Cambodia - Poverty assessment

Report No. 19858-KH Cambodia Poverty Assessment November 22, 1999 Poverty Reduction and Economic Management Sector Unit and Human D)evelopment Sector Unit East Asia and Pacific Region Document of the World Bank  Currency Evaluations Currency unit = Cambodian riel Average value of US$1.00 in 1995 1996 1997 1998 Sept 30, 1999 2,451 2,624 2,938 3,750 3,900 Weights and Measures Metric system Government Fiscal Year January 1-December 31 Abbreviations and Acronmys AIDS Acquired Immune Deficiency Syndrome CR Cambodian riel DPT Diphtheria, pertussis, and tetanus FAO Food and Agriculture Organization GDP Gross domestic product HIV Human Immuno-deficiency Virus ICRC International Committee of the Red Cross IRIC Indochinese Refugee Information Center NGO Nongovernmental organization NIS National Institute of Statistics UNDP United Nations Development Programme UNICEF United Nations Children's Fund UNHCR United Nations High Commission for Refugees UNTAC United Nations Transitional Authority in Cambodia USAID United States Agency for International Development WFP World Food Programme WHO World Health Organization Vice President Jean Michel Severino, EAP Chief Economist Masahiro Kawai, EAP Country Director Ngozi Okonjo-Iweala, EASCM Sector Manager Homi Kharas, EASPR Task Managers Nicholas Prescott, EASHD, and Victoria Kwakwa, EASPR  Acknowledgments This report is the result of collaboration between the Royal Govern-ment of Cambodia, the United Nations Development Program (UNDP), the Swedish International Development Agency (SIDA) and the WVorld Bank. Preparation of the 1997 Cambodia Socio-Economic Survey which forms the basis for the report was funded by the UNDP and by SIDA under the Capacity Development for Socio-Economic Surveys and Planning Project. The work at the Bank was led by Nicholas Prescott and Victoria Kwakwa with contributions from James Knowles (Poverty Profile), Anil Deolalikar (Social Sectors), Sophal Ear (Post Conflict and Rural Development), Markus Kostner (Military Demobilization) and Ky Tran (Data and Graphics). The peer reviewers were Lionel Demery and Giovanna Prennushi. Comments were also received from Martha Ainsworth. The report was prepared under the guidance of Masahiro Kawai (Chief Economist, East Asia and Pacific Region), Ngozi Okonjo-Iweala (Country Director, EACSM), Homi Kharas (Sector Manager, EASPR), and Tamar Manuelyan Atinc (Lead Specialist, EASPR). The manuscript was edited by Meta de Coquereaumont and processed by Muriel Greaves and Nancy Mensah (EASPR). Bonaventure Mbida- Essama, Acting Chief of the World Bank Office in Phnom Penh led the discussion of the report with the Royal Government of Cambodia in September 1999.   Contents Executive Summary 1 1. A Poverty Profile of Cambodia 9 The 1993-94 Baseline Poverty Profile 9 The 1997 Socioeconomic Survey 10 Updated Poverty Estimates 10 Socioeconomic Characteristics of the Poor 11 Household Size and Composition 11 Age and Gender of Household Heads 12 Education and Literacy 12 Access to Housing and Consumer Durables 13 Where the Poor Live 13 Key Sources of Livelihood for the Poor 13 Inequality 14 Poverty Indicators 14 Conclusion 17 Notes 17 2. Legacy of War and Social Upheaval 19 Internally Displaced Persons and Returnees 19 Repatriation 19 Reintegration Success and Poverty 20 Large Numbers of People with Disabilities 22 Assistance to People with Disabilities 23 Land Mines and Unexploded Ordnance 23 Large Public Expenditures on Defense 24 D)emobilization and a Shift from Defense to Social Spending 25 Key Issues for Government Action 26 Notes 26 3. Economic Recovery and Poverty 27 Recovering from War, 1991-96 27 Domestic Political Upheaval and the Asian Crisis, 1997-98 29 I-las Poverty Declined during the 1990s? 31 Prospects for Growth and Poverty Reduction 33 Conclusion 35 iii  ;V CAMBODIA POVERTY ASSEssMENT 4. Schooling and the Poor 37 School Enrollment 37 Rural and Urban Disparities 39 Disparity across Economic Groups 39 Gender Disparity 39 Role of the Public Sector 40 Determinants of Enrollment among the Poor 40 High Private Costs of Schooling 40 Proximity to Schools 43 School Quality 43 Parents' Schooling 44 Returns to Schooling 45 Multivariate Analysis of School Enrollment 46 Raising Enrollment of the Poor 47 Notes 48 5. Access to Healtlh Services 51 Some Improvements in Health Status 51 The Growing HIV/AIDS Epidemic 52 Patterns of Use and Access of the Poor 53 Curative Services 53 Preventive Services 54 Hospital Admissions 55 Role of the Public Sector 55 What Determines Health Service Use? 56 Private Costs of Health Service Use 57 Cost of Hospital Admission 58 Fee Exemptions 60 Physical Access to Health Providers andc Facilities 60 Other Determinants of Health Service Use 61 Multivariate Analysis of Health Services Utilization 62 What Government Can Do to Improve the Health of the Poor 63 Notes 65 6. Issues in Rural Poverty 67 Raising Rural Incomes 67 Agricultural Productivity 68 Inadequate Rural Infrastructure 70 Marketing and Distribution 70 Access to Credit and Land 71 Ensuring Food Security 73 7. Achieving Cambodia's Poverty Reduction Goals 75 Elements of a Poverty Reduction Strategy for Cambodia 75 Expanding Economic Opportunities 76 Ensuring Macroeconomic Stability 76 Promoting Private Sector-Led Growth 76 Promoting Agriculture and Rural Sector Growth 77 Building Human Assets of the Poor 77 Providing Safety Nets 78 The Government's Socioeconomic Development Plan 79  CONTENTS v Appendix A. Statistical Tables 81 Appendix B. Delayed Entry into School and Ministry of Education and Survey Data Differences 103 Appendix C. Multivariate Analysis of Schooling Outcomes 105 Appendix D. Multivariate Analysis of School Enrollment 115 Appendix E. Health Outcomes 119 Appendix F. Multivariate Analysis of Health Services Use 121 References 127 Boxes 2.1 Access to social networks 22 3.1 Major structural and macroeconomic reforms in Cambodia 28 6.1 Income structure in three rural villages 68 6.2 Some natural causes of low agricultural productivity in Cambodia 69 6.3 Some agricultural successes 70 6.4 Rice indebtedness in rural communities 74 Figures 1.1 Cumulative distribution of per capita consumption, Phnom Penh, other urban areas, and rural areas, June 1997 11 1.2 Poverty by household size, 1997 11 1.3 Poverty by gender of head of household, 1997 12 1.4 Poverty by education of head of household, 1997 13 1.5 Poverty by literacy of head of household, 1997 13 1.6 Poverty by employer of head of household, 1997 14 1.7 Poverty by industry of employment of head of household, 1997 15 3.1 Domestic exports and investment, 1991-96 28 3.2 Consumer price index and exchange rate 30 4.1 Mothers' level of completed schooling, by consumption quintile and region, 1997 45 5.1 Infant mortality rate, Cambodia, 1960-96 52 5.2 Life expectancy and infant mortality in Asia, 1995 52 5.3 Budget shares devoted to medical care, by consumption quintile and region, 1997 59 5.4 Treatment of illness in previous four weeks, for people 15 or older, by level of schooling, 1997 62 Tables 1.1 Poverty measures, June 1997 11 1.2 Consumption inequality measures, June 1997 15 1.3 Gini coefficients for selected countries, most recent year 16 2.1 Sectoral composition of expenditures, 1996-98 24 2.2 Actual expenditures as percentage of budgeted expenditures 24 3.1 GDP growth, 1992-98 29 3.2 Government operations, 1996-98 31 3.3 Summary indicators of per capita consumption, 1993-94 and June 1997 32 3.4 Social indicators, various years 33 3.5 Reduction of poverty incidence under different growth scenarios 33 3.6 Comparison of poverty gap index for Cambodia, Lao PDR, and Vietnam 34  vi CAM80DIA POVERTY AssEssMENT 4.1 Enrollment ratios for selected Asian countries, 1995 38 4.2 School enrollment by consumption quintile and region, 1997 38 4.3 Gross primary enrollments in Cambodia,Vietnam, and Lao PDR 39 4.4 Distribution of enrolled students and school-age population by consumption quintile, 1997 39 4.5 Male and female enrollment rates, 1997 39 4.6 Private schooling enrollment by consumption quintile and region, 1997 40 4.7 Affordability of public schooling by consumption quintile and region, 1997 42 4.8 Budget shares devoted to education, by consumption quintile and region, 1997 42 4.9 School quality indicators by consumption quintile and region, 1997 44 4.10 Net, gross, and age-specific enrollment rates, by mothers' level of schooling, 1997 46 4.11 Real monthly earnings, by gender, age, and schooling level, 1997 46 5.1 Immunization coverage, 1992-97 51 5.2 Cases of disease reported to the ExpaLnded Program of Immunization, 1992-97 51 5.3 Percentage of children born in the previous 24 months who where immunized, by consumption quintile and region, 1997 55 5.4 Inpatient admissions in the previous 12 months by consumption quintile and region, 1997 55 5.5 Public and private sector shares of health service utilization by consumption quintile and region, 1997 56 5.6 Affordability ratios for health services by consumption quintile and region, 1997 58 5.7 Average out-of-pocket spending on inpatient admissions in the previous 12 months by consumption quintile and region, 1997 59 5.8 Availability of health facilities and programs in rural areas by consumption quintile, 1997 61 5.9 Average distance to health facilities for the rural population by consumption quintile, 1997 61 5.10 Mothers' schooling and child immun:ization, 1997 62 5.11 Concentration of monthly health service contacts at public health facilities by consumption quintile, 1997 64 6.1 Rice paddy yields in selected Asian countries, 1990-98 69 6.2 Rice supply, 1993/94 to 1998/99 '73 7.1 Major targets for social development by 2000 75 Al.1 Social indicators by per capita consumption quintile 81 A1.2 Quantitative poverty indicators, June 1997 82 A1.3 Qualitative poverty indicators 83 A1.4 Comparison of CPI and Socioeconomic Survey retail prices in Phnom Penh, June 1997 84 A1.5 Comparison of Consumer Price Index and Ministry of Agriculture retail prices in Phnom Penh, June 1997 84 A1.6 Distribution of poverty by region, 1993-94 and June 1997 85 Al.7 Distribution of poverty by employment of household head, June 1997 85 A1.8 Distribution of poverty by employer of household head, June 1997 86 A1.9 Distribution of poverty by education of household head, June 1997 86 A1.10 Distribution of poverty by literacy of household head, June 1997 86 A1.11 Distribution of poverty by gender of household head, June 1997 86 A1.12 Distribution of poverty by household size, June 1997 87 A4.1 Average out-of-pocket expenditures on school fees per student, 1997 87 A4.2 Average out-of-pocket expenditures on school improvement fees per student, 1997 88 A4.3 Average out-of-pocket expenditures on school uniforms per student, 1997 88 A4.4 Average out-of-pocket expenditures on school books per student, 1997 89 A4.5 Average out-of-pocket expenditures on school supplies and materials per student, 1997 89 A4.6 Average out-of-pocket expenditures on transport per student, 1997 90 A4.7 Average out-of-pocket expenditures on private tutoring per student, 1997 90 A4.8 Average out-of-pocket expenditures on accommodation costs per student, 1997 91  CONTENTS vii A4.9 Average out-of-pocket expenditures on other schooling costs per student, 1997 91 A4.10 Average non-zero expenditures for students in public schools, 1997 92 A4.11 Average number of students, 1997 92 A4.12 Average private expenditure on education, 1997 92 A4.13 Sources and uses of private education expenditure, 1997 93 A4.14 Public sector gross enrollments per capita, 1997 93 A4.15 Budget of Ministry of Education, Youth, and Sports, 1996-97 94 A4.16 Public subsidies and cost recovery for public education, 1996-97 94 A4.17 Public subsidies by quintile, 1996-97 94 A4. 18 Effectiveness of targeting public subsidies for education, 1996-97 94 A4,19 Efficiency of targeting public subsidies for education, 1996-97 95 A5 .1 Average out-of-pocket expenditure on fees per health service contact, 1997 95 A5.2 Aggregate out-of-pocket expenditure on fees, 1997 96 A5.3 Average out-of-pocket expenditure on drugs per health service contact, 1997 96 A5.4 Aggregate out-of-pocket expenditure on drugs, 1997 97 A5.5 Average out-of-pocket expenditure on transport per health service contact, 1997 97 A5.6 Aggregate out-of-pocket expenditure on transport, 1997 98 A5.7 Average out-of-pocket expenditure on fees, drugs, and transport per health service contact, 1997 98 A5.8 Aggregate out-of-pocket expenditure on fees, drugs, and transport, 1997 99 A5.9 Average out-of-pocket expenditure on fees, drugs, and transport per health service contact, 1997 99 A5.10 Total number of monthly health service contacts by quintile, 1997 100 A5.11 Sources and uses of private health expenditure, 1997 100 A5.12 Utilization rates for health services by quintile, 1997 101 A5.13 Distribution of people reported ill in the last month by health care provider and expenditure quintile, 1997 101 A5.14 Percent of users who pay zero per public sector contact, 1997 102 B3.1 Age at entry into first grade, 1997 103 B3.2 Repetition rates of currently enrolled students by school level and by age at entry into grade one, 1997 103 B3.3 Average comprehensive scores of currently enrolled students by current grade level and by age at entry into grade one, Cambodia, 1997 103 B,3.4 Gross and primary enrollment according to the 1997 survey and the Ministry of Education, Youth, and Sports, 1996-97 104 E).1 Maximum likelihood probit equations for the probability of children ages 6-11 being currently enrolled in school, 1997 116 D.2 Maximum likelihood probit equations for the probability of children ages 12-14 and 15-17 being currently enrolled in school, 1997 117 E. 1 Persons reporting ill in previous month by consumption quintile and area, 1997 119 E.2 Percent of individuals reporting ill in the previous month by age group and gender, 1997 119 E'.3 Mortality of children, 1997 120 E.4 Access to drinking water and sanitation by consumption quintile, 1997 120 F. 1 Maximum likelihood probit equations for the probability of an individual obtaining any medical treatment for an illness episode during the preceding month, 1997 122 F.2 Maximum likelihood probit equations for the probability of an individual obtaining treatment from a public provider for an illness episode during the preceding month, 1997 123 P.3 Maximum likelihood probit equations for the probability of an individual obtaining treatment from a private provider for an illness episode during the preceding month, 1997 124 F.4 Maximum likelihood probit equations for the probability of an individual obtaining treatment from a drug vendor for an illness episode during the preceding month, 1997 124   Executive Summary frphis first poverty assessment for tion quintile have access to piped water, Cambodia is based on the 1997 only 4 percent in the poorest quintile do. _ update of Cambodia's baseline In rural areas the poorest quintile has net poverty profile of 1993-94. Since the early primary school enrollment of 50 percent, 1990s Cambodia has made progress in considerably below the 75 percent for the rebuilding its economy following more richest quintile. Less than 5 percent of than two decades of war. Trade and invest- children ages 12-14 in the poorest quin- ment flows have risen to unprecedented tile in rural areas are enrolled in lower- levels, supporting average annual growth secondary schools, but 25 percent in the of 6 percent. In 1997 and 1998 recon- richest quintile are. About 21 percent of struction and recovery were set back by people in the poorest quintile have to renewed political instability and by the travel more than 5 kilometers to reach a Eaist Asian crisis. With the assumption of health clinic; only 14 percent in the rich- office of a new reform-minded govern- est quintile have to travel that far. About ment in late 1998, and the virtual end of 6 percent of the poorest two quintiles live the Khmer Rouge movement in early more than 5 kilometers from the nearest 1999, there is renewed optimism that road; only 1.4 percent in the richest quin- growth and recovery will resume, tile do. While almost 20 percent of the Cambodia is nevertheless still a very richest rural quintile have access to pub- poor country, with GDP per capita esti- licly provided electric lighting, under 1 rmated at only $280 in 1998 and with percent in the poorest quintile receives other nonincome indicators of poverty the same service. comparing poorly with those in other Differences in the design and coverage countries in the region. The 1997 pover- of the 1993-94 and the 1997 surveys ty estimates confirm that poverty is per- mean that the two profiles are not vasive in Cambodia-an estimated 36 directly comparable. Nevertheless, the percent of the population is poor-and modest decline in poverty during rural households, especially those for 1993-97 (from 39 percent to 36 percent) whom agriculture is the primary source suggested by the profiles is corroborated cf income, account for almost 90 percent by the estimated 3 percent increase in cif all the poor. real per capita private consumption, as The poor are more likely than the bet- captured in the national accounts, and by ter-off to live in households that are larg- broad improvements in other nonincome er, have a larger share of children, and indicators of poverty such as enrollment have a head of household who is less edu- rates and maternal and infant mortality cated than the average. They also have rates. The profiles also suggest that much less access to public services than rural poverty declined less than urban the better off. For example, while 17 per- poverty. This is plausible given the cent of people in the richest consump- strongly urban bias in growth during the  2 CAMBODIA POVERTY ASSESSMENT period. Growth was unbalanced, centered Creating Economic in Phnom Penh and other urban areas, Opportunities through driven by reconstruction efforts, and led Rapid Growth by phenomenal increases in such activi- ties as construction and tourism. In con- Creating new economic opportunities trast, growth of rural activities, based through rapid and sustained growth, par- primarily on agriculture, showed con- ticularly in rural areas, must be the core siderable variability and significantly of a successful poverty reduction strate- lagged growth in the rest of the economy. gy for Cambodia. Almost half of the poor Despite its pervasiveness poverty in in Cambodia would be lifted out of pover- Cambodia is relatively shallow, suggesting ty within the next eight years if the econ- great scope for poverty reduction omy grew at 6 percent annually. But this through rapid growth. In fact, income growth must be more broad based than poverty rates could be cut by as much as in the recent past. This will require main- 50 percent by 2005 if Cambodia is able to taining a stable macroeconomic envi- resume and maintain average annual ronment and deepening reform in order growth of 6 percent. But such gains in to tackle the remaining impediments to poverty reduction will depend strongly on growth. Fiscal policy will continue to play the pattern of growth. If growth contin- an important role in stabilization efforts, ues to be urban-focused, poverty reduc- given the extensive dollarization of the tion gains can be expected to be much Cambodian economy. The objective of less and rural poverty in particular will policy should be to avoid levels of fiscal remain high-espec:ially since the depth deficit that could be destabilizing because of poverty is greater in rural areas. In con- they lead to rapid growth of the money trast, poverty in Phnom Penh would be supply, to high levels of domestic bor- almost eradicated. rowing that could crowd out the private This report recommends three com- sector, or to unsustainable levels of ponents to a broad poverty reduction external debt. Cambodia has in the strategy for Cambodia: recent past successfully used fiscal policy * Expanding economic opportunities to maintain macroeconomic stability. through rapid growth-particularly However, the quality of fiscal adjustment growth of the rural sector. needs to be improved through greater * Helping the poor build human assets. reliance on revenue generation-espe- * Providing an adequate safety net to cially by addressing governance issues address the needs of vulnerable groups related to forestry revenues-and less on in society. expenditure cuts, particularly in the social The government's Five-Year Socio- sectors and in operations and mainte- economic Plan (1 996-2000) gives prior- nance. Success in the government's ity to poverty reduction and recognizes demobilization efforts will be important these three elements as critical to its in achieving sustainable fiscal adjustment antipoverty strategy. However, much while meeting social sector objectives. greater attention to the growing Cambodia has made impressive HIV/AIDS epidemic is needed given the progress in its transition to a market econ- signiflcant potential it holds for increased omy. Further reforms are needed, how- poverty-both new poverty as well as a ever, to remove remaining distortions in deepening of existing poverty. Extensive the economy. This is especially important public education on thLe epidemic and on since most countries in the region have prevention should figure prominently in reinvigorated their reform efforts in the government's health and overall order to speed their recovery from the poverty reduction strategy. regional crisis and improve their long-  EXECUTIVE SUMMARY 3 term competitiveness. Cambodia cannot fisheries production is also important, afford to be left behind, The new gov- and will require improved technical ser- ernment has already introduced fiscal vices for animal health, fisheries research, reforms. Other key areas for policy conservation, and management. attention include further public enter- Investments are needed to upgrade and prise reform; private sector development, maintain secondary and tertiary roads, to and banking sector reform. Public enter- strengthen rural infrastructure and pro- prise reforms should focus on commer- vide better access to input and output cializing enterprises remaining in the markets. Support for the development of public portfolio in line with the law reg- formal rural finance institutions will be ulating public enterprises. Rehabilitation important for expanding access to rural plans should be formulated for the six credit. And easing the land constraint will rubber plantations and a privatization require addressing the issue of military plan outlined and implemented. Private land holding and speeding up land-titling sector development can be further fos- efforts-particularly in rural areas-as tered by adopting a comprehensive part of the creation of a transparent and cormmercial code and by strengthening secure land market. the rule of law through better enforce- Government efforts to promote growth ment of existing laws and modernization must be combined with efforts to increase of the judiciary. Banking sector reforms the assets of the poor, thus allowing the should focus on strengthening the pru- poor to take full advantage of the eco- dential and regulatory framework, includ- nomic opportunities created through ing adequate regulations on provisioning growth. Increasing the physical assets- for loans, legislation that facilitates the land, credit, rural infrastructure-of the taking and enforcement of collateral, and poor is an integral part of promoting enactment of a bankruptcy law. In addi- rapid growth of the rural sector and hence tion, the supervisory capacity of the of the economy as a whole. Building the National Bank of Cambodia should be human assets of the poor, including edu- strengthened, particularly its capacity to cation and health, is equally important for enforce capital adequacy standards and helping them access economic opportu- force the exit of banks that cannot com- nities and for sustaining growth. ply, especially the numerous weaker smaller banks. But growth must also be more bal- Expanding Access to Education anced than it has been in the recent past for the Poor to ensure that economic opportunities are widespread. In particular, special Cambodia's achievements in education in emphasis must be placed on stimulating the 1 960s and early 1 970s were system- growth and opportunities in the rural atically erased by the Khmer Rouge in the economy, where the overwhelming second half of the 1970s, when schools, rnajority of Cambodia's poor live. This will equipment, and books were destroyed, require both raising agricultural incomes effectively abolishing education. It is esti- and promoting diversification into other mated that 75-80 percent of all teachers rural activities. Key areas for policy and secondary school students fled or attention include increasing crop pro- were murdered in 1975-79 (Asian diuctivity through greater access to Development Bank 1996). Post-Khmer improved seed varieties; appropriate Rouge, Cambodia has made impressive soil and pest management, and better progress in expanding schooling. There water control through small-scale irriga- was a concerted effort to rebuild the edu- tion facilities. Increasing livestock and cation infrastructure, but this had to be  4 CAMBODIA POVERTY ASSESSMENT done under tight budgetary constraints. such factors as availability of teachers and The government made rehabilitation of proportion of trained teachers) is anoth- primary education its main focus in the er contributing factor. The probability of 1980s, but despite these best of inten- enrollment is also strongly associated with tions, primary school enrollments gender, with boys more likely than girls to plateaued at around 1.3 million between be enrolled. 1985 and 1991, which actually meant declining enrollments given the rapidly expanding primary school-age popula- What Government Can Do to Raise tion (Asian Development Bank 1996). Enrollment Rates among the Poor Since then, however, enrollments have Improving physical access to schools by expanded more rapidly than population, constructing more schools in poorer- and gross primary enrollments in 1997 particularly rural-areas and by improv- increased to almost 95 percent, accord- ing rural infrastructure will be critical to ing to the Ministry of Education, Youth, raising enrollments, but policy must also and Sports. This is impressive by any stan- address issues of price and quality. In par- dard. But problems r emain. ticular, the relative costs of schooling will have to be lowered for the poor. Fewer than 10 percent of the poorest students Why Enrollment Rates Are So Low among are exempted from school fees. Fee the Poor exemptions, coupled with a clearly estab- Enrollment rates vary considerably lished and transparent system of needs- between the poor and the better-off in based scholarships for the poorest Cambodia. Only 55 percent of all children students will be necessary to raise the of primary school age in the poorest quin- enrollment rate of this group. Another tile of the population attend school, com- option is for the government to provide pared with 79 percent for the top quintile. (or to insure) loans to needy students to These disparities increase at higher levels help them pay for education costs (as in of education, with virtually no children Thailand and the United States, for from the poorest quintile enrolled beyond example). The advantage of a loan pro- primary school. Both supply and demand gram is that it relies mainly on private factors are importan;t in explaining low resources in a situation in which gov- enrollment among the poor and the sig- emient funds are severely limited. The nificant disparities in enrollment across funds to finance both scholarships and income groups. Half of the rural popula- loans can be obtained by reducing high tion does not have a primary school in the levels of public subsidies to post-sec- village and more than 95 percent have no ondary education, a practice that bene- secondary school. Children in the poorest fits mainly the rich. quintile also have to walk longer distances As well as reducing schooling costs, a to school-on average almost 7 kilometers pro-poor schooling policy should address to the nearest lower-secondary school out- school quality. The extraordinarily high side the village, compared with about 4 pupil-teacher ratios in rural areas (80 and kilometers for children in the richest quin- above) will need to be brought down at tile. Limited and poor quality rural roads least to the national average (around aggravates poor physical access to schools. 60)-and probably lower. This will require Schooling is also extrernely expensive for a more rational policy of deploying the poor-the cost of one child in primary teachers across provinces and villages. school takes up a quarter of all nonfood One way to do this is to pay teachers a pre- spending per capita in the poorest quin- mium to work in underserved areas (a tile. Quality of schooling (as measured by policy used with some success in  EXECUmVE SUMMARY 5 Indonesia and elsewhere). Funds to ImprovingAccess to Health finance these quality improvements could Services for the Poor also be obtained, at least in the short run, fronm savings generated by reducing sub- Many health facilities around the coun- sidies to post-secondary education. try were completely destroyed, often Data on earnings suggest that women deliberately during Cambodia's more have higher returns to primary and lower- than 30 years of war and conflict. Even secondary schooling than do men, espe- today many parts of Cambodia still have cially at younger ages. Yet female no health facilities, and in other places enrollments, especially at the lower-sec- facilities are too dilapidated to be of any ondiary level, lag behind those of men. use. Yet despite these handicaps, the gov- This is not only inequitable but ineffi- emnent has taken concrete steps to cient. Targeting at least some additional reconstruct and revitalize the public schooling subsidies to poor girls would health system. The Ministry of Health has boost efficiency and equity. Some coun- also been instrumental in developing and tries (Bangladesh and India, for instance) expanding programs to tackle infectious ha,ve found scholarship and stipend diseases, such as the National programs targeted to girls an effective Tuberculosis Program and the National means of raising enrollments of girls. Malarial Program, and in vastly expand- Data also indicate that the private sec- ing coverage of child immunization pro- tor still has only a limited role in school- grams. In addition, since 1996 the ing in Cambodia. Fewer than 1 percent Ministry of Finance has been imple- of students are enrolled in private insti- menting a health services plan to expand tutions, most of them in post-secondary access to health facilities. This improved schools (typically vocational training), access to health services for the popula- ThLe support of the private sector will be tion could not have been achieved with- needed to expand secondary school cov- out the assistance of donors and erage in Cambodia. One strategy would nongovernmental organizations (NGOs). be for the government to target its limited educational spending to providing basic and secondary education to poor stu- Why Heakh Service Use Is So Low among dents and to let the affluent turn to the the Poor ernerging private sector for schooling High costs of health services, low house- needs. Alternatively, vouchers can be used hold incomes, limited education, and to provide access to private schools for the inadequate access to health facilities and poor. Indeed, broader use of vouchers to health personnel are all important fac- might increase efficiency in education by tors in explaining the low use of health ser- fcistering competition between public and vices by poor Cambodians. The cost of private schools. Household survey data health care (measured by health spending already show a strong shift of upper- per capita relative to household spending income groups to private schools. This per capita on nonfood items) is much should be encouraged, so that the gov- greater for the poor than for the nonpoor. ernment can better target its expendi- One outpatient visit to a commune clinic tures to the poor. It would be useful, or district health center would use up half however, to identify constraints and of all nonfood spending for someone in problems, including any formal or hid- the poorest quintile. Moreover, there is no den restrictions imposed by the govern- formal and transparent mechanism for ment, that may put a brake on the private exempting the poor from user fees. For sector expanding its role in the provision them, health care is simply unaffordable. of schooling in Cambodia. The poor also have less immediate phys-  6 CAMBODIA POVERTY ASSESSMENT ical access to health facilities than the non- in parallel with a transparent mechanism poor. Better-off Cambodians have much for exempting the poor from such fees greater local (within-village) access than and providing a free supply of drugs tar- the poor to the four most frequently used geted exclusively to Cambodia's poorest health providers--drug vendors, com- will be important in ensuring greater equi- mune clinics, private clinics, and trained ty in access to health services while reduc- midwives. Moreover, the poor have to trav- ing affordability constraints for the poor. el much longer dis:ances to all types of These measures need to be implemented health facilities than the better-off when in parallel with greater cost recovery in no health provider is available in their secondary and tertiary health facilities home village. such as provincial and central hospitals. But other factors, especially education, Several of Cambodia's health facilities are also important in determining health need rehabilitation, but in the long run services use. Immurnization rates among additional primary health facilities are children up to 24 months old are signifi- needed to increase access. Unlike most cantly lower for the joor than for better- other developing countries, Cambodia off Cambodians even though more than 95 does not have a nationwide network of percent of the poor reported a child immu- primary health care facilities. The gov- nization campaign in their village in the ernment's Health Coverage Plan aims to previous 12 months. Indeed, since child create a new network of 909 health cen- immunization has such wide coverage and ters, 65 referral hospitals, and 8 national is free in Cambodia, iit is difficult to explain hospitals in 72 operational districts. The the large disparities i-iX immunization cov- location of the new facilities will have to erage based on differences in economic be carefully selected, with an emphasis on standing or affordability. Cambodians providing facilities in remote and under- with more schooling are more likely to served regions, to meet the objective of treat their illnesses than those with no providing equitable access to basic health schooling. Moreover, women with prima- and referral services for the entire pop- ry and post-primary schooling are much ulation. In the shorter to medium term more likely than those with no schooling increased access in such areas could be to use preventive health services such as provided through mobile health posts. immunization for their children. The low quality of health care in pub- While there is no hard data, anecdo- lic facilities needs to be addressed. An tal evidence suggests that health services important first step is to raise staffing stan- to the poor are of much lower quality dards so that commune clinics and district than those available to the better-off. The health centers used by the poor have low quality of health services, combined trained medical workers. Better salaries and with high relative cost, must act as a benefits for health workers could also pro- strong deterrent to the use of health ser- vide increased incentives for higher qual- vices by the poor. ity service. In addition, stocks of drugs and medical supplies in commune clinics and district health centers need to be increased. What Government Con Do to Improve the The quality of health services could also be Health of the Poor enhanced by allowing health facilities to As with education, government efforts to retain a share of fees collected. increase use of health services by the poor These efforts to raise health services uti- should include both demand and supply lization by the poor will entail a restruc- measures. Reducing out-of-pocket costs of turing of public health expenditures to health services for the poor will be criti- benefit the poor. Public expenditures need cal. Expanding and enforcing user fees, to be targeted more toward proving ade-  EXEcluTivE SUMMARY 7 quate funding for primary health facilities, functioning system of transfers and safe- a process that began in 1996 as spending ty nets. NGO and other aid agencies are was shifted from hospitals toward prima- active in this area, but there is very little ry health care. This trend should contin- in the way of a functioning government ue, shifting more resources to preventive system. A strategy for developing such a health interventions (such as communi- system should emphasize timely delivery cab'Le disease control programs) and of assistance and effective and efficient curative services targeted to the poor. This targeting to ensure that resources reach shift also represents more efficient use of only the intended beneficiaries. Given public resources, given the substantial pos- Cambodia's largely rural population itive externalities of preventive health care and low level of development, policies and private underspending on such care. that are likely to work include well- Private sector provision should be designed rural public employment encouraged to complement government schemes that can provide insurance efforts and to ensure adequate access to against agricultural uncertainties. For health services as living standards those unable to work, options to consider improve with growth. Appropriate gov- include food-based interventions such as ernment regulation will be important in food distribution and feeding programs, setting quality standards that ensure pub- special operations to provide basic food lic safety. This is especially important in items to the poor in remote and other the case of private drug vendors who are high risk areas during times of insecuri- mainly unregistered and used heavily by ty, and feeding programs and transfer poorer groups for self treatment. schemes administered through the pub- lic health system. While the need for safe- ty nets is great, the effectiveness and Providing an Adequate Program successful targeting of such a system in ofTransfers and Safety Nets addressing poverty will depend on fur- ther progress in identifying indicators of Vulnerable groups in Cambodia can ben- poverty in Cambodia that have strong efit from the protection offered by a well predictive ability.   1. A Poverty Profile of Cambodia rT his report is the first poverty Poverty is multifaceted and can be assessment for Cambodia. It pro- analyzed from a variety of angles. The vides an updated poverty profile approach adopted here is quantitative, for Cambodia based on a socioeconomic focusing on the economic dimensions of survey of households conducted in 1997. poverty. The measure of economic wel- The assessment covers four areas: edu- fare used is household expenditure cation, health, macroeconomic perfor- expressed in per capita terms. In this mance, and rural development. This approach poverty is measured relative to focus is motivated by the strong linkages a level of expenditures that allows some between these factors and poverty in minimum standard of living-typically Cambodia. defined as food consumption that pro- Measured by both income and broad- vides at least 2,100 calories of energy per er human development indicators, person per day and a small allowance for Cambodia ranks among the poorest nonfood consumption to cover basic countries in the world. In 1998 per capi- items like clothing and shelter, A base- ta income was estimated at only $280, and line poverty profile for Cambodia was Cambodia was ranked 140th of 174 coun- developed from the results of the tries on the United Nations Development 1993-94 socioeconomic survey, the first Programme's Human Development such survey for Cambodia. The 1997 sur- Index (UNDP 1998). Thus the promi- vey forms the basis for the current pro- nence given to poverty reduction in the file. government's development program is fitting. Cambodia's first Five-Year Socio- economic Development Plan (1996- The 1993-94 Baseline Poverty 2000) states that "Eradicating poverty is Profile the single most important long-term objective of the Royal Government, and The baseline poverty profile was devel- reducing it is the central thrust of the oped using data from the 1993-94 plan for 1996-2000." Reducing poverty is socioeconomic survey.' It showed that also viewed as a means of achieving peace poverty was relatively low in Phnom Penh and national reconciliation. To be effec- (only 11 percent of inhabitants were tive, however, the government's poverty below the poverty line) but much high- strategy needs to be informed by a thor- er in other urban areas (37 percent poor) ciugh understanding of the nature of and highest in rural areas (43 percent). poverty in Cambodia. How many poor Nationwide, 39 percent of the population Cambodians are there? Who are they? was below the poverty line (similar to Where do they live? How do they make neighboring Vietnam and Lao PDR) -22 a living? These questions are addressed percent with per capita spending lower in this chapter. than the food poverty line.2 9  10 CAM0ODIA POVERTY ASSESSMENT Poverty rates were highest in house- The 1997 survey was administered to holds headed by farmers (46 percent) and randomly selected households in a sam- lowest in households headed by civil ser- ple of randomly selected villages in 20 of vants (20 percent); more than three-quar- Cambodia's 23 provinces.3 A single ters of the poor lived in households round of interviewing ran from the end headed by farmers. Poverty rates fell with of May to the end of June 1997. In the level of schooling and were lower in first stage of sampling, 474 villages were female-headed households (35 percent) selected using systematic random sam- than male-headed household (40 per- pling (with probability proportionate to cent). These results parallel findings of population size) from Phnom Penh (120 other poverty studies in East and Southeast villages), other urban areas (100 vil- Asia (Ahuja and others 1997). lages), and rural areas (254 villages). In the second stage, 10 households were selected using systematic random sam- The 1997 Socioeconomic Survey pling from each urban village and 15 from each rural village, yielding a sam- The 1997 survey was conducted using three ple of 6,010 households. The sample is questionnaires: a core household question- not self-weighting due to inaccurate naire, rotating special purpose household household counts in the sampling frame. questionnaires, and a village questionnaire. Thus all estimates are weighted to The core household questionnaire col- reflect actual sampling probabilities cal- lected data on demographic characteristics culated by the National Institute of of household members, school enrollment, Statistics (NIS 1998). Poverty estimates employment and earinings, use of health reported apply only to villages included care services, morbidiiy, housing charac- in the sampling frame. No attempt has teristics, household expenditures, and been made to extrapolate estimates to ownership of consumer durables. The include villages excluded in the sampling questionnaire was designed to collect data frame. (Details of data adjustments for needed to monitor poverty rates, including updating baseline poverty lines are pro- calculation of a wide range of poverty indi- vided in the companion technical report cators. Knowles 1998.) Three special purpose questionnaires collected more detailed information in alternating years on access to, use of, and Updated Poverty Estimates spending on social ser7vices, income and employment, and private consumption. A plotting of June 1997 real per capita The village questionnaire also collected daily consumption by area shows that the information on land use and access to cumulative distributions do not cross, community and social services (roads, although distributions for rural and electricity, markets, schools, health facil- other urban areas overlap at low levels ities, and so on). This information can be of per capita consumption (figure 1.1). combined with household data from the This implies that the ranking of the core and special purpose questionnaires three areas for the incidence of pover- to derive additional poverty indicators ty should not be affected by the choice and to research the causes of poverty. The of poverty line. The most commonly village questionnaire also collects data on used poverty measure is the headcount 75 retail prices for selected food and non- index, or the proportion of individuals food items (including 10 medicines). with per capita consumption at less than These data are used to update poverty the poverty line. Another measure, the lines for retail price chlanges. poverty gap index, is the average deficit  A POVERTY PROFILE OF CAMBODIA 11 between a person's per capita con- suretween andth person'sperty c with an Figure 1. I Cumulative distribution of per capita consumption, sur.nption and the poverty line (with a Phnom Penh, other urban areas, and rural areas, June 1997 zero deficit for those with per capita con- surnption above the poverty line). A Curr,Wtwve PerentO fpoplodn third measure is the poverty severity incdex, similar to the poverty gap index Poverty line except that the deficit is squared before (I993-94 Ph/or Penh pnces) it is averaged. Unlike the other two, this 60 measure is sensitive to the distribution of per capita consumption among those below the poverty line. The updated 0Pnon Penn indicators of poverty in Cambodia are 2 Perh presented in table 1.1). 20 6 7 8 9 (0 Socioeconomic Characteristics Log per capita daily consumpnon (del) ol the Poor Surce: NIS 1998. Key social indicators vary across the full Table 1.I Poverty measuresjune 1997 range of per capita consumption groups (table Al .1). In some cases (for example, Heodcount index Fbverty gap index Pove"r severity index education and literacy, access to safe Food pover line water), the differentials reflect additional Phnom Penh 3.4 0.5 0.1 dimensions of welfare and basic need sat- Other urban areas 15.4 3.3 1.1 isfaction that are not adequately reflect- Rural areas 20.0 3.9 1.2 ed in the private consumption data used Totala 17.9 3.5 1.1 to develop poverty estimates. Poverty lineb Phnom Penh I 1.1 2.2 0.6 Other urban areas 29.9 7.5 2.7 Rural areas 40.1 9.7 3.4 rlousehold Size and Composition Totar 36.1 8.7 3.1 The poor are predominantly members a Sampled ar on. of larger households. The average b. Includes nonfood allowance (table 1.2). household size is 5.8 people for the SOure: NIS 1998. poorest per capita consumption quintile aknd 4.2 people for the richest quintile. Thus the incidence of poverty increases Figure 1.2 Poverty by household size, 1997 with the number of people living in a Percentoge h-iousehold (from 25 percent for a 60 household of 4 people to 58 percent for 50 one with 10); beyond those numbers, it is a little lower (figure 1.2 and table 40 A1.12) . 4 Poorer households alsc) tend to 30 totota have a larger share of children. Children P3ey constitute 48 percent of households in 20 Povernt the lowest quintile and 33 percent in the index richest quintile of households. The I * elderly vary little by quintile, although 0 IL they represent a higher percentage of 2 3 4 s 6 7 8 9 lO 11 12 or more smaller households in the richer quin- Household size tiles. Similarly, the percentage of house- Source: NHIS 998.  12 CAMBODIA POVERTY AssESSMENT hold members who are disabled changes when other variables, such as education little across per capita expenditure and place of residence, are held constant quintiles.5 (Prescott and Pradhan 1977). Age and Gender of Household Heads Education and Literacy Poorer households tend to have younger Differences across per capita expenditure heads than do richer households-the quintiles are strong in literacy-and even age difference is r oughly two years more so in years of schooling. For exam- between the poorest and richest quintile. ple, 56 percent of adults (ages 15 and Poorer households are also less likely to above) were literate in the poorest quin- be headed by a woman. Women head 21 tile and 74 percent in the richest quintile. percent of households in the poorest Average schooling in the poorest quintile quintile and 26 percent in the richest was only 2.7 years; it was 5 years in the quintile. Thus poverty appears to be richest. Poverty rates are also high (and higher among male-headed households, similar) for people in households in at 37 percent, than among female- which the head has had no schooling (42 headed households, at 33 percent (figure percent) or only some primary schooling 1.3 and table Al. I l} One explanation (40 percent; figure 1.4 and table A1.9).7 may be that the average age of women These two groups accounted for 81 per- heading households (50.1 years) is cent of all poverty. Poverty rates were also greater than that of men heading house- large among those whose highest level of holds (42.3 years), and in Cambodia education is technical or vocational (40 poverty rates decline with the age of the percent), but they accounted for only 2 household head beyond age 39,6 But percent of all poverty. In contrast, there even after adjusting for age differences, was no poverty among households in poverty is still significantly lower among which the head is a graduate (less than 1 households headed by women. Most percent of the population), and poverty women are not household heads, how- was low among those with higher sec- ever, and most do not live in female- ondary and lower secondary schooling headed households (only 21 percent). (12 percent and 24 percent). Still, "femaleness" (the percentage of Using the poverty headcount index by household members age 15-60 who are reported literacy of the household heads female) is positively and significantly shows that poverty was lower among the related to per capita consumption, literate (34 percent) than among the illit- erate (42 percent; figure 1.5 and table Figure 1.3 Poverty by gender of Al. 10). The difference is relatively small, head of household, 1997 however, compared with that between pri- Percentwge mary and secondary school leavers. 00 Almost two-thirds of the poor still live in Contnhbont to total poverty households whose head is reported to be illiterate. 60 What these findings on poverty and Pover-ty education and literacy imply for policy- he.dco-nt 40 idex makers is that adult literacy campaigns are unlikely to have a big impact on 20 poverty (although they undoubtedly o _ improve quality of life), while expanding Male Female primary and secondary education should Source: NIS 1998. secure major reductions in poverty.  A POvERTY PROFILE OF CAMBODIA 13 Access to Housing and Consumer Durables Figure 1.4 Poverty by education of head of household, 1997 Some housing characteristics vary sharply Percentage across per capita expenditure quintiles. 50 For example, only 4 percent of house- holds in the poorest quintile have access 40 to in-house toilet facilities while about 37 percent of households in the richest quin- 30 tile enjoy such facilities. Access to piped Poverty water does not seem to vary much 20 headcount between the poorest three quintiles (all index toibut,o just under 20 percent) but increases to * 0 E total close to 33 percent for the richest quin- 0 1 I= tile. Differentials in ownership of con- None Primary Lower Higher Graduate Other Not sumer durables such as television sets and secondary secondary (including reported vocational radios are particularly striking. About 21 and technical) percent of households report ownership Noter Educaton refers to highest grade level started. of television sets, but the share in the rich Somce NIS 1998. est quintile is more than four times that Figure 1.5 Poverty by literacy of in the poorest. Similarly the share of head of household, 1997 households with radios in the richest Percentage quintile is double that in the poorest. 80 70 to total Where the Poor Live Poverty 50 headcoi,t Unfortunately, the 1997 survey is only 40 index large enough to support regional pover- 30 ty comparisons at a highly aggregated 20 level. While this provides a broad sense of the appropriate policy orientation in 10 regional targeting, it is of limited practi- 0 !` ~~~~~~~~~~~~~Literate Illiterate Not reported call value in selecting the geographic Sorce. NIS 1998. placement of projects. The incidence of poverty, however Cambodia clearly need to be targeted to measured, is considerably lower in the rural poor. Phnom Penh than in either other urban or rural areas (see table 1.1). For exam- ple, the poverty headcount index is only Key Sources of Livelihood for the Poor I 1 percent in Phnom Penh, whereas it is The poverty status of most people in any 30 percent in other urban and 40 percent country is determined by their earnings in rural areas. These differences are even and the income of other household more marked when using the food members. The household head is usual- poverty line or the poverty severity ly the most important income earner, and index. The poverty severity index in employment of other household mem- Phnom Penh, for example, is only about bers is often linked to the employment a sixth that in either other urban areas or characteristics of the household head (for rural areas. The majority of the poor live example, when the household head is in rural areas: rural poverty accounts for self-employed and other household about 88 percent of the poverty gap members work for the household head). index, and about 89 percent of the pover- By employment of the household ty severity index. Antipoverty policies in head, the poverty headcount is highest in  14 CAMBODIA POVERTY ASSESSMENT agriculture-44 percent. Indeed, most of unpaid family workers. Poverty incidence Cambodia's poor--71 percent-are is also relatively low for public sector work- employed in agriculture. In no other ers (20 percent). industry is there mor-e than 10 percent of Poverty rates are also estimated for the total poor, and only people in house- other industry and employer categories, holds where the heaid is not in the labor including those working but for whom force accounted for mrnore than 5 percent industry or employer was not reported, of poverty (8.4 percent). But other the unemployed, those not in the labor industries with high headcount rates force, and those for whom no employ- include construction and utilities (38 per- ment information was reported. For all cent), manufacturing and mnining (29 per- these groups the incidence of poverty is cent), and other services (27 percent). lower than average (figures 1.6 and 1.7 Relatively low poverly rates are found in and tables Al.7 and A1.8). In particular, education and health services (17 per- the percentage of household heads who cent), government services (18 percent), report that they are unemployed is less trade (19 percent), and transportation than 1 percent. The largest group, and communications (20 percent). heads of household who are not in the We look next at headcount by type of labor force (10 percent of the total), employer together with the relative con- account for 8 percent of all poverty.8 tributions of each to the total poverty rate The policy implication clearly is that (figure 1.6 and table Al.8 show similar poverty alleviation needs to focus on the estimates using the poverty gap and pover- self-employed, particularly those in agri- ty severity index). The highest incidence culture. Only about 4 percent of the poor of poverty is among household workers work in private sector nonfamily busi- (43 percent) and the self-employed (40 nesses, so policies such as minimum wage percent). The self-employed, however, legislation will have little effect on the accounted for 74 percent of all poverty, poor. Rather, the poor will benefit most whereas household workers accounted for from policies that improve the func- only 1 percent. The lowest poverty inci- tioning of rural markets, including those dence (16 percent) is found among for agricultural commodities, inputs and rural credit. Figure 1.6 Poverty by employer of head of household, 1997 Percer waee 80 Inequality Measured by several alternative indicators, 70 inequality is consistently higher in Phnom 60 l Penh, declines in other urban areas, and PC ,-ty is lowest in rural areas (table 1.2).9 50 headcount Inequality in Cambodia as measured by index 40 ^ the Gini coefficient is higher than in most other countries in Asia at Cambodia's level 30 Contibutno of economic development (table 1.3). 20 20 Poverty Indicators 0 Seif- Public Unpaid Private House- Employer Un- The core household questionnaire of the employed sector family sector hold not employed 1997 Cambodia Socioeconomic Survey worker worker reported collected information on a wide range of Source: NIS 998. poverty indicators. These indicators  A POVERTY PROFILE OF CAMBODIA 15 Figure 1.7 Poverty by industry of employment of head of household, 1997 Percentage 80 Contribubon to total 70 60 50 Povety headcount 40 index 30 20 l 0 0 Agriculture Manufac- Construction Trade Transportation Gov- Education Other Industry Unemployed Not in Not turing and emment and services not labor reported utilities service health reported force SourcE NIS 1998. Table 1.2 Consumption inequality measures, June 1997 Per capita Voriance Consumption share Consumption share Population- consumption- of the of the poorest of the richest weighted weighted logarithm of 10 percent 10 percent Gini Theil Theil per capita of the populotion of the population coefficient measure measure consumption (percent) (percent) Plnom Penha 0.46 0.78 0.38 0.41 2.6 40.3 Other urban areas 0.44 0.49 0.33 0.47 2.7 37.6 Rural areas 0.33 0.21 0.17 0.30 3.7 27.1 Tbtalb 0.42 0.48 0.29 0.41 3.0 35.3 a Based on adjusted consumption data. b. Sampled areas only. Source 1993-94, Prescott and Pradhan 1997; 1997, NIS I99& serve three main purposes. First, they where it does exist). Third, they can help help policymakers understand the dimen- targeting (for example, means testing, sions of poverty, thus enabling them to geographic targeting). An effective set of clraw up more effective poverty alleviation poverty indicators needs to meet certain policies. Second, they can be used to criteria, however. They need to be: monitor poverty alleviation even when * Valid-that is, they need to distinguish detailed information on household con- between the poor and nonpoor, and sumption is not available (and they can those close to the poverty line and act as a check on that information those further away.  16 CAMBODIA POVERTY ASSESSMENT Table 1.3 Gini coefficients for selected look most promising as overall poverty countries, most recent year indicators-for example, floor area per Gini 1998 GNP household member, distance to the Country coefficient per capia main source of drinking water, house- Bangladesh (1992) 28.3 350 hold's main source of lighting (electric- Cambodia (1997) 42.0 280 ity), household's main source of drinking China (1995) 41.5 750 water (piped, public tap, or tanker), toi- India (I1992) 33.8 430 let located in dwelling, households using Indonesia (1993) 31.7 680 fuel other than firewood as primary cook- Lao PDR (I1992) 30.4 330 Mongolia (1995) 33.2 400 ing fuel, wall of dwelling made of con- Pakistan (1991) 31.2 480 crete or plywood, roof of dwelling made Philippines (1994) 42.9 1050 of material other than thatching or Thailand (19'92) 46.2 2200 mixed thatch material, and floor of Vietnam (1993) 35.7 330 dwelling made of material other than Source:World Bank staff !stimates. earth or wood. Some indicators appear more useful in distinguishing the poor * Applicable to the whole population- from the nonpoor (floor area per house- or to important segments of it, such as hold member, for instance), and others the rural population. help distinguish between more severe * Easy to observe and verify. and less severe poverty (distance to the * Not too closely correlated with each main source of drinking water, for other, so that together they provide a instance). In addition to indicators relat- more accurate ind(lication of poverty ed to housing characteristics, other than any single indicator. promising indicators of poverty at the * Not subject to opportunistic respons- household level include, value of land es by the household or individual. owned, value of outstanding loans, Once poverty inidlicators have been household's main source of income in identified, it is critical to verify their effec- the previous 12 months (other than agri- tiveness in future surveys. culture), household ownership of motor- Poverty indicators derived from the 1997 bike, scooter, or television set. At the survey can be classified into two groups. The more aggregate level, enrollment (both first consists of aggregate poverty indicators gross and net), annual outpatient visits (such as net primary school enrollment or per person, and number of annual hos- number of outpatient lealth care visits per pital admissions per 1,000 people are 1,000 people) summarized across four cat- promising poverty indicators. egories-population below the food pover- Among demographic indicators, nei- ty line, population between the food ther percentage of household members poverty line and the overall poverty line, the who migrated nor percentage of house- nonpoor, and the total population (table hold members who are female appears to A1.2). Sharp differences in the means of be related to poverty status. Although per- these indicators acrcss the first three centage of household members who have groups (or between the poor and nonpoor) worked in the past week is moderately suggest effective poverty indicators. The related to poverty status, number ofjobs second group consists of indicators at the per household member and percentage household level such as whether the of children under age 15 working in the household owns a television set or whether past week are only weakly related. the head of household speaks a foreign lan- Percentage of children who have died is guage fluently. actually lower among the food poor than Among both grouLps of indicators, among either of the remaining poor or those referring to housing characteristics the nonpoor (probably because mortal-  A POVERTY PROFILE OF CAIaODIA 17 ity is underreported in this group). mary level; and it casts some doubt on the Percentage of children under age two poverty-alleviation value of adult literacy who have a vaccination card is only weak- campaigns (although literacy is a basic ly related to poverty, which speaks well of need). Carnbodia's immunization coverage. Although more work needs to be done Among the asset and liability indicators, to identify the most effective set of pover- livestock owned (pigs, buffaloes, cows, ty indicators for Cambodia, this analysis poultry) appears to be a weak poverty identifies potentially useful poverty indi- indicator, even when restricted to rural cators. Some of the most promising indi- households. Among characteristics of cators at the household level are related heads of households, being in an ethnic to housing characteristics; for example, liv- minority does not appear to be related to ing in households where a fuel other than poverty status. firewood is used for cooking-I 3 percent The large number of promising pover- of the nonpoor, 1.2 percent of those below ty indicators suggests that this is an area the food-poverty line, and 1.8 percent of that merits more research. Further work the remaining poor. Another is the value could attempt regression-based decom- of outstanding loans (an average CR position of the effects on the welfare mea- 277,000, or $110, among the nonpoor, CR sure of changing indicators over time and 130,000, or $52, among the food poor, and changes in the returns to the indicators. CR 137,000, or $55, among the remaining The ultimate goal would be to reduce poor) and the qualitative indicator, house- poverty indicators to a manageable set of holds owning a television set (30.5 percent robust, easily collectable and verifiable of the nonpoor, 9.1 percent of the food- inclicators that could be applied in the poor, and 13.1 percent of the remaining field for targeting social service subsidies poor). At a more aggregated level, gross anrd other poverty-alleviation interventions. secondary school enrollment rate-27.4 percent for the nonpoor, 5.4 percent for those below the food poverty line, and 11.6 Conclusion percent for the rest of the poor is a promis- ing poverty indicator. The 1997 poverty estimates show that rural households-particularly those where agriculture is the primary source of Notes inczome-continue to account for almost 1. For details on the construction of the base- 90 percent of Cambodia's poor, and line profile, see Prescott and Pradhan (1997). poverty alleviation needs to be targeted 2. The baseline poverty profile was limited mainly to them. Because three-quarters of to the surveyed population. No attempt was the poor are self-employed, policies made to extrapolate the poverty estimates to designed to make markets function more the uncovered population. efficiently (particularly markets for agri- 3. Provinces excluded were Mondul Kiri cultural commodities, inputs, and rural (included in the sampling frame, but not rep- credit) will be more useful as poverty alle- resented in the randomly selected sample), viation measures than policies targeted to Preah Vihear, and Oddar Meanchey. employers and employees, such as mini- 4. The household size patterns could also be m,um wage legislation, social security, and purely the result of the choice of welfare mea- miandatory fringe benefits. Education is sure (consumption per capita). lilely to be an effective poverty alleviation 5. The disability rates in table Al.l for intervention, too. The 1997 poverty pro- 1997 are almost three times higher than file, however, suggests that the greatest those reported in the 1993-94 poverty pro- gains accrue to education beyond the pri- file. One explanation for this difference may  18 CAMBODIA POVERTY ASSESSMENT be that the data on disabilities in the 8. Of heads of household not in the labor 1993-94 profile were obtained from the list- force, 47 percent were women. The reasons ing form for households (that is, the form most often given were "too old, infirm, or used to select sample hiouseholds from a cen- retired' (52 percent, of which 38 percent were sus of households in the villages selected in female): "housekeeping' (20 percent, 92 per- the first stage of sampling) rather than as cent female); and "believes no work is avail- part of a direct household interview. able' (12 percent, 29 percent female). Only Disability rates obtained by household 5 percent gave "disabled" as the reason for interview in the 1996 survey were even high- nonparticipation (41 percent female). er than those in the 1997 survey. 9. The area measures of inequality in table 6. The incidence of poverty increases 1.2 are unaffected by adjustments made to steadily from 29 percent among those living total consumption for the underreporting of in households where its head is under age 25 medical expenditures, since the same per- to reach a maximum of 42 percent for those centage adjustment was applied to all house- whose head is 35-39 years of age, after which holds within each stratum. The aggregate it declines steadily to :30 percent for house- measures of Inequality for Cambodia as a holds whose head is age 55 or more. whole were little affected by these adjustments, 7. Classifications by grade level in figure 1.5 the combined effect of which was to reduce are according to definitions introduced in slightly the degree of overall inequality. For 1997-that is, primary schooling is classes 1-6, example, in the absence of any adjustments, lower secondary is classes 7-9, and upper sec- the Gini coefficient for Cambodia would be ondary is classes 10-12. The 1993-94 classi- 0.43 (instead of 0.42 after adjustments), and fication was primary (classes 1-5), lower the share of total consumption of the richest secondary (classes 5-8), and upper sec- 10 percent of the population would be 36.3 ondary (classes 9-11). (instead of 35.3 after adjustments).  2. Legacy of War and Social Upheaval () '~ambodia's pervasive poverty must were internally displaced in fighting be viewed in the context of more during 1989-91. Many Cambodians were than two decades of war and civil also forced to flee the country when the strife. Some of the most pernicious 1978 Vietnamese invasion drove the results include the widespread displace- Khmer Rouge from power. Thousands merit of families, the large and still grow- fled to Thailand seeking food and secu- ing numbers of disabled persons, and rity and were eventually settled in camps high levels of government spending on along the border. By 1991 the population defense and security to the detriment of of these camps had swollen to about social expenditures that could benefit the 360,000. poor. Each of these outcomes has impli- cations for poverty, though causal link- ages are hard to substantiate empirically Repatriation given the weak database. The purpose of Following the Paris Peace Accord in 1991, this chapter is not to establish such link- the UN High Commissioner for Refugees ages. Our interest is to show the human (UNHCR) arranged for the return of story behind the poverty numbers by refugees from the Thai camps in time for highlighting these important post-conflict the UN-sponsored elections in 1993. Most issues and to draw out their implications returnees-an estimated 60 percent- for antipoverty policy. chose to settle in the northwestern provinces, some 25 percent of them choosing to return to Battambang and Internally Displaced Persons and Banteay Meanchey alone. The resurgence Returnees of fighting and insecurity in 1997 gen- erated a new wave of refugees, displacing The rise to power of the Khmer Rouge in some 80,000-100,000 Cambodians inter- 1975 resulted in the internal displace- nally or to new camps along the Thai ment of hundreds of thousands of border. Sincel997 UNHCR has returned Cambodians. The inhabitants of Phnom about 46,000 Cambodians from Thai-bor- Penh and other cities were forced into the der refugee camps. countryside as part of a strategy to build The 1992-93 repatriation was well an agricultural revolution in support of financed, with each refugee receiving an Khmer revival. This massive population estimated $330. Since then, several mil- displacement followed on earlier internal lion dollars more have been spent on displacements caused by bombings along rehabilitation projects to assist returnees the eastern part of Cambodia in the late and to develop areas where large num- 1960s and early 1970s and fighting in bers of them have settled. But many of Cambodia during 1970-75. More recent- the assumptions on which the repatria-' ly, an estimated 190,000 Cambodians tion plan was based proved to be 19  20 CAMBODIA POVERTY ASSESSMENT unsound. In particular, the plan had not working in forests and on agricultural envisaged the extreme loss of usable land land where there were land mines. as a result of land mines, decayed trans- These findings all suggest a greater inci- portation and irrigation infrastructure, dence of food poverty among returnees and the speculative rise in the value of than in the general population, where an land after reforms in 1989. As a result, estimated 17 percent of people are instead of the initialy promised land for unable to meet their basic food needs. agricultural use, most returnees were pre- The studies highlight several factors sented with a series of new options involv- that appear to distinguish returnees ing housing or cash. This was a big shift who are making it from those who are in the repatriation strategy: instead of not: access to land, capital and skills, suf- providing returnees with enough assets ficient labor, and diverse sources of (in particular land) to become self-suffi- income. cient, the new strategy was based on the hope that returnees could depend on rel- Access to land. Access to arable mine-free atives in Cambodia who would support land in a secure area with basic services them till they became self-sufficient and sufficient water appears to be the best (Studdert 1995). While many returnees guarantee of successful economic inte- did find relatives (an estimated 85 per- gration, conditions that apply to very few cent ), that did not guarantee them access returnees. Only 12 percent of respon- to assets or a livelihood, dents in the WFP survey had access to land to grow rice or crops. This is in stark contrast to local villagers. An Indochinese Reintegration Success and Poverty Refugee Information Center survey How have returnees fared? Has economic found that 15 percent of returnees had reintegration been achieved?l Not all access to cultivable land, well below the returnees have found new homes and 79 percent share for local villagers livelihoods. Five surveys of returnee (Court 1994). The survey also found that communities conducted between 1993 the average size returnees' land was one- and 1999 all confirm Lthat many returnees third smaller than that of local villagers. still face considerable economic diffi- In the Battambang area returnees con- culties (Geiger 1993; WFP 1993 and 1999; stituted a disproportionate share of the Court 1994; Davenport, Healy, and landless poor). Landlessness was low Malone 1995). While few Cambodians among local residents, but nearly the have an easy life, the studies suggest that entire sample of returnee-more than 99 returnees are economically more vul- percent-was landless (Davenport, Healy, nerable than those who had not been dis- and Malone 1995). placed. According to those surveys, Less than half of returnees had titles 30-40 percent of refugees are "not mak- for their agricultural land. This was a ing it." Many are livin-ig hand to mouth. reflection partly of the heavy pressure on With no assets to sell or resources to fall land because of the prevalence of land back on in case of illness or other mis- mines and unexploded ordnance and fortune, and without enough assets to partly of a resentment toward returnees, improve their situation in future, many who were often perceived as wealthy or can do no more than focus each day's viewed with suspicion because of their efforts on finding food for that day.2 The political background. Since 1994 an World Food Progra-mme (WFP) 1998 estimated 15 percent of returnees' titled Survey of Target Zones also found evi- agricultural land and as much as 70 per- dence that househoids in these zones cent of their untitled agricultural land has used riskier coping strategies, such as been confiscated by the military, local offi-  LEGACY OF WAR AND SOCIAL UPHEAVAL 2 1 cials, and local villagers (Pendleton and Diverse sources of income. The majority of Teas 1998). returnees support themselves through multiple activities, including fishing, Capital and skills. In the absence of land, seasonal farm work, home goods pro- skills and the capital to set up a busi- duction, day labor, and vegetable gar- ness-a shop, a motorcycle taxi or repair dens. Well-planned diversification may be shop-have enabled some returnees to one of the few ways a returnee family with malke a reasonably secure living. Again no land or capital can improve its situa- however, this group makes up a minori- tion. While some families become ty of returnees. Capital has come from trapped in a daily search for food, others savings, from wealthy relatives, or from have been able to save and improve their assistance from NGOs. Credit appears to situation. This kind of success requires have played almost no role in the rein- able-bodied family members, good man- tegration of returnees. Credit is generally agement, the ability to build relationships unavailable, and when it is available inter- and a certain amount of luck. est rates are extremely high and few retarnees are willing to borrow in the Support from relatives. The surveys found amounts necessary to successfully launch that a significant proportion of returnees a business. In addition, even with capital, (45 percent in the WFP study) live with many of the skills taught by internation- relatives and that this group tended to ally funded programs are not useful in fare better than others. The Indochinese Cambodian villages. For example, there Refugee Information Center study found are simply not enough motorcycles to that 27 percent of returnee families living repair. with relatives lived hand to mouth com- pared with 46 percent of those living with- Sufficient labor. In addition to land or cap- out relatives (Court 1994). While relatives ital, most returnee families required at can be an important source of help for least two or three able-bodied adults in retumees, the studies also suggest that the order to make it. Families need labor to amount of help depends on the closeness till the land and to help with other pro- of the relationship and on the econom- ductive activities. Sometimes kinship con- ic situation of relatives. Relatives could nections can be used to acquire labor provide a temporary base and a source of through traditional labor exchange, or resources (shelter, capital, labor) at rea- (less frequently) capital can be substi- sonable rates but could rarely provide tuted to hire labor. Labor is especially newcomers with enough to establish a crucial for families that do not have land new source of livelihood. or capital. Sufficient labor, if well man- aged, can provide families with reason- Other factors. In addition to these factors, ably secure income even when they lack social integration-access to social net- land or capital. Some families have been works of information and resources able to secure a living through a strate- (box 2. 1)-and outside support have also gy of diverse labor-intensive activities- been important in determining returnee farming small pieces of land for success. There has been no overall vegetables or cash crops, buying or sell- analysis of the importance of outside sup- ing goods, and the like. The crucial fac- port from NGOs and other internation- tors are a critical mass of labor and the al organizations Some returnees have organizational ability to manage a diverse received substantial support that has set of strategies. For these reasons widows changed their lives, while many have had with small children have been particu- little or no contact with an aid agency larly vulnerable. since returning from the border. While  22 CAMBODIA POVERTY ASSESSMENT the aid community until 1990 when there Box 2.1 Access to social networks were approximately 130,000 scattered For families without land or capital, access to social networks of infor- around the country. At that time a nation- mation and resources is crucial. One commonly cited form of resource al committee, with provincial counter- sharing is provas give with one hand, take with the other." This can parts, was formed to take responsibility be applied to the exchange of labor or information about job oppor- tunities and government services. The most commonly cited example for internally displaced persons and the is the sharing of cows. Profitable farming requires draft animals, but Cambodian Red Cross was given respon- draft animals are prohibitively expensive for most families. Cambodian sibility for coordinating relief efforts. villagers have a tradition of giving their cows to poorer families to take Currently, internally displaced persons care of, letting the poorer family keep one of the calves every other are concentrated in seven provinces year in exchange. The question is how easily returnees can access these kinds of net- around Tonle Sap. In October 1995 the works. At the beginning of repatriation, there were reports ofjealousy Cambodian Red Cross recorded 57,000 by locals toward returnees who had large sums of cash and received internally displaced persons (excluding rice without having to work for it. However, jealousy seems to have long-time displaced persons, who have stopped with the end of the rations, as it became evident that many more or less resettled). There have refugees, without land or animals, were poorer than locals. There is some evidence to suggest that reintegration was slowest in new sites been few studies on the reintegration of and faster in areas where returnees made up a lower percentage of the internally displaced persons, but the lim- population. However NGOs report that deliberately situating projects ited work available suggests that land to include old and new villages seemed to help build relations, as did mines are the biggest obstacle to return- drawing new settlers from a mixed group of returnees and local land- ing home. Those who settle in new loca- less. Situating schools so that children from both villages study togeth- tions find themselves in a similar situation er is another way to facilitate integration. to the returnees. NGO programs for returnees or the land- less have probably helped those with Large Numbers of People with whom they have had contact, tradition- Disabilities al development projects (rice banks, cow banks) have probably been less effective The protracted armed conflict has left a since most returnees have neither land large and still growing number of nor the minimum level of resources Cambodians disabled. The 1997 socioe- needed to participate. conomic survey found that 25,805 The WFP's Food for Work Program Cambodians are disabled because of war seems to have had the broadest impact or conflict, and anotherl6,290 are dis- since it reaches the most people. Most vil- abled because of land mines. This means lagers have stressed that without Food for that one in every 250 Cambodians has a Work they would not have been able to disability due directly to the country's war stay long enough to make the land pro- history. The vast majority (93 percent) of ductive. Most studies agree that quick those with disabilities are men. With 1 impact projects ease the burden of amputee for every 384 people, Cambodia returnees on local facilities. However, an also has one of the highest proportions emphasis on quickness means that pro- of amputees in the world-close to jects often have a limited development Angola's 1 per 334, but far more.than impact. An often cited example is the con- Afganistan's 1 per 631, Mozambique's I struction of schools without makingsure per 2,414, or Ethiopia's 1 per 6,663 that teachers were available to staff them. (ICRC 1997). Cambodians disabled by Also, with the withdrawal of the UNHCR, war and land mines represent close to 60 some villages are taking back the land they percent of the total number of people gave to returnees in return for projects. with disabilities. Cambodia's 1997 Internally displaced persons were "Human Development Report" found generally ignored by the government and that though the rate of disability is lower  LEGACY OF WAR AND SOCIAL UPHEAVAL 23 for the poor than for the better off, there tance programs. For a country as poor as is a higher concentration of amputees Cambodia this could hinder the proper among the poor than the rich. prioritizing of services and the long-term Land mines and other unexploded sustainability of programs, both financially ordnance are still widespread and con- and operationally. Overall coordination of tinue to maim and kill. At the end of 1998 NGO assistance in this area is weak despite land mines were claiming 100 to 200 vic- government attempts to promote such tims every month, most of them civilians. coordination. Betwreen March 1998 and February 1999, 659 of the 1,076 reported casualties from mines and unexploded ordnance were Land Mines and Unexploded Ordnance civilian. The economic and social costs of Cambodia is one of the most heavily land- this legacy are likely to be significant. mined countries. The Cambodian Mine Action Center estimates that there are 4-6 million land mines, nearly one for Assistance to People with Disabilities every rural inhabitant in Cambodia. An Canibodia has no specific government estimated 1.7 percent of Cambodian ter- programs to assist people with disabilities ritory is mined or infested with unex- in general or people disabled from war. ploded ordnance-much of it in the most Much of the small budget of the Ministry productive parts of the country. A United of Social Affairs, Labor, Vocational Nations Transitional Authority in Training, and Youth Rehabilitation, which Cambodia mine survey in 1992 in is responsible for vulnerable persons, goes Rattanak Mondul district (southwest to p,ensions for public servants, assistance Battambang) found that some 65 percent programs for widows and orphans, and a of highly productive land-land that had rehabilitation program for about 1,000 sustained one of the country's most pros- famrkilies that come to to Phnom Penh perous centers of commercial agriculture every year to return to their original in the mid-1960s-was unusable because homes (usually in the countryside). of land mines. In some communities in Assistance for people with disabilities this area people have entered known (including those disabled by war) comes mine-risk areas to gather food, to hunt mainly from the several NGOs active in and fish, and to gather wood, bamboo this area. NGOs provide training and and other natural resources, These activ- physical rehabilitation services such as ities, once a useful supplement to assistive devices, physiotherapy, and treat- incomes gained primarily from agricul- ment for children with disabilities and ture, have become for many in this area economic assistance in the form of loans, the only source of livelihood, but also the mostly in-kind but also in cash. It is clear, prime cause of casualties. however, that current programs reach There is a fundamental link between only a small proportion of those needing infestation by mines and unexploded ord- assistance, leaving significant needs nance and rural economic impoverish- unmet. For example, according to min- ment. While the costs to individual istry records, only 3,132 people with dis- households and to Cambodia as a whole abilities received training between 1987 have not been quantified, they are sure- and 1998, barely 10 percent of the of the ly large. Mines are a major cause of hand- total. Veterans International produces icap, limiting victims' ability to find about 100 prostheses in Cambodia per employment that will allow them to sup- month, but most of their clients are not port themselves and their families. Mines newly amputated. There is also limited hinder the resettlement and reintegra- analysis of the cost effectiveness of assis- tion of internally displaced persons and  24 CAMBODIA PoVERn ASSESSMENT returning refugees, limiting land avail- Large Public Expenditures on able for cultivation and for homesteads. Defense Each of these factors adds to vulnerabil- ity within Cambodian society and helps Another legacy of Cambodia's war history to perpetuate a vicious cycle of inade- is the high levels of public spending on quate human and physical assets, defense and security, to the detriment of increased vulnerability, and poverty. social spending and spending to rebuild The Cambodian Mine Action Center physical and economic infrastructure. is engaged in mine awareness, marking, Spending on defense and security and clearance, worlc it carries out in con- accounted for more than 30 percent of junction with two N(JOs-Mine Advisory total expenditures during 1996-97 and Group and HALO trust. The priorities increased to nearly 50 percent in 1998. for mine action are land for resettlement Shares of spending on social services of internally displaced persons and (education and health) declined sharply returnees for agriculture, community during the same period. In 1996 domes- purposes, and for .-econstruction and tic spending on social services was 22 per- infrastructure development. The average cent of total government-financed demining unit is able to clear about 3 expenditures but fell to only 13 percent square kilometers each year. Some 2,000 in 1998 (table 2.1). square kilometers remain to be surveyed The overriding priority given to and demined. At t:ne current rate of defense and security spending can be clearing, this would take 25 to 30 years seen clearly in the difference between to complete. The Cambodian Mine budgeted (planned) spending and actu- Action Center's activities for 1999-2001 al spending in 1994-97 (table 2.2). From will cost an estimated $70 million. Weak 1995 to 1997 domestic revenue collection management procedures have, however, fell short of budget estimates by 5-7 per- made it hard to monitor the Center's cent, yet in all three years changes in pri- activities and the productivity of its dem- orities during budget implementation ining platoons. resulted in actual spending on defense and security substantially exceeding bud- Table 2.1 Sectoral composition of expenditures, 1996-98 get estimates. And in 1994, when actual (percent) revenues exceeded budget estimates by -- 996 1997 1998 about 20 percent, actual spending (on i 996 1 997 1 998 nonwage recurrent expenditures) was Defense and security 32 33 47 about 31 percent less than budgeted for Education 1 3 12 10 social services and 27 percent less for eco- HeOath 46 7 3 nomic services. Changes in priorities dur- ing 1994 resulted in defense and security Source: World Bank staff estimates based on government of Cambodia data. expenditures exceeding original budget Table 2.2 Actual expenditures as percentage of budgeted allocations by more than 100 percent. expenditures Not only has social spending been low 1994 1995 1996 1997 overall, but spending on activities known to benefit the poor the most, such as pri- Defense and security 204 177 123 132 mary education and primary health care, Administration 86 102 9 1 88 suffered as well. This implies limited gov- Economic services 73 55 88 57 Social services 69 94 81 42 ernment assistance to help the poor Health 61 66 72 46 acquire the human assets needed to Education 84 155 99 33 enable them to participate in growth. In Note: Data are for nonwage recurrent expenditures. 1996 budget outlay on primary educa- Source World Bank staff estimates based on govemment of Cambodia data tion from the government's own  LEGACY OF WAR AND SOCIAL UPHEAVAL 25 resources was just 0.5 percent of GDP. assistance, including special counseling, Even more telling, this spending medical rehabilitation, and reintegration declined in 1994-96. Government spend- support through specialized NGOs. ing on primary health services has been The underlying objective of the vet- just as low. erans assistance program is to promote the reallocation of public funds from defense to social and economic sectors, Demobilization and a Shift from capturing an important peace dividend Delense to Social Spending as the country emerges from war. With a much smaller army, public resources There is clearly plenty of room for would be freed up for other uses. By the improving the poverty impact of public end of the program the share of defense spending by changing the composition of in government recurrent spending is budgetary spending in favor of pro-poor expected to have dropped to about 20 spending and away from defense spend- percent and the share of defense in GDP ing. This will require, first and foremost, from 3 percent to about 2 percent. This successful demobilization of the military. thus provides a unique opportunity to Toward this end, the government has restructure public spending in favor of established the Cambodian Veterans pro-poor activities such as primary edu- Assistance Program to manage the demo- cation and health. But this restructuring bilization of about 55,000 soldiers from will not happen automatically. It will take an estimated 148,000 and to reintegrate the concerted effort of key line ministries thern into civilian life. The program cov- to ensure that freed-up resources are ers around 37,300 soldiers from the reg- channeled into pro-poor expenditures. ular forces and about 17,300 from five Two points are worth noting. First, special target groups, including the such restructuring, important as it is, will spouses of deceased soldiers, chronical- be only a first step toward reducing pover- ly ill or disabled soldiers, and female sol- ty by helping the poor build up their diers. Given an average family size of assets. Increased funds for social sector about 5.2 people, this program would spending will need to be used in ways that directly affect some 290,000 people. achieve the maximum positive results. The program has four stages-regis- Doing this requires several other condi- tration, demobilization, reinsertion, and tions, including improved public expen- rein-tegration. It is expected to achieve its diture management capacity-especially objectives in four phases over a five-year greater transparency and improved mon- period, beginning with a trial run of 2,000 itoring and a capacity to design and solidiers in 2000. Lessons from this trial implement effective poverty reduction run would be incorporated into the intervention. Second, expenditure remnaining rounds of demobilization. The restructuring alone is unlikely to resolve last- phase of demobilization is expected the underfunding of pro-poor social ser- to be carried out in 2003. The program vices in Cambodia. Cambodia will need woald provide transitional safety net assis- to greatly improve its revenue effort to tance to each family to facilitate the most allow a substantial increase in public difricult phase of transition into civilian resources that will permit sustained society. Reintegration support would be improvements in pro-poor spending provided through a community eco- while maintaining budgetary balance. nomic and social reintegration fund to Cambodia's return to political stabil- support community-based reintegration ity and national rebuilding began in the activities. In addition to this fund, special early 1 990s, but progress has been slow. groups would be provided with targeted Efforts were set back in 1997 and 1998  26 CAMBODIA POVERTY ASSESSMENT with the events leading up to and fol- nificant vulnerability created by dis- lowing the overthrow of Co-Premier placement and war- and land Prince Norodom Rannaridh. Since mine-related disabilities. There is democratic elections and the assumption currently little in the way of a well- of office of a new government in 1998, functioning government system of some measure of political stability has transfers and safety nets. A strategy in been achieved, but it remains to be seen this area will need to emphasize time- whether this will be translated into last- ly delivery of assistance and effective ing peace and social stability. targeting to ensure that resources reach the intended beneficiaries. * Efforts are needed to ensure the full Key Issues for Government integration of internally displaced Attention people and returnees. This will require addressing the lack of access to land Cambodia's war past raises several relat- and to credit that seems to be espe- ed issues for government attention in its cially acute within these communities. antipoverty strategy. A first is Cambo- * Demining activities should be accel- dians' vulnerability to poverty. These erated, and management problems in include the incrieasing numbers of the Cambodian Mine Action Center Cambodians who may not be able to par- need to be addressed. ticipate fully in growth because of war- * Demobilization should be accelerated related injuries and disabilities, the large as part of the effort to restructure pub- numbers of war orplhans, and the lack of lic expenditures toward investments in access to key assets-notably land and helping the poor build human capital capital-among returnees. Each of these assets. groups is likely to become trapped in poverty. A second issue is the limited access to good quality, affordable social Notes and economic infrastructure, given the 1. Reintegration is defined as attaining par- heavy destruction dluring the war period ity with the local population in terms of living and the related excessive spending on conditions and means to self-sustainability (see defense. Robinson 1994). Policymakers should take these issues 2. All of the studies used slightly different into account when clesigning antipover- criteria. The WFP survey in 1993 used wealth ty strategies and interventions. Most indicators to classify 74 percent of returnees important: as needy or at risk. A year later, using differ- * A first and critical step would be to ent criteria, the IRIC study described 34 per- consolidate the recent peace and cent of returnees as living hand to mouth, begin to build social and political compared to 17 percent of local villagers. The cohesion. Vulnerable in the Village survey estimated that * A strong system cf transfers and safe- 40 percent of returnees interviewed were "not ty nets is needed to address the sig- managing" or "destitute."  3. Economic Recovery and Poverty EI 'mpirical evidence shows that good tional reforms were put in place. I economic performance, though However, the loss of aid from Cambodia's J not sufficient, is critical for pover- former Soviet bloc partners in the early ty reduction. An analysis of economic 1990s, just as the system of transfers from policies and outcomes in Cambodia is state enterprises was being replaced by a thus important for understanding pover- more market-oriented tax system, severe- ty trends. Cambodia has made good eco- ly hurt public sector finances. In the nomyiic progress since the Paris Peace absence of significant external assis- Accord of 1991, though economic recov- tance, monetization of the growing pub- ery suffered a setback in 1997 and 1998 lic sector deficit became the root cause of from political upheaval and from the East severe macroeconomic instability. During Asian crisis. Since early 1999 a new coali- 1990-92 inflation hovered around 140 tion. government has reinvigorated percent on an annualized basis. reform, and the Khmer Rouge movement In 1992 a package of stabilization mea- is effectively dead. These developments sures was adopted to arrest further fiscal should help secure peace and restart deterioration. Given the extensive dol- growth and recovery. The question for larization of the economy and the lack of this chapter is whether poverty has monetary policy instruments, fiscal poli- dec]lined in parallel with generally good cy was the centerpiece of Cambodia's economic performnance during the 1990s. macroeconomic stabilization efforts. In In particular, has the pattern of growth 1993 the newly formed Royal government over the 1990s benefited the rural poor resumed comprehensive reform with who form the bulk of the poor in the strong support of the international Cambodia? donor community. Reforms have reduced the size of the public sector with the pri- vatization of public enterprises and other Recovering from War, 1991-96 state assets, opened up the economy to international trade with the rationaliza- Barely a decade ago Cambodia was just tion of the trade and tariff regime and the emerging from more than two decades of reduction of subsidies and protection, and civi'l war. Growth had plummeted to just promoted domestic and foreign invest- 1.2 percent a year in 1990. Investment was ment through liberal investment legisla- estimated at just 8.2 percent of GDP and tion (box 3. 1). Agricultural incentives total exports at 6 percent of GDP. have improved with private ownership of Market reforms had begun much earlier land and reforms in pricing, taxation, and in the 1980s, but political instability had marketing. limited overall positive impacts. With the The improved political climate and return of peace and security earlier renewed reform and reconstruction reforms began to take hold and addi- activity helped stabilize the economy and 27  28 CAMBODIA POVERTY ASSESSMENT the dollar from 1995 to 1996. Reserve Box 3.1 Major structural and macroeconomic reforms in the doll y fro m abo 1. Res Cambodia cover rose slightly from about 1.5 months of imports in the early 1990s to about 2 * Liberalization: Prices liberalized; private property rights restored; months in 1996. state intervention in commodity markets abolished The supply response was generally pos- * State enterprise sector: Enterprises given greater autonomy; process of privatization launched; budget constraints facing enterprises itive. Trade and foreign investment flows hardened; line ministries banned from commnercial activities; rose significantly (figure 3.1). The dollar sector share in economic activity reduced to 10 percent of GDP value of domestic exports grew 58 per- * Public finance: Organic Budget Law enacted; revenue transfers cent annually, increasing almost fivefold consolidated; New Financial Law passed between 1991 and 1996. Domestic invest- * Trade: Private sector permlitted to establish tradcing companies; ment rates rose to 26 percent of GDP in licensing requirements eliminated for most goods; import tariff sys- tem streamlined-creation of four-band system applying to more 1996, financed by the resumption of offi- than 93 percent of tariff items; all quantitative restrictions eliminat- cial external assistance and by strong ed; export restrictions enforced only for four commodities growth in foreign direct investment * Foreign investment: Liberal Foreign Investmeni: Law adopted flows. These flows, mainly from the * Exchange rate system: Greater market orientation of exchange rate region, rose from $40 million in 1992 to system: official market rate kept generally within a 2 percent about $240 million by 1996. Increased band of parallel market rate * Banking: Two-tier banking system set up; interest ceilings elimi- external flows made up for low or stag- nated; minimum deposit requirements introduced for commer- nant domestic savings-estimated at cial bank deposits; Central Bank Law strengthening autonomy of only 10.6 percent of GDP in 1996. the National Bank enacted. Recovery in exports and investment spurred rapid growth averaging 6 percent F:igure 3.1 Domesti exports and investment, 1991-96 annually between 1991 and 1996. Per Pe-eetageofGDP capita income is estimated to have dou- 30 bled from about $150 in 1991 to nearly 25 $300 in 1997. 20 But growth was unbalanced, varying 20 ~ significantly across key sectors, in part a l 5 consequence of the fact that much of the I economy is still subject to the vagaries of weather. Industry and services grew 5 P rapidly. Construction and tourism expe- o1 rienced phenomenal growth, fueled by 191 992 1993 1994 1995 1996 reconstruction expenditures and by Soasce:WoHd Ba-sk stafF-estfrsates bas-d os go-eost oftCamoboda data Cambodia's increased attractiveness as a tourist destination as security improved. provided the beginnings of economic Growth in agriculture, the key sector, was recovery. Both the fiscal and the external less spectacular, averaging just under 3 current account dleficits widened as percent over the same period, barely reconstruction and investment activities above the population growth rate. The picked up, but more prudent financing- fastest growing activities have been through concessional external assistance forestry, livestock, and fisheries. and foreign direct investment-limited Governance problems, which have led to the risk of renewed macroeconomic insta- uncontrolled illegal logging, helped bility. Monetization of the fiscal deficit was fuel the rapid growth in forestry. Crop virtually eliminated by 1995. This allowed and rubber production has performed rapid deceleration in inflation to about more poorly. 3.5 percent in 1995. Exchange rate sta- During the 1980s rice production bility was also a factor in inflation control. recovered from the dramatic declines of The riel stabilized at 2,400-2,600 riel to 1975-79, but growth has since slowed,  EcoNoMic RECOVERY AND POVERn 29 averaging under 1 percent annually Table 3.1 GDP growth, 1992-98 between 1991 and 1996 (table 3.1). (percent) Droughtandfloodsdamagedricecropsin 1992 1993 1994 1995 (996 1997 1998 parts of the country over much of the peri- od, but constraints to productivity GDP 7.0 4.1 4.0 7.6 7.0 1.0 0.0 improvements also had a role, despite Agriculture 1.9 -1.0 0.0 6.4 2.4 1.3 0.4 Industry 15.7 I 3.1 7.5 9.9 18.2 -2.8 I.1 broad sectoral reforms to improve factor Services 11.1 7.2 7.5 7.9 6.9 2.8 -1.1 incentives. Key constraints are lack of water conktrol, limited access to high-yielding Source: World Bank staff estimates based on govemment of Cambodia data. crop varieties, and inadequate access to infiastructure and credit. Rice production Apart from its sectoral concentration, represents the single most important eco- growth was also concentrated in Phnom nomic activity for rural Cambodians, Penh, which accounts for only a small accounting for more than 80 percent of share of the population, and a few other estimated agricultural labor input. Rice urban centers. This growth was initially also provides an estimated 75 percent of driven by the increase in consumption total energy intake for Cambodians and is demand with the arrival of the United thus a key factor in food security. Its less Nations Transitional Authority in rapid growth could thus have important Cambodia (UNTAC). UNTAC's pres- implications for rural incomes and pover- ence, along with the enhanced activities ty overall. While Cambodia has recorded of the United Nations Development exports of rice since 1996, a large number Programme, related UN agencies, bilat- of Cambodians are food insecure, unable eral donors, and NGOs, stimulated a sig- to access food that may be available in nificant increase in demand for trade and other parts of the country because of mar- services, particularly in the hotel, cater- keting and transportation constraints. ing, and housing sectors. This led to a Given this sectoral pattern of growth, construction boom, largely in Phnom the structure of the Cambodian economy Penh where the bulk of activities were has shifted in favor of industry and ser- based. Tourism services and related con- vices. Industry's share has risen froml5 struction activity were also an important percent of GDP in 1990 to 20 percent in part of rapidly growing flows of foreign 1998, while services have risen from 33 direct investment. percent to 37 percent. Nevertheless, small-scale subsistence agriculture, includ- ing livestock, fisheries, and forestry, still Domestic Political Upheaval and remains the most important sector, con- the Asian Crisis, 1997-98 tributing about 43 percent of GDP in 1998 and providing employment for an Cambodia's economic recovery was set estimated 77 percent of the employed back by the domestic political upheaval population. Manufacturing and con- leading up to and following the change struction represent the largest shares of in government in July 1997 and by the industry. However, most manufacturing regional financial and economic crises. It is conducted on a small-scale, informal is hard to disentangle the individual basis. The fortunes of manufacturing effects of these two crises, which occurred activities such as brick making and around the same time, particularly ceramics are linked to the fortunes and because they can be expected to have performance of the construction sector. qualitatively similar effects on economic The service sector is heavily dominated performance and outcomes. Never- by trading activities, geared predomi- theless, the preliminary evidence suggests nEntly toward a foreign clientele. that the political events may be more to  30 CAMBODIA POVERTY ASSESSMENT blame for CambodLi's poor economic heavy dependence on foreign financing. performance in 1997 and 1998 than the Domestic investment fell as official regional crisis. The extensive dollariza- donors cut back on aid disbursements and tion of the economy appears to have as foreign direct investment flows helped moderate adverse effects on declined sharply following the regional macroeconomic stability, liquidity crunch and increased insecuri- The twin crises eroded confidence in ty in Cambodia. New commitments of for- the economy and disrupted economic eign direct investment also declined activity, leading to exchange rate depre- about 35 percent in 1997. There was a ciation, inflation, and a significant slow- slight recovery in 1998, but it is not clear down in output growth. The riel that it will be sustained since some of the depreciated by about 28 percent in nom- underlying factors for the slowdown still inal terms relative to the dollar between remain. Overall, growth fell to 2 percent July 1997, when both crises hit, and June in 1997 and then to zero in 1998. 1998 (figure 3.2). Inflation initially mod- With the formation of a new govern- erated in 1997 but then became more ment in late 1998 and an improved polit- pronounced in 1998, rising to 19 percent ical climate, confidence is returning and on an annualized basis by mid-year. economic activity has begun to recover, Significantly reduced external aid inflows particularly in tourism and retail trade. forced declines in both the public sector Macroeconomic instability has also been deficit and the external current account moderated. Inflationary pressure eased in deficit. In both caises, however, the the last quarter of 1998 as the exchange declines masked important imbalances. rate stabilized. At the end of December Economic activity slowed dramatically. 1998 the riel had appreciated 5.3 percent Sectors that were most severely hit of its value against the dollar in June 1998, include forestry, construction, and and inflation had fallen to 13 percent on tourism. Forestry was hurt by depressed an annualized basis from about 19 per- regional demand due to the economic cri- cent in June 1998. In the first quarter of sis. Tourism suffered a double blow, from 1999 inflation has continued to decline, the deteriorating security situation and falling to 8.6 percent on an annualized the increased financial attraction of basis in February. other tourist destinations in the region fol- However, major difficulties remain. lowing the deeper depreciation of other Fiscal adjustment has relied excessively on regional currencies. The slowdown in con- cuts to nonwage recurrent spending, struction activity reflected the sector's while expenditure categories such as defense spending have continued to Figure 3.2 Consumer price index and exchange rate de. Reenuefo has remained (Index,jan 97 I 00) ~~~~~~explode. Revenue effort has remained (Index.Jon 97 =100) abysmally low. In addition, about a third 150 of the deficit in 1998 was financed through government borrowing from the 40 central bank-a significant and danger- :30 Exchange rate - - ous deviation from practice in the recent past (table 3.2). The reduction in the 20 external current account deficit has 110 ,, :Consum, erpnce index been driven by cutbacks in imports for 100 externally financed investment and a boom in garment export growth, while 90 other major export categories have suf- Jan 1997 May 1997 Sept 1997 Jan 1998 May 1998 Sept 1998 fered. Neither trend is sustainable over Source:World Bank staffestimates based on government of Cambodia data. the medium term.  EcoNOl,oc RECOVERY AND POVERTY 31 Tabile 3.2 Government operations, However, significant differences between 1996-98 the 1993-94 and 1997 sample surveys and (percentage of GDP) coverage mean that the two profiles are 1996 1997 1998 not directly comparable. The 1997 sam- Revenues 9. 9.7 8.5 pling frame included many more villages Expenditure 16.3 13.9 12.5 than did the 1993-94 survey. The 1997 Current 9.9 9.0 8.9 survey included all villages in Phnom Capital 6.4 5.0 3.6 Penh but excluded 73 villages (9 percent) Current balance -0.8 0.6 -0.7 in the other urban areas and 1,571 vil- Overall balance -7.2 -4.3 -3.9 lages (14 percent) in the rural areas. Donestic financing 0.2 -0.7 1.3 Foreign financing 7.0 4.9 2.7 The 1997 survey covered a more S-uxe: VVbrldBank sfestmatbsedongovemmntof aggregated and smaller consumption set Sounc:e World Bank sraff estimates based on government of~(0fo ad1 ofo iescmae Cambodia data. (20 food and 13 nonfood items compared with 177 food and 266 nonfood items in WNhile the recent improvement in the 1993-94). In some cases the consumption political climate augurs well for the data in the 1997 survey were based on resumption of economic activity and longer recall periods than the 1993-94 growth, Cambodia's heavy reliance on survey. For example, the 1997 survey East Asia for trade and investment flows asked for education expenditures over the leaves it vulnerable to adverse medium- previous 12 months, whereas the 1993-94 ancl longer-term effects from the ongoing survey asked for education expenditures slowdown in the region. In addition, sig- during the previous month. Both factors nifiicant depreciation of regional cur- would tend to reduce the 1997 con- rerLcies has resulted in strong exchange sumption estimates and therefore bias the rate competitiveness gains for most 1997 poverty estimate upward. In addi- economies in the region. Most countries tion, because of concerns about security, are also pursuing comprehensive reforms the 1993-94 survey was unable to cover that will improve their overall competi- some remote rural areas where poverty tiveness, expand exports, and make incidence was likely to be significantly them more attractive destinations for for- higher. This suggests that the baseline eign investment flows. Cambodia will profile underestimated rural and there- have to make significant improvements in fore overall poverty. Finally, the 1997 sur- its economy to ensure that it is not left vey was conducted in one round, whereas behind. This will be important for pro- the 1993-94 survey was conducted in four mciting strong, broad-based growth that rounds over a 12 month period. To the will provide opportunities for sustained extent that there are seasonal fluctuations poverty reduction. in household consumption, the two sets of estimates are less comparable. One way around some of these prob- Has Poverty Declined during the lems is to combine different types of infor- 1990s? mation to search out the answer. These include national accounts estimates of pri- Did poverty decline in Cambodia in par- vate consumption, which tell us about aver- allel with the improved growth perfor- age levels of living; the consumption maLnce between 1991 and 1996? The most estimates of the two surveys, which tell us direct way to address this question would more about the bottom end of the distri- be to compare the baseline poverty pro- bution and, allowing for all possible bias- file (based on the 1993-94 socioeco- es, could confirm a decline in poverty; and nomic survey) with the 1997 poverty direct measures of welfare, such as infant profile (based on the 1997 survey). mortality rates and enrollments. All these  32 CAMBODIA POVERTY ASSESSMENT tell the same story, which is that welfare lev- was only 67 kilograms. Free emergency els have improved firom a very low base. relief food has declined sharply from over Between 1993-94 and 1997 real per 103,000 tons in 1993 during the repatri- capita private consumption increased ation of 370,000 Cambodians from about 3 percent, suggesting modest Thailand to 1,211 tons in 1997, despite improvement in the well-being of the aver- continuing problems of displacement. age Cambodian over this period. The con- The poorest quintile of Cambodians also sumption estimates from the two poverty had a significant increase in their own- profiles also show aL modest increase in ership of consumer durables such as real per capita consumption of close to 2 radios, televisions, bicycles, and motor- percent for the poorest quintile of the cycles. population (table 3.3) This trend is con- The profiles also suggest that the inci- firmed by results from a participatory dence of poverty nationwide, as measured household survey conducted by the by the headcount index, fell moderately World Food Programme (WFP) in 1996, between 1993-94 and 1997, from 39 per- which found a clear pattern of decreasing cent to 36 percent. The decline in pover- reliance on food aid among the vulnera- ty appears to have been largest in urban ble and the poor. I[n 1993 vulnerable areas, where poverty incidence fell from households and internally displaced per- close to 37 percent to just under 30 per- sons each received about 80 kilograms of cent. The severity of poverty also appears rice, fish, and oil in emergency relief. By to have remained the same in rural areas, 1996 the average yearly amount of food while falling in Phnom Penh and other aid received by Food Ibr Work participants urban areas. The recent trends and pat- terns of growth are also consistent with Table 3.3 Summary indicators of per capita consumption, the smaller decline in rural poverty. 1993-94 and June 1997 Agriculture-the main source of liveli- Per capita consumption per doy (nel) Food share hood for rural Cambodians-has grown Nominal Reola (percent)b at a much slower rate than industry and - services. Because the impetus for growth Consumption quintile hscm rmPnmPn n te I-poorest 1993-94 734 963 0.75 has come from Phnom Penh and other 1997' 796 979 0.75 urban areas, the urban population has 2 1993-94 1,029 1,356 0.71 benefited the most from growth. 1997 1,160 1,413 0.73 Nonincome poverty indicators suggest 3 1993-94 1,314 1,121 0.69 1997 1,543 1,849 0.70 improved welfare and declining poverty 4 1993-94 1,803 2,307 0.66 since the early 1990s. Key indicators such 1997 2,183 2,557 0.68 as infant mortality, primary enrollment 5-richest 1993-94 4,281 4,962 0.57 rate, and life expectancy at birth are esti- 1997 5,435 5,852 0.62 mated to have improved (table 3.4). Area Better security and the decline in social Phnom Penh 1993-94 4,367 4,367 0.56 1997 5,453 4,693 0.53 confict would also be expected to con- Other urban areas 1993-94 2,412 2,873 0.67 tribute to improved general welfare. 1997 2,859 3,(76 0.68 Despite the generally positive trend of Rural areas 1993-94 1,403 1,887 0.69 declining poverty, preliminary evidence 1997 1,735 2,187 0.72 on the effects of the recent political and Totald 1993-94 1,832 2,262 0.68 1997 2,223 2,530 0.69 economic crises suggests that some of these gains are at risk. The depreciation a Expressed in 1993-94 Phnom Penh prices using the Laspeyres price deflator reflected in the o t r food poverty lines for each area (table I 1). of the riel and risig iflation have erod- b. Food consumption as a proportion of total consumption. ed real incomes, particularly for those c. Based on adjusted consumption data. who receive income in riel-mainly the d. Sampled areas only. Source: NIS 1995, 1998. poorer groups in society. The contraction  EcoisTowc RECOVERY AND POVERTY 33 of the economy has also led to employ- Table 3.4 Social indicators, various years ment cuts. But employment effects have varied by region and sector. Employment 985 _ has fallen sharply in construction and tourism, while it has more than doubled Gross primary enrollment rate Total (percentage of school age) 42 - 47 in the garment industry. Insecurity caused Mal (percentage of school age) 48 - 48 Male (percentage of school age) 48 -48 by the political crisis led to new waves of Female (percentage of school age) 35 - 46 internally displaced persons and created Lde expectancy at birth (years) 40 46 52 newi refugees. Both developments are like- Infant mortality rate (per I.000 jive births) 181 160 1 10 ly to have contributed to increased Maer mortaliy rtio Maternal mortality ratio poverty, though some of the impact may (per 100,000 live births) - 500 473 be temporary. - Nat availae fhere are also indications that food Source: World Bank, World Development Report and World Development Indicators, various years. shortages emerged in some parts of the country in 1997 and 1998 as a result of Table 3.5 Reduction of poverty incidence under different growth drought and an increase in illegal rice scenarios exports to neighboring Vietnam. Efforts 1997 2005 to cope with declining real incomes and Poverty _(poverty headcount index) food shortages put some households at Gini headcount Law i3se High increased risk of malnutrition and ill coeffient index case case case heaJth. A small segment of the Cambodian Phnom Penh 46.0 1 1.1 5.8 3.6 2.3 society-those who receive incomes in Other urban areas 44.0 29.9 22.5 15.2 12.3 dollars-has fared well under the crises. Rural areas 33.0 40.1 31.9 21.6 18.0 Cambodia 42.0 36.1 28.3 19.1 15.8 Note: Average growth rate assumptions for per capita consumption during 1998-2005 are 1.6 per- cent for the low case, 3.4 percent for the base case, and 4.2 percent for the high case (corre- Prospects for Growth and sponding to average GDP growth rates of 3.5 percent, 5.3 percent, and 6.1 percent). Poverty Reduction Source: World Bank staff estimates. Different growth paths will have different percent average annual growth (4.2 per- implications for poverty reduction in cent per capita income growth), which is Cambodia. If recent progress toward consistent with the goals of the growth peace and security is consolidated and if objectives of the Socioeconomic the competitiveness of the economy Development Plan. In the base case the improves, the government expects eco- incidence of poverty is cut by close to 50 nomic recovery and growth to resume. percent by 2005. Poverty reduction The Five-Year Socioeconomic Develop- would be even more rapid in the high ment Plan (1996-2000) envisages growth growth scenario, falling about 56 percent of 7 to 8 percent annually over the decade. (table 3.5). This translates into annual per capita These projections suggest that growth growth in income of about 5 to 6 percent. can yield high payoffs in terms of reduc- If C(ambodia does grow at this rate, what ing income poverty. Close to half of the wiLl be the implications for a reduction in poor in Cambodia would be lifted out of income or consumption poverty? poverty within the next eight years if the A base case scenario of 5.3 percent economy grew at about 6 percent annu- average annual growth between 1999 and ally over the period. This fairly high 2005 (3.4 percent per capita income degree of responsiveness of poverty to growth) is contrasted with a low case sce- growth likely reflects the fact that pover- nario of 3.5 percent average annual ty in Cambodia is relatively shallow growth (1.6 percent per capita income (table 3.6). The poverty gap ranges from growth) and a high case scenario of 6.1 a low of 2.2 in Phnom Penh to 9.7 in rural  34 CAMBODIA POVERTY ASSESSMENT Table 3.6 Comparison of poverty gap index for Cambodia, Lao scope to resume and sustain rapid growth PDR, and Vietnam through appropriate domestic policies. Year Index Key areas for government attention include maintaining a stable macroeco- Cambodia 1997 8.7 nomic environment, tackling remaining 1993 9.2 impediments to private sector growth, and Lao PDR 1993 12.1 Vitetnam 1993 15.1 promoting growth of agriculture and the rural sector (see chapter 6). Source: World Bank saff estimates. Given the extensive dollarization of the economy, fiscal policy will remain the cen- areas and is estimatecl at about 8.7 for the tral element of stabilization efforts in the country as a whole. short term. Govemment should continue Poverty reduction will also depend on to maintain reasonably low deficit levels the degree to which the poor participate that avoid the use of potentially destabi- in growth. Since there are no data on GDP lizing finance methods and loss of mon- growth by regions, this analysis assumes etary control. Recourse to domestic that growth rates are the same in the three financing of the deficit-as occurred in regions (Phnom Penh, other urban areas, 1998-should be avoided. and rural areas). However, the recent sec- Cambodia's stabilization gains are not toral pattem of growth suggests that rural sustainable without improvements in the growth (based mainly on agriculture) has quality of fiscal adjustment. More empha- lagged behind urban growth. That would sis should be placed on raising domestic mean that the proje,ctions tend to over- revenue effort-Cambodia's revenue state the poverty reduction in rural areas effort is extremely low-by addressing gov- and to understate it in urban areas. Given emance issues related to management of the large weight of the rural population, forestry revenues and by strengthening tax the projected overall poverty reduction administration. The new government may be too optimistic. has already announced measures on If the current pattem of urban-focused management of forestry revenues. growth persists, poverty is likely to be less Domestic expenditures should be ratio- responsive to growt]nh and will not fall as nalized and restructured away from mili- much a projected. R.ural poverty will also tary and defense and toward the provision remain a major problem. Even when it is of social services and economic infra- assumed that rural growth is as rapid as structure. These revenue and expenditure urban growth, poverly still remains quite changes will help achieve the social high in rural areas. Rural poverty would development objectives without compro- remain close to 22 percent in the base case mising macroeconomic stability. The and 18 percent in the high growth while recent Cambodia Public Expenditure being almost eradicated in Phnom Penh, Review (anuary 1999) provides more an outcome that could lead to a reemer- detailed policy recommendations for gence of social tensions and civil strife. A improving fiscal management. clear policy implication is that promoting Market reforms over the past decade more rapid growth in rural areas and have stimulated impressive foreign private broader poverty reduction efforts target- investment flows. Private sector activity is ed to rural areas wi:ll be important for nevertheless still constrained by public sec- achieving poverty recluction goals. tor inefficiencies, inadequate regulations Cambodia's medium-term growth of private sector activity and poor enforce- prospects will invarialbly be influenced by ment of regulations, a weak financial sec- performance in the r,egion. Nevertheless, tor, and inadequate infrastructure. In a the government still has considerable 1998 survey of 200 Cambodian enterpris-  ECCNOMIC RECOVERY AND POVERTY 35 es (conducted by the Association of hold survey data that can quantify the Cambodian Local Economic Development magnitude of adverse poverty effects from Agencies under World Bank supervision), recent events are not yet available, the respondents overall saw government as use- effects of recent crises are likely to have ful, Nevertheless, they identified as serious been considerable. For a country like constraints to private sector activity sever- Cambodia that depends heavily on agri- al problems in government management culture, negative macro-level shocks could of economic policies and rules affecting also be caused by adverse weather condi- businesses, in its delivery of services, and tions (see chapter 6). Some macro-level in the integrity of many agencies. Firrms also shocks are unavoidable (for example nat- identified inadequate access to financing ural disasters and external shocks due to and weak infrastructure as important Cambodia's growing integration into glob- impediments. al markets). The antipoverty strategy will Government action is needed to tack- help reduce, to the extent possible, the fre- le these constraints and create an envi- quency and degree of such shocks through ronment conducive to rapid private prudent macroeconomic and sectoral sector-led growth. Specific measures policies and appropriate policy responses. include identifying and taking a tough A well-designed, well-functioning system of stand against all forms of corruption, transfers and safety nets to meet the needs improving public sector efficiency by of the most vulnerable groups will help completing the privatization program limit the adverse poverty effects of macro- and commercializing enterprises remain- economic shocks when they do occur. ingy in the state portfolio, adopting a com- prehensive commercial code, and strengthening the rule of law through Conclusion better enforcement of existing laws and modernization of the judiciary. To Growth in the 1990s has led to some im,prove banking and financial services, poverty reduction in Cambodia. thie government should enact adequate Resumption of growth can provide con- regulations on provisioning for loans, on siderable payoffs in reducing income takcing and enforcing collateral, and on poverty. But growth must increase in the bankruptcy. The central bank's capacity rural sector to allow such payoffs to occur to enforce these regulations-particularly and to ensure that the rural poor do not capital adequacy standards and the exit get left out of future prosperity. Resuming of noncomplying banks-should be growth will require actions to achieve strengthened. These policy efforts should more sustainable macroeconomic balance be complemented by appropriate invest- and the creation of a business environ- ments in transport and other economic ment that fosters rapid private sector-led infrastructure. growth-including improved public sec- The government's antipoverty strategy tor efficiency and governance, a stronger should take account of adverse macro-level legal and regulatory framework, and shocks that could constrain growth. The efforts to build a better functioning finan- recent political and economic crises illus- cial sector, Providing security through trate the risks to poverty reduction from transfers and safety nets against the such shocks. Growth fell to only 1 percent adverse poverty effects of macro-level in 1997 and to zero in 1998. While house- shocks to output is also important.   4. Schooling and the Poor rj|F he poverty profile for Cambodia tion (Asian Development Bank 1996). shows that poor education is both Since then, however, enrollment has a cause and a result of persistent expanded more rapidly than population: poverty. As shown in chapter 1, differences gross primary enrollment increased to across per capita income are high for lit- almost 95 percent in 1997, according to eracy and even higher for years of school- the Ministry of Education, Youth and ing. Adult (ages 15 and above) literacy was Sports.' This improvement is impressive 56 percent in the poorest quintile of by any standard. But problems remain. households and 74 percent in the richest The quality of primary education is quintile. Average schooling was only 2.7 extremely low. There are large economic years in the poorest quintile but 5 years in disparities in primary schooling opportu- the richest quintile. Households in which nities, with enrollment rates among the heads had no schooling or only some pri- poor significantly lower than among the mary schooling had high poverty inci- nonpoor. Cambodia's secondary enroll- dence, accounting for over 81 percent of ment rate is among the lowest in the world total poverty. Constraints faced by poorer and will be a drag on economic growth and households led to lower school enrollment industrialization in the future. Disparities levels for poor children. between regions, economic groups, and Cambodia's achievements in educa- boys and girls in secondary enrollments are tion in the 1960s and early 1970s were sys- even greater than in primary education. ternatically and bloodily erased by the Cambodia's gross primary enrollment Khmer Rouge in the second half of the ratio is comparable to that of most coun- 1970s, through the destruction of tries in the region (table 4.1). While its net schools, equipment, and books. It is esti- primary enrollment ratio is lower than that mated that 75-80 percent of all teachers in other Asian countries (except for Lao and secondary school students fled or PDR), most of those have significantly were murdered between 1975-79 (Asian higher per capita incomes. Cambodia's Development Bank 1996). gross secondary enrollment ratio, howev- Post-Khmer Rouge, there was a con- er, is the lowest of any country, lower even certed effort to rebuild the education than -Lao PDR, Myanmar, and Nepal. infrastructure. The government made Cambodia has considerably underinvest- rehabilitation of primary education its ed in secondary education expansion com- main focus in the 1980s, but despite these pared with other countries in Asia. intentions, under tight budgetary con- straints primary school enrollment plateaued at around 1.3 million between School Enrollment 1985 and 1991, which actually meant declining enrollments given the rapidly Data from the Cambodia Socioeconomic expanding primary school-age popula- Survey of 1997 (NIS 1998) show that 66 37  38 CAMBODIA POVERTY ASSESSMENT Table 4.1 Enrollment ratios for selected percent of children ages 6-11-the official Asian countries, 199S primary school age group in Carnbodia- are enrolled in primary school. Beyond this Pnimary Secondary age, enrollment drops off sharply, with only Gross Net gross 14 percent of children ages 12-14 enrolled Armenia 82 79 in lower-secondary school and 7 percent of Azerbaijan 104 78 children ages 15-17 enrolled in upper-sec- Brunei 110 91 78 ondary school (table 4.2). Post-secondary Cambodia 104 66 21 net enrollment is less than I percent. China 118 99 67 India 100 49 As in most developing countries at Indonesia 114 97 4 Cambodia's per capita income level, dis- Kazakhstan 96 83 crepancies between net and gross enroll- Kyrgyz Rep. 107 97 81 ments are large, reflecting over-age Lao PDR 110 60 25 enrollment (table 4.3). For instance, gross Malaysia 91 91 57 primary enrollment for the whole of Mongolia 88 80 59 Cambodia is 113 percent (more than two- Myanmar 103 30 thirds larger than net enrollment) and Nepal 110 37 Pakistan 74 gross lower secondary enrollment is 35 Philippines 116 100 79 percent (almost two and a half times the Sri Lanka 113 75 net ratio). Talikistan 89 82 About 510,000 of the 1.59 million chil- Thailand 87 55 dren ages 6-11 are not enrolled in Vietnam 101 78 47 school. If secondary school-age children Source: UNESCO 1998, except net enrollment Lao are included, Cambodia has about 1.1 mil- PDR,World Bank 1995, and Vietnarn,World sank lion children ages 6-17 not enrolled in 19.4; and Cambodia net primary and gross secondary, NIS 1998. school, a huge waste of human resources. Table 4.2 School enrollment by consumption quintile and region, 1997 (percentage oftorget group (at each level) Expenditure quintile Nation- Phnom Other Rural l-poorest 2 3 4 S-richest wide Penh urban areas areas Net enrollment Primary 5'5.4 64.7 70.5 73.2 77.5 67.0 81.3 69.8 64.9 Lower-secondary 5.6 6.0 13.0 17.1 31.0 14.1 37.9 23.8 9.5 Uppermsecondary 1.6 1.2 5.0 6.9 18.1 6.6 31.6 7.7 3.1 Post-secondary 0 0 0.1 0.6 2.3 0.7 5.8 0.6 0.1 Gross enrollment Primary 94.0 112.0 119.5 122.8 124.2 112.8 118.3 109.2 112.7 Lower-secondary 1:3.6 20.4 31.8 39.6 72.9 34.7 76.5 48.2 27.1 Upper-secondary 2.4 2.5 8.0 12.3 32.0 11.6 49.4 15.1 6.1 Post-secondary 02 0.0 0.4 0.8 4.0 1.2 10.7 1.1 0.2 Age-specific enrollment Ages 6-11 55.5 65.3 71.0 74.4 80.8 68.0 86.6 71.6 65.3 Ages 12-14 69.3 74.1 80.5 79.1 85.0 77.4 84.8 79.9 76.1 Ages 15-17 33.4 37.3 39.3 44.3 55.2 42.0 68.9 44.0 38.1 Ages 18-24 6.9 8.3 8.1 8.8 15.7 9.9 21.7 10.3 8.6 Note: The age-specific enrollment ratio is the proportion of children in the specified age group who are enrolled in school. Zero values in the table do nor: imply that there is no enrollment in that category, only that there were no such cases in the 1997 survey sample. Quintiles are based on real per capita consumption expenditure (deflated using the food poverty lines for Phnom Penh, other urban areas, and rural areas). Source: NIS 1998; Cambodia Ministry of Planning 1998.  SCHOOLING AND THE POOR 39 Rural and Urban Disparities Table 4.3 Gross primary enrollments in Cambodia,Vietnam, and There are large disparities in enrollment Lao PDR between urban and rural areas, especially Cambodia 1997 survey Vietnam 1993 Lao PDR 1994 at higher levels of schooling. For Gross Net Gross Net Gross Net instance, the net primary enrollment raionista 81hperent prinmPan Penh,rorlm25 Primary 112.8 67.0 100.8 78.2 110.0 60.0 rat.io is 81 percent in Phnom Penh, or 25 Lower-secondary 34.7 14.1 48.2 36.1 32.0 1 5.0 percent higher than in rural areas, while Upper-secondary 11.6 6.6 1 S.0 11.2 - - its net lower-secondary enrollment ratio - Not available. is 4 times higher than in rural areas and Source. Cambodia, NIS 1998; Vietnam,World Bank 1994; Lao PDR,World Bank 995. its net upper-secondary enrollment ratio is 10 times higher. Almost 86 percent of Table 4.4 Distribution of enrolled students and school-age the 510,000 children ages 6-11 not in population by consumption quintile, 1997 school live in rural areas, 10 percent in (percentage of total population in each quintie) urban areas outside Phnom Penh, and 4 -poorest 2 3 4 5-richest percent in Phnom Penh.2 Thus policies Enroled students to raise enrollments should be targeted Primary 21.0 21.4 21.2 19.5 16.9 mainly to rural areas. Lower-secondary 8. 12.3 19.0 22.6 38.2 Upper-secondary 4.1 4.2 14.4 21.1 56.3 Post-secondary 2.3 0 6.1 15.2 76.4 Disparity across Economic Groups Age-specific population Disparities in enrollment by income are Ages 6-1 1 25.1 21.6 20.0 17.9 15.4 Disparities ~~~~~~~~~~~ ~~Ages 12-14 20.5 21.0 20.6 19.8 18.2 even greater than regional disparities. Only Ages 15-17 19.4 19.3 21.0 19.9 20.4 55 percent of all children ages 6-11 in the Ages 18-24 16.1 19.1 19.3 22.3 23.3 poorest quintile attend primary school, Sourc NIS 1998. compared with 79 percent of children in thie richest 20 percent (table 4.4). As with Table 4.5 Male and female enrollment rates, 1997 regional disparities, disparities across eco- Grass enrollment Net enrallment Age-specific ncimic groups widen at higher levels of edu- Males Females Males Females Males Females cation. Virtually no children in the poorest Primary 1 9.5 106.0 68.3 65.7 69.2 66.7 20 percent continue beyond primary Lower-secondary 40.9 27.7 15.5 12.5 84.2 69.8 sclhool; their net enrollment is 6 percent Upper-secondary 15.2 8.2 8.4 5.0 54.4 30.4 in lower-secondary schools and 2 percent Post-secondary 2.2 0.4 1.2 0.3 13.3 7.0 in upper-secondary schools. The corre- Source:NIS 1998. sponding ratios are 31 percent and 18 per- cent for the richest quintile. education will benefit primarily more The population of primary school-age affluent groups.4 Subsidies to primary children in Cambodia is more or less education tend to be distribution- equal across the per capita expenditure neutral, however. quintiles, except for the richest quintile, where there are fewer primary school- age children.3 At higher levels, the dif- Gender Disparity ferences are even starker. For instance, At the primary level net enrollment for only 8 percent of lower-secondary stu- boys is only 3.5 percent greater than net dents are from the poorest quintile, enrollment for girls (table 4.5). The while 38 percent are from the richest. discrepancy in lower-secondary educa- And the poorest quintile accounts for tion is much larger, but the biggest dis- only 4 percent of upper-secondary stu- parity is in higher schooling. Roughly 69 dents, the richest quintile for 56 percent. percent more boys are in upper-sec- This implies that government subsidies ondary school and 430 percent more are tci upper-secondary and post-secondary in post-secondary school. Although  40 CAMBODIA POVERTY ASSESSMENT these are large relative differences, needs to set appropriate curriculum both males and females have low net standards, ensure that they are upheld enrollment ratios in absolute terms through school inspections, and institute (0.3-8.4 percent). national standardized primary- and sec- ondary-leaving examinations. Role of the Public Sector Determinants of Enrollment Until 1991 the state ran all schools. Since among the Poor then, there has been some growth in pri- vate schools, although they still account Identifying the causes for low enrollment for less than 1 percent of all students at among the poor is the first step in tack- all levels (table 4.6). Among post-sec- ling the problem. School enrollment is ondary students in the 1997 survey sam- influenced by several factors, including ple, there is a generally strong, positive schooling costs, proximity to schools, relationship between private schooling quality of schools, parents' schooling, and and per capita expenditure (the sample returns to education. size is quite small, however). This means that the proportion of students in private school can be expected to High Private Costs of Schooling increase significantly with economic The public sector runs all but a handful of growth and rising incomes, especially schools in Cambodia but private spending from families in the r-ichest quintiles. To accounts for a large share of schooling satisfy the enormous growth in demand expenditures. Total annual spending of for secondary schooling that will surely the Ministry of Education in 1996-97 was come with economic prosperity, the CR 81.95 billion ($33 million), of which Cambodian governmient will have to CR 8.91 billion ($3.6 million) was financed look increasingly to the private sector to through school fees paid by parents. So expand secondary school coverage. the public subsidy to education was CR As in other developing countries, 72.35 billion ($28.9 million). Moreover, there are many explicit and hidden donors and NGOs spent CR 143.15 billion restrictions on private sector participation ($57 million) in the education sector in in education in Cambodia. These need to 1996-97. The 1997 survey data indicate be recognized and rationalized or that household spending on schooling removed if partnership with the private (books, materials, transportation, uni- sector is to be of benefit. Even in private forms, and so on) was CR 141.06 billion schools, the governme nt has an obligation ($56 million) in 1997 (table A4.13). This to ensure the quality of education. It means that the government provides Table 4.6 Private schooling enrollment by consumption quintile and region, 1997 (percentage of oal students) Consumption quintile Nation- Phnom Other Rural I-poorest 2 3 4 5-richest wide Penh urban areas areas Primary 0.0 0.4 0.S 0.6 2.1 0.7 3.7 0.7 0.3 Lower-secondary 1.2. 0.0 0.0 0.8 1.5 0.8 3.3 0.4 0.0 Upper-secondary 0.0 0.0 0.0 2.0 2.6 1.9 3.2 0.0 1.0 Post-secondary 0.0 0.0 0.0 1.5 1.2 1.5 0.0 0.0 All 0.! 0.4 0.4 0.7 2.0 0.7 3.5 0.6 0.3 Note: There are only 62 post-serondary students represented in the 1997 survey sample (61 in public and one in private insti- tutions). Source: NIS 1998.  SCH-JOOLING AND THE POOR 41 only a fifth of all education resources in to household nonfood spending per Cambodia. Households finance about two- capita, with the latter serving as a proxy for fifths, and donors and NGOs account for discretionary household income (table the rest. 4.7). For example, the quintile-specific Though primary schooling is officially affordability ratio of 26 percent for pri- free in Cambodia, most parents have to pay mary schools for the poorest 20 percent of a considerable amount for their children's Cambodians is calculated by dividing their primary education. In addition to the cost average schooling cost per student of CR of school uniforms and textbooks, there 23,925 ($9.50) by nonfood expenditure are admission fees and various supple- per capita of CR 91,200 ($36). ments. Private tutoring, often by the Two things are clear. First, schooling sarne teacher as at school, is a major is simply unaffordable for the poor. The exjpense. It is frequent and nearly obliga- cost of one child in primary school takes tory, because it is seen as a favor to teach- up a quarter of all nonfood spending per ers, who have to supplement their low capita in the poorest quintile, while a salaries. Finally, students and their families child in lower-secondary school takes 57 have to contribute almost the entire costs percent of nonfood spending. of school construction, equipment and fur- Second, schooling costs are much niture, and maintenance. The household higher relative to discretionary con- survey collected information on expendi- sumption for the poor than for the bet- tures in 12. months (to June 1997) on ter-off. Average schooling cost per student tuition, school improvement fees, books in primary schools amounts to 26 percent and other supplies, school uniforms, pri- of nonfood consumption spending per val:e tutoring, transportation, and lodging capita for the poorest quintile, but only (for boarding students) for each child 12 percent for the richest quintile. between the ages of 6 and 24 currently in Differences in affordability ratios across school (tables A4.1-A4.10). the poorest and richest quintiles are even greater at the secondary level. Total household schooling costs. Per student What is the budget share of actual household costs, even in public schools, schooling expenses? Although average aria significant, amounting to an average household schooling costs per student rel- of CR 48,328 ($19) in primary schools, CR ative to nonfood consumption may be 142,466 ($57) in lower-secondary, CR higher for the poor, the share of total 267,055 ($107) in upper-secondary, and schooling expenditure in total household CR 210,163 ($84) in post-secondary insti- consumption expenditure could well be tutions.5 At all levels tuition fees and lower for the poor because of their lower sclhool improvement fees amount to only rates of enrollment. Across all quintiles about 8-10 percent of household school- only 1-1.5 percent of total consumption ing costs, with uniforms, books, school sup- expenditure is devoted to education plies, transportation, and other costs (table 4.8). The budget shares of educa- accounting for 90 percent or so. tion spending increase directly (but not Household spending per student dramatically) with per capita consump- increases sharply with per capita con- tion, from 1.05 percent for the poorest sumption spending, but this may indicate quintile to 1.46 percent for the richest. that better-off households are purchasing The absolute differences in expendi- better quality schooling inputs than the tures are much larger. poor. Affordability of public schooling can Composition of schooling costs. Virtually all best be assessed by looking at the ratio of school expenditures-fees, uniforms, average schooling expenditure per student books, supplies, and so on-increase  42 CAMBODIA POVERTY ASSESSMENT Table 4.7 Affordability of public schooling by consumption quintile and region, 1997 (ratio of schooling cost to nonftbod expenditure) Consumption quintile Nation- Phnom Other urban Rural f-poorest 2 3 4 5-richest wide Penh areas areas Quintile average cost Primary 26.2 21.0 18.9 17.7 11.9 14.6 12.6 16.1 15.4 Lower-secondary 56.9 50.2 41.5 34.0 26.2 43.7 26.0 37.5 40.4 Upper-secondary 83.2 86.3 66.5 66.0 38.4 80.9 33.7 80.8 75.1 Post-secondary 0.0 38.9 8.3 28.2 63.7 20.4 152.3 0.0 Cambodia average cost Primary 52.8 34.2 23.5 15.7 5.3 14.6 5.1 11.9 20.0 Lower-secondary 155.5 100.9 69.4 46.4 15.7 43.2 14.9 35.0 59.0 Upper-secondary 291.5 189.1 130.0 86.9 29.5 80.9 27.9 65.6 110.5 Post-secondary 229.4 148.8 102.3 68.4 23.2 63.7 22.0 51.6 87.0 Memorandum item: Per capita nonfood consumption spending (thousands of riel) 91.2 141.2 205.4 307.4 905.5 330.1 956.7 407.1 241.6 Source: NIS 199. Table 4.8 Budget shares devoted to year. All of this suggests that students education, by consumption quintile and belonging to the poorest quintile face dis- region, 1997 advantages in access to essential school (thousands of riet) inputs. To the extent that these inputs lead Share of to better student learning and perfor- Annual education in mance, test scores of poor students would household Annual consumption be expected to be worse than those of bet- expenditure household expenditure on schooling consumption (percent) ter-off students. It is surprising that school fees diverge Consumption quintile widely between rich and poor groups even I-poorest 24.8 2,356.8 1.05 2 33.6 3,220.1 1.04 inpublicschools (tableA4.1).Thiscould 3 44.5 3,871.8 1.15 reflect informal fees and other supple- 4 64.6 5,192.5 1.24 ments (say, for running electricity gener- 5-richest 162.1 11,113.7 1.46 ators or paying a supplement to teachers) Nationwide 67.9 5,150.6 1.32 over which schools have more discretion Phnom Penh 179.9 9,933.6 1.81 than they do over official tuition fees. Other urban areas 92.9 6,976.0 1.33 School uniforms, which are obligato- Rural areas 44.3 4,308.2 1 .03 ry, are one of the biggest schooling costs, Source. NIS 1998. especially for poor households. The poorest quintile spends CR 8,354 ($3) a year on school uniforms for children in sharply with per calpita consumption public primary schools-much more spending for all schooling levels (tables than the CR 2,924 ($1.20) they spend on A4.1-A4.9). For exarnple, the richest 20 tuition and CR 2,851 ($1.15) for books. percent of Cambodians spend more than Whether abolishing compulsory school twice as much as the poorest 20 percent on uniform rules in public schools will books, uniforms, and other school supplies. reduce costs to the poor, however, While the average prirnary student in the depends upon the extent to which poorest quintile spends next to nothing on school uniforms simply substitute for private tutoring, the average student in the other clothes that would have been pur- richest quintile spends CR 19,946 ($8) a chased anyway.6  SCIHOOLING AND THE POOR 43 Schoolfees. Exempting the poor from pub- school. A little more than half of lic: school fees could improve school Cambodians have a village primary access for poor children. Households school, but only 5 percent have lower- reported paying no fees for only about 9 secondary schools and less than 2 percent percent of primary school students (there have upper-secondary schools in their vil- are no official fees at this level). Most fees lage. Only 20 percent of provincial paid by primary school students in pub- Phnom Penh's residents have a primary lic: schools were likely voluntary contri- school in their village, compared with 60 butions and other special supplements. percent in other urban areas and 54 per- While more primary school students from cent in rural areas. Overall, half of the the poorest quintile (12 percent) were rural population does not have a prima- exempted than those from the richest ry school in their village, and more than quintile (7.3 percent), the difference is 95 percent have no secondary school. small. In lower- and upper-secondary Average distance to the nearest schools schools a higher percentage of the stu- outside the village increases with education dents from the poorest quintile pay no level; this distance is 1.2 km to the nearest fees compared to all other quintiles. But, primary school, 5.6 km to the nearest then, many from the richest quintile also lower-secondary school, and 10.8 km to pay no fees. Clearly there is no systemat- the nearest upper-secondary school. ic pattern of fee exemptions. These distances are longer for children School improvement fees appear to be from the poorest quintile, who, on aver- more discretionary, particularly at pri- age, walk 6.9 km to the nearest lower-sec- mary and upper-secondary levels. Almost ondary school and 11.9 km to the nearest 40 percent of primary school students in upper-secondary school. (Children from Cambodia reported paying no school the riches quintile walk an average of 4.3 improvement fees, with a much larger km to the nearest lower-secondary school percentage coming from the poorest and 8.4 km to the nearest upper-secondary quintile (49 percent) and rural areas (43 school.) With such long travel distances, percent). In upper-secondary schools 34 it seems scarcely feasible for day students percent of students paid no improvement from the rural population, 95 percent of fees (79 percent in the poorest quintile). which is without a village secondary Moreover, almost 23 percent of school, to attend secondary school. Cambodian primary school students Physical access to schools is also affected paid nothing for school books. This, of by the poor quality and limited availabil- course, could mean that students used no ity of roads. Most rural roads are impass- textbooks, not that they received them able in the rainy season (chapter 6). free. There does not appear, however, to be a link between payment for books and economic status.7 School Quality Assessing school quality is more difficult both because the notion of quality is more Piroximity to Schools complex and because data on quality indi- Another reason for the lower enrollment cators are not readily available. There is among Cambodia's poor children may be an extensive literature on school effec- the distance they have to travel to tiveness, which suggests that inputs (qual- schools. There are two measures of ity of curriculum, availability of learning physical access to schools: the percentage and instructional materials, pupil-teacher of the population having a school in their ratios, availability of trained teachers, and home village and average distance to the so on) are critical in determining educa- nearest school for those without a village tional effectiveness and student learning.8  44 CAMBODIA POVERTY ASSESSMENT Village data can be merged with regions. The Ministry of Education reports household data from ithe 1997 survey to that most Cambodian teachers have ped- analyze the quality of school inputs avail- agogy training. Similarly, there is no clear able to the population by consumption pattern in the number of schools with quintile and region (table 4.9). On one enough books. The highest grade taught widely accepted indicator of school qual- increases with consumption levels and is ity, pupil-teacher ratios, Cambodia's aver- higher in urban areas, but the differences age of 63 students f'or each primary are not dramatic. school teacher is much higher than for Despite the lack of discernible patterns other countries in the region.9 There is in some indicators on school quality, the also a big rural-urban difference in pri- evidence suggests that Cambodia's poor mary school pupil-teacher ratios-35 in and rural children are worse off in the Phnom Penh and 39 in other urban quality of schools available to them areas, but 69 in rural areas. (particularly in availability of teachers). This indicator also varies significantly This may be a factor in lower enrollment across consumption quintiles. In primary rates among the poor. Given the high schools attended by the poorest quintile, costs of schooling, parents might not con- the ratio is 98.6-more than twice as high sider it worthwhile to send their children as in primary schools attended by the rich- to poor-quality schools. Moreover, poor- est 20 percent.10 Secondary school pupil- quality schools mean poor student per- teacher ratios are much lower-19 formance, increasing the likelihood of nationwide at both lower- and upper-sec- repeated grades and dropouts and reduc- ondary levels. Ratios ,generally increase ing the likelihood that poor children will with per capita consumption at the lower- be able to move on to higher education. secondary school level and decline at the upper-secondary level. For primary schools, other quality indicators (such as Parents' Schooling the percentage of trained teachers) show Evidence from around the world shows less stark differences between quintiles and that parents (especially mothers) with Table 4.9 School quality indicators by consumption quintile and region, 1997 Consumption quintile Notion- Phnom Other urban Rural I-poorest 2 3 4 5-richest wide Penh areas areas Average pupil-teacher ratios Primary 983.6 64.4 55.90 54.30 45.60 63.40 38.70 35.10 68.80 Lower-secondary 16.6 15.1 17.90 19.40 25.90 18.70 27.90 14.80 18.70 Upper-secondary 22.0 21.4 18.70 18.10 16.20 19.00 21.40 15.00 20.30 Percent of teachers trained Primary 88.1 87.9 89.50 88.7 90.7 89.0 92.0 93.7 88.10 Lower-secondary 9S8.1 96. 93.1 94.6 94.4 95.4 81.2 100.0 95.90 Upper-secondary 97.4 97.7 99.3 98.8 99.0 98.4 100.0 98.5 98.3 Highest grade taught' Primary 4.7 5.1 5.0 5.1 5.1 5.0 5.2 5.4 4.9 Lower- secondary 8.4 8.9 8.9 8.8 8.8 8.7 9.4 9.1 8.6 Upper-secondary 10.9 11.7 11.6 11.9 12.0 11.6 12.0 12.0 11.5 Percentage in schools with "enough" books Primary IS.3 18.7 16.9 17.8 22.2 18.7 25.0 31.6 16.4 Lower-secondary 1;2.8 14.5 16.0 8.6 17.7 14.0 14.3 9.7 14.8 Upper-secondary 413.3 39.7 30.3 21.4 10.7 31.1 18.7 21.8 33.7 a. Many Cambodian primary schools are annex or satellite schools that do not offer the full six years of schooling. Some sec- ondary schools do not offer all grades at either the lower- or upper-secondary levels. Source: Merged village data and household survey data from the Cambodia Socioeconomic Survey of 1997.  SCHOOLING AND THE POOR 45 more schooling are more likely to send have upper-secondary (or above) school- their children to school and keep them ing. So, as elsewhere, maternal education there longer."1 A larger proportion of is strongly associated with child enroll- mothers in the poorest quintile (60 per- ment in Cambodia. cent) have no schooling, compared with those in the richest quintile (39 percent: figure 4.1). And, not surprisingly, fewer Retums to Schooling women in the poorest quintile have sec- If wages do not reward people according ondary or post-secondary schooling. to their levels of education, enrollments There is a strong positive link between will likely be low, since substantial eco- mothers' schooling and enrollments at nomic benefits are needed to justify the most levels (table 4. 10). In particular, chil- high costs of schooling. In Cambodia men dren with mothers who have had post-pri- with primary schooling earn a 15 percent mary education are much more likely premium over those with no schooling, than other children to attend secondary and women a premium of 36 percent and post-secondary schools. This is to be (table 4.1 1). For lower-secondary school- expected since secondary school enroll- ing the premium over primary schooling ment is more discretionary in Cambodia. is 18 percent for men and 24 percent for For instance, while the gross post-sec- women. And the wage premium for ondary enrollment rate is a meager 1 per- upper- and post-secondary education cent for children whose mothers have no (over lower-secondary) is 21 percent for schooling, it runs as high as 8 percent for men and 15 percent for women. Earnings children whose mothers have upper-sec- initially increase with age, then flatten (or onclary (and above) education. Likewise, even fall) at older ages. This is typical and gross lower-secondary enrollment is 24 reflects positive but diminishing returns percent for children whose mothers to education up to age 50 or so followed have no schooling, 39 percent for chil- by negative effects on earnings in later life. dren whose mothers have primary edu- Women earn less than men at all ages cation, 72 percent for children whose and all schooling levels. However, the mothers have lower-secondary schooling, wage premium for primary and lower sec- and 90 percent for those whose mothers ondary education (over no schooling and Figure 4.1 Mothers' level of completed schooling, by consumption quintile and region, 1997 Percentage of mothers in qu tile 100 80 60 Upper secondar-y and above 40 miii Lower-secondary 20 ;1 Pnmary 20 None 0 1 2 3 4 5 Nation- Phnom Other Rural wide Penh urban areas areas Consmupton quintile 5oun,eN1S 1998.  46 CAMEODIA POVERTY ASSESSMENT Table 4.10 Net, gross, and age-specific enrollm lent rates, significant determinants of school enroll- by mothers' level of schooling, 1997 ment. The probability of school enroll- (percentage at each level) ment is significantly related to age, Mothers'schooling _ increasing steadily up to age 11, but Lower- Upper-secondary declining continuously from ages 13 to None Primary secondary and above 17. Net enrollment rates There are also strong gender effects Primary 60.8 71.5 76.8 90.5 on school enrollment probabilities. At vir- Lower-secondary 8.0 18.3 30.7 31.9 tually all ages male children are more Upper-secondary 2.2 8.0 19.8 37.3 Post-secondary 0.4 0.3 2.8 3.6 likely to be enrolled in school than female children. The male-female dis- Gross enrollment rates Primary 110.0 115.4 111.8 137.7 parity is relatively small at younger ages Lower-secondary 23.6 39.3 73.4 89.8 but larger at older ages. During the pri- Upper-secondary 4.9 11.1 34.8 72.9 mary school years (ages 6-11) males are Post-secondary 0.7 0.7 4.1 8.1 only 3-4 percent more likely than Age-specific enrollment females to be enrolled in school. Ages 6-1 1 61.4 72.3 80.1 95.4 However, this difference increases to 14 Ages 12-14 71.4 83.1 85.9 95.1 Ages 15-17 35.5 45.2 58.8 77.2 percent among children ages 12-14 Ages 18-24 9.4 9.5 1 0.8 25.5 and to 25 percent among children ages 15-17. SssrceNIS i99B. Household standard of living, as prox- Table 4.11 Real monthly earnings, by gender, age, and schooling ied by household consumption spending level, 1997 per capita, also has a strong positive effect (riel) on school enrollment. School enrollment Lower- Upper- and post- is more income-elastic for older children None Primary secondary secondary All levels than for younger ones. For example, the -2en elasticity of school enrollment with respect Ages 15-29 91,660 103,755 136,259 171,610 114,623 to per capita spending is 0.10 for children Ages 30-39 112,221 130,363 130,465 157,361 127,788 ages 12-14, but 0.18 for children ages Ages 40-49 117,896 153,880 200,691 306,764 167.247 15-17. Interestingly, for the 6-11 age Ages 50 + 119,386 128,677 131,693 All ages 110,554 127,495 149,938 180,825 132,266 group the association between per capita Women r-educed substspending and school enrollment is Ages 15-29 52,351 86,861 103,866 136,136 84,668 reducedsubstantiallywhenprbnasyschool Ages 30-39 71,175 101,687 121,281 134,320 93,790 quality indicators are introduced in the Ages 40-49 79,744 111,543 126,500 102,018 equation. This suggests that part of the Ages 50 + 74,115 80,741 140,679 182,188 81,877 positive correlation between per capita All ages 70,422 95,814 119,067 137,097 90,059 spending and school enrollment observed Source: NIS 1998. in cross-tabulations is due to a positive asso- ciation between school quality and per primary schooling, respectively) is greater capita expenditure (that is, better-off for women, especially at younger ages. households live in communities with This should make it easier to target higher-quality schools). Indeed, since schooling opportunities to women. the equations for children ages 12-14 and 15-17 do not include school quality indi- cators, it is possible that the higher esti- Multivoriate Analysis of School Enrollment mated income effects on enrollment for The results of multivariate regression these age groups simply reflect an unob- analysis of determinants of school enroll- served school quality effect. ment (appendix D) broadly confirm that The education of adults in a household most of the factors dlescribed above are is also among the most significant factors  SCHOOLING AND THtE POOR 47 affecting children's school enrollment. not across households. There were too few Female adult education seems to be more cases to calculate meaningful village- important than male adult education in specific fees for lower- and upper- explaining enrollment for children ages secondary school. 12--14, but the reverse is true at older ages. Quality variables are statistically sig- Hcusehold size is not a significant deter- nificant and suggest that children ages minant of the probability of school 6-11 are much more likely to be enrolled enrollment for children ages 6-11, but has when village schools have more teachers a surprising positive association with the and a larger proportion of trained teach- probability of enrollment for children ers. The estimated elasticity of school ages 12-14 and 15-17. This positive effect enrollment for children ages 6-11 is 0.09 is large in magnitude; the elasticity of with respect to the number of teachers school enrollment with respect to family per primary school and 0.18 with respect size is 0.33 for children ages 15-17. This to the proportion of trained teachers, in res,ult may reflect the fact that older chil- primary schools. dren, especially girls, are often kept from att,ending school because they are respon- sible for household chores, including the Raising Enrollment of the Poor care of younger siblings. In larger house- holds there are likely to be other adult The results of the analysis of determinants members who can take care of younger of school enrollment suggest that effective children, relieving secondary-school-age policies to increase school enrollment children of these responsibilities. among the poor need to focus on both Distance to school has the expected supply- and demand-side interventions.12 negative effects on school enrollment, Improving physical access to schools by although the numerical magnitude of the constructing more schools in poorer effects is quite small. For instance, an areas and by improving rural infrastruc- increase of one kilometer in the distance ture will be critical to raising enrollments, to the nearest primary school is associat- but policy must also address issues of price ed with a decline of only 0.2 percent in and quality. Relative costs of schooling for enrollment for children ages 6-11. The the poor will need to be reduced. School effects are even smaller for older children. fee exemptions alone are inadequate in The results indicate no significant this regard since fees constitute only 8-10 eflfect of primary school fees on the enroll- percent of total schooling costs. A clearly ment of children ages 6-11. However, it established and transparent system of is unlikely that these results indicate the need-based scholarships for the poorest true price responsiveness of schooling students will be needed to raise enrollment demand because of the conceptual diffi- for this group. Such a system would ensure culty of calculating the "price" of school- that poor students could purchase essen- ing. The amount actually paid by parents tial inputs, such as textbooks. Another for their children's schooling includes option is for the government to provide both the true price of schooling and the (or insure) loans to needy students to help quality of schooling chosen by parents. them cover education costs. The advan- For this reason, the primary school fees tage of a loan program is that it relies on reported by village leaders were used as private resources in a situation in which the relevant price variables. For villages government funds are severely limited. where no such fees were reported, the Funds for scholarships and loans could be median fee reported by parents was used. obtained by reducing high levels of pub- Th-us the primary school fee variable used lic subsidies to post-secondary education, in this analysis varies across villages, but a practice that benefits mainly the rich.  48 CAMBODIA POVERTY ASSESSMENT Along with reducing schooling costs, a increase efficiency in education by fos- pro-poor schooling policy should address tering competition between public and the problem of school quality.13 The extra- private schools. Household survey data ordinarily high pupil-teacher ratios in already show a strong shift of upper- rural areas (80 and above) need to be income groups to private schools. This brought down at least. to the national aver- should be encouraged, so that the gov- age (around 60)-and probably lower. 14 ernment can better target its expenditures This will require a more rational policy of to the poor. First, however, it would be use- deploying teachers across provinces and ful to identify constraints and problems, villages. One way to do this is to pay teach- including any formal or hidden restric- ers a premium to work in underserved tions imposed by the government, that areas (a policy used with some success in may put a brake on the private sector Indonesia and elsewhere). Funds to from expanding its role in the provision finance these quality improvements could of schooling in Cambodia. also be obtained, at least in the short run, from savings generated by reducing sub- sidies to post-secondary education. Notes Targeting at least some additional 1. The gross primary enrollment ratio is the schooling subsidies to poor girls would total enrollment in primary schools, regardless raise efficiency and equity in the sector by of age, divided by the population age 6-11. promoting female enrollment. Some The net enrollment ratio is the enrollment of other developing countries (Bangladesh children ages 6-11 in primary schools divided and India, for instance) have found schol- by the population ages 6-11. Enrollment ratios arship and stipend programs targeted to for other levels of education are defined sim- girls an effective means of raising female ilarly, using different age groups. The age enrollments. groups corresponding to the official school Female enrollment could also be ages in Cambodia are 6-11 years for primary, increased by creating night schools for sec- 12-14 for lower secondary, 15-17 for upper ondary-school-aged girls. This would allow secondary, and 18-24 for post-secondary. girls ages 12-17 who are normally respon- 2. Of all children ages 6-11, 76 percent live sible for household chores, including care in rural areas, 12 percent in other urban areas, of younger siblings, to attend school with- and 12 percent in Phnom Penh; the share of out depriving their families of their labor. the rural areas in the non-school-going child The role of the private sector in school- population is much greater than its popula- ing is still limited in Cambodia. Fewer than tion share. 1 percent of students are enrolled in pri- 3. The poor are still underrepresented rela- vate institutions, with. most private enroll- tive to their share of the primary-school-age ments in post-secondary schools (typically population, while the rich are overrepresented. vocational training). The support of the 4. According to the most recent estimates, private sector will be needed to expand sec- about 36 percent of the Cambodian popula- ondary school coverage in Cambodia. One tion is poor (Cambodia Ministry of Planning strategy would be for the government to 1998)-corresponding roughly with the bot- target its limited education spending to tom two quintiles. providing primary and secondary educa- 5. Since there were only 62 post-secondary tion to poor students and to let the afflu- students in the 1997 survey sample (61 in ent turn to the emerging private sector for public institutions and 1 in a private institu- schooling needs. tion), the mean expenditures shown in table Alternatively, vouchers could be used A4.12 for post-secondary schools are unlike- to provide the poor vvith access to private ly to be representative of the true population schools. Broader use of vouchers might of those students.  SCHOOLING AND THE POOR 49 6. The household expenditure data col- reduces children's time spent in class and lected in the 1997 survey show no relationship undoubtedly compromises schooling quality. between levels of household expenditure on 11. Mothers of children over 5 years of age clothing and household expenditure on cannot be identified directiy in the household school uniforms (with other factors, such as survey. Most children are identified as chil- household size, composition, and total expen- dren of the household head. Where mothers diture held constant), suggesting that there are listed either as household head or as the is rio (or little) substitution of uniforms for spouse of the head, mothers can be identified other clothing expenditure. for most children (97.1 percent; although in 7. The 1997 village questionnaire asked a few cases the mother so identified may be whether schools in the village provided books a stepmother). free to students. Free books were provided in 12. As noted earlier, it is possible that the primary schools attended by 75 percent of stu- insignificance of school fees in the enroll- dents from the poorest quintile, and 90 per- ment regressions may simply reflect sub- cent of students from the richest quintile. stantial measurement errors, since the school 8. Student learning is also affected by the fee data were not collected directly from amount of time that a student invests in schools. schoolwork, as well as a student's innate abil- 13. Although the multivariate analysis does ities and socioeconomic background. not provide strong evidence of the importance 9. For instance, in 1995 the primary school of school quality, it is possible that the limit- pupil-teacher ratio was 24:1 in China, 23:1 in ed (and possibly unreliable) information on Indonesia, 30:1 in Lao PDR, 25:1 in Mongolia, schools in the 1997 household survey may be 48:1 in Myanmar, 39:1 in Nepal, 35:1 in the responsible for these findings. Better and Philippines, 28:1 in Sri Lanka, 20:1 in more extensive Ministry of Education data are Thailand, and 25:1 in Vietnam. India was the available on the characteristics of schools only Asian country with a pupil-teacher ratio throughout Cambodia; these data should be as large as Cambodia's (UNESCO 1998). merged with the household survey data to 1(). Many Cambodian schools faced with facilitate a more complete analysis. such high pupil-teacher ratios operate in two 14. The national average pupil-teacher (or even three) shifts. Although this reduces ratio for primary schools (not weighted for class size to more manageable levels, it population) is about 44.   5., Access to Health Services I s in education, Cambodia has of reported cases of tuberculosis, polio, made impressive progress in diphtheria, measles, pertussis, and neona- improving the health status of its tal tetanus between 1992 and 1997 population since the end of the Khmer (table 5.2).1 Rouge period. The health status of Because the 1998 census was the first Cambodians is, however, still among the in more than 30 years, and because the worst is the region. There are still no vital statistics registration system in health facilities in several parts of the Cambodia is virtually nonexistent, it is dif- country, and in other areas health facil- ficult to get consistent estimates of ities are too dilapidated to be of any use. infant mortality over time. The limited Since 1996 the Ministry of Finance, evidence, however, suggests that infant with the assistance of donors and mortality has declined sharply since NGOs, has been implementing a health 1980 (figure 5.1). However, it is still the services plan to expand access to health highest in the region at 90-115 deaths facilities. The Ministry of Health has per 1,000 live births (Cambodia Ministry also been instrumental in developing and expanding programs to tackle Table 5.1 Immunization coverage, 1992-97 infectious diseases-such as the (percent) National Tuberculosis Program and Vaccine 1992 1993 1994 1995 1996 1997 the National Malarial Program-and in Tuberculosis (BCG) 50 57 78 95 88 80 vastly increasing coverage of child Pbliomyelitis (OPV3) 32 36 54 80 75 72 immunization programs. Diphtheria, pertussis, tetanus (DPT3) 32 35 53 79 74 72 Measles 33 37 53 75 71 70 Tetanus for pregnant women (TT2+) 6 22 28 36 36 30 Some Improvements in Health Note: Percentage of children immunized before the age of one and pregnant women imnmunized Status against tetanus. Source: Cambodia Ministry of Health 1997. Between 1992 and 1997 immunization of Table 5.2 Cases of disease reported to the Expanded Program of infants up to 12 months old against polio Immunization, 1992-97 measles, diphtheria, pertussis, and Vaccine 1992 1993 1994 1995 1996 1997a tetanus more than doubled, from 32-33 percent to 70-72 percent (table 5.1). Poliomyelitis 146 135 296 103 s50 b Im munization of pregnant women Diphtheria 13 22 5 nr nr 3 against tetanus increased sixfold, from 6 Measles 2,759 1,226 946 2,038 697 134 percent to 36 percent. Coverage of Pertussis 995 611 63 nr nr nr immunization against tuberculosis also Neonatal tetanus nr 88 147 a 2 3 rose, from 50 percent to 80 percent. The nr Not reported. a.january-September. increase in immunization coverage is b.The last confirmed case was In March 1997. refl'ected in big declines in the number Source: Cambodia Ministry of Health 1997 51  52 CAMBODIA POVERTY ASSESSMENT Figure 5.1 Infant mortality rate, Cambodia, 1960-96 strife. More critical than this difference 6f- dedth, pe, !,OGC We biths is the fact that life expectancy remains 350 very low for both men and women in Cambodia relative to other countries in Asia (figure 5.2). High mortality and low life expectan- cy are partly the result of low levels of health service use by Cambodians. The health ministry estimates that the annu- al number of outpatient consultations in all health facilities was a meager 0.35 per 1960 180 .985 1987 !9P9 1996 capita in 1996 (Cambodia Ministry of SceCamboci Mi-t,yM.He.;th 1997. Health 1996)-well below both the World Health Organization (WHO) of Health 1997; Huguet 1997). It is esti- standard of 0.60 for rural areas and the mated that 40 percent of such deaths yearly contact rates elsewhere in the occur in the first month of life due to region. Only 15.9 percent of births maternal health and delivery condi- (measured using a crudely estimated tions, including low birth weight (USAID birth rate) took place at hospitals or 1996). health centers (Cambodia Ministry of Average life expectancy at birth is Health 1996). Most deliveries took place estimated at 54.4 years. Men can expect at home, probably with the assistance of to live an average of 50.3 years and traditional birth attendants. women 58.6.2 The more than eight year difference between these figures may reflect both the incomplete reporting of The Growing HIV/AIDS female deaths (relative to male deaths) Epidemic in the Demographic Survey of Cambodia 1996 and the higher mor- Increasing rates of HIV/AIDS infection is tality of males because of civil war and also a major concern. Cambodia has Figure 5.2 Life expectancy and infant mortality in Asia, 1995 1Ifn-; deo!ho per I000 live binhs 120 Ifla,t dea.hs (per ,000 ive oits) 100 Yeors _80 Average life _0 80 _ epectancy __ 80 60 60 40 ~~~~~~~~~~~~~~~~~~~~~40