Poliky, Pes.archiand Exlemal Affalrs WORKING PAPERS Welfare and Human Resources Population and Human Resources Department The World Bank May 1990 WPS 416 Improving Data on Poverty in the Third World The World Bank's Living Standards Measurement Study Paul Glewwe An account of the World Bank's ambitious effort to collect household-level data on poverty in developing countries - and of what that data are beginning to say about the effects of gov- ernment policies on living conditions of the poor. The Policy, Research, and lExtsmal Affairs Compilex disinbutes PRIE Working Papers to disseminate the findings of work in progrsss and to encourage the exchange of ideas among Bank suff and all others interested in dvevloprent issue.s Trhes papess carry the names of the authors, reflct only their Stcws, and should he used and cited accordingl% T'he findings, interpretationss and c ,nclusiisns are the authors' own [hey should not he attrihuted to the World 11ink. i 11i rd of Di>rec:ors its management. or a:;y if its mcrnr'sr cCLntnes Policy, Res"rch, and Externl Affalis 'i'L* :i i l leU Iąs :i U:: Wellaro and Human Roeourcos This papcr-- a product of thc Welfare and Human Resources Division, Population and Human Resources Department i.. part of a largcr effort in PRE to examine the causes and consequences of poverty in developing countries. Copics are available free from the World Bank, 1818 H Strcet NW, Washington DC 2(433. Please contact Angela Murphy, room S9-137, extension 33750 (36 pages with tables) Thic starting point for reducing world poverty Most of the poor are in rural areas; the was to provide accurate, up-to-date data on pov- fraction of the poor population in rural areas is erty in developing countries. The sparse, always substantially higher than the fraction of outdated data previously available were often of the total population in rural arcas. dubious accuracy and usually unavailable in a form uscful for policy analysis. Most of the poor are in households in which the hcad works in agriculture. The heads of One o lthe most ambitious attempts to poor households arc most likely to be self- improve the quality of data collected at the employed, especially in Africa. (Very few heads househol(d level from developirng countries is thC of poor households work for the govemment, so Living Standards Mcasurement Study (LSMS) freezes on govemment wages are unlikely to program, which Lhe World Bank launched in hurt many of the poor.) 1 9X(. The thcads of poor households have low hleiC main objective of LS.IS surveys is to lcvels of education - most of them elementary proN ide household-level data for evaluatinig tle education or less, and in some (particularly eflect ol various governient policies on the African) countries no education at all. populition's living coniditiois -in studics, for cxaniple. of the impact of education on nutrition, Glewwe also prescnits selected results of thc clTect of hcalth on employment, and the studies on the persistence of poverty, and of relationship hcbween incoine and fertility. studies of the effeccs on the poor of structural adjustment, of food stamns and food subsidies, Alter describing how 01c LSMS bcgan and and of raising user fees for health care and edu- how data are collected, CGlevw\kc presents se- cation. lectcd results. Sonme gencral trenlds have cmerge(d in studics Of five of thc six countries for "Ihichi LSMS data arc available: 1Ih. I'RI. Wvorking Piaper Series dissetninltes thc iendings of work under vav in the Rank's Policv, Rcserch, and ExternaI Affairs Complex. An objective of the series is to get thcsc findings out quickly, even if presentations arc less than fully k)ihshod. Thc fFindings. int.rpretationF andl conclusions in thcse papers do not necessarily represent ofricial Bank policy. Produced at the l'RE Dissemination Center Improving Data on Poverty in the Third World: The World Bank's Living Standards Measurement Study by Paul Glewwe Table of Contents I. Introduction 1 II. The History of LSMS 2 A. The State of Data on Poverty in the Third World 2 Previous to !1SMS B. LSMS from 1980 to 1985: rhe Development of 4 LSMS Methodology C. Present LSMS Survey Activities 5 D. Future Survey Implementation 7 III. LSMS Methodology 9 A. Objective of LSMS Surveys 9 B. Consumption as an Indicator of Household 9 Welfare C. Distinctive Characteristics of LSMS Surveys 13 D. Organization and Implementation 15 IV. Selected Results from Studies on Poverty Using 19 LSMS Data A. Poverty Profiles 19 B. Comparability of Different Poverty Definitions: 21 Results from CMte d'Ivoire C. Structural Adjustment and the Poor 23 D. The Effect on the Poor of Raising User Fees for 27 Health Care and Education E. Food Stamps, Food Subsidies, and the Poor in 30 Jamaica F. Persistence of Poverty in Developing Countries 31 V. Conclusion 33 References 35 I. Introduction The extent of poverty in developing countries, an.d che choine of policies to ieduce poverty, have long been of interest to scholars, governments, international organizations and many other groups and individuals. The starting point for work in this area is data on poverty in developing countries that is both accurate and up to date. Unfortunately, the data actually available from these countries are often sparse, outdated, of dubious accuracy and not available in a form useful for policy analysis. In recent years several attempts have been made to increase both the quantity and the quality of data from the developing world. Much of tnis data is useful for poverty analysis, and indeed some has been gathered specifically with this purpose in mind. One of the most ambitious attempts to improve the quality of data collected at the household level from developing countries is the Living Standards Measurement Study (LSMS) program at the World Bank. This program is specifically aimed at improving data collection efforts so .s to better understand the exrent of poverty and the determinants of livings standards. This paper provides a description of LSMS activities since its inception in 1980. It is organized as follows. Section II sketches a brief history of the LSMS project ac the World Bank. The next section presents an overview of the statistical and economic methodology underlying the LSMS data collection efforts. Section IV provides several examples of the use of LSMS data for evaluation of the extent of poverty in developing countries and the analysis of the effect of particular policies on the poor. Section V summarizes the results and concludes the paper. -2- II. The History of LSMS A. The State of Data on Poverty in the Third World Previous to LSMS At the close of World War II three international organizations were founded which would, among other responsibilities, play a major role in developing data collection systems throughout the world: the United Nations, the World Bank (International Bank for Reconstruction and rev2lopment), and the International Monetary Fund (IMF). Also after the close of World War II, the European powers began to undertake the process of decolonization, so that by the 1960's most developing countries were newly independent rnations. Perhaps the first systematic attempt to ensure that data collected from both developed and developing countries were comparable was the introduction of the United Nation's System of National Accounts. 3eginning in late 1940's and early 1950's the UN developed an accounting proceduri for calculating economic aggregates such as Gross Domestic Product, which presented a picture of national economies. These prccedures were substantially revised in the early 1960's, and data collection efforts spread to other kinds of information, such as data on education (UNESCO), labor force activity (ILO), health (WHO), and agriculture (FAO). While these steps were a major achievement in the international collection of economic and social data, the published numbers were often national averages, and as such provided little information on the distribution of income and consumption. This in turn meant that they had limited use in the measurement of poverty. To improve on these efforts data had to be collected at the household level, either in the form of a census or a household sample survey. Such types of data collection had long been in existence in many -3- developed countries (the United States has collected census data for nearly 200 years) an.i even in some developed countries as well (e.g. Sri Lanka undertook a household expenditure survey in 1953 and a census in 1901). However, for most developing countries data of this type were hard to come by, and the data that were available suffered from a variety of problems, such as non-representative samples, doubtful reliability, and very long time lags between data collection and the production of published reports. The international community undertook several programs to improve data collection efforts at the household level in developing countries. One of the most important is the United Nations Household Ourvey Capability Program (UNHSCP), which began in the late 19?0's. An early description is given in UNSO (1980). In contrast, data colle. Lon at the household level has never been done at the IMF and before the 1980's was undertaken only for special purposes at the World Bank (e.g. an urban household survey in Bogota, Colombia, in 1973). Parallel to these developments, in the early 1970's there was an increasing recognition, both inside and outside the World Bank, that development efforts needed to pay more attention to the scope and nature of poverty in the Third World. Robert MacNamara, the president of the World Bank form 1968 to 1981, announced a major change in the focus of World Bank activities in Nairobi, Kenya, in September, 1973: the World Bank would promote projects designed to directly improve the living conditions of the poor. To support this work, the World Bank's Development Research Center, under the direction of Hollis Chenery, turned much of its attention to questions of poverty and inequality. This increased emphasis on the poor brought out the need for better data at tht household level, and it eventually Led to a new -4- entity within che World Bank which would examine ways to improve data collection efforts in developing countries, .he Living Standards Measurement SL.udy. B. LSMS from 1980 to 1985: The Development of LSMS Methodology In 1980 the World Bank established the Living Standards Measurement Study (LSMS) to explore ways of improving the type and 4uality of household survey data collected by statistical offices in developing countries.- Its ultimate goal is to promote the use of household data as a basis for making policy decisions in these countries. Soon after its inception, LSMS undertook a major review of the state of data collection as it then existed in developing countries with the objective of develoning a new system of data collection at the household level. This work was the major focus of LSMS activities in the early 1980's. This early LSMS work can be divided into two types: (1) review of past experience in the collection of household data; and (2) theoretical work on various aspects of living standards, with special emphasis on how data should be collected for purposes of doing empirical research. Some of the earliest LSMS reports were assessments of recent experience in collecting household level dati in d ve'Luping countries. These papers, written in the early 1980's, consisted of critical reviews of efforts in Africa, Asia and Latin America (Visaria, 1980; Altimir and Sourrouille, 1980; Scott, de Andre and Chander, 1980; Booker, Singh and Savane, 1980; Wahab, 1980; Musgrove, See Chander, Crootaert and Pyatt (1980) for the original statement of the goals of LSMS. -5- 1982). The second area of work was at the more theoretical level. To judge whether people are poor requires conceptual definitions of poverty. To reflect the multi-dimensional nature of poverty, special studies were undertaken on different aspects of living standards. Among the papers written were a study on the economic theory of welfare measurement (Deaton, 1980), studies on employment and living standards (Mehran, 1980; Grootaect, 1982, Acharya, 1982), reports on the collection of nutrition and health data in household surveys (Martorell, 1982; Ho, 1982; Sullivan, Cochrane and Kalsbeek, 1982), a similar study on the collection of education data (Birdsall, 1982), and several other papers looking at topics ranging from collection of price data to the usefulness of collecting panel data. The end product of five years of work was the development of a household survey methodology which was first implemented in Cate d'Ivoire in 1985. The questionnaires used are explained in Grootaert (1986) and the implementation of the survey in Cate d'Ivoire is explained in Ain- rth and Munoz (1986). C. 2resent LbM' Survey Activities As seen in Table 1, six LSMS surveys have been put in the field, and there are plans to begin three more in the next few months. The Cate d'Ivoire Living Standards Survey (Enquete Permanente Aupres Des Meages) began in February, 1985, and by 1989 4 years of data had been collected. The Peru Living Standards Survey (Encuesta Nacional de Hogares Sobre Medici6 de Niveles de Vida) went into the field in July, 1985 and collected one year of data. Future surveys in Peru are now under consideration. Field work for the Ghana Living Standards Survey began in September, 1987, and so far 2 years of - 6 - TABLE 1: LSMS Surveys I. Data Now Available Country Years of Data Collection Cote d'Ivoire 1985, 1986, 1S87, 1988 Peru 1985-86 Ghana 1987-88, 1988-89 Mauritania 1987-88 Bolivia 1988, 1989 Jamaica 1988, 1989 II. Soon to be Implemented Country Expected Date in Field Laos Fall, 1990 Morocco Spring, 1990 Pakistan Fall, 1990 III. Discussions Underway: Latin America/Carribean Africa/Middle East Asia Brazil Algeria Nepal Colombia Burkina Faso Guatemala Egypt Trinidad and Tobago Jordan Venezuela Kenya -7- data have been collected. The Mauritanian Living Standards Survey (Enquete Permanente sur les Conditions de Vie Des Menages) became operational in December, 1987. The Bolivian survey (Encuesta Integrada de Hogares) began in May, 1988. Finally, the Jamaican Survey of Living Conditions began operations in August, 1988. Three more LSMS surveys are due to go into the field by late 1990. The first is the Moroccan Living Standards Survey (Enquete Sur le Niveau de Vie des Moages au Maroc) will begin field work in Lne spring of 1990. The second is a living standards survey in Pakistan (Pakistan Integrated Household Survey), which is scheduled to begin field work in late 1990. The last is the Laos Living Standards Survey, which will also begin field work in late 1990. Several more countries have indicated interest in undertaking LSMS surveys, and discussions are now underway to see whether they will go forward. In Latin America and the Carribean discussions have been held in Brazil, Colombia, Guatemala, Venezuela and Trinidad. In Africa and the Middle East discussions have been initiated in Algeria, Burkina Faso, Egypt, Jordar and Kenya. In Asia, Nepal has shown interest in LSMS surveys. The three African surveys (C6te d'Ivoire, Ghana and Mauritania) are now being managed by the SDA (Social Dimensions of A-justment) Project in the Africa region of the World Bank. This project aims to strengthen the capacity of African govcenments to design and monitor poverty programs and projects. It is being managed jointly by the World Bank, UNDP, and the African Development Bank. D. Future Survey Implementa.ion The LSMS program began as an experimental research project but has -8- now evolved into a permanent World Bank development effort. The original Living Standards Unit of the Development Research Department has become the Welfare and Human Resources Division of the Population and Human Resources Department, which in turn is part of the Policy, Research .nd External Affairs (PRZ) complex of the World Bank. The transition from. a research project to a permanent entity highlights the need for a inore systematic process for starting new LSMS surveys. Future LSMS work will focus on: (1) refinements and diversification of the LSMS survey system, and (2) expanded resea-ch activities. Regarding the former, new questionnaires are being developed based on LSMS experience in several countries in the late 1980's. Attention will be given to adding flexibility, so as to allow for adaption to the special characteristics of individual countries, and to refinements in computer software and hardware systems, to allow for easier and faster implementation of surveys. In terms of the latter, emphasis is being given to developing re5carch capability in the countries in which LSMS su-veys are done and in undertaking new types of research at the World Bank in Washington. III. LSMS Methodology A. Objective of LSNS Surveys The main objective of LSMS 3urveys is to provide household level data for evaluating the effect of various government policies on the living conditions of the population. Accordingly, LSMS surveys collect data on all major aspects of household well-being. Tn this sense they are multi-topic surveys, gathering data on income, consumption, savings, employment and -nemplcyment, health, education, fertility and contraceptive prevalence, nutrition, housing and migration. Collecting data on these topics from the same households has the added advantage of allowing for the enalysis of the relationship between these different aspects of the quality of life. Examples of this include studies of the impact of education on nutrition, of the effect of health on employment, and the relationship between income and fertility. B. Consumption as an Indicator of Household Welfare Empirical research on the effect .f government policies on households often requires a broad indicator of household welfare. In most LSMS research on poverty, household ielfare is measured by consumption.21 When one selects a certain level of consumption as the minimum amount necessary, one has a poverty line. 2/ Of course, the LSMS data are rich enough to allow for the use of otner indicators of household welfare (cf. Glewwe and van der Gaag, 1988). Household consumption is used in mos; LSMS studies because of its intuitive appeal and rigorous theoretical framework. - 10 - Most people would agree that, other things being equal, increased consumption of goods and services raises individuals' levels of welfare. Much of what we observe in human behavior supports this assumption. Of course, there may be many factors other than the consumption of goods and services that affect welfare, but since these tend to be much more difficult to measure economists usually restrict themselves to that "portion" of human w^lfare which is attributable to consumption.3/ In welfare ec.'nomics, the starting point for measuring economic welfare is the utility function, which states that welfare rises as the consumption of various goods and services increases. To compare the welfare of different individuals, it is assumed thrt each individual or household possesses the same utility function. If one had data on the consumption of individuals, as distinct from the consumption of households, one could analyze the data using a utility function at the individual level. Unfortunately, most consumption data are collected at the household level and as such require analysis using a household level utility function. Thus one assumes that household utility is a function of the consumption of goods and services and the composition of household members. The composition "adjustment" is needed to account for the fact that households with differenc compositions require different consumption levels to attain the same level of welfare (e.g. larger households need more Loods and services to attain the 3/ Household surveys usually collect data on income as well as consumption, and some studies of welfare focus on the income data. Yet economic theory assumes that it is consumption, not income (which may be saved or given away), that raises welfare, and in most surveys consumption data appear to be more reliable than income data. - 11 - same welfare level as smaller households). Again, it is neccesary to assume that all households possess the same household utility function. Another consequence of the presence of consumption data at the household rather than the individual level is that one does not know the distribution of welfare within the household; one has little choice but to assume that all household members enjoy the same level of welfare. While one would like to observe the actual utility levels of households, one only observes their levels of consumption. Yet these are monotonically related. Specifically, "money-metric" utility, which is defined as the amount of money required (given a set of prices) to attain a specified level of utility, is equal to observed levels of consumption under the assumption of utility maximization.4/ Given consumption data at the household level, adjustments must be made for household size and differences in prices. Additional household members, particularly children, are less "costly", in the sense of requiring additional consumption to maintain the welfare level of the household, relative to the initial cost of attaining that welfare level in a household composed of a single person or a childless couple. This idea is supported by both common sense and economic reasoning. Clothing and other items can be handed down from older to younger children, durable goods such as radios and refrigerators can be enjoyed by additional members at no extra cost, and even in the case of food children consume less then adults. The method for adjusting for this phenomenon is the estimation of "adult equlivalence scales", 4/ For a thorough presentation see Deaton and Muellbauer (1980). - 12 - which measure the "cost" of additional household members in terms of fractions of adults (cf. Deaton and Muellbauer, 1980, Ch. 8). In addition, money-metric measures of utility also need to be adjusted for differences in prices. This can be done by dividing the value of household consumption by a price index. These theoretical considerations imply that household surveys must collect the necessary data for creating a comprehensive measure of household consumption, including data needed to adjust for differences in household size and in prices faced by households. LSMS surveys begin with direct consumption data, which include all explicit expenditures in the last 12 months as well as the value of food produced and consumed by the household. In addition to this, there is also a consumption component from the ownership of housing and durable goods (e.g. cars, televisions, bicycles, cameras), which are not consumed when they are built (housing) or purchased (durables) but are used over a long period of time. Household welfare derived from the ownership of such goods can be based on estimated yearly rental values of those goods. For housing, the best approach is to estimate hedonic rent equations (i.e. to predict the rental value of housing based on the characteristics of the dwelling) for those households which are renters. From these estimates imputed rents for owner-occupied housing can be calculated. For other durable goods, the rental value can be estimated based on depreciation in the real value of those goods over time. LSMS surveys collect data both on dwelling characteristics and on the ownership of durable goods which are sufficient to calculate the appropriate rental values both. They also collect data on household composition and local market prices so that adjustments can be made for their variation across households. - 13 - C. Distinctive Characteristics of LSMS Surveys Aside from collecting the data needed to get a comprehension measure of consumption, LSMS surveys are different from other household surveys in several other respects. First, while other surveys are primarily designed to measure different aspects of living standards, LSMS surveys collect information which allows one to analyze the determinants of the various outcomes that one observes. For example, school attendance of children may depend on the distance to the nearest school. The need for data on determinant factors explains why LSMS surveys collect data on prices, local schools and health services (including distance from the households and fees charged), conditions of local infrastructure (roads, sources of fuel and water, availability of electricity, communications, etc.), local agricultural conditions and practices, and historical information on household members themselves (characteristics of parents, employment history, migration history, etc.). Further, information at the household level is gathered on special government poverty alleviation efforts such as feeding programs, food subsidy schemes, employment generation efforts, food stamp programs, etc. In fact, many parts of the questionnaire have been designed by those who will use the data: economists, demographers, nutritionists and other researchers. This insures that the data collected are of the sort that can be used effectively to analyze the impact of government programs on household welfare. Second, since information is collected rather intensively at the household level, LSMS surveys tend to have a smaller sample sizes in order to focus on data quality, rather than quantity. The amount of information collected from households is quite large, so attempts are made to minimize the - 14 - interview time of individual household members (e.g. the household questionnaire is filled out in two interviews two weeks apart). The quality of the data collected is enhanced by intensive supervision of all aspects of the survey work, as explained below. Third, the need for policy relevant data implies that the data must be made available quickly. With this in mind LSMS surveys have pioneered the use of personal computers at all levels of survey operations, from design of questionnaire pages to data entry in the field to analysis of the data. The use of the latest computer technology allows for better quality control (data collected from a household's first interview is entered on personal computers and checked tor internal consistency before the household is .nterviewed again) and rapid data analysis. In the first survey in Cote d'lvoire, a preliminary statistical abstract on the first 6 months of data was available within 2 months of the last interviews in the field. In Jamaica, a report covering issues of health, education, and the impact of food subsidies, food stamp and school feeding programs on the poor was prepared in Kingston in October 1988 using data collected in August 1988, despite a devastating hurricane that occurred in early September of the same year. Fourth, LSMS surveys are flexible and adaptable to the particular characteristics and policy issues of any given country. The computer-based technology of the LSMS allows for flexibility and speed of implementation for particular countries. The basic questionnaire can easily be supplemented with special modules focusing on specific information needs. Examples of this are the collection of additional health care information (e.g. health facility questionnaire) in the third year of the C6te d'Ivoire survey and supplemental - 15 - education data (testing of household respondents and local school questionnaire) in the second year of the Ghana survey. D. Organization and Implementation The standard LSMS survey gathers data on three types of questionnaires. The largest and most time-consuming of these is the household questionnaire. It consists of 16 sections; sections 1-8 are filled in on the first visit to the household, sections 9-15 are filled in on the second visit, which occurs two weeks after the first, and section 16 (anthropometrics) is filled in in both visits. The information gathered in each of these sections is given in Table 2. For details see Grootaert (1986), Ainsworth and Munoz TABLE 2: Sections in LSMS Household Questionnaire First Visit 1. Household Roster 2. Housing Amenities and Expenditures 3. Education 4. Health S. Employment and Personal Activities 6. Migration 7. Selection of Respondents for Second Visit 8. Dwelling Characteristics Second Visit 9. Agricultural Activities 10. Non-Agricultural Household Enterprises 11. Expenditures on Non-Food Intras 12. Expenditures on Food and Consumption of Food Produced by the Household 13. Fertility 14. Other Income 15. Savings and Borrowing Both Visits 16. Anthropometric Data (Height and Weight) - 16 - (1986) and Ainsworth and van der Gaag (1987). An average visit to fill out half of the household questionnaire takes about 2-3 hours, though the questionnaire is designed so that no individual need be interviewed for more than an hour. The OteLc two romponents of a typical LSMS survey are the price and community questionnaires, the latter of which is administered in rural areas only. In order to measure the true purchasing power of household incomes, it is necessary to have data on prices faced by households. The LSHS price questionnaire gathers cata on consumer prices from local markets for both food and non-food items. In some cases prices are also gathered for medicines and agricultural inputs as well. The community questionnaire collects information on local conditions in rural areas, including the nearest schools and medical facilities, common agricultural practices (including agricultural wage rates) transportation and communications, and other "infrastructure" data. These data are crucial for studying the effects of a wide variety of government policies on various aspects of living standards. For example, they have been used to analyze the effects of user fees on the demand for medical care and for schooling, and the effect of commodity prices on agricultural productivity. The LSMS questionnaires are completed by several mobile survey teams, each of which contains one supervisor, two interviewers, one anthropometrist, and one data entry operator. Each team can cover two communities in four weeks, where 16 households are interviewed in each community. During the first week the first half of the household questionnaire is filled out in one community, and in the following week the first half of that questionnaire is filled out in the other community. In the third and fourth weeks the second - 17 - half of the questionnaires are filled out in the first and second communities, respectively. The supervisor is in charge of filling out the community questionnaire, while the anthropometrist weighs and measures the height of all household members and fills out the price questionnaire. After the first week's work is finished in the first community, the team takes the half- completed household questionnaires to the data entry operator, who is located in the regional capital. The data from those half-completed questionnaires are then entered on diskettes using personal computers (PC's) which are programmed to detect inconsistencies and ceding errors in the data. Before the team returns to the first community at the beginning of the third week, they pick up computer printouts from the data entry operator and use them to correct any inconsistencies or errors from the first week's work by going over the questions again with the households during the second visit. This greatly increases data accuracy. After a four week period has ended, the teams have a one week break, during which they often return to the capital city and bring the computer diskettes with the "fresh" data. The quality of LSMS survey data is further enhanced by heavy supervision at all levels, most of the work of the supervisor is to check the work of the other team members. Every household questionnaire is checked by the supervisor both before and after data entry, and the work by the data entry operator and the anthropometrist is also constantly being checked by him. The supervisor also visits some of the households after interviewers have left to see if they performed their work correctLy and were polite to the respondents. In addition, higher level officials make unannounced visits to the teams in the field to inspect their work, including that of the supervisor. Team members whose work is deficient are replaced by standbys who - 18 - have received the same training as the team members. Once six months or a full year of interviews have been completed, statistical abstracts are jointly put together by the World Bank and the statistical offices of tne countries. As merntioned above, these have been completed within just a few months thanks to the computerized nature of data collection. Further studies are undertaken both at the World Bank and in the country itself. The data always remain the property of the statistical offices of the respective countries. - 19 - IV. Selected Results from Studies on Poverty Using LSMS Data Even though the data from LSMS surveys have only been available in the past three years, there has already been a large number of studies using these data. Many of these studies have focused on poverty in developing countries. This section provides a sample of the results that have implications for poverty policies. A. Poverty Profiles Perhaps the first information one would like about the poor is a description of who they are. Once this has been learned one can do more sophisticated analyses using the LSMS data. Table 3 presents some characteristics of the poor in the five of the six countries for which LSMS data are available. More detailed information are available in Glewwe (1987a), Clewwe (1987b) and Glewwe and Twum-Baah, 1990). There are some general trendq vident in all five countries in Table 3. First, most of the poor are found in rural areas. This is consistent with virtually every study of poverty in developing countries. Although the numbers are not shown here, the fraction of the poor population found in urban areas is always substantially higher in all 5 countries than the fraction of the total population found in rural areas. Second, most of the poor are found in households in which the head has an agricultural occupation. Similarly, the heads of poor households are most likely to be self-employed. Note an interesting implication of this - very few of the heads of poor households work for the government. This suggests that freezes on government wages are not likely to hurt a substantial number of the poor. Third, the heads of poor households have relatively low - 20 - TABLE 3: Characteristics of the Poorest 302 in five Countries C6te d'Ivoire Peru Chana Jamaica* Mauritania Location Urban 14.3% 31.82 18.7Z 25.5% 9.8 Rural 86.7 68.2 81.3 73.5 90.2 Occupation of Head None 0.1% 4.7Z 4.82 31.3Z 45.9 Agricultural 87.5 61.2 77.3 41.0 43.7 Sales/Services 7.8 13.6 5.2 20.3 5.5 Industrial/Craft 2.3 18.3 9.9 1.2 3.6 Management/White Collar 1.5 2.2 3.0 0.0 1.1 Other 0.9 0.0 0.0 6.2 0.2 Education of Head None 84.6Z 25.4Z 69.82 99.0 184.0 Primary 13.9 62.9 9.4 0.7 Secondary 1.4 10.8 19.7 15.3 0.3 Post-Secondary 0.1 0.9 2.2 1.2 0.0 Other 0.0 0.0 0.0 3.2 0.0 Employer of Head None 0.12 4.72 5.2% 31.3% 45.9 Government 1.6 3.6 7.6 2.2 2.2 Private 3.2 20.1 3.8 13.6 4.3 Self-Employed 95.1 71.6 83.5 50.3 47.6 Other/Unknown 0.0 0.0 0.0 2.7 0.0 * Jamaica figures refee to poorest 20%. - 21 - levels of education - the vast majority have an elementary education or less, and in some countries most have no education at all. Yet there are also some differences across these five countries. Most strikingly, the heads of poor households in Jamaica and Mauritania are quite likely to be without jobs, either unemployed or out of the labor force. This is not the case in the other 3 countries. Second, in Jamaica and Peru about 10 to 20 percent of the poor live in households where the head works for a private employer, but in the three African countries this percentage is substantially smaller. Third, in Peru most poor households are headed by someone who has had at least an elementary ievel of education, but in the African countries heads of poor households usually have no education at all. B. Comparability of Different Poverty Definitions: Results from Cate d'Ivoire As poverty is a multi-dimensional concept, there are many ways of defining poverty. Some definitions focus on income, others on expenditures, and for both cases some look at household figures and others look at per capita figures. Others focus on food intake using data such as food expenditures, the fraction of total expenditures going towards food, and nutritional definitions. Some look to the living conditions of households (basic needs), using concepts like "adequate housing" and "adequate health care." One could even define poverty in terms of productive assets, such as agricultural land or education of household members. Do these definitions usually classify the same people as poor, or do they each point to a separate group as "the poor." A recent paper by Glewwe and van der Caag (1988) using the LSMS data from C6te d'Ivoire addresses these - 22 - TABLE 4: Correlation of Alternative Definitions of Poverty vith the Adjusted Per Capita Consumption Definition: CUte d'Ivoire Percentage of Population 2 Accurately Identified x Statistic Definition Poor Non-Poor Total (d.f. = 1) Urban Areas Per capita income 16.80 56.85 73.65 105.2754* Household consumption 17.67 57.81, 75.51 102.2547* Per capita consumption 26.08 66.12 92.20 413.9545* Per capita food consumption 22.64 62.67 85.31 262.0501* Food ratio 14.23 54.24 68.47 30.8790' Height for age 8.70 49.47 58.17 0.2791 Weight for height 8.95 49.01 57.96 0.4649 Per capita floor area 13.26 53.48 63.74 32.9857* Adult school attainment 13.41 53.69 67.10 31.9716* Rural Areas Per capita income 17.70 57.60 75.30 166.7868* Household consumption 19.93 59.90 79.83 209.0987* Per capita consumption 26.43 66.38 92.81 595.2563* Per capita food consumption 22.84 62.85 85.69 359.0950* Food ratio 10.19 50.17 60.36 3.4448 Height-for-age 10.82 51.21 62.03 1.3537 Weight-for-height 9.56 50.07 59.63 0.0833 Per capita floor area 7.57 47.51 55.28 1.5571 Adult schooling 11.89 51.71 63.60 9.9035* Agricultural land per capita 11.94 52.00 63.94 17.5915* Note: One asterisk denotes that the hypothesis of no correlation is rejected by the chi-square test at the 1 percent level. - 23 - issues. The paper takes several commonly used definitions of poverty in examines whether they select the same group as poor as does an equivalence scale adjusted per capita consumption measure. In both urban and rural C6te d'Ivoire, the answer is often "no." As seer in Table 4, definitions of poverty based on income data, agricultural land, and adult schooling are somewhat questionnable, and definitions based anthropometric measures (weight for height or height for age), the fraction of total expenditures devoted to food or the floor area of the dwelling can be completely urcorrelated or only weakly correlated with a (equivalence scale adjusted! consumption-based definition of poverty. Whether this is true in other countries is a topic for future research. The paper by Glewwe and van der Gaag also points out that different definitions of poverty have different biases. For example, per capita income and per capita expenditure tend to select large households, while total family expenditure selects small households. Using agricultural land as a definition in rural areas often selects relatively well to do households whose income derives from sources other than agriculture. Finally, some definitions are more likely to classify urban residents as poor than others (such as ones that tend to pick smaller households as poor). C. Str ctural Adjustment and the Poor Two recent papers by Clewwe and de Tray (1988, 1989) focus on issues of structural adjustment and the poor in two countries with LSMS data, Cote d'Ivoire and Peru. Both papers discuss many aspects of structural adjustment, but here the discussion will be limited to the effect of structural adjustment - 24 - policies on producer and consumer prices. The impact on prices is important because structural adjustment programs often focus on "getting the prices right," i.e. allowing domestic prices to approach international price levels. In many cases this means that consumer and producer price subsidies will be red'zced or halted, thus increasir3 prices. Some prices may decrease if taxes or import restrictizns are removezd. The data in Table 5 provide some idea cf the impact of price rhanges on the poor in their role as agricultural producers. Turning first to the case of C6te d'Ivoire, we see that both poor and non-poor households grow the two major export crops, coffee and cocoa, in the same proportion as the rest of the population; thus changes in the prices of either would have roughly the same impact on the poor as they would on the non-poor. In contrast, cotton is TABLE 5: Crops Grown by the Poor and Non-Poor in Cote d'Ivoire and Peru Percent Within Croup Who Cultivate the Crop Country Crop Poorest 10% Poorest 30% Wealthiest 70% All Cdte d'Ivoire Cocca 25.0% 34.2% 26.9% 34.4% C6te d'Ivoire Coffee 34.1 41.4 35.8 37.5 C6te d'Ivoire Cotton 27.6 19.8 3.9 8.7 Cdte d'Ivoire Sugar 0.4 1.2 2.6 2.2 Peru Maize (yellow) 14.3 13.9 8.3 10.0 Peru Maize (white) 28.9 22.7 7.8 12.3 Peru Wheat 29.1 23.4 8.6 13.1 Peru Cotton 0.3 0.5 1.0 0.8 - 23 - much more likely to be grown by the poor in Cote d'Ivoire than by the non- poor, which implies that higher prices for that crop woull mainly benefit the poor, while lower prices would be especially harmful to them. Finally, sugar is grown by very few poor houceholds, and indeed is more likely to be grown by non-poor households. Thus reductions in producer subsidies to sug3r producers in C6te d'Ivoire would have almost no negative effect on the poor. In Peru producers of yellow maize benefit from very high levels of effective protection. Yet from the viewpoint of the poor, it would be much better to subsidize production of white maize since that crop is much more grown by them than yellow maize. On the other hand, wheat production also receives a high level of effective protection, but this is more favorable for the poor; if this protection were reduced in order to "get prices right" a relatively large proportion of the poor would be hurt. Finally, unlike C6te d'Ivoire, cotton production is very rare among the poor in Peru - any changes in the price of cotton would have almost no effect on the poor in their role as producers. Of course, "getting the prices right" will affect the poor in their role as consumers as well. Table 6 provides some data on the fraction of total expenditures devoted to specific food items among both poor and non-poor households in C6te d'Ivoire and Peru. Turning first to the C6te d'Ivoire data, one sees that although rice is an important component of food consumption among the poor, they are partially protected against any price increases by the fact that they produce more than half of what LhQy consume. However, this is not true among those poor who are found in urban areas - they produce only a small fraction of what they consume. Unlike wheat, sugar does not constitute such a large fraction of the budgets of the poor, so that - 26 - TABLE 6: Budget Shares in Household Consumption in C6te d'lvoire and Peru Poorest 30% Country Food Item Poorest 10% All Urban Only Wealthiest 70S All CUte d'Ivaire Rice: Purchasedi 3.2% 3=91 7.62 3.9% 3.9% O!,ni Produce 3.8 4.2 0.7 1.5 1.8 T,tal. 7.0 8.1 8.3 5 .5 5.7 Cote d'Ivoire Sugar 0.9 0.9 1.1 0.7 0.7 C6te d'Ivoire Bread 1.6 1.4 1.7 1.9 1.9 Peru Wheat: Purchased 0.8 0.7 0.5 0.3 0.4 Own Produce 4.3 3.2 0.4 0.5 0.8 Total 5.1 3.9 0.9 0.8 1.2 Peru Bread 3.8 4.0 5.7 2.1 2.3 Peru Maize: Purchased 0.9 0.6 0.4 0.4 0.4 Own Produce 4.1 3.5 0.5 0.9 1.1 Total 5.0 4.1 0.9 1.3 1.5 reducing producer subsidies to sugar, which would raise the market price, will not have a strong negative effect on poor Ivoirian households. Finally, in C6te d'Ivoire the non-poor spend a somewhat large fraction of their budgets on bread than do the poor, which implies that any policies that lead tc increased - 27 - bread prices would not be biased against the poor. Turning to Peru, the domestic consumer price of wheat is nearly twice as high as the international price, but this does not effect the poor very much because they grow most of what they consume. However, to the extent these high prices lead to higher prices for bread, the poor are hurt because they spend a lag:ecr propcrtion of expenditures un bread than do the non- poor. Finally, the consumer price of maize Us kept artificially low by the Peruvian government. Despite the fact that the poor spend a larger share of their money on maize than do the rich, attempts to raise that price to international prices should not hurt poor households very much because they produce most of what they consume. D. The Effect on the Poor of Raising User Fees for Health Care and Education For many years developing countries have attempted to provide health care services and schooling at little or no cost to the public. The low fees for these services are sometimes Iustified by appealing to equity considerations. Yet, in many countries these services were provided in urban areas first, while many rural areas did not have adequate schools or facilities, or when they have been built the supplies and staff have been inadequate. Finally, many developing countries are now facing financial problems which may make it difficult, even impossible, to continue providir.g medical and education services at almost not cost to those who use them. This has led many economists to call for increases in fees paid for schooLs and medical services in order to expand these services to the rural areas and to ease the financial burden of providing these services. Yet several critics of these proposals have argued that they will hurt the poor and lead to reduced - 28 - use of health facilities and schools. The resolution of this dispute depends on empirical evidence, and the LSMS data can be used to answer these questions. In the area of health care, Certlcr and van der Caag (1989) hive used the Cote d'Ivoire and Peru data to answer the question: What will happen to the poor if higher fees are charged for publLcly provided health care services? Using a general model of household health care decisions, they esrimate price elasticities for different types of health care. These TABLE 71 Effects of Increased User Fees on Health Care Utilization in Cote d'Ivoire and Peru Fraction of Ill Population Seeking Health Care From Hospital Clinic Private Doctor Increased Increased Increased Base Fees Base Fees Base Fees C6te d'Ivoire West Forest 22% 13% 23Z 22% - - Savannah 15% 1% 18% 17% - - Peru Coastal 9% 8% 7% 7% 12% 12% Sierra 11% 4Z 10% 9% 3% 6% Note: 1. Base case represents present fee structure. Increased fees represent substantial cost recovery. See Gertler and van der Caag (1989) for details. 2. For Peru assume private doctor prices remain unchanged. - 29 - elasticities can then be used to predict the impact of user fees on use of health care facilities. The results are presented in Table 7. In Cate d'Ivoire, when fees are raised at both clinics and hospitals to cover a substantial fraction of the cost of providing these services, use of hospital services by the ill in the relatively well off West Forest rural areas declines from 22% to 13%, whereas use of clinics declines only by a small amount. In contrast, in the poor rural Savannah areas use of hospital services virtually disappears. In Peru a similar disparity between rich and poor areas emerges: in the relatively well off Coastal areas increased user fees have almost no effect, but in the poor Sierra rural areas the use of government hospitals declines dramatically. Turning to the area of education, Gertler and Glewwe (1989) examine whether the poor in rural Peru are willing to pay the costs cf building new secondary schools in villages where there is no school at present. Of course, this depends on how far away the nearest secondary schools are. Their results on the willingness to pay for new schools are shown in Table 8. Given an estimated cost if 400 intis per student per year to pay teacher's salaries, Gertler and Glewwe find that in villages where the nearest school is only one hour away (ii terms of walking distance) both poor and non-poor households are not willing to pay for a new school in the village. Yet when the nearest school is 2 hours away the willingness to pay rises dramatically - both the poor and the non-poor are willing to bear the cost of paying teachers in return for a new secondary school. This comes about even though the poor are more price-sensitive to increases in school fees than the non-poor. - 30 - TABLE 8: Willingness to Pay to Reduce Travel Time to Secondary School In Rural Peru: June 1985 Intis per year and Percent of Income INCOME QUARTILE Travel Time to Poorest 25% Next 25% Next 25% Ri, ast 25% All Income Nearest Groups School % of x of X of % of Z of Intis Income Intis Income Intis Income Intis Income Intis Income 1 Hour 169 2.3% 182 1.4% 203 1.0% 374 0.8% 232 1.1% 2 Hours 412 5.7% 455 3.5% 508 2.A% 934 1.9% 577 2.8% E. Food Stamps, Food Subsidies and the Poor in Jamaica One issue that often arises in poverty alleviation efforts is: To what extent do a program's benefits reach the poor? This is often referred to as the targeting issue. Food subsidies have often come under attack because they seem to benefit the wealthier income groups more than they do the poor. Some kind of food stamp scheme is often put forth as a more effective way of channeling assistance to the poor. The Jamaica Survey of Living Conditions allows one to compare the two different programs since it collects information on boch food stamps and the consumption of subsidized food items. In a report recently put out by the Statistical Institute of Jamaica, the beneficiaries of these two programs were identified according to per capita expenditure levels. The results are reproduced in Table 9. It turns out that although the value of food subsidies in Jamaica is a larger portion of the budgets of poor households than of rich households, in money terms wealthier households receive more benefits than poorer ones; the - 31 - monthly value of food subsidies to househoLds in the weaLthiest 20% of the population was about 53 Jamaican doLlars per capita while the same figure for the poorest 20% of the population was 29 Jamaican doLlars. In contrast, food stamps are much more effective in reaching the poor - 51% of the poorest 20% were receiving food stamps compared to only 6% of the wealthiest 20%. TABLE 9: Incidence of Food Stamp and Food Subsidy Benefits: Jamaica, 1988 Population Quintiles Poorest Next Middle Next Richest All 20% 20% 20% 20% 20% Jamaica Value of Food Subsidies Received (Jamaican Dollars per 28.6 39.6 39.5 42.5 52.5 40.5 person per month) Percent of Households Receiving Food Stamnps 50.6 37.1 26.3 17.0 6.0 23.4 F. Persistence of Poverty in Developing Countries In developed countries, some households remain in poverty for many years, if not decades, while others move in and out of poverty over relatively short periods of time. What is the pattern found in developing countries? Most LSMS surveys are permanent surveys, i.e. they repeat every 12 months. In such cases it is useful to interview the same households in succeeding years to see how their economic status changes from year to year. In those LSMS surveys which are permanent, half the households interviewed in one year are reinterviewed in the following year. At the most simple level of analysis, - 32 - this allows one to see whether there is much movement in and out of poverty over a one year period. The data in Table 10 present data on the persistence of poverty for the first two LSMS surveys for which multi-year (panel) data are available. It is evident that in both C6te d'Ivoire and Chana there is substantial movement in and out of poverty in a relatively 3hort period of time. Of the poorest 30X (approximately 22.6% + 7.1%) of the population in Cote d'Ivoire in 1985, nearly one fourth (7.1%) were not poor in 1986, and of the poorest 29% (16.41 + 12.3%) in Ghana in 1987-88, about 40% (12.3%) were not poor the following year. While this movement suggests that many of the poor are only poor for a year or two, a careful analysis is necessary before any further conclusions can be drawn. TABLE 10: Persistence of Poverty in Developing Countries Cote d'Ivoire 1986 Poorest 30% Other 70% 1985 Poorest 30% 22.6% 7.1% Other 70% 16.1% 54.2% Ghana 1988-89 Poorest 30% Other 70% 1987-88 Poorest 30% 16.4% 12.3% Other 70% 15.7% 55.7% - 33 - V. Conclusion In order to formulate effective policies to reduce poverty in developing countries, the nature and causes of poverty must be well understood. This requires data from these countries which are relevent, detailed, accurate and up to date. The LSMS program was developed at the World Bank to meet this need. The surveys undertaken so far have provided a wealth of information on poverty in developing countries and have been used in a number of studies which examine the impact of various government policies on the poor. Much more can be done with the existing data, and as more countries undertake LSMS surveys a better understanding of poverty in developing countries will emerge. Of course, there is room for improvement in the design and implementation of LSMS surveys, and new techniques will continue to be explored both in terms of data collection and research. Many lessons can he learned from the experience of other household survey programs, such as the United Nations Household Survey Capability Program '(IUNHSCP) and household survey activities in the developed countries. At the same time this experience must be conveyed to the statistical and research organizations in developing countries so that their participation in this process can be increased. When Robert MacNamara declared in 1973 that the World Bank would direct its attention toward improving the plight of the poor in developing countries, it was not always clear how this could be done, and indeed some skeptics may have honestly thought that very little could be done. The development of the LSMS program at the World Bank constitutes on important - 34 - step in transforming this commitment from an abstract idea to a concrete agenda. In concert with other poverty related work, both at the World Bank and at other institutions around the world committed to poverLy alleviation, it would not be overly optimiscic to hope that significant progress can be made in improving the quality of life in ieveloping countries in the 1990's and on into the 21st century. - 35 - REPERENCES Acharya, Meena. (1982). "Time Use Data and the Living Standards Measurement Study". LSMS Working Paper No. 18, The World Bank, Washington, D.C. Ainsworth, Martha, and Juan Munioz. (1986). "The C6te d'Ivoire Living Standards Survey: Design and Implementation". LSMS Working Paper No. 26, The World Bank, Washington, D.C. Alcimir, Oscar, and J. Sourrouille. (1980). "Measuring Levels of Living in Latin America: An Overview of Main Problems". LSMS Working Paper No. 3, The World Bank, Washington, D.C. Birdsall, Nancy. (1982). "Child Schooling and the Measurement of Living Standards". LSMS Working Paper No. 14, The World Bank, Washington, D.C. Booker, William, Parmeet Singh and L. Savane. (1980). "Household Survey Experience in Africa". LSMS Working Paper No. 6, The World Bank, Washington, D.C. Chander, Ramesh, Christiaan Grootaert and Graham Pyatt. (1980). "Living Standards Surveys in Developing Countries." LSMS Working Paper No. 1, The World Bank, Washington, D.C. Deaton, Angus. (1980). "The Measurement of Welfare: Theory and Practical Guidelines". LSMS Working Paper No. 7, The World Bank, Washington, D.C. Deiton, Angus and John Muellbauer. (1980). Economics and Consumer Behavior. Cambridge Universitv Press. Cambridge, U.K. Gertler, Paul, and Jacques van der Gaag. (1989). The Willingness to Pay for Medical Care: Evidence from Two Developing Countries. Forthcoming. The World Bank, Washington, D.C. Gertler, Paul, and Paul Glewwe. (1989). "The Willingness to Pay for Education in Developing Countries: Evidence from Rural Peru". LSMS Working Paper No. 29, The World Bank, Washington, D.C. Glewwe, Paul. (1987a). "The Distribution of Welfare in the Republic ot Cate d'Ivoire". LSMS Working Paper No. 29, The World Bank, Washington, D.C. Glewwe, Paul. (1987b). "The Distribution of Welfare in Peru in 1985-86". LSMS Working Paper No. 42, The World Bank, Washington, D.C. Glewwe, Paul, and Jacques van der Gaag. (1988). "Confronting Poverty in Developing Countries: Definitions, Information and Policies". LSMS Working Paper No. 48, The World Bank, Washington, D.C. Clewwe, Paul, and Dennis de Tray. (1988). "The Poor During Adjustment: A Case Study of the Cota d'Ivoire". LSMS Working Paper No. 47, The World Bank, Washington, D.C. - 36 - Clewwe, Paul, and Dennis de Tray. (1989). "The Poor in Latin America During Adjustment: A Case Study of Peru". LSMS Working Paper No. 56, The World Bank, Washington, D.C. Clewwe, Paul and Kwaku Twum-Baah. (19°9). "The Distribution of Welfare in Ghana". Mimeo. The World Bank, Washington, D.C. Crootaert, Christiaan. (1982). "The Labor Market and Social Accounting: A Framewoork of Data Presentation. LSMS Working Paper No. 17, The World Bank, Washington, D.C. Crootaert, Christiaan. (1986). "Measuring and Analyzing Levels of Living in Developing Countries: An Annotated Questionnaire". LSMS Working Paper No. 26, The World Bank, Washington, D.C. Ho, Teresa J. (1982). "Measuring Health as a Component of Living Standards." LSMS Working Paper No. 15, The World Bank, Washington, D.C. Martorell, Reynaldo. (1982). "Nutrition and Health Status Indicators: Suggestions for Surveys of the Standard of Living in Developing Countries". LSMS Working Paper No. 13, The World Bank, Washington, D.C. Mehran, Farhad. (1980). "Employment Data for the Measurement of Living Standards". LSMS Working Paper No. 6, The World Bank, Washington, D.C. Musgrove, Philip. (1982). "The ECIEL Studv of Household Income and Consumption in Urban Latin America: An Analytical History". LSMS Working Paper No. 12, The World Bank, Washington, D.C. Scott, Christopher, P.T.A. de Andre and Ramesh Chander. (1980). "Conducting Surveys in Developing Countries: Practical Problems and Experience in Brazil, Malaysia, and the Philippines". LSMS Working Paper No. 5, The World Bank, Washington, D.C. Sullivan, Jeremiah, Susan H. Cochrane and William D. Kalsbeek. (1982). "Procedure for Collecting and Analyzing Mortality Data in LSMS". LSMS Working Paper No. 16, The World Bank, Washington, D.C. United Nations Statistical Office. (1980). "Towards More Effective Measurement of Levels of Living", and "Review of Work of the United Nations Statistical Office (UNSO) Related to Statistics of Levels of Liv:.ng". LSMS Working Paper No. 4, The World Bank, Washington, D.C. Visaria, Pravin. (1980). "Poverty and Living Standards in Asia: An Overview of the Main Results and Lessons of Selected Household Surveys". LSMS Working Paper No. 2. The World Bank, Washington, D.C. Wahab, Mohammed Abdul. (1980). "Income and Expenditure Surveys in Developing Countries: Sample Design and Execution". LSMS Working Paper No. 9, The World Bank, Washington, D.C. PRE Workina Paner Series Contact Iha Alhnhr A for pagar WPS402 The GATT as International Discipline J. Michael Finger March 1990 N. Artis Over Trade Restrictions: A Public 38010 Choice Approach WPS403 Innovative Agricultural Extension S. Tjip Walker for Women: A Case Study of Cameroon WPS404 Chile's Labor Markets in an Era of Luis A. Riveros April 1990 R. Luz Adjustment 34303 WPS405 Investments in Solid Waste Carl Bartone April 1990 S. Cumine Management: Opportunities for Janis Bernstein 33735 Environmental Improvement Frederick Wright WPS406 Township, Village, and Private William Byrd April 1990 K. Chen Industry in China's Economic Alan Gelb 38966 Reform WPS407 Public Enterprise Reform: Ahmed Galal April 1990 G. Orraca-Tetteh A Challenge for the World Bank 37646 WPS408 Methodological Issues in Evaluating Stiln Claessens May 1990 S. King-Watson Debt-Reducing Deals Ishac Diwan 33730 WPS409 Financial Policy and Corporate Mansoor Dailami April 1990 M. Raggambi Investment in Imperfect Capital 37657 Markets: The Case of Korea WPS410 The Cost of Capital and Investment Alan Auerbach April 1990 A. Bhalla in Developing Countries 37699 WPS411 Institutional Dimensions of Poverty Lawrence F. Salmen May 1990 E. Madrona Reduction 37496 WPS412 Exchange Rate Policy in Developing W. Max Corden May 1990 M. Corden Countries 39175 WPS413 Supporting Safe Motherhood: L. M. Howard May 1990 S. Ainsworth A Review of Financial Trends 31091 Full Report WPS414 Supporting Safe Motherhood: L. M. Howard May 1990 S Ainsworth A Review of Financial Trends 31091 Summary WPS415 How Good (or Bad) are Country Norman Hicks April 1990 M. Berg Projections? Michel Vaugeois 31058 PRE Workino Paoer Series Contact LIQAuthor Dat e for pagper WPS416 Improving Data on Poverty ;n the Paul Glewwe May 1990 A. Murphy Third World: 1 he World Bank's 33750 Living Standards Measurement Study WPS417 Modeling the Macroeconomic William Easterly May 1990 R. Luz Requirements of Policy Reform E. C. Hwa 34303 Piyabha Kongsamut Jan Zizek WPS418 Does Devaluation Hurt Private Ajay Chhibber May 1990 M. Colinet Investment? The Indonesian Case Nemat Shafik 33490 WPS419 The Design and Sequencing of Brian Levy May 1990 B. Levy Trade and Investment Policy Reform: 37488 An Institutional Analysis WPS420 Making Bank Irrigation Investments Gerald T. O'Mara May 1990 C. Spooner More Sustainable 30464 WPS421 Taxation of Financial Intermediation: Christophe Chamley May 1990 W. Pitayato- Measurement Principles and Patrick Honohan nakarn Application to Five African Countries 37666 WPS422 Civil Service Reform and the World Barbara Nunberg May 1990 R. Malcolm Bank John Nellis 37495 WPS423 Reiat;ve Price Changes and the M. Shahbaz Khan May 1990 WDR Office Growth of the Public Sector 31393 WPS424 Mexico's External Debt Sweder van Wijnbergen Restructuring in 1989-1990: An Economic Analysis WPS425 The Earmarking of Government William A. McCleary Revenues in Colombia Evamaria Uribe Tobon WPS426 Growth-Oriented Adjustment Riccardo Faini Programs: A Statistical Analysis Jaime de Melo Abdel Senhadji-Semiali Julie Stanton WPS427 Exchange Reform, Parallel Markets Ajay Chhibber May 1990 M. Colinet and Inflation in Africa: The Case Nemat Shafik 33490 of Ghana WPS428 Perestroyka and Its Irmplications Bela Balassa May 1990 N. Campbell for European Socialist Countries 33769