POLICY RESEARCH WORKING PAPER 1489 Educational Attainm ent Expectations are that educational attainment v,,1' in Developing Countries eowtmostnteMeE. and North Afica and Ya;a New Estimates and Projections Sub-Saharan Afrca improve greatly in Soufh A Disaggregated by Gender where the level of atnIr rv is lowest. The gender iJ1 Vinod Ahuja education may have rise Deon Filmer the past decade This it- will continue unless taunn a intensify their efforts o educate girls. Background paper for World Development Report 1995 The World Bank Office of the Vice President Development Economics July 1995 POLICY RESEARCH WORKING PAPER 1489 Summary findings Ahuja and Filmer present new estimates of educational in Sub-Saharan Africa. attainment in 71 developing countries for the years 1985, * South Asia - currently the least educated part of 1990, and 1995. They also project levels of educational the world - is expected to substantially augment its attainment through the year 2020 by using the United stock of human capital by the year 2020. Nations Educational, Scientific, and Cultural * In most regions, enrollment levels are expected to Organization's projections of enrollment and the remain lower for girls than for boys. International Labour Organization's projections of * The gender gap in education may have risen in the population by age and sex. past decade. This trend toward a widening of the gender The projections suggest interesting trends: gap may continue unless countries intensify their efforts * Growth of stock in human capital is expected to be to educate girls. highest in the Middle East and North Africa and lowest This paper -a product of the Office of the Vice President, Development Economics -was prepared as abackground paper for World Development Report 1995 on labor. Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Michael Geller, room T7-079, telephone 202-473-1393, fax 202-676-0652, internet address MGELLER@WORLDBANK.ORG (23 pages). July 1995. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers cany the names of the authors and should be used and cited accordingly. The findings, interpretations, and conclusions are the authors' own and should not be attributed to the World Bank, its Executive Board of Directors, or any of its member countries. Produced by the Policy Research Dissemination Center Educational Attainment in Developing Countries: New Estimates and Projections Disaggregated by Gender A Background Paper for the World Development Report 1995 Vinod Ahuja and Deon Filmer Address for correspondence: Vinod Ahuja Deon Filmer Department of Agriculture The World Bank and Resource Economics 1818 H St. NW University of Maryland Washington DC 20433 College Park, MD 20742 Tel: (301) 422 1873 Tel: (202) 473 1303 Email: ahujav@arec.umd.edu Email: dfilmer@worldbank.org We wish to thank Michael Walton, Ishac Diwan and David Lindauer for useful discussions at various stages of the analysis. EDUCATIONAL ATTAINMENT IN DEVELOPING COUNTRIES: NEW ESTIMATES AND PROJECTIONS DISAGGREGATED BY GENDER Vinod Ahuja and Deon Filmer Human capital has come to be regarded as the primary source of long term economic growth along with physical capital (Azariadis and Drazen, 1990; Barro, 1991; Mankiw, Romer, and Weil, 1992). However, most of the empirical studies of growth have relied on measures of the accumulation of human capital such as enrollment ratios, or proxies of the stock such as illiteracy rates (Romer, 1989; Mankiw, Romer and Weil, 1992). Inadequacy of these indicators to represent the stock of human capital embodied in the population needs no explanation. Lack of data on appropriate measures of human capital stock has been the main motivation for recent studies attempting to estimate the level and distribution of the global stock of human capital across countries (Barro and Lee, 1993; Nehru, Swanson and Dubey, 1993; Dubey and King, 1994, Psacharopoulos and Arrigada, 1986 and 1992; Kyriacou, 1991). While these studies have, to some extent, narrowed the knowledge gap, the estimates presented in these studies suffer from several shortcomings. For example, Barro and Lee (BL) estimate the proportions of the population with primary, secondary and higher education. But, their estimates pertain to individuals aged 25 years and above. Thus, their estimates do not reflect the enormous expansion of education that has occurred in the developing countries since 1960. Nehru, Swanson and Dubey (NSD) construct the estimates of mean schooling (years) at primary, secondary, and tertiary levels for the working age population but their estimates are not disaggregated by sex. Dubey and King (DK), estimate 2 mean school years disaggregated by sex and age groups, but because of data limitations they drop the higher education category. In addition, the latest available year in these studies is 1987. Although stocks of human capital change slowly, recent rapid expansion of education implies relatively large changes which need to be considered. Because of these reasons, this paper, (i) presents new estimates of educational attainment for the years 1990 and 1995, (ii) quinquenially projects these estimates until the year 2020, and (iii) updates BL's estimates for 1985 by including the stock of human capital embodied in individuals below 25 years. Methodology The main procedure for constructing human capital stock in the previous studies has been the 'perpetual inventory' method. This method requires very long data series on enrollments, repeaters and dropouts. For example, for constructing human capital stock in 1990 the perpetual inventory method would require data starting at least 1925. But, available data series on enrollments, repeaters and drop outs do not go that far back in time. The analysts are thus forced to backcast these series which introduces errors in their estimates. In order to avoid that problem, we use somewhat different methodology described below. Because long data series are needed to estimate the educational achievements of adult population only, we obviate the need to use them by dividing the population into two broad groups: adults (those aged 25 years and more) and youngsters (those between 6 and 24 years), and using BL's estimates of educational achievements for adult population for the base year. Since their estimates combine census/survey data' and estimates based on these as benchmark, we believe that using their estimates as the base for educational achievements Approximately 40 percent observations in BL dataset are obtained from available censuses and survey estimates of educational achievements from various UNESCO publications. 3 of adult population minimizes the potential error that may be introduced by using perpetual inventory method. By combining BL's estimates with the estimate of educational achievements for youngsters, we obtain the educational attainment of the population between 6 and 60 years (excluding those currently in school) across primary, secondary, and tertiary educational levels. The no education category is the residual. These new estimates for 1985 form the 'base' for our estimates in 1990 and 1995 and projections thereafter. From 1985 onwards we perform the 'stock-flow' analysis. That is, we follow each age cohort year after year and, after adjusting for age and gender specific mortality rates, recompute the proportion of population across no schooling, primary, secondary, and tertiary schooling categories. In order to estimate the educational achievement of youngsters in the base year, we use age and gender specific enrollment rates at each educational level. Let Y. be the young it population (less than or equal to 24 years) in j age-cohort where j=1,2,3 corresponds to 6- 11, 12-17, and 18-24 years age categories in t time period. Further, let E be the enrollment rates at kt education level in t time period where k=1,2,3 corresponds to primary, secondary, and tertiary levels. The number of students enrolled at each level at time t is then given by SAt= Yjt*Ek for j = k. where Sjk = the number of students at kth educational level at time t. To estimate the educational achievement of youngsters not in school at time t (number of individuals in age cohort 2 who dropped out of school after primary education, 2 Since this population has not yet completed its educational achievement we do not count them as part of the stock. 4 and the number of individuals in age cohort 3 who dropped out of school after primary or secondary education) we follow the age cohorts who entered the educational system 6 and 12 years prior to time t. Since we have estimates of those in school at time t, the number of individuals who dropped out of school after primary and secondary education can be easily estimated. Finally, for 1990 and 1995 we decompose the estimates for four categories into six categories: no schooling, primary incomplete, primary complete, secondary incomplete, secondary complete, and some tertiary using the available drop-out rates for 1985 and 1990. We do not decompose the tertiary education category into complete-incomplete because of non-availability of drop-out rates at this level. Also, the projections beyond 1995 are not decomposed into complete-incomplete categories because of our unwillingness to hazard a guess about the behavior of drop-out rates in the future. The data This section describes the source of each data series and presents a regional overview of some of the educational indicators. 1. Educational composition of adult population As noted before, this data is obtained from BL. They provide estimates of the distribution of adult population for seven educational categories: no schooling, primary incomplete, primary complete, secondary incomplete, secondary complete, higher incomplete, and higher complete for the years 1960-85. For 1985, this data is available for 106 countries. However, out of these, 25 are OECD countries. Since the availability of enrollment projections is limited to developing countries, these 25 countries are excluded. 5 This leaves us with 81 developing countries. A regional summary of the educational attainment of adult population is presented in Table 1. Table 1: Educational attainment of adult population: 1985 Percent of adult population with Region No Some Some Some Mean schooling primary secondary higher Schooling schooling schooling schooling (Years) East Asia and the Pacific 23.6 51.3 18.8 6.3 5.19 Latin America and the Caribbean 22.4 56.6 13.9 77.1 4.47 Middle East and North Africa 52.8 26.5 16.0 4.8 3.51 OECD 3.3 37.7 40.8 18.2 8.88 South Asia 69.0 13.7 14.1 3.2 2.49 Sub Saharan Africa 48.1 41.7 9.3 1.0 2.67 Source : Barro and Lee, 1993. 2. Population Age and gender specific population estimates and projections are obtained. from ILO (1986) and subsequent ILO data updates. In order to combine UNESCO provided enrollment ratios and BL attainment statistics, and apply the stock-flow model of educational projections, the population data are modified as described below. First, the age breakdown is redefined so that net enrollment ratios can be applied to the appropriate age groups. ILO population data are broken down into nine, five year age groups between 10 and 64 (i.e. 10-14, 15-19, ...), ages 0 to 9, and a group of those 65 and above. These are reshaped by assuming: (1) the size of the age cohort aged 5 to 9 in a given year is equal to the size of the same cohort when aged 10 to 14 (i.e. five years later) and (2) the age distribution within each five year cohort in a given year is uniform. Using these simplifying assumptions, the data are transformed into population data for ten, six year age groups (i.e. 0-5, 6-11, ..) between 0 and 59, and those 60 and above. 6 Second, the years need to be redefined in order to be able to carry out the stock-flow model of educational attainment. ILO data are estimated and projected for every five years until 2025. Data for intervening years are imputed by applying the average annual growth rate for each age cohort over the five years to the stock in the base year. 3. Enrollment rates. repetition rates. and drop out rates The gender specific gross enrollment ratios for years before 1990 are obtained from the Economic and Social Database of the World Bank (BESD). The projections of enrollments are from UNESCO (1993). This publication presents decennial projections of gross enrollment ratios at Ist, 2nd, and 3rd level of education for 107 developing countries. The projections for the intervening years are derived by log-linear interpolations. Finally, after combining this data with BL dataset, and retaining only those countries for which data is available in both data sets, we are left with 71 countries34. Appendix 1 lists the countries included in the analysis. The repeat rates for primary and secondary school are derived from UNESCO data on the number of students enrolled at each level and among those, the number who have repeated a grade. Using the available data, a simple imputation and projection model is used to fill in data for between 1960 and 20201. If there is no data for a country, no adjustment is made. If there is only one observation, this is assumed to be the repeater rate. If there is more than one observation for a country, the annual average growth rate for the repeat rate is used to infer missing data and project these rates to 2020. Since the rates can vary greatly 3 BL dataset does not contain estimates for China. For China the distribution in the base year is obtained from Chinese Social Statistics Publisher (1993). However, the disaggregation across gender could not be available. Thus, China's estimates and projections are not disaggregated by gender. 4 Combined together these countries account for over 80 percent population of non-OECD countries. 5 See Appendix 2 for the countries and time periods for which data on repeat rates is available. 7 (even from one year to the next), some of these growth rates imply what may be implausible values. Therefore, data on repeat rates are constrained from below by the value at the 10th percentile, and from above by the value at the 90th percentile. Using these repeat rates, we adjust the gross enrollment ratios including the projections. The adjusted gross enrollment ratios are then used in further estimations and projections. Except for Sub-Saharan Africa (SSA), all regions seem to have made significant progress in expanding the coverage of primary and secondary education (Figures lA-IC and Figure 2)6. In SSA, primary enrollments are still much below 100 per cent and have already started declining. Further, UNESCO has projected continuation of these declining trends at all educational levels. Drop out rates are constructed using data on enrollment and repeat levels for each grade within the educational attainment categories. Denoting NEi as the net enrollments at grade j in year t, the drop out rate at grade j is given by D1 = (N E1.1.1 - NE ti) / N Et These grade specific drop-out rates are then used to calculate the overall drop out rate at each educational level7. The drop-out rates at the regional level for 1985 and 1990 are presented in Figures 3 and 4. These drop-out rates may appear high but are consistent with the regional drop-out rates reported elsewhere (see, for example, Chowdhary, 1995; UNESCO, 1984). 6 The trends presented in Figures IA-IC and Figure 2 are computed using the data for 71 countries in our sample. 7 The data used to construct the drop out rates were obtained from BESD. It should be noted however that these data are scarce specially at the secondary level. For secondary level, therefore the decomposition across complete-incomplete categories is based on limited data and should be used with caution. 8 Figure IA: Gross Enrolment Ratios (adjusted for repeaters) at the Primary Level : Trends and Projections China East Asia & the Pacific 1201 130 110 T å 120- 100 110 90/100 l" --:-W -: .:g: 80 9 I80 70 -0- Males -- Females -O- Both Sexes 70 60 60 -o- Males -a- Females -O- Both Sexes 50 450 40 c. 40 -o.- mal Latin America & the Caribbean Middie East & North Africa 130__ _ _ _ _ _ _ _ _ _ _ 120 - ·- Males -å- Females --- Both Sexes 120-- -0- Males -- Females -0- Both Sexes 110- 100 100 g è No- --fao - -a 80- 70 60 60 504 1 W) 20 South Asia Sub Saharan Africa 130 120.- -o--Males - Females -O-Both Sexes 120 110 -0--Males -- Females --- Both Sexes 100- 40O ~100 -0 o- - C> - -0- - 70~ ---O- 60 60 50 40 40- 30 20 9 Figure IB: Gross Enrolment Ratios (adjusted for repeaters) at the Secondary Level : Trends and Projections China East Asia & the Pacific 800 -- Males -a- Females 70 70 -o- Both Sexes 50 ,60 30 30 2020 10 -0-*Males -- Females -<- Both Sexes 0 10 0 ; i i i i i i W) in 10 83 Latin America and the Caribbean Middle East & North Africa 80 70 - --- Males -- Females -O-- Both Sexes 8 -o- Males -&- Females -0- Both Sexes 70.- 50 -. e ·· 30 30 20 20 10 10 0 1 0 South Asia Sub Saharan Africa 80. 80· -7-Mals -å- Females -O- Both Sexes 70· 80 60 - --Malis -6- Females - Both Suxes 30- 30- 20 20- 1 0 . 1- --- - 0 • O • O "" 10 Figure IC: Gross Enrolment Ratios at the Tertiary Level : Trends and Projections 25 China East Asia & the Pacific -o- Males -a- Females -0-- Both Sexes 5 c 5 . o-- -2-- Males -å-Females --0- Both Sexes 0 0 0 &0 90 Years YearS Latin America and the Caribbean Middie East & North Africa 25 25 20 :.a r··- 9 '0 ---Males -å- Feiales -0- Both Sexes j15 .. 5«0 - - - -- --o - -- -0-Males -a-Females -O-Both Sexes 0 5- 0 0~ 0 C: | C|| O M40 - - .- . - - fl fl Years Years South Asia Sub Saharan Africa 25 25 -O-Males -6-Females -0-Both Sexus 120 ~20 · - -o - Males --a-- Females 15 I·· -0-- Both Sexes 110. -o0 . 0-- o 1 o- -- O-- 5 .- in 16 19 5 e 0 O 1965 1975 1985 1995 2005 2015 Years Years 11 Figure 2: Repeat Rates at Primary and Secondary Levels China East Asia and the Pacific 35 35 - 30 30 - -0- Males (Primary) 25 -4- Males (Primary) 25 --Females (Primary) -4-Females (Primary) --0- Males (Secondary) 20 Males (Secondary) 20 - --Females (Secondary) Females (Secondary) 15 15 10 10 0 1111: . 0. II Illi 35 - Latin America and the Caribbean I Middle East and North Africa 35I S30 - 3 0 - Males (Primar) 0 25! - i- Males (Primary) j25 -+-- Females (Primary) --FmlSo(rmay ~ I -0-Males (Secondary) L20 - -U- Females (Secondary) 20 0--%Females (Secondary) 15 15 1V0 st & .% Ch a, g . . z. a. .I 0 , 1 1 1 : : : .. 0 . l I i l l I IlI l C- - -4 1-4 ---4 1- - - C 4 4 - 14 Sub Saharan Africa South Asia 35 30 3o -*- Males (Primary) --Males (Primary) w 25 -- Females (Primary) 25 --Females (Primary) M-0-Males (Secondary) -0-Males (Secondary) . 20 - --Females (Secondary) 20 -4-Females (Secondary) 115 15 10 1 OK m. .m . . m. . m -*-m a 0 : I : I 12 Figure 3 50 Drop Out Rates at the Primary Level: Both Sexes 45 10E31985 01990 30 25 20la 15 10 5 East Asia Latin Middle South Asia Sub & the America & East & Saharan Pacific the North Africa Canbbean Africa Drop Out Rates at the Primary Level: Males 50 45 1 1985 40 01990 35 - 30 - 25 20 -- 1 l 15 10 5 0 *** --- I East Asia Latin Middle South Asia Sub & the Amienca & East & Saharan Pacific the North Africa Caribbean Africa Drop Out Rates at the Primary Level: Females 45-- 40 - 11985 35 -01990 30 25 20 - 15 10 0 - East Asia Latin Middle South Asia Sub & the America & East & Saharan Pacific the North Africa Caribbean Africa 13 Figure 4 Drop Out Rates at the Secondary Level: Both Sexes 80 70 N1985 60 O01990 50 40 30 20 10 0 Drop Out Rates at the Secondary Level: Males 70 60- 01985 50 01990 40 30 20 10 0-I East Asia Latin Middle South Asia Sub & the America & East & Saharan Pacific the North Africa Caribbean Africa Drop Out Rates at the Secondary Level: Females 80 70 60 * 1985 50 01990 40 30 20 10 0 East Asia Latin Middle South Asia Sub & the America & East & Saharan Pacific the North Africa Caribbean Africa 14 Like previous studies attempting to estimate the education stock, this study also had to deal with several methodological and data problems. Wherever feasible, we have attempted to minimize the use of assumptions and extrapolations. Nevertheless, the results are subject to a number of caveats. The first issue is that of measurement error in the raw data. In the literature there has been some concern that for some countries UNESCO's data may contain significant measurement error which, in turn, would influence the stock estimates (for a discussion of this issue, see Jimenez, 1994). To the extent that UNESCO's data on enrollments and repeaters and BL's estimates of stock for the base year may be influenced by the measurement error, the bias would also be reflected in our estimates. Second, like previous studies, we had no alternative but to ignore the fact that age ranges for various educational levels are not necessarily uniform across countries although in our sample the differences are not wide. Third, the estimates do not account for the differences in quality of education across countries. While we recognize the importance of quality of education, the fact that quality is a multi-dimensional concept and no single indicator is available to represent this variable it is nearly impossible to adjust the estimates for the quality of education'. Similar is the case of non-schooling education such as 'on the job training' etc. Finally, the estimates beyond 1995 are merely 'projections' of educational attainment and should not be interpreted as the 'forecasts'. A forecast model has to take into account the Some previous studies have used teacher-pupil ratio as a proxy for quality of education. Once again, this variable does not reflect the phenomena such as absenteeism (not only of pupils but also that of teachers) which is not uncommon in the developing countries. 15 effects of several economic variables such as returns to education, direct and indirect costs of education, and the policy variables such as public provision of resources for education, etc. Our estimates, on the other hand, simply reflect (i) the effects of demographics, and (ii) UNSECO's projections of enrollment levels. It may be noted, however, that the effect of demographics is expected to be very strong especially in the regions with low educational levels such as South Asia (SA) and SSA. That is simply because the educational gap between younger and older age cohorts in these regions is wide and thus, when the older age cohorts retire and younger age cohorts move into adult population the educational level of the population is bound to rise significantly even if these regions fall short of the enrollment levels projected by the UNESCO. Results The country level estimates are presented in an Appendix which is not included as part of this paper due to space limitations. An electronic copy of the appendix can be obtained from the authors. In this section we confine our attention to the regional level estimates. Before presenting the results, however, it is important to provide some indication of the reliability and the quality of estimates. We do this by comparing our estimates with the estimates in previous studies for the year 1985. The estimates obtained in this study show very high correlation with BL's estimates9. That is, however, no surprise because we used BL's estimates as the base case for educational attainment of adult population. However, our estimates of educational levels are 9 All correlation coefficients greater than 0.9. 16 higher than BL's estimates because younger age cohorts are generally more educated. In order to compare our estimates with those of NSD, DK, and Kyriacou, we estimated the mean schooling levels (years) by assigning half the number of years in the educational cycle to those who did not complete the cycle. The resulting estimates of mean schooling at the regional level are presented in Table 2. These estimates show high correlation with the estimates of NSD, DK and Kyriacou (Table 3). Since we have used a different methodology than the previous studies, such high correlation of our estimates with those in previous studies reinforces our confidence in the new estimates and projections. Table 2: Estimated Mean Schooling Levels (years) Regions BL's estimates Our estimates 1985 1985 1990 1995 Both Sexes China 5.3 5.3 5.6 East Asia & the Pacific 5.2 5.8 6.3 6.9 Latin America & the Caribbean 4.9 5.3 5.9 6.4 Middle East & North Africa 3.6 4.6 5.3 6.2 South Asia 2.4 3.4 3.9 4.3 Sub Saharan Africa 2.1 3.0 3.5 4.0 Females East Asia & the Pacific 4.5 5.6 6.0 6.4 Latin America & the Caribbean 4.7 5.3 5.8 6.3 Middle East & North Africa 2.7 4.6 5.1 5.8 South Asia 1.6 3.1 3.4 3.7 Sub Saharan Africa 1.4 2.8 3.1 3.5 Males East Asia & the Pacific 5.8 5.8 6.1 6.8 Latin America & the Caribbean 5.1 5.2 5.8 6.3 Middle East & North Africa 4.4 4.7 5.5 6.4 South Asia 3.1 3.5 4.1 4.7 Sub Saharan Africa 2.8 3.2 3.7 4.3 17 Table 3: Correlation coefficient of mean schooling with estimates in other studies This study BL NSD DK Kyriacou Both sexes This study 1.00 .. BL 0.95 1.00 NSD 0.92 0.91 1.00 DK 0.88 0.81 0.92 1.00 Kyriacou 0.88 0.89 0.88 0.79 1.00 Females This study 1.00 .. BL 0.94 1.00 DK 0.91 0.88 .. 1.00 Males This study 1.00 .. BL 0.94 1.00 DK 0.81 0.72 .. 1.00 Note: All correlation coefficients significant at 1 percent level. According to our estimates, all regions have made significant progress in augmenting the educational attainment of population (Table 4). The highest gain appears to be in the Middle East and North Africa (MENA) region where primary enrollment ratio is close to 100 percent, and post primary enrollments are also quite high". If the countries in MENA achieve the enrollment levels projected by the UNESCO, MENA region would emerge as one of the highly educated regions amongst the developing countries with the proportion of population with no schooling falling below 1 percent and the proportion of population with at least some secondary education exceeding 50 percent by the year 2020 (Table 5). On the other hand, the least educated region is SA where more than 50 percent population had no schooling in 1985. SA is closely followed by SSA where the comparable figure was nearly 10 The two major MENA countries missing from the sample are Egypt and Morocco. In Egypt, which accounts for nearly 25 percent of the regional population, the enrollment ratios at all level are relatively high and rising. In Morocco, which accounts for another 10 percent of population, the enrollment ratios are low and stagnating. Inclusion of Morocco in the sample would, therefore, somewhat diminish the estimates and projections for MENA. 18 40 percent. During the last decade, however, the enrollments ratios in SA have increased steadily and the primary enrollment ratio was close to 100 percent in 1995. As a result, the proportion of population with no schooling in SA fell by more than 18 percent points during 1985-95. In contrast to that the primary enrollment ratios in SSA are much below 100 percent and are declining. Thus, the fall in proportion of population with no schooling was only about 12 percent points. If these trends continue, the education map of the developing regions could significantly alter during the next 25 years (Table 5). Table 4: Estimates of Educational Attainment at Regional Level: Both Sexes Region Proportion of population under 60 years (not in school) with No Primary Secondary Higher Schooling Schooling Schooling Schooling Complete Incomplete Complete Incomplete Year= 1985 China 22.8 22.5 22.6 31.1 0.9 East Asia & the Pacific 16.7 35.3 25.5 9.0 9.3 4.1 Latin America & the Caribbean 16.7 20.8 42.3 6.8 8.5 4.9 Middle East & North Africa 37.6 26.4 16.6 8.0 9.0 2.4 South Asia 50.6 15.0 16.9 6.0 8.9 2.4 Sub Saharan Africa 39.3 24.2 29.7 1.3 5.0 0.5 Year = 1990 China 21.8 22.7 23.0 31.5 0.9 East Asia & the Pacific 12.4 37.4 23.1 10.4 10.9 5.6 Latin America & the Caribbean 13.4 23.6 38.7 7.9 9.7 6.6 Middle East & North Africa 28.9 32.3 15.0 9.0 11.5 3.2 South Asia 41.1 19.5 19.3 7.5 9.6 2.9 Sub Saharan Africa 33.5 29.9 28.0 1.9 6.2 0.6 Year= 1995 China 18.4 23.8 24.1 32.6 1.0 East Asia & the Pacific 7.3 39.1 20.5 12.3 13.1 7.6 Latin America & the Caribbean 9.3 25.4 36.6 9.1 11.1 8.5 Middle East & North Africa 19.7 37.7 14.1 10.3 14.0 4.3 South Asia 32.4 23.0 21.6 8.9 10.4 3.6 Sub Saharan Africa 26.9 35.2 27.0 2.5 7.6 0.8 19 Table 5: Projections of Educational Attainment at the Regional Level: Both Sexes Proportion of population under 60 years (not in school) with Region No schooling Some Some Some primary secondary higher schooling schooling schooling Year-2000 China 15.1 49.6 34.1 1.2 East Asia and the Pacific 6.2 57.4 27.4 9.0 Latin America & the Caribbean 7.4 59.6 22.6 10.4 Middle East & North Africa 12.0 54.3 28.4 5.3 South Asia 28.1 47.1 20.8 4.0 Sub Saharan Africa 22.2 65.1 11.7 0.9 Year-2005 China 10.5 51.8 36.4 1.3 East Asia and the Pacific 2.5 55.4 31.1 10.9 Latin America & the Caribbean 5.9 56.4 25.3 12.3 Middle East & North Africa 4.6 55.9 33.1 6.3 South Asia 21.0 51.1 23.1 4.7 Sub Saharan Africa 19.2 66.7 13.0 1.1 Year-2010 China 6.8 52.2 39.4 1.5 East Asia and the Pacific 2.0 50.7 34.5 12.8 Latin America & the Caribbean 5.9 52.6 27.4 14.0 Middle East & North Africa 0.7 55.1 37.0 7.2 South Asia 15.3 54.3 25.1 5.3 Sub Saharan Africa 17.1 67.7 14.0 1.2 Year-2015 China 2.4 51.8 43.9 1.8 East Asia and the Pacific 1.4 45.6 38.1 14.8 Latin America & the Caribbean 5.7 48.9 29.7 15.8 Middle East & North Africa 0.2 51.1 40.7 8.0 South Asia 9.2 57.3 27.5 5.9 Sub Saharan Africa 15.9 67.9 14.9 1.3 Year=2020 China 2.3 49.9 45.7 2.1 East Asia and the Pacific 1.7 42.2 40.0 16.0 Latin America & the Caribbean 6.1 46.6 30.6 16.7 Middle East & North Africa 0.2 48.3 43.0 8.4 South Asia 9.2 55.3 29.2 6.3 Sub Saharan Africa 16.2 67.3 15.2 1.3 20 While overall level of human capital stock in the developing countries have increased significantly, its distribution across gender has been far from uniform. In all regions except Latin America and the Caribbean (LAC), female primary enrollments continue to be lower than those of males although East Asia & the Pacific (EAP) and MENA have made significant progress in that respect. As a result of female enrollments falling short of males, the gender gap in education seems to have widened during the last decade. For example, in SA, the proportion of males with no schooling declined by more than 20 percent points during 1985-95 compared to less than 14 percent for females. Comparable figures for EAP, MENA, and SSA were 11 and 8, 21 and 15, and 15 and 9 percent points, respectively (Tables 6 and 7). Further, the expansion of post primary education has also been disproportionate. During 1985-95, the proportion of males with post primary education increased significantly faster than that of females in all regions. Also, UNESCO's projections suggest that (i) female enrollments at the primary level will continue to be lower than male enrollments in SA and SSA, and (ii) at the post primary level the female enrollments will be lower than male enrollments in all regions. In view of the recent research finding that (i) the social benefits of reducing the gender gap in education are large (World Bank, 1991; King and Hill, 1993), and (ii) the effect of female secondary education on social indicators such as fertility rate and infant mortality rate are very strong (Subbarao and Raney, 1993; see also Figures 5 & 6), the widening of the gender gap is a source of concern. Conclusions This paper has presented new estimates of educational attainment in 71 developing countries for the years 1985, 1990, and 1995. For 1985, our estimates compare well with 21 other available estimates. Since no estimates of human capital stock are available for 1990 and beyond, we could not perform any comparative analysis for those years. However, these estimates relate very well with other indicators of social and economic development such as GNP per capita, fertility rate and infant mortality rate. This reinforces our confidence in the new estimates. The paper also projects the levels of educational attainment until the year 2020 by using UNESCO's projections of enrollments and ILO's projections of population by age and sex. The projections uncover some interesting trends. For example, growth in stock of human capital is expected to be highest in MENA and lowest in SSA. SA, which is currently the least educated region, is expected to substantially augment its stock of human capital by the year 2020. The findings also suggest that the gender gap in education may have risen during the last decade. Since in most regions the female enrollment levels are expected to continue to be lower than those for males, the trend in widening of gender gap may continue unless countries intensify their efforts in educating girls. 22 Table 6: Estimates of Educational Attainment at Regional Level: Females Region Proportion of population under 60 years (not in school) with No Primary Secondary Higher Schooling Schooling Schooling Schooling Complete Incomplete Complete Incomplete Year= 1985 East Asia & the Pacific 17.0 22.3 38.3 8.6 9.6 4.1 Latin America & the Caribbean 16.8 18.4 45.4 7.0 7.5 5.0 Middle East & North Africa 39.5 23.9 18.5 7.7 7.8 2.5 South Asia 53.4 13.5 16.4 5.4 8.7 2.6 Sub Saharan Africa 41.3 20.2 32.3 1.4 4.3 0.5 Year= 1990 East Asia & the Pacific 13.5 20.5 39.7 8.6 12.6 6.7 Latin America & the Caribbean 14.3 20.4 40.5 6.6 12.0 6.3 Middle East & North Africa 32.9 31.0 15.2 7.7 10.3 2.8 South Asia 46.6 16.8 19.0 6.1 8.8 2.8 Sub Saharan Africa 37.5 24.7 30.0 1.2 5.6 0.5 Year = 1995 East Asia & the Pacific 8.9 19.2 40.0 9.1 16.0 6.7 Latin America & the Caribbean 10.2 21.5 38.9 5.8 15.7 7.9 Middle East & North Africa 24.4 36.7 13.9 8.1 13.5 3.4 South Asia 39.7 19.8 21.4 6.9 9.3 2.8 Sub Saharan Africa 32.2 28.9 30.0 1.0 7.3 0.5 Table 7: Estimates of Educational Attainment at Regional Level: Males Region Proportion of population under 60 years (not in school) with No Primary Secondary Higher Schooling Schooling Schooling Schooling Complete Incomplete Complete Incomplete Year = 1985 East Asia & the Pacific 16.5 22.6 38.0 8.8 9.8 4.1 Latin America & the Caribbean 16.8 18.4 45.4 7.0 7.5 5.0 Middle East & North Africa 36.1 27.8 16.3 7.9 9.6 2.3 South Asia 48.5 16.2 17.3 6.5 9.1 2.3 Sub Saharan Africa 37.5 25.1 30.0 2.1 4.9 0.4 Year = 1990 East Asia & the Pacific 11.6 21.2 39.3 9.9 11.7 6.1 Latin America & the Caribbean 13.5 19.9 43.6 8.2 7.9 6.8 Middle East & North Africa 26.1 36.3 12.8 8.8 12.4 2.8 South Asia 37.8 20.2 20.1 8.4 10.2 3.2 Sub Saharan Africa 29.9 30.5 29.7 2.9 6.3 0.7 Year= 1995 East Asia & the Pacific 5.9 20.1 39.5 11.5 14.3 8.7 Latin America & the Caribbean 9.2 21.0 43.2 9.3 8.5 8.8 Middle East & North Africa 15.5 42.3 11.3 10.2 15.6 3.4 South Asia 28.3 23.5 20.0 10.2 11.3 4.2 Sub Saharan Africa 22.2 35.5 29.4 3.9 7.9 1.0 23 Table 8: Projections of Educational Attainment at the Regional Level: Males Proportion of population under 60 years (not in school) with Region No schooling Some Some Some primary secondary higher schooling schooling schooling Year=-2000 East Asia and the Pacific 4.6 57.4 27.8 10.2 Latin America & the Caribbean 7.2 62.6 19.3 10.9 Middle East & North Africa 6.9 56.3 30.4 6.4 South Asia 22.6 48.5 23.9 5.0 Sub Saharan Africa 17.0 68.0 13.7 1.3 Year-2005 East Asia and the Pacific 2.1 54.0 31.5 12.4 Latin America & the Caribbean 5.6 60.2 21.2 13.0 Middle East & North Africa 1.0 55.9 35.3 7.8 South Asia 14.8 52.5 26.7 5.9 Sub Saharan Africa 13.7 69.5 15.2 1.5 Year-2010 East Asia and the Pacific 1.5 49.2 34.9 14.4 Latin America & the Caribbean 5.5 57.0 22.7 14.8 Middle East & North Africa 0.4 51.3 39.5 8.9 South Asia 8.4 55.5 29.3 6.8 Sub Saharan Africa 11.7 70.2 16.4 1.7 Year=2015 East Asia and the Pacific 1.3 43.8 38.4 16.5 Latin America & the Caribbean 5.4 53.4 24.4 16.7 Middle East & North Africa 0.2 46.5 43.4 9.9 South Asia 7.8 52.5 32.1 7.6 Sub Saharan Africa 10.8 69.9 17.5 1.8 Year-2020 East Asia and the Pacific 1.8 40.6 40.0 17.5 Latin America & the Caribbean 5.6 51.6 25.1 17.6 Middle East & North Africa 0.2 43.9 45.6 10.4 South Asia 8.1 50.2 33.7 8.0 Sub Saharan Africa 11.3 69.2 17.7 1.9 24 Table 9: Projections of Educational Attainment at the Regional Level: Females Proportion of population under 60 years (not in school) with Region No schooling Some Some Some primary secondary higher schooling schooling schooling Year=-2000 East Asia and the Pacific 8.2 56.8 27.0 7.9 Latin America & the Caribbean 8.1 58.2 24.0 9.7 Middle East & North Africa 17.4 53.1 25.5 4.1 South Asia 36.5 43.4 17.1 2.9 Sub Saharan Africa 28.3 61.4 9.7 0.6 Year=2005 East Asia and the Pacific 4.5 55.6 30.5 9.4 Latin America & the Caribbean 6.5 55.3 26.6 11.5 Middle East & North Africa 9.8 55.2 30.2 4.8 South Asia 30.3 47.5 18.9 3.3 Sub Saharan Africa 25.4 63.1 10.8 0.6 Year=2010 East Asia and the Pacific 2.9 53.4 32.7 11.0 Latin America & the Caribbean 6.5 51.8 28.6 13.2 Middle East & North Africa 4.4 56.0 34.2 5.4 South Asia 25.1 50.7 20.5 3.6 Sub Saharan Africa 23.5 64.0 11.8 0.7 Year-2015 East Asia and the Pacific 2.3 49.1 35.5 13.2 Latin America & the Caribbean 6.3 48.1 30.8 14.8 Middle East & North Africa 0.5 55.4 38.0 6.1 South Asia 19.7 53.8 22.5 4.0 Sub Saharan Africa 22.7 64.1 12.5 0.7 Year-2020 East Asia and the Pacific 2.3 42.4 39.2 16.1 Latin America & the Caribbean 6.5 46.5 31.3 15.7 Middle East & North Africa 0.3 52.4 40.8 6.5 South Asia 17.3 54.0 24.3 4.4 Sub Saharan Africa 23.0 63.4 12.8 0.8 Figure 5 9 9 a 8 0 40 7 0 20 408 60 &9 10 0 2 04 5 07 8 9 2I 2a 7 0 0 20 40 60 80 100 0 10 20 30 40 50 60 70 Proportion of womee with somi primary schooling Proportion of women with some secondary schooling 8 9- 7 8 Ž4 5 1 2 2 0 -*--------4 o 20 40 60 80 100 0 l0 20 30 40 50 60 70 Proportion of males with smem primary schooling Proportion of malet with smem secondary schooling Figure 6 180 200 160 S ISO 140 g 160 g0 120 t 140-0 0o b 120- ~~~~100 2 * * , * *, 40 o* * * 20 so 20 * i 0 tI 0 20 40 60 80 100 0 10 20 30 40 50 60 70 Properflon of women with some primary schooling Proportion of women with some secondary schooling 1830 130 160 160 gS 140 140 120 120 son0 S ooo a * 10000 0 0 10 0 -0 0 30 2S a 610 0 40 4)4 )5( 2g gg20 0 4) S)* C 0 20 40 60 so 100 0 10 20 30 40 50 60 70 Proportion of males with some primary schooling Proportion of mles with some secondary schooling ISO ISO_ P一”。·“。·。‘&m·，留‘’。‘&&c&”。“&&&rr。，·，，。二‘m·‘一，&，。m·‘·‘OO&,&&&&,, 浴寫合當g弓雪召藝寫台冶g召皂8藝 0卡一一―卜一一一一十－一“0才一一一一一－一―一一 】一0 00、7瞬訕。―？斗分h哎0〞 -00、楓．A L vv、．色。 10萬b二▼9-vA不州b》▼ _}v、。O―、－ 負―、乏―、．。 Z-．、。哺l、Ckv ，幼1 .v、6，呂―一O、v 馴q》IA鳥＇〝一壟01騙j〝 作鷹二IV〝，01＇喊鳥〝△ 他0-do、內n｝一、又 為！▼一、Xi{t夕O 細一t、驕，-,d, 名―、0可―、－ 一。―〝《》目｝, 喝一．,M;―、＇tj黠 刈。｝唱Jt一訕•―A,kf 騙J一一，嗎J-l&j,j鳥 嗎―〝j『唱―，▼ 頂】tv，文Ik 編一I,,01己 響沫―&J么Oj, 中藝干v他乞父―t O《二l，巨戶→】． 1.-，勿1 一 16」，oq k力1 CI一C 《二0 一｀邊 -Pro凶r電100 ofm:Ies with,om,po,t primary 1 ProportionoffeoaIeswiths0mepostprim•ry,chooliogiol995 schooling in 1995一，j森嬌。g 一計必必J參彎■O、、瀾OC,C CI CC）《j C. －汗《十‘禿，鬥口》《g▼t 石r》〈【，仕哺二念―. 28 References Azariadis, Costas and Allen Drazen. 1990. "Threshold Externalities in Economic Development", Quarterly Journal ofEconomics, :501-26 Barro, Robert. 199 1. "Economic Growth in a Cross Section of Countries", Quarterly Journal ofEconomics, 106(2):407-443. Barro, Robert and Jhong -Wha Lee. 1993. "International Comparison of Educational Attainment", Journal ofMonetary Economics, 32:363-394. Chinese Social Statistics Publisher. 1993. Chinese Social Statistics: 1993, Beijing: Chinese Social Statistics Publisher. Chowdhary, Kowsar P. "Literacy and Primary Education", HRO Working Paper 50, The World Bank, Washington DC. Dubey, Ashutosh and Elizabeth King. 1994. "A New Cross-Country Education Stock Series Differentiated by Age and Sex", The World Bank, (Typescript). ILO (International Labor Office). 1986. Economically Active Population Estimates and Projections: 1950-2025, Geneva: International Labor Organization. Jimenez, Emmanuel. 1994. "Human and Physical Infrastructure: Public Investment and Pricing Policies in Developing Countries", Policy Research Working Paper 128 1, Ile World Bank. King, Elizabeth and M. Anne Hill (eds). 1993. Women's Education in Developing Countries, Washington DC: The World Bank. Kyriacou, G. A. 199 1. Level and Growth Effects ofHuman Capital: A Cross-Count?y Study of the Convergence Hypothesis, Department of Economics, New York University (mimeo). Mankiw, N.G., David Romer and David N. Weil. 1992. "A Contribution to the Empirics of Economic Grpwth", Quarterly Journal ofEconomics, Nehru, Vikram, Eric Swanson, and Ashutosh Dubey. 1993. "A New Database on Human Capital Stock: Sources, Methodology and Results", Policy Research Working Paper 1124, The World Bank. Psacharopoulos, George and Ana Maria Arrigada. 1986. "The Educational Composition of the Labor Force: An International Comparison", International Labor Review, 125 (5):561-574. 29 Psacharopoulos, George and Ana Maria Arrigada. 1992. "The Educational Composition of the Labor Force: An International Update", Journal ofEducational Planning and Administration, 6 (2):141-159. Romer, P. M. 1989. "Human Capital and Growth: Theory and Evidence", NBER Working Paper 3173. Subbarao, K. and Laura Raney. 1993. "Social Gains from Female Education: A Cross National Study", World Bank Discussion Paper 194. UNESCO. 1984. The Drop-out Problem in Primary Education: Some Case Studies, Bangkok: UNESCO Regional Office for Education in Asia and the Pacific. UNESCO, 1993. Trends and Projections ofEnrollment Level ofEducation by Age and . Sex: 1960-2025, Paris: UNESCO Division of Statistics. World Bank. 1991. World Development Report 1991: The Challenge ofDevelopment, Washington DC: The World Bank. 30 Appendix 1: Country coverage Region Countries East Asia and the Pacific Fiji, Hong Kong, Indonesia, Korea, Myanmar, Malaysia, Philippines, Papua New Guinea, Thailand China China Latin America and the Caribbean Argentina, Barbados, Bolivia, Brazil, Colombia, Chile, Costa Rica, Dominican Republic, Ecuador, Guatemala, Guyana, Honduras, Haiti, Jamaica, Mexico, Nicaragua, Panama, Peru, Paraugay, El Salvador, Trinidad & Tobago, Uruguay, Venezuela Middle East & North Africa Bahrain, Algeria, Iran, Iraq, Kuwait, Syria, Tunisia South Asia Afghanistan, Bangladesh, India, Sri Lanka, Nepal, Pakistan Sub Saharan Africa Benin, Botswana, Central African Republic, Cameroon, Ghana, Gambia, Guinea Bissau, Kenya, Lesotho, Mali, Mozambique, Mauritius, Malawi, Niger, Rwanda, Sudan, Senegal, Sierra Leone, Togo, Tanzania, Uganda, Zaire, Zambia, Zimbabwe ECA Turkey 31 Appendix 2: Availability of data on repeat rates Country Years for which data on repeaters is available Primary Secondary Afghanistan 1977-85 1977-85 Argentina 1970-76 1970-75 Bangladesh 1976-89 - Bahrain 1981-92 1981-92 Bolivia 1990 1990 Brazil 1970-87 1970-87 Barbados - - Botswana 1975-92 1985-92 Central African Republic 1970-89 1970-89 Chile 1970-83 1975-83 China 1988-93 1989-93 Cameroon 1975-90 1975-90 Colombia 1970-92 1970-92 Costa Rica 1970-93 1980-92 Dominican Republic 1970-80 - Ecuador 1970-87 1970-84 Fiji 1976-86 1977-86 Ghana 1970-91 1970-91 Gambia 1975-91 1975-91 Guinea Bissau 1975-87 1975-86 Guatemala 1970-86 1975-80 Guyana 1970-86 1970-83 Hong Kong 1979-84 1979-84 Honduras 1980-91 1991 Haiti 1978-90 1978-90 Indonesia 1975-92 1975-92 India 1970-87 1983 Iran 1983-92 1983-92 Iraq 1970-92 1970-92 Jamaica 1975-90 1975-90 Kenya 1970-81 - Korea 1970 1970 Kuwait 1970-93 1975-93 Sri Lanka 1970-92 1970-92 Lesotho 1975-92 1986-87 Mexico 1975-92 1976-92 Mali 1970-93 1970-93 Myanmar 1970 1970 Mozambique 1981-92 1982-92 contd... 32 Malawi 1975-92 1987 Malaysia - - Niger 1970-90 1970-90 Nicaragua 1970-92 1979-92 Nepal 1988-92 1988-92 Pakistan - - Panama 1970-89 1975-88 Peru 1875-85 1975-85 Philippines 1980-89 1989 Papua New Guinea - - Paraguay 1970-92 - Rwanda 1970-91 1981-91 Sudan 1970 1970 Senegal 1970-91 1978-91 Sierra Leone 1977-82 1977 El Salvador 1975-92 1975-89 Syria 1970-93 1970-93 Togo 1970-90 1970-90 Thailand 1975-88 1975-77 Trinidad and Tobago 1981-91 1984-97 Tunisia 1970-93 1975-93 Turkey 1983 1983 Tanzania 1975-93 - Uganda 1975-86 1976 Uruguay 1970-92 1975-92 Venezuela 1970-92 1975-92 Zaire 1970-92 1975-92 Zambia 1975-86 1976-84 Zimbabwe 1984 1984 Policy Research Working Paper Series Contact Title Author Date for paper WPS1464 How Does the North Ameican Free Edward E. Learner May 1995 S. Vallimont Trade Agreement Affect Central Alfonso Guerra 37791 America? Martin Kaufman Boris Segura WPS1465 Post Trade Liberalizatson Policy Sarath Rajapatirana May 1995 J. Troncoso and Institutional Challenges in 37826 Latin America and the Caribbean WPS1466 Ownership and Financing of Charles D Jacobson June 1995 WDR Infrastructure: Historical Joel A. Tarr 31393 Perspectives WPS1467 Beyond the Urug.ay Round: The Jeffrey D. Lewis June 1995 B. Kim mulications of an Asian Free Trade Sherman Robinson 82477 Area Zhi Wang WPS1468 Government's Role Pakistan Rashid Faruqee June 1995 C. Anbiah Agriculture: Majoi Re,orms are Neeajed 81275 WPS1469 The Role of Labor Unions in Fostering John Pencavel June 1995 WDR Economic Development 31393 WPS1470 Pension Systems and Reforms: Patricio Arrau Jule 1995 E. Khine Country Experiences and Research Klaus Schmidt-Hebbei 37471 Issues WPS1471 Pension Reform and Growth Giancarlo Corsetti June 1995 E. Khine Kiaus Schmidt-Hebbel 37471 WPS1472 Fiscal and Monetary Contraction in Klaus Schmidt-Hebbel June 1995 E. Khine Chile: A Rational-Expectations Luis Serven 37471 Approach WPS1473 The Surge in Capital intlows to Eduardo Fernandez-Arias June 1995 R. Vo Developing Countries- Prospects and Peter J. Montiel 33722 Policy Response WPS1474 Are Stable Agreements for Sharing D. Marc Kilgour June 1995 C. Spooner international River Waters Now Ariel Dinar 32116 Possible? WPS1475 Decentralization: 1 he Way Forward Andrew N. Parker June 1995 D. Housden for Rural Development? 36637 WPS1476 Public Spending and the Poor: What Dominique van de Walle June 1995 C. Bernardo We Know, What We Need to Know 37699 WPS1477 Cities Without Land Markets: Alain Bertaud June 1995 L. Lewis Location and Land Use in the Bertrand Renaud 30539 Socialist City Policy Research Working Paper Series Contact Title Author Date for paper W;VPS1478 Promoting Growth in Sri Lanka: Sadiq Ahmed June 1995 A. Bhalla Lessons from East Asia Priya Ranjan 82168 WPS1479 Is There a Commercial Case for Panayotis N. Varangis June 1995 J. Jacobson Tropical Timber Certification? Rachel Crossley 33710 Carlos A. Primo Braga WPS1480 Debt as a Control Device in Herbert L. Baer June 1995 G. Evans Transitional Economies: The Cheryl W. Gray 85783 Experiences of Hungary and Poland WPS1481 Corporate Control in Central Europe Peter Dittus June 1995 G. Evans and Russia: Should Banks Own Stephen Prowse 85783 Shares? WPS1482 A Measure of Stock Market Robert A. Korajczyk June 1995 P. Sintim-Aboagye Integration for Developed and 38526 Emerging Markets WPS1483 Costa Rican Pension System: Asli Demirg0q-Kunt June 1995 P. Sintim-Aboagye Options for Reform Anita Schwarz 38526 WPS1484 The Uruguay Round and South Asia: Nader Majd July 1995 J. Ngaine An Overview of the Impact and 37947 Opportunities WPS1485 Aggregate Agricultural Supply Maurice Schiff July 1995 J. Ngaine Response in Developing Countries: Claudio E. Montenegro 37947 A Survey of Selected Issues WPS1486 The Emerging Legal Framework for Pham van Thuyet July 1995 G. Evans Private Sector Development in 85783 Viet Nam's Transitional Economy WPS1487 Decomposing Social Indicators Using Benu Bidani July 1995 P. Sader Distributional Data Martin Ravallion 33902 WPS1488 Estimating the World at Work Deon Filmer July 1995 M. Geller 31393 WPS1489 Educational Attainment in Developing Vinod Ahuja July 1995 M. Geller Countries: New Estimates and Deon Filmer 31393 Projections Disaggregated by Gender