respectively. However this A second way to classify the employed population pattern is completely different in urban and rural areas is according to whether they are in the private or pub- Figure 3.6: Sector of employment by urban-rural divide and gender 100 80 AGRICULTURE 60 Percentage INDUSTRY 40 SERVICES 20 0 Men Women Urban Men Women Rural Source: 2002/03 HIES/LSMS. (see Figure 3.6). In the capital and aimag centers, serv- lic sector or in a state company. Nationwide, less than ices account for almost three quarters of those three quarters are in the private sector, almost a quar- employed, industry stands for one fifth and the ter in the public sector and not even one out of twen- remaining is involved in agriculture. By contrast, in ty in state companies. This result stands across urban soum centers and the countryside, livestock and farm- and rural regions, although the sector composition ing activities make up for three quarters of employ- shifts, the private share increases to five sixths in rural ment, services for a fifth and industry for the rest. The areas and the non-private rises to two fifths in urban second finding is that differences among men and areas. Being employed in a public institution or in a women are minor within each area, maybe with the state company seems to be correlated with higher liv- exception that in urban areas, men are more likely to ing standards, one third of the non-poor work there be employed in industry and women in services. compared to only a fifth of the poor. What are the differences in employment along the The occupation of those employed provides a poverty dimension? The poor are more likely to be third approach to categorize them. In Mongolia, engaged in agriculture activities, five out of nine do so, herders and farmers are by far the most important and only a third is in services (see Table D.26 in group, they account for two fifths of workers. The CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET 55 Figure 3.7: Occupation of the working population by poverty and urban-rural divide 100 90 80 HERDERS, FARMERS 70 PROFESSIONALS, TECHNICIANS, MANAGERS 60 Percentage 50 SERVICE WORKERS, SALESPEOPLE 40 CRAFT AND RELATED 30 TRADER WORKERS 20 OTHERS 10 0 URBAN URBAN RURAL RURAL NON-POOR POOR NON-POOR POOR Source: 2002/03 HIES/LSMS. three other main groups are service employees and are more related with the poor, particularly in the cap- salespeople, and craft and related trade workers. Each ital and aimag centers, where they account for more one of these groups has a share of about ten percent. than a fifth of their jobs compared to half that for the Figure 3.7 shows a division by poverty status and non-poor. urban-rural divide. In both regions the non-poor have Unemployment more than double chances than the poor to be work- ing in occupations that require more education and According to the household survey the unemploy- skills such as being managers, professionals and tech- ment rate is 6.6%. Urban areas present unemploy- nicians. The likelihood of being employed in services or ment rates significantly higher than rural regions, one in sales is similar regardless of the poverty status, but out of eleven urban dwellers participating in the labor varies by region. Craft and related trader occupations, force was looking for a job compared to less than half which includes miners, carpenters and textile workers, Table 3.17: Unemployment rates by poverty, gender and urban-rural Gender Poverty Total Men Women Non-poor Poor Urban 9.5 8.6 6.8 15.9 9.1 Rural 3.7 4.6 2.6 6.5 4.1 National 6.5 6.7 4.9 10.2 6.6 Source: 2002/03 HIES/LSMS. 56 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET that share for rural residents59. Unemployment is high- is to mitigate the adverse effects of economic, social, est among the youth, those under 25 years have an environmental and physical situations that affect the unemployment rate that is almost double the national household ability to properly cope with them. These figure. Population with tertiary education displays the shocks can be permanent, such as a disability that hin- lowest unemployment rates. Men and women have ders the faculty to work, or temporal, like unemploy- similar unemployment rates at the national level. In ment. They can also have an effect on most members urban areas men show slightly higher rates whereas of a society, such as the occurrence of natural disas- the opposite occurs in rural regions (see Table 3.17). ters, or be specific to a family, like the death of the Finally, the poor have considerably higher unemploy- main earner in the household. Different responses are ment rates than the non-poor, especially in urban designed for each one of them. Broadly speaking there areas. are two types of networks that serve as safety nets: pri- vate safety nets, which involve traditional, and gener- Figure 3.8: Characteristics of the unemployed 70 60 50 40 Percentage 30 20 10 0 Urban Rural Non Poor Men Women 16-24 25-34 35-44 45 plus None, Sec. Complete Vocational, poor primary 8th secondary tertiary grade Source: 2002/03 HIES/LSMS. ally informal, coping mechanisms based on communi- What are the characteristics of the unemployed? ty and family support; and public transfers, which are Figure 3.8 depicts this group. They are mainly urban the response of the state to protect and help those residents, seven out of ten unemployed live in the cap- that are vulnerable. ital and aimag centers, with these two domains con- tributing with equal shares. They are likely to be Mongolia possesses an extensive system of social young, two out of five are under 25 years, and three protection, mainly insurance and assistance60. A large out of ten between 25 and 34 years old. There are no role of the state in providing social welfare is a fairly differences either by gender or by poverty status. As common situation among countries that have made expected, education seems to offer some protection the transition from socialist to market economies. But against unemployment, those with vocational or terti- the population also relies in an informal support net- ary education comprise only a quarter of the unem- work. For instance, herders often exchange animals as ployed. 59 See Table D.24 in Appendix D for a characterization of the population by labor force sta- 3.4. Safety nets tus and along several variables of interest. Tables D.27 and D.28 show unemployment rates by gender and poverty status. 60 Social insurance comprises benefits provided by the state to cover specific risks such as retirement pensions, unemployment or sickness benefits. Social assistance refers to bene- Safety nets typically play a key role in reducing fits intended to provide protection to disadvantaged or vulnerable groups. These include economic insecurity and alleviating poverty. Their aim disability or special pensions, and also family assistance, which is targeted particularly to children. CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET 57 a form of private transfers. This section examines first lar coverage but the former makes up for almost three the extent and relative importance of formal and infor- quarters of the total amount transferred. Third, the mal networks in the country. Then it analyzes the inci- main component of public transfers is the retirement dence of private and public transfers received by the pension. It reaches three out of ten households in the household. Finally, it assesses the correlation between country and represents three quarters of the public transfers and poverty levels. funds. Fourth, nine out of ten Tugrug transferred from private sources to households come from relatives and Extent and importance of transfers friends. Other donors such as non-governmental and religious organizations account for the remaining. Table 3.18 summarizes information on safety nets Fifth, among households benefiting from public trans- in Mongolia according to whether the household is the fers, these make up for a fifth of their consumption. recipient or the donor of transfers and remittances. On the other hand, private transfers represent on aver- Several findings are worth highlighting. First, the age only seven percent of the consumption of house- Table 3.18: Safety nets Households Population Among those receiving/giving with with Average Share of Share of transfers transfers household consumption total transfers (%) (%) transfer (%) (%) (Tugrug per month) Remittances and aid received 68.3 68.3 26,658 21.7 100.0 Remittances and aid 45.2 44.8 10,936 8.0 27.1 Family and friends 36.0 35.3 12,097 8.6 23.9 Others a/ 12.7 12.8 4,632 4.0 3.2 Social welfare 46.1 46.1 28,735 24.2 72.9 State pension 29.1 27.7 33,199 27.4 53.1 Disability pension 8.9 10.1 17,783 15.7 8.7 Survivor pension 3.9 4.3 18,311 18.1 3.9 Maternity benefit 4.8 5.5 6,947 5.7 1.8 Child allowance 5.5 6.6 6,448 6.0 1.9 Others b/ 4.2 4.5 14,948 11.5 3.4 Remittances and aid given 52.2 53.6 7,359 5.0 100.0 Family and friends 46.5 47.6 7,220 4.9 87.3 Others c/ 13.0 13.9 3,749 2.5 12.7 Received or given 79.1 79.4 18,145 * 15.4 * - a/ Includes persons that are neither relatives nor friends, local or state governments, NGO's, and religious organizations. b/ Includes special pension, unemployment benefits, illness payments, funeral payments and other benefits. c/ Includes persons that are neither relatives nor friends, and religious or charitable organizations. * Refers to net transfers: total remittances received by the household minus total transfers given. Source: 2002/03 HIES/LSMS. holds that receive them. Lastly, the principal recipients extent of these networks is impressive, four out of five of remittances given by households are family and households either give or receive some sort of transfer. friends, which receive almost nine tenths of the value Seventy percent of households are recipients, while of these transfers. every other family is a donor. Second, public and pri- vate transfers received by the households have a simi- 58 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET Incidence of the transfers received by tance, which accounts for 30% of public transfers, is the household largely neutral. Private transfers display a regressive pattern too, better-off households capture the most of What is the incidence of the public and private them. Remittances coming from relatives and friends transfers received by households? Figure 3.9 plots the are highly regressive, while other private transfers are mildly regressive. Figure 3.9: Public and private incidence of transfers received by households Public Private 100 100 80 80 60 60 40 40 Cum. share of benefits/beneficiaries 20 20 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Components of Public Transfers Components of Private Transfers 100 100 80 Others 80 Others 60 60 40 40 Disability pension 20 Retirement 20 pension Family and friends 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Cum. percentage of population (rank by per capita consumption) Source: 2002/03 HIES/LSMS. cumulative share of the remittances against the cumu- lative share of the population. Public transfers are Poverty and transfers received by the regressive but there are differences in their composi- household tion. Retirement pensions are highly regressive, the bottom 40% of the population only receives 20% of One of the main objectives of safety networks is to these pensions, whereas the top 20% of Mongolians provide households with the means to avoid economic obtained 40% of these benefits. It shall be kept in insecurity and help some groups that may be vulnera- mind that retirement pensions are not social assis- ble. The correlation between the incidence of poverty tance, they reflect the contributions made by workers and whether or not the household receives a private or to their retirement funds, hence this finding should not public transfer is shown in Table 3.19. Nationwide, sim- be understood as if the state is wrongly targeting these ilar levels of poverty are observed among those living in pensions. The rest of the social insurance and assis- households getting transfers and those in households CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET 59 Table 3.19: Poverty and transfers received by the household Private Public Urban Rural Urban Rural No Yes No Yes No Yes No Yes Headcount 30.0 30.6 43.7 43.0 25.2 35.3 44.8 41.3 (2.3) (2.2) (3.0) (3.2) (1.9) (2.4) (2.7) (3.1) Poverty Gap 8.5 9.9 13.1 13.4 7.4 11.0 12.8 13.8 (0.9) (1.0) (1.2) (1.3) (0.7) (1.0) (1.1) (1.3) Severity 3.4 4.6 5.4 5.7 3.2 4.7 5.1 6.1 (0.4) (0.6) (0.7) (0.7) (0.4) (0.5) (0.5) (0.8) Memorandum items: Household size 4.4 4.3 4.3 4.2 4.3 4.5 4.4 4.1 Dependency ratio (%) 39.2 44.5 44.1 46.8 35.8 48.2 41.0 50.7 Children (% household size) 30.5 28.8 32.7 33.6 35.0 24.0 40.2 23.5 Age of household head 44.7 47.8 42.0 42.9 39.8 52.9 36.7 49.9 Male household head (%) 83.3 75.9 87.2 84.3 88.2 70.5 92.7 77.4 Share below PL (%) 23.4 23.1 32.7 20.9 19.4 27.1 32.5 21.0 Population share 28.2 27.3 27.1 17.5 27.7 27.7 26.2 18.4 Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. that do not get them. The split into an urban-rural of poverty, but the opposite is found in rural regions. divide shows that in the case of private transfers, the However, in rural areas the result is reversed again regional pattern follows the national trend. All poverty when looking at the other two poverty measures. indicators are alike, regardless of whether or not the Retirement pensions household receives private remittances. But for public transfers, there are some regional disparities. For Given the importance of public transfers, the link instance, in urban areas the population living in house- between retirement pensions, by far the most impor- holds that received public transfers display higher levels Table 3.20: Poverty and retirement pensions National Urban Rural No Yes No Yes No Yes Headcount 37.3 33.0 29.7 31.4 45.3 36.2 (1.6) (2.3) (1.9) (2.8) (2.6) (4.0) Poverty Gap 11.3 10.2 9.4 8.8 13.4 12.7 (0.6) (1.0) (0.8) (1.0) (1.0) (2.0) Severity 4.8 4.4 4.2 3.5 5.4 6.1 (0.3) (0.5) (0.5) (0.5) (0.5) (1.2) Memorandum items: Household size 4.4 4.1 4.4 4.4 4.4 3.7 Dependency ratio (%) 38.5 55.1 36.4 53.0 40.8 58.5 Children (% household size) 37.6 15.6 35.5 17.5 39.7 12.7 Age of household head 38.6 58.7 40.1 58.7 37.0 58.5 Male household head (%) 87.7 70.0 84.8 68.9 90.8 71.9 Share below PL (%) 74.6 25.4 30.7 15.8 44.0 9.6 Population share 72.3 27.7 37.3 18.2 35.0 9.6 Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. 60 CHAPTER 3. SOCIAL SECTORS, LABOR MARKET AND SAFETY NET tant component of those transfers, and poverty is also Poverty and the level of transfers examined (see Table 3.20). At the national level, peo- ple living in households receiving these pensions are Another issue to take into consideration is less poor than those who do not receive them. But this whether or not there is an association between the hides different regional patterns. In fact, while in rural incidence of poverty and the level of the transfer areas poverty is significantly lower among those receiv- received by the households. Figure 3.10 displays this ing these benefits, in urban areas there are no differ- relationship for urban and rural areas and with trans- ences in poverty levels between recipients and non- fers measured in per capita net terms61. It clearly shows recipients. A possible implication of this finding is that two findings that hold across both regions. People liv- having a retirement pensioner in soum centers and the ing in households that are net donors, i.e. those with countryside improves the living standards of the other negative net transfers, display a negative correlation household members, which is possibly related to the between the amount of the transfer given and its fact that this is a regular source of income and it does poverty incidence. The more they transfer, the less not depend on seasonal patterns. The distribution of poor they are. By contrast, among the population liv- Figure 3.10: Poverty and net transfers received by the household 40 Rural Urban 25 Headcount (%) 10 0 -5,000 0 10,000 20,000 30,000 Per capita net transfer received per month ( Tugrug) Source: 2002/03 HIES/LSMS. the poor is closely aligned with that of the population, ing in households that are net recipients, there is a around a quarter of the poor live in recipient house- negative association between the amount of transfer holds, this share increases to a third in urban areas but received and its level of poverty. The more they falls to less than a fifth in rural regions. Demographic received, the less poor they are. The implication of indicators show clear trends too. Children represent a these results is that although on average individuals in lower share among those receiving transfers but households receiving transfers are not better-off than dependency ratios are higher, reflecting a larger share those who do not get remittances, among those of elders within the household. Heads are much older receiving, the amount received does matter. and more likely to be female in households benefiting from these remittances. 61 Net transfers are defined as the difference between both private and public transfers received by the family minus all remittances given to other households. REFERENCES 61 References Labor Force Survey with Child Activities Module 2002- 2003, Survey Report of All Four Survey Rounds con- Deaton, Angus, 1997, The Analysis of Household ducted during October 2002 – September 2003, Draft, Surveys: A microeconometric approach to develop- Ulaanbaatar. ment policy, published for The World Bank, The John National Statistical Office of Mongolia and United Hopkins University Press, Baltimore and London. Nations Development Programme, 1999, Living Deaton, Angus and John Muellbauer, 1986, “On Standards Measurement 1998, Ulaanbaatar. measuring child costs: with applications to poor coun- Nutrition Research Center et al., 2003, Final report tries�, Journal of Political Economy, 94, 720-44. of a survey assessing the nutritional consequences of Deaton, Angus and Salman Zaidi, 2002, the Dzud in Mongolia, available at www.un-mongo- “Guidelines for Constructing Consumption lia.mn/reports. Aggregates for Welfare Analysis�, LSMS Working Ravallion, Martin, 1996, “Issues in Measuring and Paper 135, World Bank, Washington, DC. Modeling Poverty�, The Economic Journal, 106, 1328- Griffin, Kenneth, 2001, A Strategy for Poverty 1343. Reduction in Mongolia, available at www.un-mongo- Ravallion, Martin, 1998, “Poverty lines in theory lia.mn/reports. and practice�, LSMS Working Paper 133, World Bank, Hentschel, Jesko and Peter Lanjouw, 1996, Washington, DC. “Constructing an Indicator of Consumption for the United Nations Development Programme Analysis of Poverty: Principles and Illustrations with Mongolia and Government of Mongolia, 2004, Human Principles to Ecuador�, LSMS Working Paper 124, Development Report Mongolia 2003, Urban-Rural World Bank, Washington, DC. Disparities in Mongolia, Ulaanbaatar. Howes, Steven and Jean Olson Lanjouw, 1997, World Bank, 1996, Mongolia Poverty Assessment “Poverty Comparisons and Household Survey Design�, in a Transition Economy, East Asia and Pacific Regional LSMS Working Paper 129, World Bank, Washington, Office, World Bank, Washington, DC. DC. International Monetary Fund, 1999, Country report No. 99/4, available at www.imf.org. International Monetary Fund, 2002, Mongolia: Selected Issues and Statistical Appendix, Country Report No. 02/253, available at www.imf.org. Lanjouw, Peter, Branco Milanovic and Stefano Paternostro, 1998, “Poverty and Economic Transition: How Do Changes in Economies of Scale Affect Poverty Rates of Different Households?�, Policy Research Working Paper 2009, World Bank, Washington, DC. Ministry of Finance and Economy of Mongolia, 2003, Effectiveness and Contributions of Official Development Assistance for Mongolia, “Implementing the Economic Growth Support and Poverty Reduction Strategy�, Mongolia Consultative Group Meeting, Tokyo, Japan. National Statistical Office of Mongolia, 2002, Mongolian Statistical Yearbook, Ulaanbaatar. National Statistical Office of Mongolia, 2003, Internal Migration and Urbanization in Mongolia: Analysis based on the 2000 Census, Ulaanbaatar. National Statistical Office of Mongolia, 2004, A. APPENDIX A: SAMPLE DESIGN AND DATA QUALITY 64 A. APPENDIX A: SAMPLE DESIGN AND DATA QUALITY This appendix provides some details on the gener- months (quarters)62. Each month, the interviewer left a al characteristics of the HIES-LSMS survey, the sample diary with the household to be used to record all types design and an overall assessment of the quality of the of expenditures and consumption deriving either from data. purchases or from own production, gifts, and barter exchanges. A.1. An overview of the HIES-LSMS The LSMS households are a subset of the house- hold interviewed for the HIES: one third of the HIES The 2003 Living Standard Measurement Survey households were contacted again and interviewed on (LSMS) design has the peculiarity of being a sub-sam- the LSMS topics. The subset was equally distributed ple of a larger survey, namely the Household Income among the four quarters. At the planning stage the and Expenditure Survey (HIES). Instead of administer- time lag between the HIES and LSMS interviews was ing an independent consumption module, the LSMS expected to be relatively short. However, for various depends on HIES information on household consump- reasons it is on average of about 9 months, and for tion expenditure. This is why the survey is referred as some households more than one year. Households HIES-LSMS. The HIES-LSMS is the only source of infor- interviewed in the first and second quarter of 2002 mation of income-poverty, and the questionnaire is were generally re-interviewed in March and April designed to provide poverty estimates and a set of use- 2003, while households of the third and fourth quar- ful social indicators that can monitor more in general ter of 2002 were re-interviewed in May, June and July human development, as well as more specific issues on of 2003. The considerable time lag between HIES and key sectors, such as health, education, and energy. LSMS interviews was the main responsible for a consid- Table A.1 provides a summary of the contents of the erable loss of households in the LSMS sample, house- LSMS questionnaire, together with the relevant sec- holds that could not be easily relocated and therefore tions from the HIES. Table A.1: The HIES-LSMS questionnaire HIES (relevant sections) LSMS Food expenditure and consumption 1 General information Non-food expenditure 2 Household roster 3 Housing 4 Education 5 Employment 6 Health 7 Fertility 8 Migration 9 Agriculture 10 Livestock 11 Non-farm enterprises 12 Other sources of income 13 Savings and loans 14 Remittances 15 Durable goods 16 Energy The HIES interviewed 11,232 households which re-interviewed. Due also to some incomplete question- were equally distributed in four quarters over the peri- naires, the number of households that were used for od of one year (from February 2002 to January 2003). the final poverty analysis is 3,308. In fact the HIES collected monthly consumption infor- 62 An important exception is the 'first quarter' made up of February 2002, March 2002 and mation for each household in three consecutive January 2003. A. APPENDIX A: SAMPLE DESIGN AND DATA QUALITY 65 In conjunction with LSMS household interviews A.3. Data quality the NSO also collected a price and a community ques- tionnaire in each selected soum. The latter collected If we exclude the problems encountered in some information on the quality of infrastructure, and basic field operations in the selection of households65, the education and health services. overall data quality is to be considered of good stan- dard. In fact, the data entry program implemented a A.2. The sample design considerable number of in-built consistency checks that alerted the data entry operator whenever some The HIES, and consequently the LSMS, used the 2000 clear inconsistency was found in the data. This helped Census as sample frame. 1,248 enumerations areas were to prevent errors and raised the overall quality of the part of the sample, which is a two-stage stratified random data. At the analysis stage the dataset was also sample. The strata, or domains of estimation, are four: checked for internal consistency and the number of Ulaanbaatar, Aimag capitals and small towns, Soum cen- corrections were overall of a limited amount: excessive tres, and Countryside. At a first stage a number of Primary expenditure values were checked against the paper Sampling Units (PSUs) were selected from each stratum. questionnaire and corrected whenever a data entry In the selected PSUs enumerators listed all the households mistake was found. residing in the area63, and in a second stage households Figure A.1: Population by age group (Census and HIES-LSMS) 14 HIES-LSMS 12 Census 10 Percentage 8 6 4 2 0 0-4 5-9 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70+ Age groups Source: 2002/03 HIES/LSMS and 2000 Census. More generally some comparisons have been made were randomly selected from the list of households iden- to check whether the HIES-LSMS sample is indeed rep- tified in that PSU (10 households were selected in urban resentative of Mongolia. The age-group population dis- areas and 8 households in rural areas)64. The use of this tribution and the sex ratio for these groups have been sampling procedure means that households living in dif- compared with those of the 2000 Census data (see ferent areas of the country have been selected with differ- ing probabilities. Therefore, in order to obtain representa- tive statistics for each of the strata and for Mongolia, it is 63 However, in some instances, there are indications that the listing operations may not have been exhaustive. Probably, in some cases only officially registered households were listed. necessary to use sampling weights. These weights are This might well explain the low proportion of migrants estimated using the LSMS sample (see section 1 of the main report). applied to each household and correspond to the inverse 64 Again, in some cases there might have been some problems in the field operations, as of the probability of selection, calculated taking into there is evidence that in about 10% of the cases households were not selected using infor- mation from the listing operation, but some other criteria. account the sampling strategy. 65 Unfortunately, it is impossible to assess what is the actual implication of the non-compli- ance with the sample selection instruction, but one clear and quantifiable effect is defi- nitely the reduced sample size (3,308 households from the originally planned 3,744). 66 A. APPENDIX A: SAMPLE DESIGN AND DATA QUALITY Figure A.1, and Figure A.2). Overall discrepancies seem to be within an acceptable range. Even though the sam- ple was not designed to provide estimates at the region- al level, population shares of the HIES-LSMS sample are very close to those of the Census (see Table A.2). Figure A.2: Sex ratio by age group (Census and HIES-LSMS) 120 HIES-LSMS Census 100 80 Sex ratio (%) 60 40 20 0 0-4 5-9 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70+ Age groups Source: 2002/03 HIES/LSMS and 2000 Census. Table A.2: Population by geographical region HIES-LSMS Census Urban Rural Total Urban Rural Total West 5.9 11.1 17.0 5.2 12.6 17.8 Highland 7.5 16.5 24.1 7.4 15.6 23.0 Central 7.9 11.6 19.5 8.5 10.2 18.7 East 3.9 5.4 9.2 3.6 5.0 8.5 Ulaanbaatar 30.2 0.0 30.2 32.0 0.0 32.0 National 55.4 44.6 100.0 56.6 43.4 100.0 Source: 2002/03 HIES/LSMS and 2000 Census. B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR 68 B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR Poverty analysis requires three main elements. The first theoretical consideration is that both con- First, a measure of welfare that is both measurable and sumption and income can be approximations to utility, acceptable, and that will allow us to rank all popula- even though they are different concepts. Consumption tion. Second, an appropriate poverty line to be com- measures what individuals have actually acquired, pared against the chosen indicator in order to classify while income, together with assets, measures the individuals in poor and non-poor. Lastly, a set of meas- potential claims of the person. Second, the time peri- ures that combine individual welfare indicators into an od over which living standards are to be measured is aggregated poverty figure. important. If the interest is the long-run, as in a lifetime period, both should be the same and the choice does This appendix explains all the steps involved in the not matter. In the short-run though, say a year, con- construction of the consumption measure, the deriva- sumption is likely to be more stable than income. tion of the poverty line and the poverty measures. Households are able to smooth out their consumption, Section 1 reviews the arguments to choose consump- which may reflect access to credit or savings as well as tion as the preferred welfare indicator. Section 2 information on future streams of income. describes the estimation of the nominal household Consumption is also less affected by seasonal patterns consumption. Section 3 and 4 explain how to arrive to than income, for example, in agricultural economies, an individual measure of real consumption by correct- income is more volatile and affected by growing and ing for differences in location, interview dates and harvest seasons, hence relying on that indicator might demographic composition of households. Section 3 is over or underestimate significantly living standards. concerned with the spatial and temporal price adjust- ment and Section 4 deals with the household compo- On the other hand, there are practical arguments sition adjustment. Section 5 clarifies the derivation of to take into account. First, consumption is generally an the poverty line. Finally, Section 6 presents the poverty easier concept to grasp for the respondents rather measures used in this report. than income, especially if the latter is from self- employment or own-business activities. For instance, B.1. The choice of the welfare indicator workers in formal sectors of the economy will have no problem in reporting accurately their main source of Poverty involves multiple dimensions of depriva- income i.e. their wage or salary. But people employed tion, such as poor health, low human capital, limited in informal sectors or in agriculture will have a harder access to infrastructure, malnutrition, lack of goods time coming up with a precise measure of their and services, inability to express political views or pro- income. Often is the case that household and business fess religious beliefs, etc. Each of them deserves sepa- transactions are intertwined. Besides, as it was men- rate attention as they summarize different components tioned before, seasonal considerations are to be of welfare, and indeed may help policy makers to focus included to estimate an annual income figure. Finally, attention on the various facets of poverty. we also need to consider the degree of reliability of the Nonetheless, often there is a high degree of overlap- information. Households are less reluctant to share ping: a malnourished person is also poorly educated information on consumption than on income. They and without access to health care. may be afraid than income information will be used for Research on poverty over the last years has different purposes, say taxes, or they may just consid- reached some consensus on using economic measures ered income questions as too intrusive. It is also likely of living standards and these are routinely employed that household members know more about the house- on poverty analysis. Moreover, income-based poverty hold consumption than the level and sources of house- indicators are the basis to monitor the first of the hold income. Millennium Development Goals. Although they do not cover all aspects of human welfare, they do capture a B.2. The construction of the consumption central component of any assessment of living stan- measure dards. The main decision is to make the choice between income and consumption as the welfare indi- Creating an aggregate of consumption is also cator. Consumption is the preferred measure because guided by theoretical and practical considerations. it is likely to be a more useful and accurate measure of First, it must be as comprehensive as possible given the living standards than income. This preference of con- sumption over income is based on both theoretical and 66 See Deaton and Zaidi (2002). practical issues66. B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR 69 available information. Omitting some components expenditures and quantities just for food purchases, assumes that they do not contribute to people’s wel- whereas for all other transactions only quantities are fare or that they do no affect the rankings of individu- recorded. Instead of collecting reference prices to als. Second, market and non-market transactions are value this consumption, unit values (expenditures to be included, which means that purchases are not divided by quantities) from purchases were calculated the sole component of the indicator. Third, expendi- and used to estimate the monetary value of non-pur- ture is not consumption. For perishable goods, mostly chased items. Most food items are disaggregated food, it is usual to assume that all purchases are con- enough to be regarded as relatively homogeneous sumed. But for other goods and services, such as hous- within each category, however unit values are not ing or durable goods, corrections have to be made. prices, they will also reflect differences in the quality of Lastly, the consumption aggregate comprises five main the good. To minimize this effect, and to consider spa- components: food, non-food, housing, durable goods tial and seasonal differences too, median unit values and energy. The specific items included in each com- were computed at several levels: by household, cluster, ponent and the methodology used to assign a con- aimag, strata and quarter. Hence if a household pur- sumption value to each of these items is outlined chased a food item, the same unit value would be used below. to value its self-produced and in-kind consumption. If the household did not make any purchase but con- Food component sumed a food item, unit values from the immediate upper level were used to estimate the value of con- The food component can be readily constructed sumption. by simply adding up all consumption per food item, normalized to a uniform reference period, and then Non-food component aggregating all food items per household. HIES records information on food consumption at the household As in the case of food, non-food consumption is a level for 92 items, organized in 10 categories: meat simple and straightforward calculation. Again, all pos- and meat products, milk and dairy products, flour and sible sources of consumption must be included and flour products, vegetables, fruits, sweets, tea, coffee normalized to a common reference period. This com- and beverages, spices, alcohol and tobacco, and meals ponent draws on data from both HIES and LSMS. As it eaten away from home. The information on HIES was was mentioned before, HIES collects information based collected through a diary left to the household for on a diary kept by the households during 3 months. three consecutive months, enumerators went to the Data on an extensive range of non-food items is avail- household at the end of each month and based on the able, 242 items arranged in 14 different groups: cloth- diaries, they filled out the questionnaires. Theoretically ing and footwear for men, women and children, jew- speaking then, the food component uses factual data elry and souvenirs, clothing materials, education, from a 3-month period as opposed to the typical last health, recreation, beauty and toilet articles and servic- week or last month recall period. es, cultural expenses, household goods, durable goods, housing expenditures, transportation, and A few general principles are applied in the con- communication. Even tough most non-food items are struction of this component. First, all possible sources too heterogeneous to try to calculate unit values, HIES of consumption should be included. This means that does gather data on expenditures and quantities for the food component shall comprise not only expendi- most of them, yet only expenses were taken into tures on purchases in the market or on meals eaten account for the estimation of consumption. LSMS away from home but also food that was own pro- records information on education, health, rent of the duced, received as a gift or as part of payment, or dwelling, durable goods and energy expenses, using bartered. Second, ideally only food that was actually mostly a last year recall period. With the exception of consumed, as opposed to food purchases or total durable goods, housing and energy, which will be home-produced food, must enter in the consumption dealt with later, this section covers the consumption of aggregate. HIES provides a detailed account of all all the other non-food items. transactions for each food item and also information on initial and final stocks, therefore an exact measure Practical difficulties arise often for two reasons: of actual consumption can be calculated. the choice of items to include and the selection of the recall period. Regarding the first issue, the rule of Third, non-purchased food items need to be val- thumb is that only items that contribute to the con- ued and included in the welfare measure. HIES collects 70 B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR sumption are to be included. For instance, clothing, health care could be considered as investments. footwear, beauty articles and recreation are included. Differences in access to publicly provided services may Others such as taxes are commonly excluded because distort comparisons across households. If some sectors they are not linked to higher levels of consumption, of the population have access to free or significantly households paying more taxes are not likely to receive subsidized health services, whereas others have to rely better public services. Capital transactions like purchas- on private services, differences in expenditures do not es of financial assets, debt and interest payments correspond to differences in welfare. But there are should also be excluded. The case for lumpy or infre- other factors to take into account. First, health expen- quent expenditures like marriages, dowries, births and ditures are habitually infrequent and lumpy over the funerals is more difficult. Given their sporadic nature, reference period. Second, health may be seen as a the ideal approach would be to spread these expenses “regrettable necessity�, i.e. by considering in the wel- over the years and thus smooth them out, otherwise fare indicator the expenditures incurred by a house- the true level of welfare of the household will probably hold member that was sick, the welfare of that house- be overestimated. Lack of information prevents us to hold is increased when in fact the opposite has hap- do that, so they are left out from the estimation. pened. Third, health insurance can also distort compar- Finally, remittances given to other households are bet- isons. Insured households may register small expendi- ter excluded. The rationale for this is to avoid double tures when some member has fallen sick, while unin- counting because these transfers almost certainly are sured ones bigger amounts. We decided to include already reflected in the consumption of the recipients. health expenses. As with education, excluding them Hence including them would increase artificially living would imply making no distinction between two standards. households, both facing the same health problems, but only one paying for treatment. Besides, a positive Two non-food categories deserve special atten- relationship was found between health expenses and tion: education and health. In the case of education the rest of the consumption aggregate. there are three issues to consider. First, some argue that if education is an investment, it should be treated The second difficulty regarding non-food consump- as savings and not as consumption. Benefits from tion is related with the election of the recall period. The attending school are distributed not simply during the key aspect to consider is the relationship between recall school period but during all years after. Second, there periods and frequency of purchases. Many non-food are life-cycle considerations, educational expenses are items are not purchased frequently enough to justify a concentrated in a particular time of a person’s life. Say weekly or monthly recall period, exceptions being for that we compare two individuals that will pay the same instance toiletries, beauty articles and payment of utili- for their education but one is still studying while the ties. Generally recall periods are the last quarter or the other finished several years ago. The current student last year. For most of non-food categories information might seem as better-off but that result is just related comes only from HIES, thus just one option can be used, to age and not to true differences in welfare levels. data based on a 3-month period, or in other words, a One way out would be to smooth these expenses over quarter. Still, a few non-food categories are available the whole life period. Third, we must consider the cov- from both HIES and LSMS: mainly education and health. erage in the supply of public education. If all popula- Aside from the fact of different recall periods, the other tion can benefit from free or heavily subsidized educa- significant difference to keep in mind is that, for those tion (as it happens in Mongolia) and the decision of two expenses, HIES collects expenditures at the house- studying in private schools is driven by quality factors, hold level, while the LSMS at the individual level. When differences in expenditures can be associated with dif- the reference is the household, questions are normally ferences in welfare levels and the case for their inclu- more aggregated than when the same topics are asked sion is stronger. Standard practice was followed and to each household member. Generally households are educational expenses were included in the consump- known to provide a more accurate account of expenses tion aggregate. Excluding them would make no dis- when they are asked in more detail, which would favor tinction between two households with children in the use of the LSMS modules. That is indeed the case of school age, but only one being able to send them to health expenses, where LSMS records a higher level than school. that of HIES. For education though, expenditures are very similar. Since the LSMS modules might capture bet- Health expenses share some of the features pre- ter the long-term welfare of the household, it was sented for education. Expenditures on preventive decided to use them. B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR 71 Durable goods could be sold for. The implicit rental value can in prin- ciple be used in the consumption aggregate whenever Ownership of durable goods could be an impor- actual rents are not reported. Implicit rents are a hypo- tant component of the welfare of the households. thetical concept though and the estimates may not Given that these goods last typically for many years, always be credible or usable69. An additional complica- the expenditure on purchases is not the proper indica- tion is that almost half of the population lives in gers, tor to consider. The right measure to estimate, for con- for which establishing a rent value appears to be even sumption purposes, is the stream of services that more difficult70. households derive from all durable goods in their pos- Two sets of hedonic housing regressions were run, session over the relevant reference period. This flow of one with the imputed rent value as the dependent vari- utility is unobservable but it can be assumed to be pro- able and the other with the imputed value of the portional to the value of the good. A usual procedure dwelling. The set of independent variables included involves calculating depreciation rates for each type of characteristics of the dwelling such as main type of good based on their current value and age, which in floor, walls, roof, number of rooms, access to water, this case is provided by the LSMS along with the num- electricity, heating, location, etc. This exercise was con- ber of durables owned by the household67. ducted separately for gers, detached houses and apart- The estimation of this component involved three ments. Results show that the value of the dwelling has steps. First, a selection of durable goods was done. The a more consistent correlation with its characteristics LSMS supplies data on 47 durable goods, ranging from and this is intuitively explained by the fact that even home appliances to furniture. However, a third of though households do not rent dwellings, they do buy them were excluded because they were goods used for and build them, so they report more accurately the household businesses or fell under jewelry, dwelling or overall value of the dwelling rather than a hypothetical “other� categories. Second, to calculate implicit depre- rent. A second factor that favors the use of the prop- ciation rates a non-linear regression for each of the erty value is its higher response rate (more than 90% remaining goods was run with the current unit value as of the households reported this value compared to the dependent variable on a constant and the age of around 55% reporting imputed rents), which would the durable. This technique allows also for the possibil- suggest, as it was mentioned before, that households ity of applying multiple depreciation rates, for instance do have a better sense of the property value of their a higher one when the durable good is new. Finally, dwellings. However, the use of property values the stream of consumption is computed by multiplying requires an additional assumption to arrive to an esti- the current value of the good times its depreciation mation of the services provided from housing and that rate, and aggregating these consumptions by house- is the depreciation rate of the dwelling. It was assumed hold. that the annual rates were 3% for houses and apart- ments, and 6% for gers, in other words, houses and Housing apartments will fully depreciate after 33 years and gers after 17 years. Two alternative sets of depreciation Housing conditions are considered an essential rates (2 and 5%, and 4 and 7%) produced very similar part of people’s living standards. Nonetheless, in most poverty measures. Therefore for the consumption developing countries limited or nonexistent housing aggregate, we used the estimated imputed rents rental markets pose a difficult challenge for the estima- derived from the imputed property values as estimates tion and inclusion of this component in the consump- for the flow of services from housing, and otherwise tion aggregate. As in the case of durable goods, the actual rents if available. objective is to try to measure the flow of services received by the household from occupying its dwelling. Energy When a household rents its dwelling, and provided rental markets function well, that value would be the The final non-food component that justified spe- actual rent paid. In Mongolia, housing value for non- cial attention was energy, meaning basically expendi- renters households cannot be determined based upon on information from renters because very few cases 67 Further refinements can be made using the inflation rate and the nominal interest rate. 68 Only 24 out of the 3,308 households. reported renting their dwellings68. Yet the LSMS asked 69 Indeed, after careful inspection, some imputed rents, as well as property values, consid- households for estimates of how much their dwelling ered as outliers were dropped. 70 Although in the definition of household expenditure the System of National Accounts rec- could be rented for and also how much their dwelling ommends the inclusion of imputed rents, in the case of Mongolia several attempts to impute them failed, so that at the present time they are not included. 72 B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR tures on heating and electricity. Mongolia is a country from either central or local systems or simple heating that endures extreme weather conditions, during win- units fueled by firewood, coal or dung. While informa- ter temperatures can easily reach –40 degrees Celsius tion on the former was appropriately captured, the lat- and in the summer 30 plus degrees. While summer ter presented a few complications. The questionnaire may pose fewer inconveniences, winter is indeed a collected data on average purchases (expenditures and serious matter. Winters are long, they last on average quantities) and collection (quantities) per winter and 6 months and with usual below zero temperatures. For non-winter month for those three main sources of instance, average temperatures in January and fuel. First, to value consumption coming from collect- February in the capital are –25C. This means that heat- ed fuel, unit values for each one of the three main ing becomes a basic and essential necessity for house- fuels were applied to their respective collected quanti- holds all over the country, and in some cases it could ties. In urban areas, where most fuel is purchased, unit be a very significant and important component of their values were estimated from actual purchases recorded consumption. in the LSMS following a similar procedure as in the case of valuing food collection. In rural areas tough, Both surveys provide information on energy but where most fuel is collected and there is no market for the LSMS is the one that contains a very comprehen- fuel, the same method will likely overestimate the sive and detailed module, hence it is likely to be much value of consumption (Since no transactions are regis- more accurate than the corresponding HIES section. tered at the cluster level and very few at the aimag Electricity and lighting expenses offered no problems level, unit values are probably drawn from urban for their inclusion in the welfare indicator. Heating was areas). Information on household fuel consumption a different case. Heating is provided to households Table B.1: Maximum monthly fuel consumption during winter Wood Coal Dung (m3) (tons) (kgs) Quantities Ulaanbaatar 1 1.2 800 Aimag centers 2 1.0 1,250 Soum centers 2 0.7 1,500 Countryside 2 0.4 1,800 Expenditure (Tugrug) Ulaanbaatar 10,000 24,000 2,000 Aimag centers 14,000 15,000 3,125 Soum centers 5,000 4,333 3,750 Countryside 2,900 2,200 4,500 Note: Households interviewed for the LSMS appear to have reported fuel consumption by calendar season, i.e. 3 months for winter and 3 months for non-winter, rather than by month. In order to arrive to a monthly figure, the estimated monthly household consumption was compared to the established maximum cut-off point. If either the quantities or expenditure reported by the household were higher than the cut-off points, the reported expenditure would be divided by three. Values for maximum expenditure were derived by multiplying the maximum quantities times their median unit value by stratum (as in valuing fuel collection, unit values in urban strata came from actual purchases recorded in the LSMS whereas those in rural strata were the same as reported in the previous footnote). Non-winter cut-off points were assumed to be 40% of those in winter. Finally, the LSMS recorded information on quantities of wood, coal and dung in kilograms, tons and cubic meters. All quantities were transformed into a single unit for each fuel using the following equivalence: one cubic meter was equivalent to 600kg of wood, 850kg of coal and 400kg of dung. B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR 73 was gathered from several aimag statistical offices and The Paasche price index for the primary sam- unit values were obtained from there71. Second, given pling unit is obtained with the following for- that the recall period was the last year, we needed to mula: make an assumption on the duration of winter and non-winter seasons in order to arrive to a monthly fig- −1 ure. It was assumed that each season lasts on average  n  pik   −1 = ∑ wik    (1) P 6 months. pi    k =1  p0 k     However monthly figures appeared to be too high, especially in the case of purchases. A close look at them revealed that, although questions referred to where wik is the budget share of item k in the pri- a monthly reference period, households apparently mary sampling unit i; reported in many cases seasonal rather than monthly pik is the median price of item k in the primary sam- expenditures. An explanation for this is the fact that pling unit i; people often buy these fuels once or twice for the whole season and it was easier for them to report the pok is the national median price of item k. expenditure as such72. The solution to this data prob- Budget shares were computed from the house- lem consisted in establishing a reference table with hold surveys, as well as food prices. However, it is average and maximum fuel consumption for winter important to note that the household survey does not and non-winter seasons (see Table B.1 above). These collect information on prices themselves, but on cut-off points allowed us to distinguish cases in which implicit prices, obtained dividing expenditure by quan- the household reported seasonal instead of monthly tities purchased. Inevitably, implicit prices represent figures. Table B.1 was set in consultation with aimag also differences in quality of the item purchased. statistical offices and considering the different sources Quality differences are generally considered acceptable of heating used by the household. for food items, but are more problematic for non-food items, which are likely to be less homogenous in B.3. Price adjustment nature (also questions on non-food items are less detailed than those for food ones). On the contrary, Mongolia shows remarkable seasonal price differ- both the soum and aimag centers questionnaires col- ences, especially for food items, with prices in the lected information on actual prices and on much more spring (April to June) commonly 10% higher than in well defined items. Nonetheless, the soum center price autumn. At the same time, across all seasons there are questionnaire was not always of the desired quality, also regional price differences. In particular in some of the items show price differences that are too Ulaanbaatar, prices are relatively higher than in the rest large, suggesting that in such cases prices of items of of the country. Therefore, in order to properly measure rather different quality were collected. This is to be living standards, expenditure values need to be cor- expected in fragmented and incomplete markets, rected for such differences using some price indexes. A where the enumerator might have been compelled to price index is made of two components: prices and substitute items that were not found. budget shares that attach the proper weights to prices. It follows that differences of price indices can come Instead, the aimag centers prices appear to be both from prices and consumption patterns. more accurate because they are the result of a perma- nent activity, prices are collected in the same outlets The household survey provides information on and with more precise guidelines about the type of budget shares as well as information on implicit prices item for which the price is sought. Both for the soum (unit values) paid by the household. Moreover, togeth- and the aimag price questionnaire, information is not er with the household survey the NSO also conducted available for each household, but is representative a price questionnaire in soum centers collecting infor- respectively for the soum or aimag. However, it is like- mation on about 250 prices, and regularly collects ly that both within the same soum, and indeed the prices for about 140 items in all aimag centers. All this provides a rich source of information, which was used 71 Unfortunately, this was not a proper and systematic survey covering all areas, so in order to construct a Paasche price index at the cluster level. to minimize the potential bias, median unit values by stratum were considered for valua- tion purposes. These values were as follows: one cubic meter of wood was Tugrug 2,500 In each cluster generally between 8 and 10 households in soum centers and 1,450 in the countryside; one kilogram of dung in both strata was 2.5; and one ton of coal was 6,500 in soum centers and 5,500 in the countryside. have been interviewed and prices they face as well as 72 The same situation arose in at least another recent LSMS, so it seems that there is a les- consumption patterns tend to be very similar. son to be learned that goes beyond the case of Mongolia. 74 B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR same aimag, prices of non-food items show a relative- that value to each household member. Common prac- ly small variation. This is because price differences for tice to do this is to assume that all members share an these items are mainly due to transportation costs equal fraction of household consumption, however as (from Ulaanbaatar), and the soum/aimag price already it will be explained later that is a very particular case. captures most of such costs. Two types of adjustments have to be made to cor- More problematic is the fact that while for food rect for differences in composition and size. The first items budget shares are immediately matched with relates to demographic composition. Household mem- ‘prices’, when information on prices is taken from the bers have different needs based mainly on their age price questionnaires, the correspondent budget share and gender, although other characteristics can also be needs to be properly identified, and in some cases, considered. Equivalence scales are the factors that where such correspondence does not exist, key items reflect those differences and are used to convert all are considered to be representative for the budget household members into “equivalent adults�. For shares of similar items. For instance, in the case of instance, children are thought to need a fraction of transportation expenditure, the only price that was what adults require, thus if a comparison is made used was the one of petrol (petroleum A-76). between two households with the same total con- Table B.2: Cluster Paasche Index by quarter and analytical domain Quarter Annual I II III IV average Ulaanbaatar 1.08 1.07 1.06 1.09 1.07 Aimag centers 0.98 0.99 0.92 0.94 0.96 Soum centers 0.99 0.99 0.93 0.96 0.96 Countryside 1.03 1.06 0.92 0.94 0.98 National 1.02 1.03 0.96 0.99 1.00 Source: 2002/03 HIES/LSMS. The average values of the price index by quarter sumption and equal number of members, but one of and analytical domains are reported in Table B.2. The them has children while the other is comprised entire- index confirms that living costs in Ulaanbaatar are ly by adults, it would be expected that the former will higher than anywhere else in the country and it also have a higher individual welfare than the latter. shows the seasonality effects: the index is higher in the Unfortunately there is no agreement on a consistent first and second quarters and then decreases in the fol- methodology to calculate these scales. Some are based lowing quarters. on nutritional grounds, a child may need only 50% of B.4. Household composition adjustment the food requirements of an adult, but is not clear why the same scale should be carried over non-food items. The final step in constructing the welfare indicator It may very well be the case that the same child involves going from a measure of standard of living requires more in education expenses or clothing. defined at the household level to another at the indi- Others are based on empirical studies of household vidual level. Ultimately the concern is to make compar- consumption behavior, although with more analytical isons across individuals not households. Consumption grounds, they do not command complete support data are collected typically at the household level either73. (usual exceptions are health and education expenses), so computing an individual welfare measure generally is done by adjusting total household consumption by 73 See Deaton and Muellbauer (1986) or Deaton (1997). the number of people in the household, and assigning B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR 75 The second adjustment focuses in the economies with respect to families with no kids or with a small of scale in consumption within the household. The number of members respectively. It is important then motivation for this is the fact that some of the goods to conduct sensitivity analysis to see how robust the and services consumed by the household have charac- poverty measures and rankings are to different teristics of “public goods�. A good is said to be public assumptions regarding child costs and economies of when its consumption by a member of the household scale. Appendix C will show those results. does not necessarily prevent another member to con- sume it too. Examples of these goods could be hous- B.5. The poverty line ing and durable goods. For example, one member watching television does not preclude another for The poverty line can be defined as the monetary watching too. Larger households may spend less to be cost to a given person, at a given place and time, of a as well-off as smaller ones. Hence, the bigger the share reference level of welfare (Ravallion, 1998). If a person in total consumption of public goods, the larger the does not attain that minimum level of standard of liv- scope for economies of scale. On the other hand, pri- ing, she will be considered as poor. But setting pover- vate goods cannot be shared among members, once ty lines could be a very controversial issue because not they have been consumed by one member, no other only people disagree on what “minimum� is but also can. Food is the classic example of a private good. It is of its eventual effects on monitoring poverty and poli- often pointed out that in poor economies, food repre- cy making decisions. sents a sizeable share of the household budget and The poverty line will be absolute because it fixes a therefore in those cases there is little room for given welfare level, or standard of living, over the economies of scale. domain of analysis. This guarantees that comparisons Both adjustments can be implemented using the over time or across individuals will be consistent e.g. following approach: two persons with the same welfare level will be treat- ed the same way regardless of the location where they AE = (A + αK)θ live. Second, the reference utility level is anchored to where AE is the number of adult equivalents of certain attainments, generally nutritional ones, for the household, A is the number of adults, K the num- instance, obtaining the necessary calories to have a ber of children, α is the parameter that measures the healthy and active life. Finally, the poverty line will be relative cost of a child compared to an adult and θ rep- set as the minimum cost of achieving that require- resents the extent of the economies of scale74. Both ment. parameters can take values between zero and one. It is The Cost of Basic Needs method was employed to been reported that in developing countries, children estimate the nutrition-based poverty line. This are relatively cheaper than adults, perhaps with values approach calculates the cost of obtaining a consump- of α as low as 0.3 while in developed ones values are tion bundle believed to be adequate for basic con- closer to one75. At the same time, in poorer economies sumption needs. If a person cannot afford the cost of food is often the most important good in the house- the basket, it will be considered to be poor. First, it hold consumption, and given that is a private good, shall be kept in mind that the poverty status focuses on the budget share of public goods is limited and so is whether the person has the means to acquire the con- the scope for economies of scale, perhaps with θ close sumption bundle and not on whether its actual con- to 1, whereas in richer countries around 0.75. sumption met those requirements. Second, nutritional It was mentioned that standard practice is to use references are used to set the utility level but nutrition- a per capita adjustment for household composition al status is not the welfare indicator. Otherwise, it will and that is also followed here. This is a special case of suffice to calculate caloric intakes and no costing the above formulation, it happens when α and θ are would be necessary. Third, the consumption basket set equal to one, so all children are treated as if their can be set normatively or to reflect prevailing con- cost relative to adults were the same and there is no sumption patterns. The latter is undoubtedly a better room for economies of scale. In other words, all mem- alternative. Lastly, the poverty line comprises two main bers within the household consume equal shares of components: food and non-food. the total consumption and costs increase in proportion to the number of people in the household. In general, 74 Actually, since the elasticity of adult equivalents with respect to "effective size" A+ K isθ, per capita measures will underestimate the welfare of the measure of economies of scale is 1-θ households with children as well as larger households 75 Deaton and Zaidi (2002). 76 B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR Food component it was possible to assign a caloric factor. Fourth, medi- an unit values were derived in order to price the food The first step in setting this component is to deter- bundle. Unit values were computed using only transac- mine the nutritional requirements deemed to be tions from the reference group. Again, this will capture appropriate for being healthy and able to participate in more accurately the prices faced by the poor. Fifth, the society. Clearly, it is rather difficult to arrive to a con- average caloric intake of the food bundle was estimat- sensus on what could be considered as a healthy and ed, so the value of the food bundle could be scaled active life, and hence to assign caloric requirements. proportionately to achieve 2,100 calories per person Common practice is to establish 2,100 calories per per- per day. For instance, the average daily caloric intake son per day as the reference for energy intake. Second, of the bottom 40% of the population in Mongolia was a food bundle must be chosen. In theory, infinite food around 1,345 calories per person and the daily value of bundles can provide that amount of calories. One way the food bundle was Tugrug 307 per person. Hence out of this is to take into consideration the existing the value of the daily poverty line is Tugrug 480 ( = food consumption patterns of a reference group in the Tugrug 307 x 2,100 / 1,345 ) per person. Table B.3 country. It was decided to use the bottom 40% of the shows the caloric contribution of the main food cate- population, ranked in terms of real per capita con- gories as well as the their respective share in the cost sumption, and obtain its average consumed food bun- of the food poverty line76. Table B.3: Food bundle per person per day by main food groups Caloric intake Value Calories Share Tugrug Share (%) (%) Meat and meat products 417 20 197 41 Milk and milk products 152 7 81 17 Flour and flour products 1,304 62 127 26 Vegetables 52 2 26 5 Fruits 5 0 4 1 Candy, sugar 92 4 21 4 Tea, coffee, beverages 9 0 9 2 Seasonings 70 3 15 3 Total 2,100 100 480 100 Source: 2002/03 HIES/LSMS. dle. It is better to try to capture the consumption pat- Non-food component tern of the population located in the low end of the welfare distribution because it will probably reflect bet- Setting this component of the poverty line is far ter the preferences of the poor. Hence the reference from being a straightforward procedure. There is con- group can be seen as a first guess for the poverty inci- siderable disagreement on what sort of items should dence. Third, caloric conversion factors were used to be included in the non-food share of the poverty line. transform the food bundle into calories. The main However, it is possible to link this component with the source for these factors was the Food Research Center, normative judgment involved when choosing the food which is a unit of the Ministry of Health of Mongolia. component. Being healthy and able to participate in Alcohol, tobacco and meals eaten outside the house- society requires spending on shelter, clothing, health hold were excluded from this calculation, the former care, recreation, etc. A usual practice is to scale up the because they can be regarded as non-essential and the food poverty line to allow for basic non-food items, latter because it is very difficult to approximate caloric which can be done by dividing the food poverty line by intakes for them. For all of the remaining food items, 76 A more detailed table by food item is provided at the end of the annex. B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR 77 some estimation of the budget share devoted to food. those households whose food expenditures lie within The advantage of this is that the non-food component plus and minus one percent around the poverty line. can be based on the prevailing consumption behavior The same exercise is then repeated for households of a reference group and no pre-determined non-food lying plus and minus two percent, three percent, and bundle is needed. up to ten percent. Second, these ten mean food shares are averaged and that will be the final food share of Table B.4: Monthly poverty lines per person Lower poverty line Upper poverty line Tugrug % Tugrug % Food 14,386 58 14,386 44 Non-food 10,357 42 17,984 56 Total 24,743 100 32,370 100 Source: 2002/03 HIES/LSMS. the poverty line. Finally, the non-food component can The initial step is to choose a reference group. be easily estimated77. Table B.4 displays the food and There are two ways in which this is usually done. The non-food components of both poverty lines. first is to determine the food share of the population whose food expenditures are equal to the food pover- The lower poverty line is applied throughout the ty line. The rationale behind is that if an individual report, while poverty estimates with the upper poverty spends in food what was considered appropriate for line are presented in Table C.3. being healthy and maintaining certain activity levels, it can be assumed that this person has also acquired the B.6. Poverty measures necessary non-food items to support its lifestyle. The resulting poverty line is called the upper or higher Even though there is an extensive literature on poverty line. The second way to calculate the food poverty measurement, attention will be given to the share is to estimate it from the population whose total class of poverty measures proposed by Foster, Greer expenditures are equal to the food poverty line. The and Thorbecke (1984). This family of measures can be justification is that these people have substituted basic summarized by the following equation: food needs in order to satisfy some non-food needs, therefore that amount can be interpreted as the mini- α  z − yi  q mum necessary allowance for non-food spending. Pα = (1 / n)∑   Two different procedures to calculate the non- i =1  z  food component can be proposed. One relies on econometric techniques to estimate the Engel curve, where α is some non-negative parameter, z is the e.g. the relationship between food spending and total poverty line, y denotes consumption, i represents indi- expenditures. Another is to use a simple non-paramet- viduals, n is the total number of individuals in the pop- ric calculation as suggested in Ravallion (1998). The ulation, and q is the number of individuals with con- advantages of the latter is that no assumptions are sumptions below the poverty line. made on the functional form of the Engel curve and The headcount index (α=0) gives the share of the that weights decline linearly around the food poverty poor in the total population, i.e. it measures the per- line i.e. the closer is the household to the food pover- centage of population whose consumption is below ty line, the higher its weight. This procedure was used the poverty line. This is the most widely used poverty to determine the non-food components for the upper measure mainly because it is very simple to understand and lower poverty lines. For instance, in the case of the upper poverty line, first food shares are estimated from 77 For the lower poverty line, the same can be applied but taking instead households whose food spending is close to the food poverty line. 78 B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR and easy to interpret. However, it has some limitations. that even if the number of the poor is the same, but It takes into account neither how close or far the con- there is a welfare reduction in a poor household, the sumption levels of the poor are with respect to the measure of poverty should increase. And fourth, the poverty line nor the distribution among the poor. The severity of poverty will also comply with the transfer poverty gap (α=1) is the average consumption shortfall axiom: it is not only the average welfare of the poor of the population relative to the poverty line. Since the that influences the level of poverty, but also its distri- greater the shortfall, the higher the gap, this measure bution. In particular, if there is a transfer from one overcomes the first limitation of the headcount. Finally, poor household to a richer household, the degree of the severity of poverty (α=2) is sensitive to the distribu- poverty should increase78. tion of consumption among the poor, transfers among Finally, along the report all poverty measures are the poor will leave unaffected the headcount or the shown with their respective standard errors. Since poverty gap but will increase this measure. It applies a those estimations are based on surveys and not on relatively higher weight to the largest poverty gaps. census data, standard errors must reflect the elements These measures satisfy some convenient proper- of the sample design i.e. stratification and clustering79. ties. First, they are able to combine individual indica- Ignoring them will risk, when carrying out poverty tors of welfare into aggregated measures of poverty. comparisons, mixing up true population differences Second, they are additive in the sense that the aggre- with differences in sampling procedures. Appendix E gate poverty level is equal to the population-weighted shows confidence intervals and sample-design effects sum of the poverty levels of all subgroups of the pop- for the poverty measures when correlated with main ulation. Third, the poverty gap and the severity of variables of interest. poverty satisfy the monotonicity axiom, which states 78 Both the monotonicity and transfer axioms were formulated by Sen (1976). 79 See Howes and Lanjouw (1997) for a detailed explanation. B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR 79 Table B.5: Food bundle per person per day Unit Calories Daily Daily Price Daily per unit quantity calories per value of (kcals) consumed provided unit the food (units) (kcals) (Tugrug) bundle a/ (Tugrug) Meat and meat products 101 Mutton kg 1,083 0.054 58 918 77 102 Beef kg 1,531 0.027 41 969 41 103 Goat kg 1,057 0.013 13 682 14 104 Horse kg 911 0.016 15 593 15 105 Camel kg 1,026 0.002 2 701 2 106 Dried meat kg 4,292 0.006 26 2,336 22 107 Pork kg 3,554 0.000 0 1,823 0 108 Chicken kg 1,908 0.000 0 1,586 0 109 Hunting meat kg 1,788 0.001 2 414 1 110 Fish kg 821 0.000 0 833 0 111 Animal interior kg 1,058 0.012 13 429 8 112 Interior fat kg 8,973 0.009 83 889 13 113 Sausage kg 2,507 0.001 2 1,765 3 114 Canned meat kg 2,250 0.000 0 1,907 0 115 Canned fish kg 1,966 0.000 0 1,813 0 116 Egg unit 79 0.006 1 94 1 117 Dry egg kg 5,441 0.000 0 1,902 0 118 Other meat kg 2,456 0.005 11 189 1 Milk and milk products 201 Milk lt 671 0.063 42 454 45 202 Yogurt lt 564 0.013 7 396 8 203 Dried curds kg 4,908 0.002 12 1,462 6 204 Horse milk lt 487 0.004 2 542 4 205 Cheese kg 4,733 0.001 4 954 1 206 Skim kg 5,788 0.003 15 2,377 9 207 Cream kg 2,495 0.000 1 1,347 1 208 Butter kg 5,323 0.002 9 1,488 4 209 Other diary products kg 2,566 0.000 1 859 0 210 Dried milk kg 3,293 0.001 2 2,897 2 211 Condensed milk lt 4,850 0.001 3 668 1 212 Other kg 3,244 0.000 0 689 0 Flour and flour products 301 Flour, highest grade kg 3,617 0.014 50 305 7 302 Flour, 1st grade kg 3,250 0.126 410 291 57 303 Flour, 2nd grade kg 3,474 0.056 194 254 22 304 Other flour, barley kg 3,742 0.001 3 414 1 305 Pasta kg 3,732 0.002 9 669 2 306 Bread 670 gr 1,590 0.030 48 240 11 307 Bakery kg 4,050 0.004 18 752 5 308 Biscuit kg 2,508 0.001 1 1,371 1 309 Millet kg 3,513 0.002 8 360 1 310 Rice kg 3,447 0.026 91 421 17 311 Other grain kg 3,455 0.000 1 735 0 312 Other cakes, etc kg 3,097 0.000 0 3,122 1 (table continues on following page) 80 B. APPENDIX B: THE CONSTRUCTION OF THE WELFARE INDICATOR Table B.5: Food bundle per person per day Unit Calories Daily Daily Price Daily per unit quantity calories per value of (kcals) consumed provided unit the food (units) (kcals) (Tugrug) bundle a/ (Tugrug) Vegetables 401 Potato kg 877 0.031 27 249 12 402 Cabbage kg 140 0.004 1 359 2 403 Carrot kg 224 0.003 1 398 2 404 Turnip kg 208 0.004 1 449 3 405 Onion kg 336 0.005 2 448 4 406 Garlic kg 1,108 0.000 0 959 0 407 Tomato kg 260 0.000 0 939 0 408 Cucumber kg 142 0.000 0 1,006 1 409 Noodles made of potato flour kg 3,272 0.000 1 878 1 410 Pickled cucumber kg 164 0.000 0 1,410 0 411 Canned vegetable salad kg 1,121 0.000 0 1,377 1 412 Other kg 714 0.000 0 1,383 0 Fruits 501 Apple kg 468 0.003 1 540 2 502 Grape kg 1,812 0.000 1 1,151 1 503 Dried fruit kg 2,721 0.000 0 1,829 0 504 Jam kg 2,867 0.000 0 1,322 0 505 Stewed fruit kg 814 0.000 0 1,715 0 506 Peanuts kg 5,980 0.000 1 1,376 0 507 Fruit kg 400 0.000 0 1,143 0 508 Other fruit kg 504 0.000 0 920 0 Candy, sugar 601 Sugar kg 3,992 0.011 43 610 10 602 Lump sugar kg 3,996 0.001 2 1,123 1 603 Caramel, domestic kg 3,697 0.001 5 1,538 3 604 Caramel, imported kg 3,837 0.002 7 1,641 5 605 Chocolate kg 5,481 0.000 2 3,006 2 606 Other marmalades kg 2,644 0.000 0 1,544 0 Tea, coffee, beverages 701 Green tea kg 1,076 0.004 4 1,090 6 702 Tea gr 1 0.033 0 6 0 703 Coffee gr 1 0.019 0 8 0 704 Beverage lt 343 0.004 1 358 2 705 Fruit juice lt 488 0.000 0 1,019 0 706 Other beverages lt 869 0.000 0 442 0 Seasonings 901 Salt kg 0 0.012 0 189 4 902 Vegetable oil lt 8,991 0.005 42 1,170 9 903 Mayonnaise kg 6,258 0.000 1 2,654 1 904 Vinegar, sauce gr 1 0.770 1 1 1 905 Other gr 4 0.097 0 5 1 TOTAL PER DAY 1,345 480 a/ Values are already scaled up to achieve 2,100 calories per person per day i.e. the daily calories provided times the price per calory (price per unit divided by calories per unit) times the scaling caloric factor (2100/1358). Source: 2002/03 HIES/LSMS. C. APPENDIX C: SENSITIVITY OF POVERTY ESTIMATES TO CRUCIAL HYPOTHESES 82 C. APPENDIX C: SENSITIVITY OF POVERTY ESTIMATES TO CRUCIAL HYPOTHESES As discussed in Appendix B in the process of esti- requirement for children is lower than the one for mating poverty, a number of assumptions and estima- adults for what concerns food, and other non-food tions have been made. Since some of these adjust- expenditure. Taking all this into consideration, reason- ments involve an unavoidable degree of arbitrariness, able values of α are unlikely to be below 0.5. it is important to test how sensitive the final results are The groups of households considered in this analy- to these assumptions. In particular, we want to analyze sis are: the effect of: 1) Elderly households (households composed exclu- 1) Different hypotheses of economies of size and sively by elderly people: women more than 54 and equivalence scale; men more than 59); 2) The exclusion of heating and imputed rents from 2) Households with high child ratio (more than aver- the consumption aggregate. age number of children, children are those aged less than 16); C.1. Alternative hypotheses of equivalence 3) Female-headed households; scale and economies of size 4) Households with high dependency ratio (higher than average dependency ratio); As discussed in section IV of appendix B, it is important to test whether the poverty profile is very 5) Households with no children; sensitive to the different possible adjustments of 6) Households with 1 child; household size, taking into account equivalence scales 7) Households with 2 children; and economies of size. The formula presented earlier 8) Households with 3 children or more. was as follows: These groups of households are used to evaluate AE = (A + αK)θ the changes in their relative levels of poverty when giv- However, it is also possible to consider the same ing different values to α, but keeping the overall head- effect considering α single parameter and express the count ratio equal to 36%. Table C.1 shows the results adult equivalent household size as follows: of such analysis considering values of α from 0.5 to 1. Although as α decreases, the head count increases sig- AE = (Household size)α nificantly for elderly households and households with Both higher economies of size and larger differ- no children, poverty rankings of these groups remain ences in needs between people of different age (equiv- the same. Moreover, it is worth to remember that alence scale parameters) will have the effect of reduc- households with only elderly people represent less ing the parameter α. This approach has been used by than 2% of the population. This result suggests that Lanjouw, Milanovic and Paternostro (1998), and it is poverty estimates within these groups are no particu- applied here to test for the effect of different values of larly sensitive to the different values of α, at least with- α on the ranking of the main demographic groups, in the considered range. The only exception is female- where it is likely that different adjustments might have headed households, where as α decreases, they an impact. In fact, these tests want to assess whether become relatively poorer than households with high different adjustments of household size affect the con- dependency ratio and high child ratio. These results are clusions reached in generating the poverty profile of reported also in two graphs Figure C.1 and Figure C.2. relevant population groups. These groups are those The same analysis can be repeated considering with high household size and with members that other groups based on other characteristics, for might have consumption needs lower than adults, instance geographical areas, but in this case rankings namely children and elderly people. are even less affected by different hypothesis of α, The source of potential economies of size is main- because there are no substantial differences in demo- ly related to the share of consumption expenditure for graphic characteristics between the various geograph- public goods or quasi-public goods: housing (rent), ical areas (strata and regions). durables, and utilities. These consumption subgroups represent respectively 5%, 1% and 9% of total con- sumption, altogether 15% of total consumption. In Mongolia it is also likely that different needs of chil- dren versus adults may be important. In fact, education is still subsidized and it is reasonable to believe that the C. APPENDIX C: SENSITIVITY OF POVERTY ESTIMATES TO CRUCIAL HYPOTHESES 83 Table C.1: Headcount within different groups of households making different assump- tions on the extent of economies of scale á = 0.5 á = 0.6 á = 0.7 á = 0.8 á = 0.9 á = 1 % of pop. Poor 36.1 36.1 36.1 36.1 36.1 36.1 Elderly households 32.4 24.8 15.7 9.6 6.5 1.5 1.7 Female-headed households 48.6 47.1 45.8 44.9 44.6 43.8 14.2 High dependency ratio 43.2 42.7 42.1 41.8 41.8 41.6 51.5 High child ratio 42.6 42.5 42.5 42.6 42.8 42.9 62.0 No. children 23.1 20.6 18.5 16.7 15.8 14.7 17.1 1 child 27.1 27.2 26.4 25.9 25.0 24.0 25.4 2 children 34.8 34.2 33.8 33.7 33.1 32.9 28.5 3+ children 53.0 54.9 57.1 58.8 60.7 62.5 29.0 Av. hhsize for the poor 4.6 4.8 5.0 5.2 5.3 5.4 Av. hhsize for the non-poor 4.2 4.1 4.0 3.9 3.9 3.9 % of children in poverty 42.7 43.0 43.1 43.4 43.7 43.8 % of elderly in poverty 32.0 29.5 26.8 24.2 23.3 21.5 Source: 2002/03 HIES/LSMS. Figure C.1: Headcount within different groups of households making different assump- tions on the extent of economies of scale 50 40 Poor Headcount (%) 30 Elderly HH Female headed HH 20 High Dep. Ratio High Child Ratio 10 0 0.5 0.6 0.7 0.8 0.9 1 Economy of scale parameter Source: 2002/03 HIES/LSMS. 84 C. APPENDIX C: SENSITIVITY OF POVERTY ESTIMATES TO CRUCIAL HYPOTHESES Figure C.2: Headcount within different groups of households making different assumptions on the extent of economies of scale 70 60 Poor 50 No. children Headcount (%) 40 1 child 30 2 children 20 3+ children 10 0 0.5 0.6 0.7 0.8 0.9 1 Economy of scale parameter Source: 2002/03 HIES/LSMS. C.2. The inclusion of rent and heating expenses good part of the lower part of the distribution. in the consumption aggregate The same analysis is conducted for the main geo- graphical regions. Looking at the cumulative distribu- The inclusion of imputed rents as well as heating tion functions in Figure C.4, the West is still the worse- expenses (central heating, wood, coal, and dung) off region followed by the Highland, but for the other required elaborated analysis, and although it is regions, the curves intersect in various points and there believed that the best use of the available data was is not a clear trend that emerges. Contrary with the made, it is important to check how the final results are result presented in section 2, when rent and central sensitive to a consumption aggregate that excludes heating expenditure are excluded Ulaanbaatar is no both rent and heating expenses. The exclusion of these longer better-off than the rest of the Central region, two consumption components is because there are and the East. Therefore, the finding that Ulaanbaatar is some important inter-linkages between the two: the the least poor depends on the inclusion of rent and imputed rent seems to be strongly associated with the heating expenditure in the consumption aggregate. heating system the dwelling uses. The conclusion is that the geographical poverty The population rankings to test are that of the rankings are sensitive to the treatment of heating main analytical domains. In fact, it is between urban expenditure and rent. Although urban areas remain (Ulaanbaatar and aimag centers) and rural areas (soum better-off than rural ones, the differences in welfare centers and countryside) that the main differences in levels between the two are sensibly reduced, and rent and heating expenditures are likely to be. In order Ulaanbaatar is no longer in-equivocally the richest area to see whether the rankings between these areas of the country. Poverty estimates with and without change when excluding rent and heating expenditures rent and heating are shown in Tables C.2 and C.3, from the consumption aggregate, the same technique which also present estimations with the lower and explained in section 2 is used to plot on the same upper poverty line. graph three cumulative distribution functions: one for Ulaanbaatar, one for aimag centers and one for rural areas. As shown in Figure C.3 urban areas are still bet- ter-off than rural areas, although the gap between the two is reduced considerably. Also the gap between Ulaanbaatar and aimag centers now is very small for a C. APPENDIX C: SENSITIVITY OF POVERTY ESTIMATES TO CRUCIAL HYPOTHESES 85 Figure C.3: Cumulative distribution functions of urban and rural areas (excluding rents and heating costs) .8 Rural areas Cumulative fraction of population .6 Aimag centers Ulaanbaatar .4 .2 0 5 10 15 20 25 30 35 40 45 Per capita real consumption (Thousands of Tugrug per month) Source: 2002/03 HIES/LSMS. Figure C.4: Cumulative distribution functions by region (excluding rent and heating costs) West .8 Highland Cumulative fraction of population .6 Ulaanbaatar .4 Other regions .2 0 5 10 15 20 25 30 35 40 45 Per capita real consumption (Tugrug per month) Source: 2002/03 HIES/LSMS. 86 C. APPENDIX C: SENSITIVITY OF POVERTY ESTIMATES TO CRUCIAL HYPOTHESES Table C.2: Lower poverty estimates All components Excluding rent and heating Headcount Poverty Severity Headcount Poverty Severity Gap Gap National 36.1 11.0 4.7 41.0 13.4 6.0 (1.4) (0.6) (0.3) (1.5) (0.7) (0.4) Analytical domain Ulaanbaatar 27.3 8.1 3.3 35.8 11.8 5.3 (2.6) (1.0) (0.5) (2.8) (1.2) (0.7) Aimag centers 33.9 10.5 4.7 39.6 13.2 6.2 (2.2) (1.0) (0.7) (2.2) (1.1) (0.7) Soum centers 44.5 14.4 6.4 46.2 16.0 7.4 (3.0) (1.5) (0.9) (2.9) (1.6) (1.0) Countryside 42.7 12.6 5.1 44.8 13.8 5.8 (3.3) (1.3) (0.7) (3.4) (1.4) (0.8) Region West 51.1 14.6 5.7 55.3 17.0 7.2 (3.5) (1.3) (0.7) (3.5) (1.5) (0.8) Highland 38.7 12.3 5.2 42.0 13.8 6.1 (2.9) (1.3) (0.7) (3.0) (1.3) (0.8) Central a/ 34.4 10.1 4.3 37.7 12.0 5.3 (3.0) (1.4) (0.8) (3.0) (1.4) (0.9) East 34.5 12.4 6.6 36.1 13.9 7.6 (4.4) (2.3) (1.6) (4.3) (2.4) (1.7) Location Urban 30.3 9.2 4.0 37.6 12.5 5.7 (1.7) (0.7) (0.4) (1.8) (0.8) (0.5) Rural 43.4 13.2 5.6 45.3 14.6 6.4 (2.4) (1.0) (0.5) (2.4) (1.1) (0.6) Memorandum items: Bottom 40% Calories 1,345 1,337 National poverty line Food 14,386 14,323 Non-food 10,357 10,245 Total 24,743 24,568 a/ Excludes Ulaanbaatar. Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. C. APPENDIX C: SENSITIVITY OF POVERTY ESTIMATES TO CRUCIAL HYPOTHESES 87 Table C.3: Upper poverty estimates All components Excluding rent and heating Headcount Poverty Severity Headcount Poverty Severity Gap Gap National 53.6 19.0 9.0 52.9 18.9 9.1 (1.5) (0.8) (0.5) (1.5) (0.8) (0.5) Analytical domain Ulaanbaatar 42.4 14.5 6.7 47.9 16.8 8.0 (3.1) (1.3) (0.8) (2.8) (1.4) (0.8) Aimag centers 51.5 18.1 8.8 51.0 18.4 9.1 (2.2) (1.2) (0.8) (2.2) (1.2) (0.9) Soum centers 61.2 23.1 11.6 55.9 21.7 10.9 (2.7) (1.8) (1.2) (2.9) (1.8) (1.2) Countryside 62.9 22.2 10.2 58.4 20.0 9.1 (3.5) (1.7) (1.0) (3.4) (1.6) (1.0) Region West 71.1 25.5 11.8 66.6 24.2 11.2 (3.0) (1.7) (1.0) (3.2) (1.7) (1.0) Highland 56.7 20.8 10.0 54.7 19.5 9.3 (3.3) (1.6) (1.0) (3.2) (1.6) (1.0) Central a/ 52.3 17.8 8.4 49.1 17.1 8.1 (2.8) (1.6) (1.1) (2.7) (1.6) (1.1) East 52.3 19.5 10.5 48.1 18.5 10.1 (4.9) (2.6) (1.9) (4.9) (2.6) (1.9) Location Urban 46.5 16.1 7.6 49.3 17.5 8.5 (1.9) (0.9) (0.6) (1.8) (0.9) (0.6) Rural 62.3 22.5 10.8 57.5 20.6 9.8 (2.4) (1.2) (0.8) (2.4) (1.2) (0.8) Memorandum items: Bottom 40% Calories 1,345 1,337 National poverty line Food 14,386 14,323 Non-food 17,984 15,029 Total 32,370 29,352 a/ Excludes Ulaanbaatar. Note: Standard errors taking into account the survey design are shown in parentheses. Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 90 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Table D.1: Inequality measures Gini coefficient Theil index National 0.329 0.183 Urban 0.331 0.185 Rural 0.313 0.165 Region West 0.306 0.166 Highland 0.320 0.171 Central a/ 0.314 0.164 East 0.317 0.173 Analytical domain Ulaanbaatar 0.332 0.187 Aimag centers 0.324 0.175 Soum centers 0.318 0.170 Countryside 0.309 0.162 a/ Excludes Ulaanbaatar. Source: 2002/03 HIES/LSMS. Table D.2: Decomposition of inequality between and within various population groups (Theil index) Within Between Total Urban/rural areas 96.7 3.3 100.0 Geographical regions 95.6 4.4 100.0 Strata (Ulaanbaatar, aimag centers, soum centers, countryside) 95.8 4.2 100.0 Dwelling type (house, apartment, ger) 92.4 7.6 100.0 Water source 87.5 12.5 100.0 Toilet (inside, outside) 90.5 9.5 100.0 Whether household has telephone 87.4 12.6 100.0 Heating system (central, wood, coal, other) 89.6 10.4 100.0 Household size 77.1 22.9 100.0 Age of household head (15-29, 30-49, 50+) 98.8 1.2 100.0 Sex of household head 100.0 0.0 100.0 Education of household head 90.4 9.6 100.0 Sector of employment of household head 97.0 3.0 100.0 Source: 2002/03 HIES/LSMS. Table D.3: Per capita daily caloric intake by main food groups National Urban Rural Analytical domains Geographical regions Ulaanbaatar Aimag Soum Countryside West Highland Central East centers centers a/ Caloric intake Meat and meat products 379 302 474 270 342 408 512 452 434 385 445 Milk and dairy products 213 127 320 122 132 202 388 155 335 186 354 Flour and flour products 1,062 1,072 1,048 1,023 1,132 1,031 1,058 1,089 1,079 1,096 1,020 Vegetables 60 82 32 89 74 45 25 47 35 59 56 Fruits 10 14 6 15 12 7 5 6 7 11 10 Candy, sugar 106 109 102 104 115 102 103 98 114 108 103 Tea, coffee, beverages 10 12 8 13 11 9 8 10 8 10 8 Spices 81 112 42 122 99 63 31 35 58 94 62 Total 1,921 1,830 2,034 1,758 1,916 1,865 2,129 1,891 2,071 1,948 2,058 Shares Meat and meat products 20 17 23 15 18 22 24 24 21 20 22 Milk and dairy products 11 7 16 7 7 11 18 8 16 10 17 Flour and flour products 55 59 52 58 59 55 50 58 52 56 50 Vegetables 3 4 2 5 4 2 1 2 2 3 3 Fruits 1 1 0 1 1 0 0 0 0 1 1 Candy, sugar 6 6 5 6 6 5 5 5 6 6 5 Tea, coffee, beverages 1 1 0 1 1 0 0 1 0 1 0 Spices 4 6 2 7 5 3 1 2 3 5 3 Total 100 100 100 100 100 100 100 100 100 100 100 a/ Excludes Ulaanbaatar. Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 91 92 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Table D.4: Per capita monthly consumption by poverty status and urban-rural divide Total Urban Rural Non-poor Poor Non-poor Poor Non-poor Poor Consumption Food 20,504 9,002 18,636 7,912 23,366 9,949 Alcohol and tobacco 1,767 557 1,851 531 1,637 579 Education 3,284 1,166 3,986 1,400 2,209 962 Health 2,561 782 2,813 801 2,176 765 Durable goods 1/ 574 120 709 132 369 110 Rent 2/ 2,722 586 4,060 832 671 372 Heating 3/ 1,349 934 1,739 1,429 751 504 Utilities 4/ 2,730 927 3,748 1,196 1,171 695 Clothing 6,206 1,684 6,308 1,460 6,049 1,878 Transportation and communication 2,659 534 2,945 602 2,220 476 Others 5/ 3,434 921 3,591 929 3,194 915 Total 47,790 17,214 50,386 17,224 43,813 17,205 Shares Food 43 52 37 46 53 58 Alcohol and tobacco 4 3 4 3 4 3 Education 7 7 8 8 5 6 Health 5 5 6 5 5 4 Durable goods 1/ 1 1 1 1 1 1 Rent 2/ 6 3 8 5 2 2 Heating 3/ 3 5 3 8 2 3 Utilities 4/ 6 5 7 7 3 4 Clothing 13 10 13 8 14 11 Transportation and communication 6 3 6 3 5 3 Others 5/ 7 5 7 5 7 5 Total 100 100 100 100 100 100 1/ Estimation of the monetary value of the consumption derived from the use of durable goods. 2/ Estimation of the monetary value of the consumption derived from occupying the dwelling. If the household rents its dwelling, the actual rent will be included instead of the imputed rent. 3/ Includes central and local heating, firewood, coal and dung. 4/ Includes electricity and lighting, water and telephone. 5/ Includes recreation, entertaiment, beauty and toilet articles, and household utensils. Source: 2002/03 HIES/LSMS. Table D.5: Per capita monthly consumption by poverty status and analytical domain Total Ulaanbaatar Aimag centers Soum centers Countryside Non-poor Poor Non-poor Poor Non-poor Poor Non-poor Poor Non-poor Poor Consumption Food 20,504 9,002 18,426 7,612 18,911 8,200 19,849 8,780 25,310 10,644 Alcohol and tobacco 1,767 557 1,861 544 1,839 518 1,819 617 1,536 557 Education 3,284 1,166 4,253 1,418 3,635 1,384 3,720 1,269 1,374 780 Health 2,561 782 2,657 806 3,018 797 3,130 843 1,649 719 Durable goods 1/ 574 120 774 140 623 125 426 111 337 110 Rent 2/ 2,722 586 5,915 1,029 1,622 644 664 390 675 361 Heating 3/ 1,349 934 1,756 1,669 1,717 1,199 942 606 645 444 Utilities 4/ 2,730 927 4,392 1,292 2,901 1,104 1,671 947 894 545 Clothing 6,206 1,684 5,476 1,161 7,402 1,749 6,402 1,704 5,854 1,982 Transportation and communication 2,659 534 3,481 866 2,241 347 2,266 461 2,194 484 Others 5/ 3,434 921 3,615 852 3,560 1,002 3,132 1,030 3,229 846 Total 47,790 17,214 52,605 17,387 47,468 17,066 44,022 16,758 43,698 17,471 Shares Food 43 52 35 44 40 48 45 52 58 61 Alcohol and tobacco 4 3 4 3 4 3 4 4 4 3 Education 7 7 8 8 8 8 8 8 3 4 Health 5 5 5 5 6 5 7 5 4 4 Durable goods 1/ 1 1 1 1 1 1 1 1 1 1 Rent 2/ 6 3 11 6 3 4 2 2 2 2 Heating 3/ 3 5 3 10 4 7 2 4 1 3 Utilities 4/ 6 5 8 7 6 6 4 6 2 3 Clothing 13 10 10 7 16 10 15 10 13 11 Transportation and communication 6 3 7 5 5 2 5 3 5 3 Others 5/ 7 5 7 5 8 6 7 6 7 5 Total 100 100 100 100 100 100 100 100 100 100 1/ Estimation of the monetary value of the consumption derived from the use of durable goods. 2/ Estimation of the monetary value of the consumption derived from occupying the dwelling. If the household rents its dwelling, the actual rent will be included instead of the imputed rent. 3/ Includes central and local heating, firewood, coal and dung. 4/ Includes electricity and lighting, water and telephone. 5/ Includes recreation, entertaiment, beauty and toilet articles, and household utensils. Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 93 94 Table D.6: Per capita monthly consumption by poverty status and region Total West Highland Central East Ulaanbaatar Non-poor Poor Non-poor Poor Non-poor Poor Non-poor Poor Non-poor Poor Non-poor Poor Consumption Food 20,504 9,002 19,097 9,521 22,717 9,664 20,966 9,183 23,594 8,870 18,426 7,612 Alcohol and tobacco 1,767 557 1,662 574 1,825 588 1,764 601 1,432 357 1,861 544 Education 3,284 1,166 3,071 1,005 2,633 977 3,006 1,398 2,237 1,016 4,253 1,418 Health 2,561 782 2,390 841 2,119 645 3,216 908 2,143 692 2,657 806 Durable goods 1/ 574 120 485 170 498 104 440 67 444 95 774 140 Rent 2/ 2,722 586 759 387 1,005 483 1,327 509 971 444 5,915 1,029 Heating 3/ 1,349 934 1,411 840 941 552 1,191 804 1,115 682 1,756 1,669 Utilities 4/ 2,730 927 1,815 760 1,528 742 2,214 958 1,980 921 4,392 1,292 Clothing 6,206 1,684 6,274 2,103 6,773 1,773 6,636 1,679 6,460 1,650 5,476 1,161 Transportation and communication 2,659 534 2,295 595 1,955 330 2,800 445 1,594 296 3,481 866 Others 5/ 3,434 921 3,033 882 3,421 1,041 3,348 917 3,548 866 3,615 852 Total 47,790 17,214 42,291 17,679 45,415 16,899 46,909 17,469 45,519 15,889 52,605 17,387 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Shares Food 43 52 45 54 50 57 45 53 52 56 35 44 Alcohol and tobacco 4 3 4 3 4 3 4 3 3 2 4 3 Education 7 7 7 6 6 6 6 8 5 6 8 8 Health 5 5 6 5 5 4 7 5 5 4 5 5 Durable goods 1/ 1 1 1 1 1 1 1 0 1 1 1 1 Rent 2/ 6 3 2 2 2 3 3 3 2 3 11 6 Heating 3/ 3 5 3 5 2 3 3 5 2 4 3 10 Utilities 4/ 6 5 4 4 3 4 5 5 4 6 8 7 Clothing 13 10 15 12 15 10 14 10 14 10 10 7 Transportation and communication 6 3 5 3 4 2 6 3 4 2 7 5 Others 5/ 7 5 7 5 8 6 7 5 8 5 7 5 Total 100 100 100 100 100 100 100 100 100 100 100 100 1/ Estimation of the monetary value of the consumption derived from the use of durable goods. 2/ Estimation of the monetary value of the consumption derived from occupying the dwelling. If the household rents its dwelling, the actual rent will be included instead of the imputed rent. 3/ Includes central and local heating, firewood, coal and dung. 4/ Includes electricity and lighting, water and telephone. 5/ Includes recreation, entertaiment, beauty and toilet articles, and household utensils. Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 95 Table D.7: Per capita monthly consumption by decile Total Urban Rural Ulaanbaatar Aimag Soum Countryside centers centers Poorest 10,991 11,422 10,589 12,333 10,456 9,646 11,257 II 16,481 17,750 15,375 18,618 17,007 14,651 15,820 III 20,407 22,356 18,800 23,607 21,173 18,185 19,234 IV 24,288 26,961 21,819 28,450 25,321 21,435 22,002 V 28,589 31,526 25,537 33,648 29,602 25,567 25,552 VI 33,150 36,691 29,368 39,107 34,162 30,086 29,002 VII 38,559 42,799 33,894 46,449 39,277 34,222 33,647 VIII 46,353 51,603 40,144 55,552 46,210 39,799 40,415 IX 58,201 63,596 50,343 67,168 58,360 48,692 51,360 Richest 90,533 99,171 77,106 105,726 90,650 77,064 77,366 Total 36,747 40,348 32,269 43,002 37,175 31,881 32,491 Note: Deciles were constructed separately for each geographical domain. They comprise 10% of the population of the respective region. Source: 2002/03 HIES/LSMS. Table D.8: Share of total consumption by decile Total Urban Rural Ulaanbaatar Aimag Soum Countryside centers centers Poorest 3.0 2.8 3.3 2.9 2.8 3.0 3.6 II 4.5 4.4 4.8 4.4 4.7 4.6 4.8 III 5.5 5.5 5.8 5.5 5.6 5.8 6.0 IV 6.6 6.6 6.8 6.5 6.9 6.5 6.7 V 7.8 7.9 7.8 7.9 7.9 8.2 7.8 VI 9.0 9.1 9.1 9.1 9.3 9.3 8.9 VII 10.5 10.7 10.6 10.9 10.4 10.8 10.4 VIII 12.6 12.8 12.4 12.7 12.5 12.5 12.4 IX 15.8 15.7 15.6 15.7 15.7 15.2 16.0 Richest 24.6 24.5 23.8 24.5 24.3 24.1 23.4 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Note: Deciles were constructed separately for each geographical domain. They comprise 10% of the population of the respective region. Source: 2002/03 HIES/LSMS. 96 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Table D.9: Poverty incidence by characteristics of the household head and urban-rural divide Headcount Share of population Share of poor Urban Rural National Urban Rural National Urban Rural National Gender Male 27.9 42.8 34.8 82.5 89.9 85.8 75.9 88.7 82.8 Female 41.6 48.4 43.8 17.5 10.1 14.2 24.1 11.3 17.2 Age Less than 30 years 23.1 29.3 27.0 7.6 15.9 11.3 5.8 10.7 8.4 Between 30 and 49 32.4 49.4 40.2 56.6 60.3 58.2 60.5 68.7 64.9 50 years or more 28.4 37.6 31.6 35.8 23.8 30.5 33.7 20.6 26.7 Educational attainment None 52.3 43.8 45.8 1.8 7.3 4.2 3.1 7.4 5.4 Primary 48.1 44.5 45.6 8.1 21.7 14.2 12.9 22.2 17.9 Secondary 8th grade 47.7 43.8 45.5 20.9 35.6 27.5 33.0 35.9 34.6 Complete secondary 29.2 44.8 34.9 21.7 15.4 18.8 20.9 15.9 18.2 Vocational 34.9 50.2 40.7 11.4 8.7 10.2 13.2 10.1 11.5 Higher diploma 19.6 34.4 23.4 18.3 7.9 13.6 11.8 6.2 8.8 University 8.8 29.0 11.6 17.8 3.6 11.5 5.2 2.4 3.7 Migration Migrant 29.0 38.7 31.2 17.2 6.1 12.3 16.5 5.5 10.6 Non-migrant 30.5 43.7 36.8 82.8 93.9 87.7 83.5 94.5 89.4 Employment Labor force participation Employed 25.5 41.4 33.6 62.9 82.1 71.5 53.0 78.3 66.5 Unemployed 43.8 60.1 48.7 3.8 2.0 3.0 5.5 2.8 4.0 Out of labor force 37.8 51.7 41.6 33.3 15.9 25.5 41.6 18.9 29.4 Among those employed, Economic activity Agriculture 40.6 41.0 41.0 5.6 60.7 30.2 7.5 57.4 34.2 Industry 28.2 57.5 33.2 13.1 3.4 8.8 12.2 4.4 8.1 Services 22.7 39.5 26.9 44.2 18.1 32.6 33.2 16.5 24.3 Sector Private 28.8 42.4 37.1 35.7 69.2 50.6 34.0 67.5 51.9 Herders 43.9 38.8 39.2 3.1 55.6 26.5 4.5 49.8 28.8 Non-herders 27.4 56.7 34.7 32.6 13.6 24.1 29.4 17.8 23.2 Public 22.4 34.7 25.9 23.3 11.2 17.9 17.3 9.0 12.8 State 13.1 45.0 21.6 3.9 1.8 3.0 1.7 1.8 1.8 Total 30.3 43.4 36.1 100.0 100.0 100.0 100.0 100.0 100.0 Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 97 Table D.10: Poverty incidence by characteristics of the household head and analytical domain Headcount Share of population Share of poor Ulaan- Aimag Soum Country Ulaan- Aimag Soum Country Ulaan- Aimag Soum Country baatar centers centers side baatar centers centers side baatar centers centers side Gender Male 22.8 33.3 44.3 42.0 78.6 87.2 89.4 90.2 65.8 85.6 88.9 88.6 Female 43.5 37.9 46.8 49.4 21.4 12.8 10.6 9.8 34.2 14.4 11.2 11.4 Age Less than 30 years 24.2 21.8 40.2 26.1 7.5 7.8 10.0 19.3 6.6 5.0 9.0 11.7 Between 30 and 49 27.1 37.5 48.5 49.9 50.9 63.3 64.5 57.9 50.6 70.1 70.3 67.7 50 years or more 28.0 29.2 36.1 38.5 41.6 28.9 25.5 22.8 42.7 24.9 20.7 20.6 Educational attainment None 48.0 56.7 33.4 45.1 1.6 1.9 2.3 10.1 2.9 3.2 1.8 10.7 Primary 47.8 48.4 56.0 42.0 8.5 7.8 10.5 28.1 14.8 11.1 13.3 27.6 Secondary 8th grade 43.2 52.1 49.8 41.1 18.8 23.5 30.6 38.4 29.8 36.1 34.2 36.9 Complete secondary 29.6 28.7 44.7 45.0 20.1 23.6 22.9 11.0 21.8 20.0 23.0 11.6 Vocational 34.5 35.3 51.4 48.5 10.2 12.9 13.7 5.8 12.9 13.4 15.8 6.6 Higher diploma 17.2 22.6 29.8 40.4 19.1 17.2 12.4 5.2 12.1 11.5 8.3 5.0 University 7.2 12.0 21.8 53.0 21.7 13.2 7.5 1.3 5.7 4.7 3.7 1.6 Migration Migrant 23.4 35.3 35.0 45.5 16.7 17.9 10.9 3.4 14.3 18.7 8.6 3.6 Non-migrant 28.0 33.5 45.7 42.6 83.4 82.1 89.1 96.6 85.7 81.3 91.5 96.4 Employment Labor force participation Employed 21.9 29.3 41.6 41.3 59.6 66.9 71.4 88.3 47.8 57.9 66.7 85.2 Unemployed 43.5 44.2 72.4 43.6 3.5 4.2 3.2 1.3 5.6 5.4 5.1 1.4 Out of labor force 34.4 42.9 49.3 55.1 36.9 29.0 25.4 10.4 46.6 36.7 28.1 13.4 Among those employed, Economic activity Agriculture 28.2 45.3 48.7 39.5 2.9 8.9 27.4 79.7 3.0 11.9 30.0 73.8 Industry 27.4 29.1 55.7 61.8 12.6 13.8 6.6 1.5 12.7 11.8 8.2 2.2 Services 19.9 26.1 33.9 56.4 44.1 44.2 37.5 7.0 32.2 34.2 28.6 9.3 Sector Private 23.6 35.0 47.9 40.7 35.5 35.8 44.4 83.3 30.8 37.0 47.8 79.3 Herders 28.5 50.5 41.1 38.5 1.7 4.8 19.2 76.4 1.8 7.2 17.7 68.8 Non-herders 23.4 32.5 53.1 64.2 33.8 31.0 25.3 7.0 29.0 29.8 30.1 10.5 Public 19.5 25.4 28.1 52.9 21.5 25.5 22.6 4.7 15.4 19.1 14.2 5.8 State 17.4 10.8 47.2 19.2 2.6 5.5 4.4 0.2 1.6 1.8 4.7 0.1 Total 27.3 33.9 44.5 42.7 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: 2002/03 HIES/LSMS. 98 Table D.11: Poverty incidence by characteristics of the household head and region Headcount Share of population Share of poor West High- Central East Ulaan- West High- Central East Ulaan- West High- Central East Ulaan- land baatar land baatar land baatar Gender Male 51.4 37.6 33.6 32.7 22.8 94.1 89.0 84.2 88.9 78.6 94.7 86.5 82.3 84.2 65.8 Female 46.3 47.5 38.5 49.4 43.5 5.9 11.0 15.8 11.1 21.4 5.3 13.5 17.7 15.8 34.2 Age Less than 30 years 41.7 26.7 22.1 19.1 24.2 12.2 12.5 13.1 15.2 7.5 10.0 8.7 8.4 8.4 6.6 Between 30 and 49 58.3 40.1 41.0 39.8 27.1 64.7 61.8 56.5 64.6 50.9 73.9 64.1 67.4 74.4 50.6 50 years or more 35.8 41.1 27.3 29.4 28.0 23.1 25.6 30.5 20.2 41.6 16.2 27.3 24.2 17.2 42.7 Educational attainment None 56.2 40.8 38.9 44.6 48.0 6.4 6.4 3.0 5.6 1.6 7.0 6.8 3.4 7.3 2.9 Primary 56.9 43.6 38.0 45.1 47.8 14.5 18.8 16.0 16.4 8.5 16.2 21.2 17.6 21.5 14.8 Secondary 8th grade 58.3 42.6 44.2 37.9 43.2 30.3 32.6 29.0 33.7 18.8 34.7 35.9 37.3 37.0 29.8 Complete secondary 47.7 43.3 29.2 24.4 29.6 17.3 17.3 19.9 19.8 20.1 16.1 19.3 16.9 13.9 21.8 Vocational 56.4 35.3 38.4 39.5 34.5 12.8 6.3 12.7 10.3 10.2 14.1 5.7 14.2 11.8 12.9 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Higher diploma 31.2 27.9 25.0 26.7 17.2 12.0 11.9 10.9 8.9 19.1 7.4 8.6 7.9 6.9 12.1 University 34.9 14.0 10.8 10.9 7.2 6.8 6.7 8.5 5.3 21.7 4.6 2.4 2.7 1.7 5.7 Migration Migrant 42.4 29.4 44.6 43.5 23.4 10.2 14.5 5.1 11.2 16.7 8.5 11.0 6.6 14.1 14.3 Non-migrant 52.0 40.2 33.9 33.4 28.0 89.8 85.5 94.9 88.8 83.4 91.6 89.0 93.4 85.9 85.7 Employment Labor force participation Employed 51.3 35.4 33.4 24.1 21.9 80.7 79.5 72.2 71.0 59.6 81.1 72.7 70.2 49.6 47.8 Unemployed 59.9 50.9 36.8 50.5 43.5 3.3 2.9 1.4 4.3 3.5 3.9 3.9 1.4 6.3 5.6 Out of labor force 48.0 51.4 36.9 61.8 34.4 16.0 17.6 26.4 24.6 36.9 15.0 23.4 28.3 44.1 46.6 Among those employed, Economic activity Agriculture 59.0 38.2 37.3 25.0 28.2 43.3 48.8 31.2 44.7 2.9 50.0 48.1 33.8 32.3 3.0 Industry 62.5 28.0 32.7 41.8 27.4 5.5 5.6 11.9 4.0 12.6 6.7 4.1 11.3 4.8 12.7 Services 38.9 31.7 29.6 19.2 19.9 32.0 25.1 29.2 22.4 44.1 24.4 20.5 25.1 12.5 32.2 Sector Private 54.7 38.7 37.2 24.5 23.6 61.9 61.9 47.3 56.8 35.5 66.3 61.9 51.2 40.2 30.8 Herders 58.4 37.7 30.5 24.0 28.5 37.0 46.2 25.1 39.8 1.7 42.3 45.0 22.3 27.7 1.8 Non-herders 49.2 41.5 44.8 25.7 23.4 24.8 15.7 22.2 17.0 33.8 23.9 16.9 28.9 12.6 29.0 Public 40.8 27.4 25.1 19.9 19.5 18.1 14.5 18.9 12.7 21.5 14.5 10.3 13.8 7.3 15.4 State 22.7 7.7 30.0 44.1 17.4 0.7 3.0 6.0 1.6 2.6 0.3 0.6 5.3 2.0 1.6 Total 51.1 38.7 34.4 34.5 27.3 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 99 Table D.12: Poverty incidence by characteristics of the dwelling and urban-rural divide Headcount Share of population Share of poor Urban Rural Total Urban Rural Total Urban Rural Total Dwelling Ger 47.5 41.9 43.4 20.9 73.0 44.2 32.9 70.6 53.1 House 33.9 48.5 38.2 44.7 23.1 35.1 50.1 25.8 37.1 Apartment 14.3 41.8 16.6 32.9 3.7 19.9 15.6 3.6 9.2 Other 1/ 31.2 20.0 30.0 1.4 0.2 0.9 1.5 0.1 0.7 Water supply Central, hot and cold 9.7 34.3 10.8 32.5 1.9 18.8 10.4 1.5 5.6 Central, only cold 35.7 27.7 34.5 4.7 1.0 3.0 5.5 0.6 2.9 Protected well 41.2 47.5 43.6 43.8 33.3 39.1 59.7 36.5 47.3 Unprotected well 32.5 30.7 30.9 2.3 20.1 10.2 2.5 14.2 8.8 Truck distribution 36.4 43.1 38.4 14.1 7.2 11.1 17.0 7.2 11.8 Other 2/ 57.3 47.5 48.3 2.6 36.5 17.7 4.9 40.0 23.7 Improved water sources 3/ Yes 28.3 46.3 33.0 81.0 36.2 61.0 75.7 38.6 55.8 No 38.8 41.7 40.9 19.0 63.8 39.0 24.4 61.4 44.2 Sewage system Yes 25.9 45.1 30.1 71.3 24.5 50.5 61.1 25.5 42.0 No 41.0 42.8 42.2 28.7 75.5 49.5 38.9 74.5 58.0 Improved sanitation 4/ Yes 26.0 45.0 30.2 73.3 25.6 52.0 63.0 26.5 43.5 No 41.9 42.8 42.5 26.7 74.4 48.0 37.0 73.5 56.5 Heating Central 13.4 18.8 13.6 38.0 2.5 22.2 16.8 1.1 8.4 Simple unit 5/ 40.6 44.1 42.5 61.8 97.4 77.7 82.8 98.9 91.4 Other 6/ 57.1 0.0 43.3 0.2 0.1 0.2 0.4 0.0 0.2 Electricity Central 29.7 47.0 33.3 89.4 29.8 62.9 87.7 32.3 58.0 Local 29.3 46.2 38.2 8.7 12.1 10.2 8.4 12.9 10.8 Other 7/ 57.7 18.7 19.7 0.2 7.6 3.5 0.3 3.3 1.9 None 64.1 44.3 45.1 1.7 50.4 23.4 3.6 51.5 29.2 National 30.3 43.4 36.1 100.0 100.0 100.0 100.0 100.0 100.0 1/ Students dormitory, public dormitory, other public apartments, others. 2/ Spring, river, snow, ice, others. 3/ It refers to the percentage of the population with access to an improved water source such as household connection, public standpipe or protected well or spring. Unimproved sources include vendors, tanker trucks and unprotected wells and springs. 4/ It refers to the percentage of the population with access to improved sanitation facilities such as adequate excreta disposal facilities (private or shared but not public). They can range from simple but protected pit latrines to flush toilets with sewerage connection. 5/ Simple heating units fueled by firewood, coal or dung. 6/ Individual electric heating unit, private low pressure stove, others. 7/ Solar or wind systems, small gen-sets, others. Source: 2002/03 HIES/LSMS. 100 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Table D.13: Poverty incidence by characteristics of the dwelling and analytical domain Headcount Share of population Share of poor Ulaan- Aimag Soum Country Ulaan- Aimag Soum Country Ulaan- Aimag Soum Country baatar centers centers side baatar centers centers side baatar centers centers side Dwelling Ger 56.7 42.1 46.9 40.4 14.3 28.9 48.1 87.2 29.7 35.9 50.7 82.4 House 33.9 33.8 44.1 57.2 45.5 43.8 42.4 12.0 56.7 43.8 42.1 16.1 Apartment 8.3 25.1 35.9 84.7 38.7 26.1 9.0 0.7 11.8 19.3 7.2 1.4 Other 1/ 33.9 27.4 0.0 100.0 1.5 1.3 0.5 0.1 1.9 1.0 0.0 0.2 Water supply Central, hot and cold 4.4 20.5 30.2 71.5 39.9 23.6 4.6 0.3 6.5 14.3 3.1 0.5 Central, only cold 45.6 30.8 27.3 36.9 2.8 6.9 2.6 0.1 4.7 6.3 1.6 0.1 Protected well 40.8 41.7 46.9 48.5 45.4 42.0 52.4 22.4 67.9 51.7 55.2 25.4 Unprotected well 100.0 30.7 35.0 29.4 0.1 4.9 12.8 24.2 0.4 4.5 10.1 16.7 Truck distribution 46.9 28.0 47.5 36.8 11.5 17.3 11.7 4.7 19.9 14.3 12.5 4.0 Other 2/ 60.8 57.1 49.4 47.1 0.3 5.3 15.8 48.3 0.6 9.0 17.5 53.3 Improved water sources 3/ Yes 24.5 33.8 44.7 48.7 88.1 72.5 59.7 22.8 79.1 72.3 59.9 26.0 No 47.7 34.1 44.3 41.0 11.9 27.5 40.3 77.2 20.9 27.7 40.1 74.0 Sewage system Yes 22.3 31.0 43.4 49.6 76.2 65.5 48.1 11.0 62.3 60.0 46.9 12.8 No 43.2 39.3 45.6 41.9 23.8 34.5 51.9 89.0 37.7 40.0 53.2 87.2 Improved sanitation 4/ Yes 22.4 30.8 42.7 50.4 77.0 68.8 49.5 12.0 63.3 62.7 47.5 14.1 No 43.4 40.5 46.3 41.7 23.0 31.2 50.5 88.0 36.7 37.3 52.5 85.9 Heating Central 7.4 23.1 19.4 10.2 43.4 31.5 6.5 0.3 11.9 21.5 2.9 0.1 Simple unit 5/ 42.3 38.9 46.4 42.8 56.3 68.3 93.2 99.7 87.3 78.5 97.2 99.9 Other 6/ 100.0 0.0 0.0 - 0.2 0.2 0.2 0.0 0.8 0.0 0.0 0.0 Electricity Central 27.1 33.6 46.6 47.9 99.1 77.9 59.1 13.1 98.4 77.4 61.8 14.7 Local - 29.3 41.8 61.0 0.0 19.1 25.7 4.4 0.0 16.6 24.1 6.3 Other 7/ 100.0 0.0 30.1 18.2 0.2 0.2 1.0 11.5 0.7 0.0 0.7 4.9 None 36.3 72.6 42.0 44.6 0.7 2.8 14.3 71.0 1.0 6.0 13.5 74.1 National 27.3 33.9 44.5 42.7 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1/ Students dormitory, public dormitory, other public apartments, others. 2/ Spring, river, snow, ice, others. 3/ It refers to the percentage of the population with access to an improved water source such as household connection, public standpipe or protected well or spring. Unimproved sources include vendors, tanker trucks and unprotected wells and springs. 4/ It refers to the percentage of the population with access to improved sanitation facilities such as adequate excreta disposal facilities (private or shared but not public). They can range from simple but protected pit latrines to flush toilets with sewerage connection. 5/ Simple heating units fueled by firewood, coal or dung. 6/ Individual electric heating unit, private low pressure stove, others. 7/ Solar or wind systems, small gen-sets, others. Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 101 Table D.14: Poverty incidence by characteristics of the dwelling and region Headcount Share of population Share of poor West High- Central East Ulaan- West High- Central East Ulaan- West High- Central East Ulaan- land baatar land baatar land baatar Dwelling Ger 54.0 40.8 33.7 31.0 56.7 65.2 68.4 38.7 51.4 14.3 68.9 72.2 38.0 46.2 29.7 House 48.3 37.1 41.0 34.7 33.9 32.7 22.3 36.1 36.5 45.5 30.9 21.4 42.9 36.7 56.7 Apartment 7.0 24.9 26.4 49.3 8.3 1.7 8.5 24.5 12.0 38.7 0.2 5.5 18.8 17.2 11.8 Other 1/ 0.0 45.8 14.4 0.0 33.9 0.5 0.8 0.7 0.1 1.5 0.0 1.0 0.3 0.0 1.9 Water supply Central, hot and cold 21.2 20.6 20.1 31.2 4.4 3.2 7.2 17.4 12.3 39.9 1.3 3.8 10.1 11.1 6.5 Central, only cold 0.0 28.0 30.2 34.6 45.6 0.2 1.6 6.8 4.8 2.8 0.0 1.1 6.0 4.8 4.7 Protected well 53.7 40.0 44.1 36.7 40.8 47.4 30.9 36.7 30.5 45.4 49.9 31.9 47.1 32.4 67.9 Unprotected well 52.3 20.8 25.1 30.1 100.0 11.5 8.4 18.4 28.4 0.1 11.8 4.5 13.4 24.7 0.4 Truck distribution 23.9 34.4 37.3 41.1 46.9 8.3 8.0 17.8 8.2 11.5 3.9 7.2 19.3 9.8 19.9 Other 2/ 57.6 45.3 48.6 37.6 60.8 29.4 44.0 2.9 15.8 0.3 33.2 51.5 4.1 17.2 0.6 Improved water sources 3/ Yes 51.5 36.0 35.7 35.0 24.5 50.8 39.6 60.9 47.6 88.1 51.2 36.9 63.2 48.3 79.1 No 50.6 40.4 32.4 34.1 47.7 49.2 60.4 39.1 52.4 11.9 48.8 63.1 36.8 51.7 20.9 Sewage system Yes 42.8 38.0 31.9 30.5 22.3 43.9 36.2 41.9 33.6 76.2 36.8 35.6 38.8 29.6 62.3 No 57.5 39.1 36.2 36.6 43.2 56.1 63.8 58.1 66.5 23.8 63.2 64.4 61.2 70.4 37.7 Improved sanitation 4/ Yes 43.2 37.3 31.5 32.0 22.4 44.6 37.4 45.6 35.7 77.0 37.7 36.1 41.8 33.1 63.3 No 57.4 39.5 36.8 35.9 43.4 55.4 62.6 54.4 64.3 23.0 62.3 63.9 58.2 66.9 36.7 Heating Central 17.2 21.5 20.8 30.7 7.4 62.1 51.8 77.8 54.0 99.1 1.2 4.9 14.5 16.5 11.9 Simple unit 5/ 52.3 40.3 38.7 35.6 42.3 2.5 5.3 4.8 8.7 0.2 98.9 95.2 85.5 83.5 87.3 Other 6/ 0.0 0.0 0.0 0.0 100.0 35.4 42.9 17.4 37.3 0.7 0.0 0.0 0.0 0.0 0.8 Electricity Central 51.0 37.5 36.2 35.9 27.1 33.5 34.2 74.2 49.3 99.1 33.5 33.1 78.0 51.2 98.4 Local 35.7 44.0 19.5 40.4 - 28.5 17.7 3.6 4.7 0.0 20.0 20.1 2.1 5.5 0.0 Other 7/ 27.2 17.2 21.8 12.1 100.0 2.5 5.3 4.8 8.7 0.2 1.4 2.4 3.0 3.1 0.7 None 65.2 40.1 33.4 37.3 36.3 35.4 42.9 17.4 37.3 0.7 45.2 44.4 16.9 40.3 1.0 National 51.1 38.7 34.4 34.5 27.3 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1/ Students dormitory, public dormitory, other public apartments, others. 2/ Spring, river, snow, ice, others. 3/ It refers to the percentage of the population with access to an improved water source such as household connection, public standpipe or protected well or spring. Unimproved sources include vendors, tanker trucks and unprotected wells and springs. 4/ It refers to the percentage of the population with access to improved sanitation facilities such as adequate excreta disposal facilities (private or shared but not public). They can range from simple but protected pit latrines to flush toilets with sewerage connection. 5/ Simple heating units fueled by firewood, coal or dung. 6/ Individual electric heating unit, private low pressure stove, others. 7/ Solar or wind systems, small gen-sets, others. Source: 2002/03 HIES/LSMS. 102 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Table D.15: Characteristics of the adult population by highest level of education attained None Primary Secondary Complete Vocational Higher University Total 8th grade Secondary diploma Location Urban 25.0 32.5 43.5 67.2 62.4 75.3 86.8 57.6 Rural 75.0 67.5 56.5 32.8 37.6 24.7 13.2 42.4 Ulaanbaatar 12.6 17.6 22.8 38.0 31.9 43.8 57.0 32.6 Aimag centers 12.5 15.0 20.7 29.2 30.5 31.5 29.8 25.0 Soum centers 9.1 11.4 16.4 16.7 21.5 16.3 9.7 15.3 Countryside 65.8 56.1 40.1 16.2 16.1 8.4 3.5 27.1 West 23.6 20.2 16.4 13.6 17.7 13.3 9.0 15.4 Highland 37.5 28.2 29.4 21.2 15.3 20.1 13.9 23.5 Central a/ 13.3 23.9 20.7 19.8 26.2 15.1 15.9 19.8 East 13.0 10.1 10.6 7.5 8.9 7.7 4.1 8.7 Gender Men Women 46.1 46.8 56.2 44.0 47.7 39.5 45.7 47.4 53.9 53.2 43.8 56.0 52.3 60.5 54.3 52.6 Quintile Poorest 22.4 21.6 24.8 14.6 18.0 8.9 3.1 16.7 Q2 21.7 20.0 22.7 19.0 19.2 14.4 7.4 18.4 Q3 19.5 20.5 19.7 21.4 19.6 18.8 15.2 19.7 Q4 18.9 19.0 17.5 23.3 22.8 24.7 28.1 21.8 Richest 17.5 19.0 15.4 21.7 20.4 33.3 46.2 23.4 Poverty Non-poor 60.6 61.6 57.3 70.5 65.6 80.9 91.6 68.7 Poor 39.4 38.4 42.7 29.5 34.4 19.2 8.4 31.3 National 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 a/ Excludes Ulaanbaatar. Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 103 Figure D.1: Public spending in lower and upper secondary 100 80 Lower secondary Cum. share of benefits/beneficiaries (5 to 8 th grade) 60 40 Upper secondary (9 to 10 th grade) 20 0 0 20 40 60 80 100 Cum. percentage of population (rank by per capita consumption) Source: 2002/03 HIES/LSMS. Figure D.2: Public spending in primary schools by urban-rural divide 100 80 Rural Cum. share of benefits/beneficiaries 60 National 40 Urban 20 0 0 20 40 60 80 100 Cum. percentage of population (rank by per capita consumption) Source: 2002/03 HIES/LSMS. 104 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Figure D.3: Public spending in secondary schools by urban-rural divide 100 National 80 Cum. share of benefits/beneficiaries 60 Rural 40 Urban 20 0 0 20 40 60 80 100 Cum. percentage of population (rank by per capita consumption) Source: 2002/03 HIES/LSMS. Figure D.4: Public spending in universities by urban-rural divide 100 80 Cum. share of benefits/beneficiaries 60 Urban 40 Rural National 20 0 0 20 40 60 80 100 Cum. percentage of population (rank by per capita consumption) Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 105 Table D.16: Enrollment rates comparison, 2002 Net enrollment rates Gross enrollment rates LSMS NSO LSMS NSO Primary 89 89 109 103 Men 89 87 108 103 Women 88 91 111 103 Secondary 75 82 82 82 Men 72 79 79 77 Women 78 84 84 87 Source: 2002/03 HIES/LSMS and National Statistics Office. 106 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Table D.17: Educational level of current students Primary Secondary University, Vocational, Total college others Location Urban 29.9 51.5 17.3 1.2 100.0 Rural 43.1 46.4 8.9 1.7 100.0 Ulaanbaatar 28.2 49.4 21.1 1.3 100.0 Aimag centers 31.8 53.8 13.3 1.2 100.0 Soum centers 34.6 51.0 13.2 1.2 100.0 Countryside 51.5 41.8 4.6 2.1 100.0 West 44.0 45.0 10.0 1.0 100.0 Highland 38.7 49.3 11.1 0.9 100.0 Central a/ 33.5 51.4 12.8 2.3 100.0 East 36.3 55.4 6.7 1.7 100.0 Gender Men 37.3 49.1 12.0 1.6 100.0 Women 32.8 50.0 16.0 1.2 100.0 Quintile Poorest 47.8 48.7 2.5 1.0 100.0 Q2 39.6 51.4 7.3 1.7 100.0 Q3 32.8 53.0 13.4 0.8 100.0 Q4 28.2 49.8 21.4 0.7 100.0 Richest 26.0 44.7 26.5 2.8 100.0 Poverty Non-poor 29.5 49.3 19.7 1.5 100.0 Poor 44.5 50.0 4.3 1.2 100.0 National 34.9 49.6 14.1 1.4 100.0 a/ Excludes Ulaanbaatar. Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 107 Table D.18: Characteristics of current students by level of education enrolled Primary Secondary University, Vocational, Total college others Location Urban 53.0 64.4 76.1 54.8 61.9 Rural 47.0 35.6 24.0 45.2 38.1 Ulaanbaatar 25.9 32.1 48.0 29.6 32.1 Aimag centers 27.1 32.3 28.0 25.2 29.8 Soum centers 18.7 19.5 17.7 16.3 18.9 Countryside 28.3 16.2 6.2 28.9 19.2 West 21.4 15.4 12.1 12.5 17.0 Highland 24.3 21.8 17.2 14.8 21.9 Central a/ 19.7 21.3 18.6 33.1 20.5 East 8.8 9.5 4.1 10.1 8.5 Gender Men 50.4 46.8 40.2 54.5 47.2 Women 49.6 53.2 59.8 45.5 52.8 Quintile Poorest 27.3 19.6 3.5 14.2 20.0 Q2 23.3 21.3 10.6 24.7 20.5 Q3 19.1 21.8 19.4 11.6 20.4 Q4 15.4 19.2 28.9 9.1 19.1 Richest 14.9 18.1 37.6 40.4 20.0 Poverty Non-poor 53.5 63.1 88.8 68.5 63.5 Poor 46.5 36.9 11.2 31.6 36.5 National 100.0 100.0 100.0 100.0 100.0 a/ Excludes Ulaanbaatar. Source: 2002/03 HIES/LSMS. 108 Table D.19: Contraceptive methods, all women 15-49 National Urban Rural Non-poor Poor Poorest Q2 Q3 Q4 Richest Ever used contraceptive methods (%) 42 41 43 43 40 40 39 43 41 46 None, primary 18 10 22 15 21 19 23 17 8 20 Sec. 8th grade 34 27 39 30 39 43 31 32 30 28 Complete secondary 44 41 48 44 44 44 45 45 44 41 Vocational, tertiary 58 55 65 59 55 57 53 64 54 60 Married 63 63 62 63 62 64 59 64 61 65 Divorced 49 52 41 52 44 39 51 57 46 53 Single 11 10 11 11 10 11 9 10 8 14 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Among women that had used, Current use of contraceptive methods (%) 91 90 92 90 93 94 92 87 91 91 None, primary 89 89 89 89 89 85 94 87 100 83 Sec. 8th grade 90 92 89 87 93 94 92 82 91 87 Complete secondary 93 92 95 92 94 96 93 93 92 91 Vocational, tertiary 90 89 92 89 92 93 91 85 91 91 Married 93 93 93 92 95 96 94 91 94 91 Divorced 74 75 70 71 78 82 76 53 76 83 Single 90 90 89 90 88 84 96 89 77 98 Which method? (%) IUD 47 44 52 44 53 56 46 48 44 43 Pill, drugs 19 21 15 20 16 14 21 17 22 20 Calendar 13 16 10 16 7 5 10 17 15 17 Injection 9 6 14 7 13 14 11 9 8 6 Condom 9 11 5 9 7 8 7 7 9 12 Others b/ 3 3 4 3 4 4 6 2 3 3 a/ Includes abstinence, withdrawal, patch, male or female sterilization, diaphragm, and spermicide. Source: 2002/03 HIES/LSMS. Table D.20: Abortions, all women 15 to 49 National Urban Rural Non-poor Poor Poorest Q2 Q3 Q4 Richest Ever had abortions? (%) 12 15 7 14 9 9 8 13 13 16 None, primary 3 4 3 3 4 4 2 5 3 2 Sec. 8th grade 7 8 6 6 8 9 5 6 6 7 Complete secondary 10 13 6 12 8 7 8 13 15 8 Vocational, tertiary 22 23 17 23 16 15 17 24 19 27 Married 19 25 12 21 14 15 13 20 21 25 Divorced 15 19 5 18 11 11 10 22 18 16 Single 1 2 0 1 1 0 1 2 1 2 Reasons for abortion (%) Due to health 29 24 43 29 29 28 29 25 27 34 Do not want a child 20 21 18 24 11 11 13 18 21 30 Too soon to give birth again 21 22 18 20 24 19 28 26 20 15 Lack of money 19 20 16 16 29 36 19 23 18 10 Others a/ 10 12 5 11 8 6 11 8 14 11 a/ Attending school, not married, others. Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 109 110 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Table D.21: Labor force participation and unemployment rates comparison Household survey Labor Administrative International Mongolian Force data standard standard Survey 2002 2002 2002 2003 Labor force participation rates National 61.6 65.2 67.7 62.7 Urban 53.0 57.1 56.8 n.a. Rural 73.2 76.0 81.2 n.a. Men 64.3 67.6 72.7 64.9 Women 59.1 62.9 62.8 60.5 Unemployment rates National 6.3 6.6 14.2 3.4 Urban 8.8 9.1 18.7 n.a. Rural 3.9 4.1 10.0 n.a. Men 6.4 6.5 14.2 3.1 Women 6.2 6.7 14.1 3.8 Source: 2002/03 HIES/LSMS, 2003 Labor Force Survey and National Statistical Office. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 111 Table D.22: Participation rates by gender Men Women Total Location Urban 58.2 56.1 57.1 Rural 79.5 72.5 76.0 Ulaanbaatar 57.2 54.8 56.0 Aimag centers 59.5 57.7 58.6 Soum centers 62.3 58.8 60.5 Countryside 89.6 81.5 85.7 West 77.5 68.0 72.7 Highland 77.2 71.7 74.4 Central a/ 62.9 59.4 61.1 East 70.6 67.4 69.0 Quintile Poorest 62.2 58.5 60.3 Q2 66.5 62.6 64.5 Q3 69.7 62.8 66.1 Q4 68.4 63.7 66.1 Richest 70.4 66.4 68.3 Poverty Non-poor 69.1 64.4 66.7 Poor 64.6 59.9 62.2 Education None 70.6 61.9 67.4 Primary 67.5 57.8 63.4 Secondary 8th grade 64.2 57.5 61.2 Complete secondary 58.9 52.2 55.1 Vocational 74.6 70.5 72.5 Higher diploma 82.3 81.1 81.6 University 82.5 83.2 82.9 National 67.6 62.9 65.2 a/ Excludes Ulaanbaatar. Source: 2002/03 HIES/LSMS. 112 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Table D.23: Participation rates by poverty status Non-poor Poor Total Location Urban 59.7 50.5 57.1 Rural 78.0 73.1 76.0 Ulaanbaatar 58.6 48.2 56.0 Aimag centers 61.2 52.8 58.6 Soum centers 60.8 60.0 60.5 Countryside 88.3 81.7 85.7 West 70.0 75.8 72.7 Highland 76.2 71.3 74.4 Central a/ 63.5 56.3 61.1 East 76.8 52.2 69.0 Gender Male 69.1 64.6 67.6 Female 64.4 59.9 62.9 Education None 68.1 66.7 67.4 Primary 67.1 59.2 63.4 Secondary 8th grade 63.5 58.0 61.2 Complete secondary 52.0 62.7 55.1 Vocational 73.9 70.0 72.5 Higher diploma 84.1 71.3 81.6 University 83.7 73.8 82.9 National 66.7 62.2 65.2 a/ Excludes Ulaanbaatar. Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 113 Table D.24: Population by labor force status As % of the variable of interest As % of the labor force status Employed Unemployed Out of the Total Employed Unemployed Out of the Total Labor Force Labor Force Location Urban 52.0 5.2 42.9 100.0 48.8 68.9 70.5 57.2 Rural 72.9 3.1 24.0 100.0 51.2 31.1 29.5 42.8 Ulaanbaatar 51.3 4.6 44.0 100.0 26.7 34.0 40.0 31.6 Aimag centers 52.7 5.9 41.4 100.0 22.1 34.9 30.5 25.6 Soum centers 55.3 5.2 39.5 100.0 14.9 19.9 18.6 16.4 Countryside 83.9 1.8 14.3 100.0 36.4 11.2 10.9 26.4 West 68.2 4.6 27.3 100.0 17.6 16.6 12.3 15.7 Highland 69.9 4.5 25.6 100.0 27.6 25.2 17.7 24.1 Central a/ 58.4 2.7 38.9 100.0 18.8 12.5 22.0 19.6 East 63.4 5.6 31.0 100.0 9.3 11.8 8.0 9.0 Quintile Poorest 53.5 6.7 39.7 100.0 15.9 28.3 20.6 18.1 Q2 58.7 5.8 35.5 100.0 18.6 26.2 19.7 19.3 Q3 61.4 4.7 33.9 100.0 20.2 21.9 19.5 20.0 Q4 63.1 3.0 33.9 100.0 22.0 14.6 20.7 21.2 Richest 66.5 1.8 31.7 100.0 23.4 9.1 19.5 21.4 Poverty Non-poor 63.5 3.3 33.3 100.0 69.4 50.8 63.7 66.6 Poor 55.9 6.3 37.8 100.0 30.6 49.2 36.3 33.4 Gender Men 63.2 4.4 32.4 100.0 51.0 50.3 45.8 49.2 Women 58.7 4.2 37.1 100.0 49.0 49.7 54.2 50.8 Age 16-24 34.0 5.0 61.0 100.0 19.0 39.3 59.6 34.0 25-34 73.2 4.8 22.0 100.0 31.8 29.6 16.7 26.5 35-44 79.2 4.5 16.3 100.0 30.7 24.8 11.1 23.6 45-54 71.0 1.8 27.2 100.0 16.1 5.8 10.8 13.8 55-59 b/ 69.8 1.0 29.2 100.0 2.4 0.5 1.8 2.1 Education None 64.1 3.3 32.6 100.0 3.7 2.7 3.3 3.5 Primary 61.1 2.3 36.6 100.0 10.3 5.5 10.9 10.3 Secondary 8th grade 56.2 4.9 38.8 100.0 26.9 33.5 32.5 29.1 Complete secondary 50.0 5.1 44.9 100.0 22.9 33.3 36.0 27.9 Vocational 68.1 4.4 27.5 100.0 9.6 8.8 6.8 8.6 Higher diploma 78.2 3.4 18.4 100.0 14.6 9.1 6.0 11.4 University 79.5 3.4 17.1 100.0 12.0 7.2 4.5 9.2 Total 60.9 4.3 34.8 100.0 100.0 100.0 100.0 100.0 a/ Excludes Ulaanbaatar. b/ Includes only men. Source: 2002/03 HIES/LSMS. 114 Table D.25: Industry, sector and occupation by urban-rural divide and gender Urban Rural National Men Women Total Men Women Total Men Women Total Industry Agriculture 8.0 7.8 7.9 77.6 75.4 76.6 45.1 41.0 43.1 Industry 23.9 14.8 19.2 3.9 2.6 3.3 13.2 8.8 11.1 Services 68.1 77.4 72.9 18.5 21.9 20.1 41.7 50.2 45.9 Agriculture 8.0 7.8 7.9 77.6 75.4 76.6 45.1 41.0 43.1 Mining 5.6 1.8 3.7 1.7 0.8 1.3 3.5 1.3 2.5 Manufacturing 6.1 8.1 7.1 0.8 1.2 1.0 3.2 4.8 4.0 Electricity/water 4.8 1.6 3.1 0.7 0.4 0.6 2.6 1.0 1.8 Contruction 7.4 3.2 5.3 0.7 0.2 0.5 3.8 1.7 2.8 Trade 10.8 17.7 14.3 1.9 3.4 2.6 6.1 10.7 8.3 Transportation 15.1 3.3 9.1 3.0 0.8 2.0 8.7 2.1 5.5 Business 6.9 6.4 6.6 1.6 0.7 1.2 4.1 3.6 3.8 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Public administration 14.0 8.4 11.1 4.6 1.7 3.3 9.0 5.1 7.1 Education 6.3 15.2 10.8 3.5 8.2 5.7 4.8 11.8 8.2 Health 2.0 10.5 6.4 1.2 4.1 2.5 1.6 7.4 4.4 Other 13.2 15.9 14.6 2.6 3.0 2.8 7.5 9.6 8.5 Sector Private 61.5 57.9 59.7 86.7 84.0 85.4 74.9 70.7 72.9 Public 32.4 38.8 35.7 11.3 14.9 13.0 21.2 27.1 24.1 State 6.0 3.3 4.6 2.1 1.1 1.6 3.9 2.2 3.1 Occupation Herders, farmers 6.7 6.8 6.7 75.0 73.4 74.3 43.1 39.4 41.3 Managers, senior officials and legislators 6.9 4.1 5.5 2.8 0.6 1.8 4.7 2.4 3.6 Professionals 12.1 24.3 18.3 2.6 7.3 4.8 7.0 16.0 11.4 Technicians and associate professionals 9.0 11.0 10.0 1.8 3.4 2.6 5.2 7.3 6.2 Clerks 1.8 5.5 3.7 0.9 2.1 1.5 1.3 3.8 2.5 Service workers, shop and market salespeople 14.3 24.1 19.3 2.5 5.8 4.0 8.0 15.1 11.5 Skilled agricultural and fishery workers 0.7 0.5 0.6 1.5 1.2 1.4 1.1 0.9 1.0 Craft and related trader workers 17.8 12.1 14.9 4.7 2.2 3.5 10.8 7.2 9.1 Plant and machine operators 20.4 1.7 10.8 5.0 0.3 2.8 12.2 1.0 6.7 Elementary occupations 7.7 7.5 7.6 2.3 3.1 2.7 4.8 5.4 5.1 Others 2.6 2.6 2.6 1.0 0.6 0.8 1.7 1.6 1.7 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: 2002/03 HIES/LSMS. Table D.26: Industry, sector and occupation by urban-rural divide and poverty status Urban Rural National Non-poor Poor Total Non-poor Poor Total Non-poor Poor Total Industry Agriculture 5.9 14.7 7.9 75.6 78.2 76.6 37.9 54.9 43.1 Industry 18.1 22.7 19.2 2.8 4.1 3.3 11.1 11.0 11.1 Services 76.0 62.6 72.9 21.6 17.7 20.1 51.0 34.2 45.9 Agriculture 5.9 14.7 7.9 75.6 78.2 76.6 37.9 54.9 43.1 Mining 3.8 3.4 3.7 1.6 0.9 1.3 2.8 1.8 2.5 Manufacturing 6.3 10.0 7.1 0.7 1.5 1.0 3.7 4.6 4.0 Electricity/water 3.4 2.4 3.1 0.3 0.9 0.6 2.0 1.5 1.8 Contruction 4.8 6.9 5.3 0.2 0.9 0.5 2.7 3.1 2.8 Trade 15.1 11.6 14.3 2.2 3.3 2.6 9.2 6.4 8.3 Transportation 9.3 8.5 9.1 2.0 1.9 2.0 5.9 4.3 5.5 Business 6.8 6.1 6.6 1.1 1.3 1.2 4.2 3.1 3.8 Public administration 12.1 7.9 11.1 3.8 2.4 3.3 8.3 4.4 7.1 Education 11.8 7.6 10.8 6.5 4.5 5.7 9.4 5.6 8.2 Health 6.3 6.4 6.4 3.1 1.5 2.5 4.9 3.3 4.4 Other 14.6 14.5 14.6 2.8 2.7 2.8 9.2 7.1 8.5 Sector Private 56.9 69.0 59.7 83.4 88.7 85.4 69.1 81.5 72.9 Public 37.8 28.7 35.7 14.8 10.0 13.0 27.2 16.8 24.1 State 5.3 2.4 4.6 1.8 1.3 1.6 3.7 1.7 3.1 Occupation Herders, farmers 4.6 14.0 6.7 73.9 74.9 74.3 36.4 52.5 41.3 Managers, senior officials and legislators 6.6 1.8 5.5 2.4 0.8 1.8 4.6 1.1 3.6 Professionals 21.6 7.5 18.3 6.1 2.7 4.8 14.5 4.5 11.4 Technicians and associate professionals 11.3 5.8 10.0 2.9 2.0 2.6 7.4 3.4 6.2 Clerks 4.2 2.0 3.7 1.5 1.5 1.5 2.9 1.6 2.5 Service workers, shop and market salespeople 18.5 22.0 19.3 4.1 3.9 4.0 11.9 10.5 11.5 Skilled agricultural and fishery workers 0.5 1.0 0.6 1.0 1.9 1.4 0.7 1.6 1.0 Craft and related trader workers 12.1 24.2 14.9 2.7 4.9 3.5 7.8 12.0 9.1 Plant and machine operators 11.6 8.2 10.8 2.9 2.7 2.8 7.6 4.7 6.7 Elementary occupations 6.6 11.0 7.6 2.1 3.6 2.7 4.5 6.3 5.1 Others 2.6 2.6 2.6 0.6 1.2 0.8 1.7 1.7 1.7 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 115 116 D. APPENDIX D: ADDITIONAL STATISTICAL TABLES Table D.27: Unemployment rates by gender Men Women Total Location Urban 9.5 8.6 9.1 Rural 3.7 4.6 4.1 Ulaanbaatar 9.0 7.5 8.3 Aimag centers 10.1 9.9 10.0 Soum centers 7.9 9.4 8.6 Countryside 2.0 2.3 2.1 West 6.6 5.9 6.3 Highland 5.0 7.2 6.0 Central a/ 3.8 5.2 4.5 East 9.1 7.2 8.2 Quintile Poorest 10.7 11.7 11.2 Q2 7.0 11.1 9.0 Q3 7.9 6.3 7.1 Q4 4.8 4.1 4.5 Richest 3.4 1.9 2.7 Age 16-24 13.1 12.5 12.8 25-34 5.1 7.2 6.2 35-44 5.5 5.3 5.4 45-54 3.4 1.7 2.5 55-59 b/ 1.4 - 1.4 Education None 6.8 1.2 4.9 Primary 3.2 4.2 3.6 Secondary 8th grade 8.5 7.5 8.1 Complete secondary 8.6 9.9 9.3 Vocational 6.1 6.1 6.1 Higher diploma 3.6 4.6 4.2 University 3.2 4.7 4.1 National 6.5 6.7 6.6 a/ Excludes Ulaanbaatar. b/ Includes only men. Source: 2002/03 HIES/LSMS. D. APPENDIX D: ADDITIONAL STATISTICAL TABLES 117 Table D.28: Unemployment rates by poverty status Non-poor Poor Total Location Urban 6.8 15.9 9.1 Rural 2.6 6.5 4.1 Ulaanbaatar 6.5 14.5 8.3 Aimag centers 7.1 17.2 10.0 Soum centers 5.1 13.8 8.6 Countryside 1.6 3.0 2.1 West 5.2 7.4 6.3 Highland 4.0 9.9 6.0 Central a/ 3.5 6.7 4.5 East 4.8 18.9 8.2 Gender Male 5.3 9.1 6.5 Female 4.5 11.4 6.7 Age 16-24 9.1 18.6 12.8 25-34 5.2 8.3 6.2 35-44 4.2 7.9 5.4 45-54 1.7 4.9 2.5 55-59 b/ 1.8 0.0 1.4 Education None 2.2 7.6 4.9 Primary 2.0 5.8 3.6 Secondary 8th grade 5.3 12.1 8.1 Complete secondary 7.6 12.7 9.3 Vocational 5.1 7.9 6.1 Higher diploma 3.7 6.6 4.2 University 3.7 8.9 4.1 National 4.9 10.2 6.6 a/ Excludes Ulaanbaatar. b/ Includes only men. Source: 2002/03 HIES/LSMS. E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS 120 E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS Table E.1: Poverty and urban-rural divide Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect Headcount National 36.11 1.43 33.31 38.92 2.92 3,308 Urban 30.27 1.70 26.92 33.62 2.52 1,851 Rural 43.38 2.37 38.73 48.04 3.36 1,457 Poverty Gap National 10.99 0.60 9.82 12.16 3.40 3,308 Urban 9.20 0.71 7.81 10.58 2.93 1,851 Rural 13.22 1.00 11.26 15.18 3.86 1,457 Severity National 4.67 0.33 4.02 5.32 3.33 3,308 Urban 3.97 0.40 3.18 4.75 2.92 1,851 Rural 5.55 0.55 4.48 6.62 3.74 1,457 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS 121 Table E.2: Poverty and geography Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect Headcount West 51.06 3.48 44.21 57.90 2.72 527 Highland 38.68 2.86 33.05 44.30 2.75 849 Central 34.40 2.99 28.53 40.27 2.55 697 East 34.54 4.36 25.97 43.12 2.58 332 Ulaanbaatar 27.27 2.55 22.25 32.28 3.27 903 Poverty Gap West 14.58 1.34 11.95 17.20 2.78 527 Highland 12.26 1.27 9.77 14.75 3.46 849 Central 10.11 1.38 7.40 12.83 3.75 697 East 12.36 2.29 7.86 16.86 3.18 332 Ulaanbaatar 8.11 0.98 6.19 10.02 3.57 903 Severity West 5.73 0.65 4.44 7.01 2.43 527 Highland 5.19 0.69 3.83 6.55 3.38 849 Central 4.31 0.81 2.73 5.90 3.96 697 East 6.58 1.56 3.51 9.64 3.22 332 Ulaanbaatar 3.32 0.48 2.37 4.26 3.37 903 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. 122 E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS Table E.3: Poverty and analytical domains Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect Headcount Ulaanbaatar 27.27 2.55 22.25 32.28 3.27 903 Aimag centers 33.86 2.19 29.55 38.16 1.79 948 Soum centers 44.53 3.01 38.62 50.44 1.96 753 Countryside 42.73 3.30 36.25 49.21 4.17 704 Poverty Gap Ulaanbaatar 8.11 0.98 6.19 10.02 3.57 903 Aimag centers 10.50 1.02 8.49 12.52 2.40 948 Soum centers 14.37 1.54 11.34 17.40 2.92 753 Countryside 12.56 1.30 10.01 15.11 4.53 704 Severity Ulaanbaatar 3.32 0.48 2.37 4.26 3.37 903 Aimag centers 4.74 0.66 3.45 6.03 2.66 948 Soum centers 6.42 0.92 4.62 8.22 3.01 753 Countryside 5.06 0.68 3.72 6.39 4.37 704 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS 123 Table E.4: Poverty and seasonality Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect Headcount Quarter I 29.09 3.03 23.14 35.05 3.59 814 Quarter II 40.34 2.66 35.12 45.56 2.25 757 Quarter III 33.54 2.79 28.05 39.03 2.95 859 Quarter IV 41.23 2.90 35.53 46.93 3.10 878 Poverty Gap Quarter I 7.97 1.01 5.99 9.95 3.32 814 Quarter II 11.71 1.11 9.53 13.89 2.68 757 Quarter III 10.34 1.16 8.06 12.62 3.37 859 Quarter IV 13.70 1.35 11.04 16.36 3.85 878 Severity Quarter I 3.09 0.54 2.02 4.16 3.31 814 Quarter II 4.91 0.59 3.75 6.07 2.40 757 Quarter III 4.43 0.59 3.27 5.60 3.01 859 Quarter IV 6.12 0.80 4.56 7.69 3.99 878 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. 124 E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS Table E.5: Poverty and gender of the household head Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect Headcount Men 34.84 1.51 31.88 37.81 2.84 2,733 Women 43.77 2.98 37.92 49.63 1.70 575 Poverty Gap Men 10.29 0.60 9.10 11.48 3.20 2,733 Women 15.21 1.43 12.41 18.02 2.05 575 Severity Men 4.29 0.33 3.64 4.94 3.15 2,733 Women 6.99 0.84 5.34 8.63 2.10 575 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS 125 Table E.6: Poverty and highest education level completed by the household head Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect Headcount None 45.75 4.88 36.17 55.34 1.34 169 Primary 45.60 3.57 38.58 52.62 2.41 435 Secondary 8th grade 45.47 2.30 40.96 49.98 1.93 841 Complete secondary 34.86 2.34 30.26 39.45 1.50 646 Vocational 40.67 3.38 34.03 47.31 1.60 335 Higher diploma 23.35 2.47 18.50 28.21 1.53 476 University 11.60 2.13 7.42 15.78 1.68 406 Poverty Gap None 12.79 1.74 9.37 16.20 1.32 169 Primary 16.35 1.75 12.91 19.79 2.75 435 Secondary 8th grade 13.81 0.92 12.00 15.61 2.01 841 Complete secondary 9.29 0.88 7.57 11.01 1.77 646 Vocational 13.07 1.47 10.17 15.97 1.71 335 Higher diploma 6.74 0.93 4.91 8.58 1.72 476 University 2.93 0.65 1.64 4.21 1.65 406 Severity None 4.84 0.88 3.11 6.56 1.27 169 Primary 7.90 1.07 5.79 10.01 2.61 435 Secondary 8th grade 5.72 0.47 4.79 6.65 1.85 841 Complete secondary 3.56 0.45 2.68 4.44 1.81 646 Vocational 6.00 0.95 4.14 7.86 1.97 335 Higher diploma 2.74 0.55 1.66 3.81 1.92 476 University 1.06 0.30 0.47 1.65 1.58 406 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. 126 E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS Table E.7: Poverty and type of dwelling Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect Headcount Gers 43.39 2.19 39.09 47.69 2.85 1,403 Houses 38.17 1.91 34.41 41.92 1.79 1,192 Apartments 16.62 2.29 12.13 21.11 2.48 679 Others 29.99 6.68 16.87 43.11 0.61 34 Poverty Gap Gers 13.48 0.92 11.68 15.28 3.17 1,403 Houses 11.31 0.86 9.62 13.00 2.43 1,192 Apartments 4.99 1.09 2.86 7.13 3.88 679 Others 9.09 2.47 4.23 13.95 0.73 34 Severity Gers 5.68 0.49 4.71 6.65 3.02 1,403 Houses 4.82 0.51 3.81 5.82 2.58 1,192 Apartments 2.25 0.66 0.96 3.55 4.50 679 Others 3.23 1.29 0.69 5.77 0.91 34 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS 127 Table E.8: Poverty, type of dwelling and urban-rural divide Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect Headcount Urban Gers 47.51 3.25 41.13 53.89 1.62 400 Urban Houses 33.90 2.16 29.65 38.14 1.71 792 Urban Apartments 14.34 2.11 10.19 18.48 2.18 632 Urban Others 31.22 7.06 17.34 45.09 0.60 27 Rural Gers 41.92 2.73 36.55 47.30 3.30 1,003 Rural Houses 48.48 3.74 41.12 55.83 1.90 400 Rural Apartments 41.84 10.41 21.39 62.30 2.44 47 Rural Others 20.05 18.42 -16.14 56.24 0.67 7 Poverty Gap Urban Gers 14.72 1.42 11.92 17.51 1.93 400 Urban Houses 10.49 1.04 8.43 12.54 2.50 792 Urban Apartments 3.92 0.74 2.47 5.37 2.25 632 Urban Others 9.65 2.74 4.26 15.03 0.75 27 Rural Gers 13.04 1.13 10.81 15.26 3.65 1,003 Rural Houses 13.30 1.50 10.34 16.25 2.25 400 Rural Apartments 16.79 8.00 1.08 32.51 5.45 47 Rural Others 4.58 4.21 -3.69 12.85 0.67 7 Severity Urban Gers 6.19 0.77 4.67 7.71 1.83 400 Urban Houses 4.67 0.65 3.39 5.94 2.73 792 Urban Apartments 1.62 0.38 0.88 2.36 2.18 632 Urban Others 3.50 1.47 0.61 6.38 0.95 27 Rural Gers 5.50 0.61 4.29 6.70 3.49 1,003 Rural Houses 5.18 0.77 3.67 6.69 2.11 400 Rural Apartments 9.24 5.28 -1.14 19.62 5.44 47 Rural Others 1.05 0.96 -0.84 2.94 0.67 7 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. 128 E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS Table E.9: Poverty and livestock holdings Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect NATIONAL Headcount Non-herders 34.56 1.58 31.47 37.66 2.28 2,205 Herders 38.73 2.61 33.59 43.86 3.55 1,103 Poverty Gap Non-herders 10.86 0.70 9.49 12.24 2.81 2,205 Herders 11.20 1.02 9.20 13.19 3.97 1,103 Severity Non-herders 4.80 0.41 3.99 5.61 2.93 2,205 Herders 4.45 0.53 3.41 5.49 3.90 1,103 URBAN-RURAL Headcount Urban Non-herders 29.92 1.76 26.47 33.38 2.46 1,680 Urban Herders 33.71 5.14 23.61 43.82 1.97 171 Rural Non-herders 53.49 3.17 47.26 59.72 1.65 525 Rural Herders 39.51 2.91 33.78 45.23 3.78 932 Poverty Gap Urban Non-herders 9.23 0.73 7.79 10.66 2.81 1,680 Urban Herders 8.88 2.31 4.35 13.42 3.26 171 Rural Non-herders 17.53 1.75 14.10 20.96 2.51 525 Rural Herders 11.56 1.12 9.36 13.76 4.08 932 Severity Urban Non-herders 4.01 0.42 3.19 4.84 2.89 1,680 Urban Herders 3.50 1.25 1.05 5.95 3.28 171 Rural Non-herders 8.03 1.09 5.88 10.17 2.69 525 Rural Herders 4.60 0.58 3.46 5.74 4.00 932 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS 129 Table E.10: Poverty and access to improved water sources Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect NATIONAL Headcount Yes 33.04 1.65 29.81 36.28 2.47 2,099 No 40.92 2.40 36.20 45.65 3.08 1,209 Poverty Gap Yes 9.91 0.66 8.63 11.20 2.67 2,099 No 12.67 1.01 10.69 14.65 3.48 1,209 Severity Yes 4.24 0.35 3.55 4.93 2.41 2,099 No 5.36 0.56 4.25 6.46 3.54 1,209 URBAN-RURAL Headcount Urban Yes 28.27 1.89 24.55 32.00 2.63 1,492 Urban No 38.77 3.33 32.23 45.32 1.63 359 Rural Yes 46.32 3.07 40.28 52.36 2.03 607 Rural No 41.72 3.05 35.73 47.71 3.59 850 Poverty Gap Urban Yes 8.56 0.74 7.11 10.01 2.73 1,492 Urban No 11.92 1.54 8.89 14.95 2.21 359 Rural Yes 13.68 1.35 11.04 16.33 2.49 607 Rural No 12.95 1.25 10.49 15.41 3.93 850 Severity Urban Yes 3.69 0.40 2.90 4.47 2.49 1,492 Urban No 5.16 0.87 3.45 6.86 2.22 359 Rural Yes 5.76 0.72 4.36 7.17 2.21 607 Rural No 5.43 0.70 4.06 6.81 4.04 850 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. 130 E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS Table E.11: Poverty and access to improved sanitation facilities Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect NATIONAL Headcount Yes 30.19 1.68 26.88 33.50 2.31 1,819 No 42.54 2.08 38.46 46.62 2.80 1,489 Poverty Gap Yes 9.05 0.69 7.70 10.40 2.69 1,819 No 13.10 0.86 11.40 14.79 3.06 1,489 Severity Yes 3.84 0.37 3.12 4.56 2.55 1,819 No 5.57 0.49 4.62 6.53 2.98 1,489 URBAN-RURAL Headcount Urban Yes 26.03 1.91 22.27 29.79 2.55 1,356 Urban No 41.88 2.78 36.42 47.34 1.55 495 Rural Yes 44.99 3.26 38.58 51.41 1.62 463 Rural No 42.83 2.73 37.46 48.20 3.35 994 Poverty Gap Urban Yes 7.67 0.74 6.22 9.13 2.84 1,356 Urban No 13.37 1.31 10.81 15.94 1.94 495 Rural Yes 13.93 1.59 10.81 17.05 2.25 463 Rural No 12.97 1.10 10.80 15.14 3.66 994 Severity Urban Yes 3.19 0.39 2.43 3.94 2.67 1,356 Urban No 6.10 0.83 4.46 7.73 2.11 495 Rural Yes 6.16 0.90 4.39 7.93 2.20 463 Rural No 5.34 0.60 4.17 6.52 3.57 994 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS 131 Table E.12: Poverty and access to electricity Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect NATIONAL Headcount Yes 34.02 1.46 31.15 36.89 2.30 2,595 No 41.82 3.22 35.50 48.13 3.78 713 Poverty Gap Yes 10.32 0.61 9.11 11.53 2.78 2,595 No 12.81 1.30 10.25 15.37 3.85 713 Severity Yes 4.35 0.33 3.69 5.01 2.77 2,595 No 5.55 0.72 4.13 6.97 3.41 713 URBAN-RURAL Headcount Urban Yes 29.64 1.70 26.31 32.98 2.49 1,815 Urban No 63.51 8.76 46.29 80.74 1.12 36 Rural Yes 46.74 2.72 41.40 52.09 1.84 780 Rural No 40.96 3.32 34.43 47.49 3.90 677 Poverty Gap Urban Yes 8.81 0.66 7.51 10.11 2.71 1,815 Urban No 29.88 7.28 15.58 44.18 1.82 36 Rural Yes 14.71 1.36 12.04 17.38 2.77 780 Rural No 12.14 1.32 9.55 14.72 4.18 677 Severity Urban Yes 3.69 0.35 3.01 4.37 2.52 1,815 Urban No 18.77 5.98 7.02 30.51 1.88 36 Rural Yes 6.28 0.80 4.71 7.85 3.07 780 Rural No 5.03 0.71 3.64 6.41 3.91 677 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. 132 E. APPENDIX E: STANDARD ERRORS AND CONFIDENCE INTERVALS OF POVERTY ESTIMATIONS Table E.13: Poverty and joint access to improved water sources, sanitation facilities and electricity Survey mean estimation Number of obs = 3,308 Number of strata = 4 Number of PSUs = 460 Population size = 2,328,812 Estimate Std. Err. [95% Conf. Interval] Design Obs. effect NATIONAL Headcount Yes 26.90 1.81 23.35 30.46 2.31 1,456 No 42.78 1.92 39.00 46.56 2.90 1,852 Poverty Gap Yes 7.87 0.68 6.53 9.21 2.41 1,456 No 13.25 0.82 11.64 14.86 3.28 1,852 Severity Yes 3.29 0.35 2.60 3.98 2.21 1,456 No 5.67 0.46 4.76 6.59 3.23 1,852 URBAN-RURAL Headcount Urban Yes 24.00 2.00 20.06 27.94 2.55 1,173 Urban No 41.00 2.43 36.23 45.77 1.64 678 Rural Yes 41.42 3.84 33.86 48.97 1.41 283 Rural No 43.75 2.65 38.54 48.96 3.55 1,174 Poverty Gap Urban Yes 6.97 0.74 5.53 8.42 2.62 1,173 Urban No 13.01 1.17 10.71 15.30 2.24 678 Rural Yes 12.34 1.73 8.95 15.74 1.82 283 Rural No 13.38 1.09 11.24 15.52 3.89 1,174 Severity Urban Yes 2.88 0.38 2.15 3.62 2.37 1,173 Urban No 5.82 0.73 4.38 7.25 2.46 678 Rural Yes 5.33 0.93 3.51 7.15 1.74 283 Rural No 5.59 0.60 4.42 6.77 3.76 1,174 Note: Poverty measures were calculated taking into account the survey design i.e. the strata and primary sampling units. Estimations were done at the household level but considering population weights. Source: 2002/03 HIES/LSMS. Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. 133 LIST OF PARTICIPANTS OF THE SURVEY Report writers: The World Bank team Ludovico Carraro The World Bank consultant Martin Cumpa The World Bank consultant The NSO team D.Oyunchimeg Deputy director of the PSSD Yu.Tuul Senior officer of the PSSD, PhD in economics Ts.Amartuvshin Officer of the PSSD B.Enerelt Officer of the PSSD Translation editor: B.Munkhjargal Senior officer of the NSO in charge of foreign affairs Technical consultants: Valerie Evans The World Bank consultant Ludovico Carraro The World Bank consultant Martin Cumpa The World Bank consultant Juan Munoz The World Bank consultant Steering commitee: P.Byambatseren Chairman of the NSO, head of the steering committee N.Gerelsuren Former parliament member S.Chinzorig National director of the project "MON/01/U01", Poverty research and employment facilitation for policy development, vice minister of the Ministry of Social Welfare and Labor Ch.Khurelbaatar State secretary of ministry of Finance B.Bayart Director of Information monitoring and evaluation department of the Ministry of Health N.Narangerel Director of Population and social department of the Ministry of Social Welfare and Labour J.Jargalsaikhan Director of General policy and planning department of the Ministry of Finance Ts.Gelegjamts Head of Monitoring, evaluation, and pressing coordination division of the Ministry of Education, Culture and Science P.Bolormaa National coordinator of the project MON/01/U01 j.Demberelsuren Director of Population and Health institute, Ministry of Health D.Narmandakh Vice President of Mongolian Trade Unions 134 Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. M.Sarantuya Expert of UNDP V.Tsolmon Officer of the World Bank Survey working group at the NSO: Head of the working group D.Oyunchimeg Deputy director of the PSSD, NSO Members: Yu.Tuul Senior officer of the PSSD, NSO, PhD in economics Ts.Amartuvshin Officer of the PSSD, NSO P.Baigalmaa Officer of the DPSDD, NSO L.Ganzaya Officer of the PSSD, NSO L.Myagmar Officer of the DPSDD, NSO B.Sarangerel Senior officer of the MBSD, NSO Ch.Enkhbayar Officer of the PSSD, NSO B.Enerelt Officer of the PSSD, NSO Field staff for the LSMS: Supervisors Ts.Amartuvshin S.Jambaldorj B.Tuul Ch.Ganchimeg S.Oyuntsetseg Yu.Tuul G.Jiidiimaa B.Sarangerel Ts.Chimeddorj Interviewers N.Altantuya P.Delgermuron Kh.Tumorsukh M.Altantsetseg Ts.Davaa-Ulzii B.Khumbaa B.Batbayar Sh.Dorjkhand G.Kherlen Ts.Batbayar A.Zolboo B.Chuluuntsetseg G.Batbuyan O.Itgel L.Elbegsaikhan S.Bat-Oyun D.Odgerel Ch.Enkhbayar M.Ganbayar B.Oyuntuya D.Enkhjargal G.Davaasuren Kh.Oyuntsetseg B.Enerelt Yu.Alt-Ochir Field staff for the HIES: Arkhangai aimag: Supervisor B.Chuluuntsetseg Interviewers: Ts.Tserensodnom Ch.Sarantuya D.Batsaikhan P.Gombodorj B.Dalaitsogt B.Batbilegt S.Rentsenbal D.Narantuya Bayan-Ulgii aimag: Kh.Umirzakh Interviewers: A.Shepen Kh.Emusiz Kh.Beisen B.Estai A.Khajiimurat Kh.Khabiil Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. 135 Bayankhongor aimag: Supervisor Kh.Oyunchimeg Interviewers: T.Elbegdorj L.Gal B.Davaasuren P.Gerelee A.Batbold N.Tsetsgee D.Amgalan Bulgan aimag: Supervisor Ch.Ariun Interviewers: Ts.Batsaikhan Ch.Buyankhishig G.Tsogt O.Otgonsuren S.Uurtsaikh Ts.Batbayar Govi-Altai aimag: Supervisor L.Ariuntuya Interviewers: M.Enebish B.Jargalsaikhan J.Chantsaldulam Sh.Myanganbayar Sh.Guliraanz Dornogovi aimag: Supervisor D.Munkhtuya Interviewers: Ch.Narangerel D.Banzragch Yo.Kolya Sh.Gantsetseg Dornod aimag: Supervisor D.Enkhbaatar Interviewers: Z.Tsendem G.Tsend-Ayush D.Tuyatsetseg M.Altantsetseg G.Tsolmon Dundgovi aimag: Supervisor G.Bayasgalan Interviewers: Ts.Dorjpagma J.Zorigt N.Tsetsegmaa B.Munkhbayar Ts.Erdenebayar Zavkhan aimag: Supervisor S.Suvdaa Interviewers: B.Damlansuren N.Janchiv J.Bazarvaani R.Davaajantsan G.Odkhuu Sh.Michiddorj D.Oyungerel Uvorkhangai aimag: Supervisor B.Delger Interviewers: Ch.Galbadrakh Ts.Biziya B.Urtnasan J.Dalantai L.Gantulga D.Dolgorsuren Ts.Tsoodol Yo.Dugersuren Umnogovi aimag: Supervisor Ts.Jugarsuren Interviewers: D.Doljin G.Tsevelsuren J.Tsoodol D.Uulii Sukhbaatar aimag: Supervisor Ts.Khad Interviewers: A.Davaasuren B.Ankhbayar Z.Ichinkhorloo L.Avaadorj N.Gombosuren Selenge aimag: Supervisor N.Oyunaa Interviewers: S.Luvsanperenlei Ts.Baigalmaa D.Enkhtuya G.Tsetsegmaa P.Chuluuntsetseg P.Gejee D.Otgon Tuv aimag: Supervisor S.Undrakhbayar Interviewers: S.Dulamsuren D.Ayush L.TSogtgerel O.Damdinjav Ch.Sukhbaatar Sh.Tsagaankhuu 136 Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. D.Otgonbaatar Ch.Battsesem Uvs aimag: Supervisor D.Erdene Interviewers: Z.Bumnanjid B.Davaa Kh.Namsrai T.Tsevegmed G.Munkhbat S.Lhagvasuren B.Tsevegmed Khovd aimag: Supervisor A.Badamgarav Interviewers: T.Chuluunbaatar Ch.Bereevenjav Ch.Batnasan Ch.Dorj Khuvsgol aimag: Supervisor Ch.Barsbold Interviewers: L.Enebish V.Enkhtuya B.Dovchindorj D.Batjargal G.Bagsamshin T.Altantsetseg S.Altanchimeg M.Saikhanbayar S.Narantsetseg L.Unorsaikhan Khentii aimag: Supervisor B.Gantugs Interviewers: E.Tsetsegdelger B.Narantsetseg E.Chimegbaatar D.Oyun-Erdene S.Tumurtogoo B.Oyunbaatar Ch.Sarantuya Darkhan-Uul aimag: Supervisor Z.Amarsaikhan Interviewers: S.Enkh-Amgalan Sh.Roza L.Jamiyansuren B.Urtnasan M.Nansalmaa Ch.Oyungerel D.Tserendulam Orkhon aimag: Supervisor B.Tuul Interviewers: Ch.Darvadorj S.Lhagvasuren B.Olzvoi Ch.Oyunbileg Govisumber aimag: Interviewer: Ts.Narantsatsral Ulaanbaatar city: Supervisor: B.Tuul Bayangol district: Supervisor Yu.Altantuya Interviewers: Yu.Altantsetseg E.Tumendemberel Ts.Mandal N.Dumbormaa T.Soyolmaa Ch.Munkhbaatar T.Narangarav Songinokhairkhan district: Supervisor D.Enkhbayar Interviewers: L.Yanjmaa Ts.Chimeddorj A.Batsaikhan S.Ulziisaikhan B.Khombogo D.Dolgorsuren S.Uranchimeg Bayanzurkh district: Supervisor D.Narangerel Interviewers: A.Adiya D.Urantsetseg T.Mungonshagai J.Tsogtsolmaa G.Ariuntuya N.Bazargardoo Yu.Alt-Ochir Chingeltei district: Supervisor Ch.Oyunjargal Interviewers: S.Sodgerel B.Tungalag Ts.Tuya J.Galya G.Enkhjargal Sukhbaatar district: Supervisor B.Khandmaa Interviewers: J.Gantumor S.Urjinbadam S.Tsogt-Ochir Main Report of "Household Income and Expenditure Survey/Living Standards Measurement Survey" 2002-2003. 137 Kh.Enkhbulag Khan-Uul district: Supervisor Ts.Enkh-Oyun Interviewers: Ts.Oyunbat Ts.Oyuntsetseg D.Mijiddorj J.Ayurzana Nalaikh district: Interviewers: A.Gantsetseg Sh.Erdenechimeg Baganuur district: Interviewers: D.Jargalmaa D.Enkhbaatar Data entry persons: R.Delgermuron S.Tserensoli G.Erdenetsetseg P.Yanjmaa