THE WORLD BANK Internal Discussion Paper ASIA REGIONAL SERIES Report No. IDP 51 Educational Development in Asia: A Comparative Study Focussing on Cost and Financing Issues Jee-Peng Tan Alain Mingat October 1989 The views presented here are those of the author, and they should not be interpreted as reflecting those of the World Bank. October, 1989 EDUCATIONAL DEVELOPMENT IN ASIA: A COMPARATIVE STUDY POCURSING ON COST AND FINANCING ISSUES Jee-Peng Tan Alain Mingat Technical Department Asia Region The World Bank Acknowledgements Many people have contributed to this study at various stages of its gestation. When work began, we were greatly assisted by colleagues who generously made available to us data and resource materials which they have collected for their own work. Among them we wish especially to thank Shigeko Asher, Richard Cambridge, Mae-Chu Chan&, Nat Colleta, Rosslyn Hees, Elizabeth King, Jack Maas, Kenichi Ohashi, Andre Salmon, Thomas Schmidt, William J. Smith, Cecilia Valdevieso, Chia-Ling Yang and Roberto Zagha. In collecting, processing, and checking the data we were assisted by three competent summer interns at the Bank, San Ling Lam, Andrea Madarassy, and Aleta Domdom. The earliest draft of the study was read with great care and insight by Estelle James and Charles Griffin, both of whom suggested improvements in the study's structure, and refinements to the data. We also received valuable comments from, among others, Birgir Fredriksen, Emmanuel Jimenez, Martin Karcher, Elizabeth King, and Jandhyala Tilak. As usual, none of the above friends and colleagues is in any way responsible for the flaws that still remain. Abstract This study attempts to provide a quantitative documentation of education in Asia, and to derive general lessons about educational policies. It emphasizes comparative analysis using data for 11 to 15 countries. The study found wide diversity across countries in educational outcomes, external constraints on the sector's development, and choice of sectoral policies. It suggests that while demographic and macroeconomic conditions have important effects on educational development, policies in the sector are perhaps even more crucial to consider. The study argues that substantial scope exists for improving efficiency and equity, and identifies three broad policy thrusts toward this end: (a) raising cohort survival rates in primary education. This is a major regional challenge, since the region-wide rate of cohort survival is currently only 62 percent; (b) freeing up resources to improve primary education, by making more efficient use of resources at all levels in the system, but perhaps most importantly, by reducing the public subsidization of higher education; and (c) in selected countries, increasing overall spending on education (channeled mostly to primary education), particularly where current levels of investment in the sector is relatively low. EDUCATIONAL DEVELOPMENT IN ASIA: A COMPARATIVE STUDY FOCUSSING ON COST AND FINANCING ISSUES Table of Contents Page No. 1. Introduction and Overview of F1;nd_=s . . . . . . . . . . . . . 1 Part A: Overall Approach and Framework . . . . . . . . . . . 1 Part B: Summary of the Principle Findings . . . . . . . . . 5 B.1 Education in Asia and Major Policy Challenges: An Overview . . . . . . . 5 B.2 Synthesis of the More Specific Findings . . . . . 8 B.3 Implications for Policy Dialogue and FurtherAnalysis .............. 37 2. Educational Develoment in Asia: Some Basic Charateristics . 39 2.1 International Comparisons . . . . . . . . . . . . . . . 39 2.2 Variation Across Asian Countries . . . . . . . . . . . 43 2.3 Sources of Differences in Educational Development Across Countries . . . . . . . . . . . . . . . . . . . .. 49 3. The Costs of Education-and Its FMnanLn:. Further Comparative Statistics . . . . . . . . . . . . . . . . . . . . . . . . . 57 3.1 The Intrasectoral Allocation of Public Spending on Education . . . . . . . . . . . . . . . . . . . . . . 57 3.2 Variation in the Unit Costs of Public Education . . . . 58 3.3 Sources of Variation in Unit Costs . . . . . . . . . . 64 3.4 The Private Financing of Education . . . . . . . . . . 73 4. EfficiencX in the Provision of Eduation . . . . . . . . . . . . 77 4.1 External Efficiency . . . . . . . . . . . . . . . . . . 78 4.2 The Internal Efficiency of the Education System . . . . 84 * ii - 5. EAquit Considerations in Education . . . . . . . . . . . . . . . 105 5.1 Global Aspects of Equity . . . . . . . . . . . . . . . 106 5.2 Social Selectivity in Education . . . . . . . . . . . . 124 6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . 141 6.1 Overview of Findings and Assessment of Policy Issues . 141 6.2 Priorities for Future Research on Education in Asia . . 146 Appendix A: Individual Country Profiles in a Regional Perspective . 148 Bangladesh . . . . . . . . . . . . . . . . . . . . . 156 Bhutan . . . . . . . . . . . . . . . . . . . . . 159 Burma . . . . . . . . . . . . . . . . . . . . . . . 162 China . . . . . . . . . . . . . . . . . . . . . . . 165 India . . . . . . . . . . . . . . . . . . . . . . . 168 Indonesia ..................... 171 Korea . . . . . . . . . . . . . . . . . . . . . . . 174 Malaysia . . . . . . . . . . . . . . . . . . . . . . 177 Nepal . . . . . . . . . . . . . . . . . . . . . . . 180 PapuaNe.Guinea.................. 183 Philippines . . . . . . . . . . . . . . . . . . . . 186 Sri Lanka . . . . . . . . . . . . . . . . . . . . . 189 Thailand . . . . . . . . . . . . . . . . . . . . . . 192 Appendix B: Sources of Data and Basic Statistics . . . . . . . . . . 195 Appendix C: Miscellaneous Tables Referenced in Text and Figures . . 226 Bibliography ........................... 232 - iii - List of Text Tables 1.1 Indicators of overall educational development in Asia, mid 1980s . . . . . . . . . . . . . . . . . . 9 1.2 Constraints and prospects for educational development, Asia, 1970-2000 . . . . . . . . . . . . . 13 1.3 Operational characteristics of education systems in Asia, early to mid-1980s . . . . . . . . . . . . . 15 1.4 Patterns of educational costs and their sources of differences, Asia, mid-1980s . . . . . . . . . . . 17 1.5 Indicators of financing and institutional arrangements in education, Asia, mid-1980s . . . . . . . . . . . . 19 2.1 Educational attainment of adults, world regions, 1980s . . . 40 2.2 Enrollment ratios, world regions, 1985 . . . . . . . . . . . 41 2.3 Public spending on education, world regions, 1985 . . . . . 42 2.4 Share of enrollments in private sector, world regions, 1985 43 2.5 Educational attainment of adults in asia, 1970-1980s . . . .44 2.6 Gross enrollment ratios (t) by level of education in Asia 45 2.7 Public spending on education as a percentage of total government spending . . . . . . . . . . . . . . . . . . 46 2.8 Public spending on education as a percentage of the GNP . 47 2.9 The percentage share of enrollments in private education, Asia, 1970-1985 . . . . . . . . . . . . . . . . . . . . 48 2.10 Levels and trends in the dependency ratio, Asia, 1970-2000 . . . . . . . . . . . . . . . . . . . . . . 50 2.11 Population and real economic growth rates (percent p.a.), Asia, 1975-2000 . . . . . . . . . . . . . . . . . . . 51 2.12 OLS regression results using grade attainment of current population as dependent variable . . . . . . . . . . . 54 3.1 Distribution of public spending on education, 1985 . . . . . 57 3.2 Unit operating costs of public education in Asia, mid-1980s 60 - iv - 3.3 Unit operating costs of public higher education, mid-1980s . 62 3.4 Index of overall the costliness of public education, Asia, mid-1980s . . . . . . . . . . . . . . . . . . . . 63 3.5 Annual teacher remuneration and pupil-teacher ratios in public primary and secondary schools, mid-1980s . . . . 65 3.6 Higher education pupil-teacher ratio, 1985 . . . . . . . . . 69 3.7 Distribution of higher education enrollments by type of institutions, mid-1980s . . . . . . . . . . . . . . . . 71 3.8 Fees for public education as a percent of unit operating costs, mid-1980s . . . . . . . . . . . . . . . . . . . 73 3.9 Estimated rate of private financing in higher education Asia, circa 1985 . . . . . . . . . . . . . . . . . . . 74 4.1 International patterns of social returns to education . . . 79 4.2 Rates of return to education in Asia, latest available year 80 4.3 International evidence on the social returns to selected secondary and university programs . . . . . . . . . . . 81 4.4 OLS regression of the deterninants of adult literacy rates, 1985 . . . . . . . . . . . . . . . . . . . . . . 83 4.5 Patterns of cohort survival in primary and secondary education, mid-1980s . . . . . . . . . . . . . . . . . 86 4.6 Intra- and inter-cycle selection in primary and secondary education, Asia, mid-1980s . . . . . . . . . . . . . . 88 4.7 Science achievement in Asia and other countries, mid-1980s . . . . . . . . . . . . . . . . . . . . . . . 94 4.8 Size distribution of Philippine public secondary schools, 1986 . . . . . . . . . . . . . . . . . . . . . 100 4.9 Economies of scale in Chinese higher education, 1980s . . . 101 4.10 Size distribution of tertiary institutions in China by type, 1987 . . . . . . . . . . . . . . . . . . . . . 102 4.11 Graduate output from various types of higher education in Thailand, 1980s . . . . . . . . . . . . . . . . . . 103 - v . 5.1 Unit operating unit costs of public education and indices of deviation from regional mean, and of bias toward higher education, Asia, mid-1980s . . . . . . . . . . . 107 5.2 Actual and predicted enrollment ratios in Asia, mid-1980s . 109 5.3 Relative degree of emphasis on higher education in Asia, mid-1980s . . . . . . . . . . . . . . . . . . . . . . . 113 5.4 Average public subsidies per pupil, Asia mid-1980s . . . . 116 5.5 Projected distribution of the educational attainment and cumulative public subsidies among members of current school-age population, Asia mid-1980s (percent) . . . . 117 5.6 Distribution of public spending on education, Asia mid-1980s . . . . . . . . . . . . . . . . . . . . 118 5.7 SES composition of school and reference populations, major world regions () . . . . . . . . . . . . . . . 125 5.8 Distribution of cumulative public spending on education by SES group, major world reions . . . . . . . . . . . 127 5.9 Percent females enrolled by level of education, Asia, 1970-1985 . . . . . . . . . . . . . . . . . . . . 127 5.10 Proportion of grade 1 entrants surviving first 10 grades of schooling, India, 1980 . . . . . . . . . . . 133 5.11 Cohort survival rates in elementary education by socioeconomic groups, Philippines, 1982 . . . . . . . . 137 5.12 Access to secondary and higher education in Thailand, 1980s . . . . . . . . . . . . . . . . . . . . . . . . . 139 - vi - List of Text Figures 1.1 (5.2) Deviation of actual enrollment ratios from those predicted on the basis of per capita GNP, Asia circa 1985 . . . . . . . . . . . . . . . . . . . . . 12 1.2 (2.2) Relationship between aggregate public spending on education and average grade attainment, Asia, mid-1980s . . . . . . . . . . . . . . . . . . . . . 23 1.3 (3.2) Relationship between overall costliness of education and per capita GNP, Asia, circa 1985 . . . . . . . . 24 1.4 (3.6) Relationship between average secondary teachers' remuneration and per capita GNP, Asia, circa 1985 . 25 1.5 (3.5) Relationship between pupil-teacher ratio in secondary education and per capita GNP, Asia, circa 1985 . . . 26 1.6 (3.7) Relationship between student-faculty ratio in conventional public higher education and per capita GNP, Asia, circa 1985 . . . . . . . . . . . . . . . . . . . . . 27 1.7 (3.8) Relationship between per capita GNP and degrees of private 'inancing in higher education, Asia, mid-1980s . . . 29 1.8 (4.5) Relationship between the costliness of conventional public higher education and overall extent of private financing in subsector, Asia, circa 1985 . . . . . . 30 1.9 (5.7) Relationship between overall extent of private financing in higher education and enrollment ratios in higher education, Asia, circa 1985 . . . . . . . . . . . . 31 1.10 (5.6) Relationship between overall extent of private financing in higher education and share of cumulative spending on education by top 10% by education, Asia, circa 1985 . . . . . . . . . . . . . . . . . . 32 1.11 (4.1) Relationship between survival rates in primary education and per capita GNP, Asia, circa 1985 . . . . . . . . 33 1.12 (5.11) Relationship between female share of enrollments in primary education and cohort survival rates in primary education, Asia, circa 1985 . . . . . . . . . . . . 34 1.13 (4.2) Relationship between survival rates and unit operating costs in primary education, Asia, circa 1985 . . . . . . . 35 1.14 (4.3) Relationship between cohort survival rates and pupil-teacher ratios in primary schooling, Asia, circa 1985 . . . 36 *vii - 2.1 Relationship between the gross enrollment ratios and per capita GNP, major world regions, circa 1985 . . . . 41 2.2 Relationship between aggregate public spending on education and average grade attainment, Asia, mid-1980s . . . . . . . . . . . . . . . . . . . . . . . 56 3.1 Relationship between the unit costs of regular public higher education and per capita GNP, Asia, circa 1985 . . . . . . . . . . . . . . . . . . . . . . . . . 61 3.2 Relationship between the overall costliness of public education and per capita GNP, Asia, circa 1985 . . . . 64 3.3 Relationship between average primary school teachers' remuneration and per capita GNP, Asia, circa 1985. . . 67 3.4 Relationship between pupil-teacher ratio in primary education and per capita GNP, Asia, circa 1985 . . . . 67 3.5 Relationship between average secondary school teachers remuneration and per capita GNP, Asia, circa 1985 . . . 68 3.6 Relationship between pupil-teacher ratio in secondary education and per capita GNP, Asia, circa 1985. . . . . 68 3.7 Relationship between student-faculty ratio in conventional public higher education and per capita GNP, Asia, circa 1985 . . . . . . . . . . . . . . . . . . . . . . 70 3.8 Relationship between per capita GNP and degree of private financing in higher education, Asia, mid-1980s . . . . 76 4.1 Relationship between survival rates in primary education and per capita GNP, Asia, circa 1985 . . . . . . . . . 89 4.2 Relationship between survival rates and unit operating costs in primary education, Asia, circa 1985 . . . . . 91 4.3 Relationship between cohort survival rates and pupil-teacher ratios in primary schooling, Asia, circa 1985 . . . . . 91 4.4 Relationship between the relative importance of inter- cycle selection in primary and secondary education and per capita GNP, Asia, circa 1985 . . . . . . . . . 92 4.5 Relationship between the costliness of public higher education and the overall extent of private financing in the subsector, Asia, circa 1985 . . . . . 97 -viii - 4.6 Relationship between unit operating costs and school size, secondary education in thb Philippines, 1986 . . 99 5.1 Deviation of the unit operating costs of public education from the regional average, Asia, circa 1985 . . . . . . 108 5.2 Deviation of actual enrollment ratios from those predicted on the basis of per capita GNP, Asia circa 1985 . . . . . . . . . . . . . . . . . . . . . . 111 5.3 Relationship between emphasis on higher education and per capita GNP, Asia, circa 1985 . . . . . . . . . . . 114 5.4 Relationship between share of cumulative public spending on education received by the 10% best educated and per capita GNP, Asia, circa 1985 . . 119 5.5 Share of cumulative public spending received by the 10% best educated, in absolute terms (y-axis) and relative to the trend line in figure 5.3 (x-axis), Asia, 1985 . 120 5.6 Relationship between overall extent of private financing in higher education and share of cumulative spending on education by top 10% by education, Asia, circa 1985 121 5.7 Relationship between overall extent of private financing in higher education and enrollment ratios in higher education, Asia, circa 1985 . . . . . . . . . . . . . . 123 5.8 Relationship between female share of enrollments in primary education and per capita GNP, Asia, circa 1985 . . . . 128 5.9 Relationship between female share of enrollments in secondary education and per capita GNP, Asia, eirea 1985 . . . . . . . . . . . . . . . . . . . 129 5.10 Relationship between female share of enrollments in higher education and per capita GNP, Asia, circa 1985 . . . . . . . . . . . . . . . . . . . . . . 130 5.11 Relationship between female share of enrollments in primary education and cohort survival rate in primary education, Asia, circa 1985 . . . . . . . . 130 5.12 Relationship between female share of enrollments in secondary education and cohort survival rate in secondary education, Asia, circa 1985 . . . . . . . 131 - Ex * 5.13 Relationship between female share of enrollments in secondary education and cohort survival rate in primary education, Asia, circa 1985 . . . . . . . . . . . . . . 132 5.14 Net enrollment ratios by grade in overall population, and among urban boys and rural girls, India, circa 1980 . . 135 5.15 Cohort survival rates in the Philippines education system, circa 1985 . . . . . . . . . . . . . . . . . . 136 5.16 Cohort survival rates in primary education according to educational attainment of pupil's father, Philippines, 1982 . . . . . . . . . . . . . . . . . . . 138 . x - List of Appendix Tables A.1 Summary of comparative data on educational development in Asia, mid-1980s . . . . . . . . . . . . . . . . . . . . 150 A.2 Qualitative assessment of educational development in Asian countries, mid-1980s . . . . . . . . . . . . . . 153 B1.1 Length of educational cycles, Asia, 1986 (years) . . . . . . 199 B1.2 Enrollment ratios by level of education, Asia, 1970-85 . . . 199 B1.3 Percent females in total enrolled by level of education, Asia, 1970-1985 . . . . . . . . . . . . . . . . . . . . 200 B1.4 Share of private enrollments by level of education, Asia, 1970-85 . . . . . . . . . . . . . . . . . . . . . 201 B1.5 Percent secondary enrollments in vocational/technical education, Asia, 1970-85 . . . . . . . . . . . . . . . 202 B1.6 Higher education enrollments by type of institution, Asia, mid-1980s . . . . . . . . . . . . . . . . . . . . 202 B1.7 Distribution of higher education enrollments by field, Asia, 1980-85 . . . . . . . . . . . . . . . . . . . . . 203 B2.1 Estimated educational attainment of current population, Asia, mid-1980s . . . . . . . . . . . . . . . . . . . . 205 B2.2 Cohort cu.vival rates, Asia, eary to mid-1980s . . . . . . . 206 B2.3 Excess demand for higher education, Asia, mid-1980s . . . . 207 B2.4 Pupil-teacher ratio in primary education, Asia, 1970-85 . . 208 B2.5 Pupil-teacher ratio in secondary education, Asia, 1970-85 . 209 B2.6 Pupil-teacher ratio in tertiary education, Asia, 1970-85 . . 210 B3.1 Dependency ratios, Asia, 1970-2000 . . . . . . . . . . . . . 211 B3.2 Population and economic growth rates, Asia, 1975-2000 . . . 211 B3.3 Percent adults literate, Asia, 1970-1985 . . . . . . . . . . 212 B3.4 Educational attainment of the adult population in Asia, early 1980s . . . . . . . . . . . . . . . . . . . . . . 213 . xi - B3.5 Distribution of GDP and labor force, Asia, 1980-1986 . . . . 214 B3.6 Overall government spending as a percent of GNP, Asia, 1970-1980s . . . . . . . . . . . . . . . . . . . 215 B3.7 Total government spending as % of GDP, Asia, 1970-1980s . . 215 B3.8 Share of public spending on education by luvel, Asia, 1970-19R0s . . . . . . . . . . . . . . . . . . . . . . 216 B3.9 Government spending on education, Asia, 1970-1980s . . . . . 218 B3.10 Public spending on education as a percentage of the GNP, Asia, 1970-1980s . . . . . . . . . . . . . . . . . . . 219 B4.1 Unit operating cost of public education (mid-1980s) . . . . 220 34.2 Teacher remuneration as ratio to per capita GNP, Asia, mid-1980s . . . . . . . . . . . . . . . . . . . . . . . 221 34.3 Fees for public education as percent of unit operating costs, Asia, mid-1980s . . . . . . . . . . . . . . . . 222 35.1 Rates of return to investment in education, Asia, latest available year . . . . . . . . . . . . . . . . . 223 B5.2 Dates of establishment of open universities, Asia . . . . . 224 B5.3 Returns by field of study in higher education, Thailand, 1985 . . . . . . . . . . . . . . . . . . . . 225 C.1 OLS regression of the relationship between gross enrollment ratios and the per capita GNP, mid-1980s . . . . . . . 227 C.2 Average per capita GNP and gross enrollment ratios, world regions, 1980s . . . . . . . . . . . . . . . . . 227 C.3 Data for text figures . . . . . . . . . . . . . . . . . . . 228 1. Introduction and Overview of Findings Asia is a region of striking diversity among countries. This study documents an interesting aspect of that diversity, namely education, and attempts to derive lessons about educational policies, particularly concerning issues of costs and financing. Its general approach and framework are set out in this chapter (Part A). For convenience, the study's principal findings and relevance to policy dialogue are also summarized here (Part B). Part A: Overall approach and framework To place the study in perspective, a brief explanation is in order. about its motivation, scope and focus. Further, in order to minimize misapplication of its findings, it is useful at the outset to take note of the study's limitations that stem from the analytical approach adopted, and constraints in the data. (a) Motivation for the study In most developing countries, the rapid growth of education in the past two to three decades has been accompanied by a concomitant rise in government spending. As a result, spending on education often claims a large (in some cases, the largest) share of the public budget, and governments are commonly the single most important source of financing for the sector, even in countries where private education is well-developed. Governments' dominant role in financing education is partly responsible for the general interest worldwide in issues relating to educational costs and finance. In a large group of developing countries, the interest arises from concern about the prospects for educational development. While the demand for education is expected to grow relentlessly owing to rapid population growth, governments are increasingly hard-pressed for resources, not only because the macroeconomic outlook is poor, but also because the keen intersectoral competition for resources makes it difficult to expand already large budget appropriations to education. In this context, a docume.tation of educational costs and financing is essential as a basis for formulating policies for the sector's development under a tightening external environment. Most Asian countries share this concern, but to a lesser extent than elsewhere. In many of them the external constraints are projected to ease considerably, with the slowdown of population growth and relatively bright prospects on the economic front. Of concern is not so much how to survive the present adverse environment, as how to improve on past achievements. This issue is obviously important on its own merit, but the fact that enormous * 2 - amounts of resources are devoted to education adds weight to the need for assessment. Such an evaluation is particularly relevant at this point as Asian countries enter a changing external environment for educational development in which interest is likely to shift increasingly from sheer quantitative expansion of the system to upgrading the quality of available services. (b) Its focus and scope The present study is an attempt to provide a statistical description and overall assessment of the current status of educational development in Asia. It focuses mainly on costs and financing issues, since a documentation of these aspects yields basic information about how resources are deployed in the sector. I/ Apart from this primary objective, a related goal is to derive general lessons about policy options by comparing educational outcomes across countries with different patterns of costs and financing arrangements. In this respect, the Asian countries provide particularly rich ground for the analysis because they differ so widely in educational policies, and in the characteristics of their education systems. The report's emphasis is on comparative analysis rather than on details relating to individual countries. For this reason, special attention was paid to the compilation of data and indicators that satisfy, to the extent possible, two requirements: comparability across countries and consistency within countries. The materials presented are therefore intended to complement the more detailed data that are relevant in sector assessments for individual countries. Several aspects of educational development are not discussed in this study. In particular, the management, pedagogical, cultural and political facets, though clearly important, receive relatively little attention as they are not easily quantified. Instead, the chief focus is on economic and financial issues. Accordingly, the educational achievement of countries is generally compared using such indicators of outcome as the formation of human capital and the promotion of social equity. With respect to human capital formation, the broad concern is with the efficiency of investments in education. How do countries differ in resource allocation to the sector as a whole, and within the sector? Are allocations such as to maximize the returns to society, measured and unmeasured, of investing in education? To obtain better results, in what direction might it be advantageous for countries to shift their educational policies? Regarding the promotion of equity, the issues are motivated by at least two considerations. First, since public spending on education is 1/ It thus complements other comparative studies, such as Unesco (1985) that deal mostly with educational outcomes. - 3 - usually substantial, its distribution is likely to have a significant impact on income distribution, and hence on some aspects of poverty. Second, beyond this cross-sectional effect, a longer term influence derives from the fact that access to education (a) is a powerful vehicle of individuals' social and economic advancement in society; and (b) is determined to a large extent by the pattern of public spending in the sector. In this context, the main questions that this study seeks to address include: How do countries differ in their relative emphasis on the various levels of education, in terms of inputs, coverage, and financing arrangements? What are the patterns of access to and exit from the education system? What impact do these characteristics have on the distribution of educational opportunities, and public spending on education? As before, what policy options exist for improvement? (c) Caveats to note Three main caveats about this study's analytical approach are appropriate to mention here. First, it is a country-level study, meaning that countries are generally taken as the relevant unit of observation. Large countries, such as China, India, and Indonesia, are treated as single units, even though disaggregated data would undoubtedly reveal a more accurate picture of actual conditions. Having recognized this defect, note, however, that each country's education system operates under a common institutional and policy framework. Thus, as long as the focus remains on broad structural issues, the reliance on aggregated data is not as serious a shortcoming as might appear at first sight. Second, the analysis is meant only to highlight policy issues of potential importance. It does not substitute for more detailed studies that are essential in designing country-specific policy interventions. For example, if unit costs in a particular country are significantly higher than those in neighboring countries, it does not immediately justify a program to reduce costs; rather this finding signals the need for additional analysis to examine the causes of high costs, and to evaluate the merits of alternative cost-reducing measures. Third, not all aspects of educational costs and financing are addressed in adequate depth. The study relies chiefly on secondary sources of data, and suffers from the scarcity of resource materials on some topics. Moreover, the nature of some issues is such that they require survey rather than aggregated data for a meaningful analysis. For these reasons, the discussion on such topics as external efficiency, student achievement and some aspects of equity, does not go as far as would be desirable, or indeed necessary. Nevertheless, the existing data can yield useful insights. Aside from offering a broad overview of policy issues in the sector, they also help identify gaps in our present understanding, thereby suggesting the focus of future work. Indeed, defining the analytical and research agenda is probably also an important product of this study. .4- (d) Remarks about the data 2/ For most countries, statistics on the standard indicators -- such as enrollment ratios, percent females in total enrolled -- are readily obtainable. In contrast, the financial data are scarce and often not comparable across countries even if reported in a single source. Most of them were thus derived from basic sources, whenever these are available, including country sector studies, government statistical publications, and budget documents. The data on a basic set of indicators were assembled for a core of 11 Asian countries (Bangladesh, China, India, Indonesia, Korea, Malaysia, Nepal, Papua New Guinea, the Philippines, Sri Lanka, and Thailand). For completeness of regional coverage, the data for other Asian countries are also presented when available. As for accuracy, the data suffer from the obvious disadvantage of having been generated mostly from secondary sources. This shortcoming was mitigated somewhat by making special efforts to check for internal consistency in the data (in one or two instances with the originators of the data), and to ensure comparability in data definitions across countries. Thus, if broad orders of magnitude are sufficient, then the statistics provide a reasonably good basis for the "overview" approach adopted in this study. It is to be stressed that while they are robust enough to offer conclusions about general policy directions, further refinements are probably called for when assessing specific countries and policy options. (e) Organization of the report The rest of this chapter summarizes the study's principal findings and its implications for policy dialogue concerning educational development in Asia. The data and analysis are amplified in subsequent chapters as follows. To place the analysis in context, chapter two highlights the main features of education in Asia in an international and regional perspective. This done, the next chapter provides a more detailed documentation of educational costs and financing arrangements in Asian countries, focussing on such items as the pattein and level of unit costs across levels of education, the distribution of public spending, the contribution of private financing, and so on. With the basic pieces of data in place, the issues of efficiency and equity are addressed in chapters four and five. Chapter six offers general conclusions based on the cross-sectional analysis, and suggestions regarding future work. Taking account of the overall findings, Appendix A provides a succinct description and comparative evaluation of the current status of education in individual Asian countries. Details on data sources and the corresponding core educational statistics appear in Appendix B. Miscellaneous data referred to in the text and figures are provided in Appendix C. 2/ For more details on data sources, see Appendix B. - 5 - Part B: Summary of the Principal Findings At the broadest level, the study yields an overview of the current state of education in Asia, and the major policy challenges it faces. These conclusions are based on more specific findings, a synthesis of which also appears below. B.1 Education in Asia and Major Policy Challenges: An Overview (a) Diversity across countries. Countries in the region differ widely in educational outcomes, external constraints on the sector's development, and choice of sectoral policies. With respect to outcomes, the spectrum stretches from Bhutan, with an adult literacy rate of only 15 percent and a primary enrollment ratio of 25 percent in the mid-1980s, to such countries as Korea and Thailand, where over 90 percent of the adults are literate, and primary education has reached, or is approaching, universal coverage. Given current structures of the enrollment pyramid, the average grade attainment of future adults ranges from less than five years in Bangladesh, Bhutan, India, Nepal and Papua New Guinea, to over nine years in Korea, Malaysia, the Philippines and Sri Lanka. In part, these differences in outcomes owe their origin to past external constraints. Low-income countries tend to perform less well than richer ones because of the twin pressures of rapid population growth and relatively slow economic expansion -- problems which are often compounded by the high cost of educational inputs, the weak demand for schooling, and the sparseness of institutional infrastructure. The squeeze on resources has been tight in countries with a high ratio of school-age population to adults (Bangladesh, Burma, Indonesia, Laos, Nepal, Papua New Guinea, and the Philippines) since larger ratios imply heavier fiscal burdens for educational finance. But the constraints were perhaps even tighter in countries where the economy grew slower than the population (Papua New Guinea), or where it grew only slightly faster (Nepal, and the Philippines), since these conditions permit only limited possibility for expanding coverage or upgrading services. Perhaps more importantly, countries' achievement in education also reflected differences in the choice of sectoral policies. The following examples suffice to illustrate the diversity in this respect in the mid-1980s: aggregate spending on education varied from less than 2 percent of GNP in Bangladesh, Burma, Nepal and the Philippines, to 6 percent and above in Malaysia and Papua New Guinea; the share of public spending devoted to primary education, reflecting choices in intrasectoral emphasis, ranged from less than 30 percent in India, to around 60 percent in Indonesia, Korea, the Philippines and Thailand; the rate of cost recovery for conventional public higher education stretched from less than 5 percent in Bangladesh, China, India, Papua New Guinea, Sri Lanka, and Thailand, to nearly 50 percent in Korea; and the mix of institutional arrangements to cope with excess demand for higher education encompassed low-cost distance systems (Sri Lanka, and Thailand), private education (Korea and the Philippines), and overseas education (Malaysia). (b) The potential for enhancing efficiency and equity Significant progress in educational development has occurred in many Asian countries, but striking inefficiencies and inequities remain. The unit costs of education appear to be relatively high in some countries, owing to various factors, including unexploited economies of scale in providing public education, and inadequate incentives for greater cost-consciousness among consumers and providers. Many education systems also suffer from poor internal efficiency; for example, dropping out in the first two cycles of education is an important mechanism of selection, albeit an implicit one, in a large number of countries. The problem is especially serious in primary education, with adverse implications for human capital formation and social selectivity. For the region as a whole, 91 percent of the population currently enter grade 1, but only 62 percent of the entrants reach the end of primary schooling. The situation is especially alarming in such countries as Bangladesh, Bhutan, India, Laos, and Nepal, where survival rates are 40 percent or less. Girls' participation in schooling appears to be especially weak under such conditions. With regard to equity, the intrasectoral allocation of education expenditures in some countries reveals a decided bias toward higher education at the expense of primary education. This diversion results from failure to tap private sources of funding for higher education, despite evidence of excess demand. The subsector's claim on resources is thus generated by the twin pressures of high unit subsidies, and expansion of coverage to cope with excess demand. Evaluated on the basis of resource-intensity per student, extent of coverage, and degree of private financing across levels of education, the emphasis on higher education appears to be especially strong in Bangladesh, India, and Papua New Guinea. Substantial scope therefore exists for improving efficiency and equity in Asian education. In most countries, the external context for potential policy shifts is likely to be quite favorable, with the projected slowdown of population growth and generally bright prospects for economic expansion. The challenge in such countries is to avoid the temptation of complacency, focussing instead on consolidating and further strengthening past achievements through appropriate policies to enhance the allocation of spending on education. In a handful of countries, however,-- including Bangladesh, Bhutan, Laos, Nepal and Papua New Guinea -- the demographic and macroeconomic conditions will probably remain difficult. For them, the choice of appropriate polices is crucial for future progress, since the wrong choices can exacerbate present inefficiencies and inequities, and may even reverse past achievements. - 7 - (c) Broad policy options Given the diversity among countries in the region, the choice of precise policy options will clearly differ according to initial conditions, and sectoral objectives. However, few governments would disagree with the broad goal of alleviating poverty. Since education is an essential ingredient of policies to enhance social and economic advancement among the poor, this objective may be taken as a primary aim of policies in the sector. In some situations, promoting equity involves some sacrifice of efficiency, but the analysis in this report indicates a fortuitous absence of conflict, at least at the aggregate level. Indeed, improving efficiency is inescapable if equity is to be enhanced, since inefficiency implies high costs, which in turn shrinks the pool of resources actually available for poverty-alleviating interventions. Given these broad goals, three major policy thrusts are indicated: (i) Increased agregate spending on education in selected countries. This option is intuitively appealing, but should be treated cautiously, in light of the finding that appreciable changes in educational outcomes result only with relatively large shifts in spending. The explanation is quite simple: outcomes are determined as much by the efficiency with which resources are used, as by the aggregate amount of resources available. Thus, for most Asian countries, the potential for improvement resides largely in promoting the efficient use of resources rather than in expanding the overall resource envelope. In a few countries, however, such as Bangladesh, Nepal, and the Philippines, current spending levels are so low that an increase would probably relieve severe pressures on the education system. The increase should, as much as possible, be directed toward primary education, particularly in the first two countries. (ii) Imnroving access and retention rates in -rimarveducation. Primary education deserves emphasis because it affects social equity and the formation of basic human capital. In some countries (Bhutan, China, India, Nepal, and Papua New Guinea), a two-pronged approach is required since not all school-age children enter grade 1, and not all who enter survive to the end of the cycle. In the remaining Asian countries, with the exception of Korea and Malaysia, reducing the incidence of non-completion is the main issue. The choice of specific interventions will obviously differ according to local conditions. Increasing the level of spending per student is probably warranted in some settings, such as Bangladesh and India. However, in view of the extensive research evidence that "throwing more money" at schools does not in itself insure against low cohort survival rates, it would also be relevant to consider how resources are spent among school inputs. Assessing the options in this respect deserves priority on the agenda of future analytical work. (iii) Freeing MR resources for orimary education. The orientation of sectoral policies in favor of primary education requires concomitant changes in other subsectors. Increased efficiency--for example, by exploiting economies of scale in the spatial organization of schools--is crucial as it would enlarge the amount of resources effectively available for reallocation to primary education. Reducing the public subsidization of higher education is, however, perhaps the most clearcut overall shift in policy called for. A menu of options is available in this regard, as indicated by the experience of several Asian countries with relatively low rates of public subsidization in higher education (for example, Korea, Thailand, and the Philippines): instituting cost recovery in conventional public institutions, encouraging the emergence of a largely self-financing private sector, and relying on relatively low-cost distance systems. As before, the optimal mix of institutional and financing arrangements depends on unique country conditions. Further analysis of this issue would probably be worthwhile. B.2 Synthesis of the More Specific Findings For convenience, these findings may be separated into two groups: those that relate to cross-country comparisons of the relative strengths and weaknesses of the sector in individual countries; and those that concern the impact of various policies on educational development. (a) Assessment of strengths and weaknesses from a regional gerspective For ease of discussion, the cross-country comparison is made on the bAsis of five major aspects of education: (i) the system's overall development; (ii) constraints on and prospects for its future evolution; (iii) the system's operational and student flow characteristics; (iv) its structure of unit costs and relative costliness; (v) and the financing and institutional arrangements in the sector. (i) Develoment of the education system. Data on some of the basic indicators -- adult literacy rates, aggregate public spending on education, and enrollment ratios -- appear in table 1.1. 1/ First, consider the data on adult literacy rates, a statistic that reflects the cumulative impact of past investments in education. Countries vary widely around the regional mean of 65 percent. In one group, comprising 2./ For brevity, the data reported in this and subsequent tables in this chapter pertain mostly to the latest available year, usually early to mid-1980s. Other years6 data are provided in later chapters, and in the appendix tables. To facilitate the reading of graphs in this report, each country is assigned a unique number in all the tables and graphs. -9- Korea, the Philippines, Singapore, Sri Lanka, and Thailand, most adults are literate. At the other extreme are such countries as Bangladesh, Bhutan, and Nepal, where only one-third or fewer of the current adult population is literate. The remaining countries are characterized by moderate rates of adult literacy, ranging from 43 percent in India, to 84 percent in Laos. In countries where there is still scope for improvement, progress depends on current investments in education, particularly at the primary level. Two factors are relevant in this regard: coverage of the education system in primary education, and its capacity to retain pupils long enough to impart permanent literacy. A/ Assessment based on these considerations suggests that the prospects for achieving universal adult literacy among the next generatiQn of adults are particularly bleak in Bangladesh, Bhutan, and Nepal, while the opposite is true in Malaysia. In China, India, Indonesia, and Papua New Guinea, progress toward that goal is likely to materialize, although its speed will depend importantly on the choice of sectoral policies, particularly those affecting the development of primary education. Table 1.1s Indicators of overall educational development in Asia, mid-1980a Public spending Per capita on education aI Gross enrollrena.atio (2) Average GNP (1985) 2 adults grade Cgugtry 9ZS _ leat 2 ian-udget 2 Ngary Secondary Higge b/ attainment oI 1 Bangladesh 159 33 10.3 1.5 60 18 5.2 3.9 2 Bhutan 151 15 7.3 3.8 25 4 0.1 1.4 3 BSurma 184 - 10.9 1.8 107 23 5.4 7.0 4 China 273 69 7.8 3.3 118 39 1.7 5.1 5 India 259 43 13.7 3.0 92 41 9.0 4.8 6 Indonesia 470 74 15.0 3.7 118 42 6.5 7.3 7 Rorea 2,040 92 16.6 3.4 96 75 31.6 11.4 8 Laos 332 84 - - 94 19 1.5 4.8 9 Malaysia 1,860 74 16.0 6.0 99 53 6.0 (9.0) 9.2 10 Nepal 142 26 9.6 1.8 82 25 4.6 3.6 11 Papua New Guinea 621 45 17.9 6.9 70 13 2.0 4.3 12 Philippines 581 86 11.5 1.8 106 65 38.0 10.2 13 Singapore 7,093 86 - * 115 71 11.8 9.9 14 Sri Lanka 374 87 8.1 2.8 103 63 4.6 (5.1) 9.5 15 Thailand 712 91 19.4 3.6 97 30 19.6 7.0 Regional average 1,017 65 12.6 3.3 92 39 9.8 6.6 So*-ces data on per capita GNP are from World Bank (1988)s other data are from sources cited in appendix B. a/ See page 2 of appendix 3 for discussion of the coverage of this indicator. b/ Data in parentheses refer to the enrollment ratio if nationals studying abroad are included in the numerator. of Variable refers to the average educational status that the current a-bool-age population is is likely to attain given the current structure of the enrollment pyramid, and the pattern of cohort survival in the system. The figure for Burma is not strictly comparable to those for other countries as it is based only on the enrollment structure. It is excluded in calculating the regional average. W The latter of these two factors will be discussed in more detail later in this chapter. - 10 - Turn next to consider the current level of public spending on education, / an indicator that reflects a country's overall fiscal "effort" to invest in human capital, and the intersectoral priorities in the allocation of public funds. On average, government spending on education, expressed as a percentage of GNP, is somewhat lower in Asia than in other developing regions of the world: 3.3 percent compared with 4.0 percent in Africa, 3.5 percent in Latin America, and 5.3 percent in the Middle East. However, there is wide variation among countries, the size of spending being particularly low in Langladesh (1.5 percent), and in Nepal and the Philippines (1.8 percent), and particularly high in Malaysia (6.0 percent) and in Papua New Guinea (6.9 percent). As a matter of simple extrapolation, it is clear that a country's educational achievement, in terms of, say, coverage, tends to improve with more resources, given the system's operational characteristics. The data for Asia reveal, however, that across countries, very different outcomes are possible with comparable levels of fiscal "effort," or conversely, very similar outcomes with quite different levels of "effort." For example, public spending on education in Bangladesh is comparable to that in the Philippines, but the coverage of its system is much smaller; in Korea and Malaysia, coverage is comparable, but public spending is much smaller in the former country, 3.4 percent of GNP compared to Malaysia's figure of 6.0 percent. These patterns are an early indication that other considerations besides aggregate spending -- for example, the efficiency with which the education system operates -- also affect educational outcomes. Finally, consider the structure of coverage as indicated by the gross enrollment ratio (columns 5-7). For most Asian countries, this statistic for primary education in 1985 exceeds, or approaches, 100 percent. The exceptions are Bangladesh, Bhutan, Nepal and Papua New Guinea, where despite past growth, coverage still lags behind the rest of the region. Bhutan belongs in its own category, however, with a gross enrollment ratio of only 25 percent, compared to the range of 60-82 percent in the other three countries. The expansion of coverage in the fifteen years to 1985 has been particularly remarkable at the post-primary levels; on average, the ratio for secondary education jumped from 26 percent in 1970 to 42 percent in 1985; and in higher education, it grew from 5 to 10 percent in the same period. It is noteworthy that the variance among countries widens with rising levels of education. The structure of coverage in the education system can be summarized by the average grade attainment of the current school-age population (column 8). This statistic projects the educational status of the next generation of adults based on current patterns of enrollment across the three levels of / Public spending usually refers to spending by the central government; however, in countries where lower levels of government also finance education to a significant extent (e.g India) their expenditure (net of transfers) is also included. For further remarks on this issue, see appendix B. * 11 - education, and cohort survival. This indicator ranges widely, from 1.4 years in Bhutan, to 11.4 years in Korea. Part of the variation reflects differences in level of economic development, since wealthier countries tend, on average, to have better developed systems of education. Note, however, that Sri Lanka has a per capita GNP comparable to Laos' but its average grade attainment is nearly twice as high, 9.5 years compared to 4.8; similarly, the Philippines' per capita GNP is comparable to Papua New Guinea's, but its grade attainment is significantly higher, 10.2 years compared to 4.3. To standardize more systematically for differences in countries' wealth, it is useful to compare actual enrollment ratios to those predicted on the basis of per capita GNP. Using data from a large sample of developing countries, it is possible to estimate the average relationship between coverage in an education system and per capita GNP. The enrollment ratios predicted from this relationship are not normative targets, but they provide a useful yardstick against which to evaluate actual enrollment ratios. b/ Figure 1.1 (5.2) Z/ shows the percentage by which actual ratios deviate from that predicted at each level of education. A/ Countries differ substantially in the relationship between the actual and predicted enrollment ratios. In Bhutan and Papua New Guinea, the education system's coverage is below expectations at all levels. Malaysia fits in this category to some extent, particularly in view of the pronounced gap between the predicted and actual rates in higher education. In a second group, consisting of India and the Philippines, the actual ratios exceed those predicted at all levels of education. Korea in fact also belongs in this category even though its primary enrollment ratio is smaller than the predicted value. As will be seen later, this shortfall stems from the country's exceptionally high rate of retention in primary education. On the other hand, India's actual coverage in primary education is probably less widespread than this comparison indicates because the corresponding cohort survival rate is relatively low. In a third group of countries, comprising Burma and Nepal, coverage at all three levels is more or less consistent with j/ Because the enrollment ratio compresses as it approaches 100 percent, comparing the actual and predicted enrollment ratios does not always provide a clear assessment of coverage in primary education. Note, too, that the comparison does not incorporate information about cohort survival rates, a defect that is more serious in primary education than at the other levels because these rates tend to be lower in primary education, and they vary more widely among countries. 2/ The figures in this chapter are referred to by two numbers. The first is sequential, the second (in parentheses) indicates where it is repeated in the main body of the report. JL/ An alternative is to use the ratio of the actual to the predicted values, but this index would obscure the comparisons since for negative deviations, its potential range is zero to one, while for positive deviations it is one to infinity. - 12 - i Primary I Secondary COHigher Bangladesh - Burma - China - _ Ind ia - Indonesia - Korea - lialaysia - Nepal - Papua New Guilea - Philippines - Sri Lanka -..- Thailand - III IIII -80 -60 -40 -20 0 20 40 60 80 % actual ratio deviates from predicted louite: see Lable 5.3. sets: 6eialits ftea set tresasted at phs aid alas 60 5. FLR"vm 1.1 (5.2t Dovation of actual enrollment ratios tro those predicted on the basis of per capita GNP, Asia, circa 1965. the country's per capita GNP. Finally, in the remaining countries, the actual ratio is greater than the predicted ratio at some levels, and smaller at others. However, the pattern and magnitude of imbalance differs among countries in this group. For example, Indonesia and Sri Lanka share China's pattern of imbalance, but the difference between their actual and predicted ratios is much smaller. Thailand's pattern is comparable to Sri Lanka's, but the bottleneck in coverage appears at the transition from primary to secondary education rather than from secondary to higher education. (ii) Constraints and DrosDects. The overall constraints on educational development are determined largely by two factors, demographic pressures and macroeconomic conditions (table 1.2). Other variables -- such as the aggregate level of public spending on education, the organization of the education system and its financing arrangements -- are clearly also - 13 - important to consider, but they reflect policy choices in education rather than exogenous conditions, particularly in a longir term perspective. A useful indicator of demographic pressures is the dependency ratio, defined here simply as the ratio between the school-age population to the adult population. It reveals the fiscal obligation of adults (which they may or may not discharge) for the education of young people in the population. Large differences across countries exist in this indicator: in 1985, it ranged from a low of 31 percent in Korea, to a high of 53 percent in Laos. To appreciate the meaning of these differences, note that with Laos' dependency ratio and other things being equal, Korea would have had to spend 70 percent more than it did in 1985 to achieve the system's actual coverage in that year. The dependency ratio is projected to decline over time for most Asian countries, following a trend from the 1970s. On average, it is forecast to drop from 44 percent in 1985 to 37 percent by 2000. Assuming constancy in other variables, this decline implies a drop of 16 percent in each adult's fiscal obligation for the public financing of education. However, the prospects are brighter in some countries than in others: the ratio is expected to drop by as much as 29 percent in Thailand, and 26 percent in Burma, but is projected to rise by 4 percent in Nepal, and 7 percent in Bhutan. Table 1.2; Constraints and prospects for educational development, Asia, 1970-2000 School-age population as Projected rate Of arerth Al nercent of adult population b/ Population GNP in X change Cgunag aged 5-14. real terms Difference IM I 8S 2000 1985-2000 1 Bangladesh 2.0 4.9 2.9 55 49 42 -14 2 Shutan 3.0 - - 46 44 47 7 3 Burma 1.0 3.5 2.5 44 47 35 -26 4 China 0.4 6.6 6.2 44 33 26 -21 S India 1.3 4.8 3.5 49 44 35 -20 6 Indonesia 1.0 3.9 2.9 50 46 35 -24 7 Yorea 0.3 6.8 6.5 52 31 24 -23 8 Laos - - - 45 53 49 -8 9 Malaysia 1.4 5.0 3.6 56 41 33 -20 10 Nepal 3.1 3.8 0.7 46 49 51 4 11 Papua New Guinea 2.0 5.1 3.1 47 30 40 -20 12 Philippines 1.3 5.3 3.8 53 47 37 -21 14 Sri Lanka 1.1 4.8 3.7 49 36 31 -14 15 Thailand 0.5 6.0 5.5 56 41 29 -29 Average 1.4 5.0 3.6 49 44 37 -16 Sourcess see appendix B. a/ In percent a year for the period 1985-2000 for population, and 1990-2000 for GNP. Figures in the third column refer to the difference In growth rates between the GNP and the population aged 5-14. b/ for the present purposes, the school-age population comprises those aged 5-141 the adult population, those aged 15-65. A related indicator of demographic pressures on educational development is the growth rate of the school-age population. It has the advantage of direct comparability with the rate of economic growth, and thus permits a contrast to be drawn between the future demand for educational resources, and the potential availability of those resources. For Asia as a whole, the school-age population is projected to grow slowly over the next decade, at about 1.4 percent a year. Since the GNP is expected to grow at roughly 5.0 percent a year, there is potentially a 3.6 percent a year growth in resources over the next decade for expanding and/or upgrading the education system. 9/ Thus, for the region as a whole, the prospects for the sector are genecally quite favorable. As before, conditions vary widely among countries. In China, Korea and Thailand the prospects are particularly bright as the difference between the population and economic growth rates is at least 5.5 percent a year. Bangladesh, Burma, and Indonesia are in a less favorable position, with a gap of less than 3 percent a year. In Nepal, the prospects for expanding or upgrading education appear quite bleak since the economy is forecast to grow only marginally faster than the school-age population. (iii) Operation of the education system. The formation of human capital through education depends, among other things, on entry rates into the system, and pupils' subsequent schooling careers. With regard to the first factor, the differences among Asian countries are relatively small. Most of them have achieved, or are close to achieving, universal entry in grade 1. However, there remains room for improvement in a few countries, notably in Bhutan, Papua New Guinea, Nepal, and to a lesser extent, also in China and India (table 1.3). The variance among countries is wider in terms of the schooling careers of people who enter the system, an important aspect of which is reflected in the patterns of cohort survival within each cycle of education. First, consider the data on survival rates in primary education. For the region as a whole, more than one-third of grade 1 entrants fail to reach the end of the cycle, an indication that low survival rates in primary education is probably an important regional issue. The wide variation around this average suggests three groupings of countries. In one group, comprising Korea, Malaysia, Singapore, Sri Lanka, and Thailand, non-completion is no longer a major problem. At the other extreme are such countries as Bangladesh, Bhutan, India and Nepal, where two-thirds or more of all first graders drop out before the end of primary schooling. In these countries, reducing dropout rates is a clear priority for public policy. Finally, in a third group of countries, including China, Indonesia, Laos, Papua New Guinea, and the Philippines, cohort survival rates are higher, but are nevertheless still significantly below 100 percent. As with the previous group of countries, improving the system's retention capacity deserves emphasis in sectoral policy. 2/ Note, however, that the scope for expansion depends on the evolution of educational costs, which in turn depends importantly on how teachers' pay rise as the economy expands. *15 - TaS 1,31 Operational characteristics of education systems in Asia, early to mid-1980a Percent of first year entrants Index of X of pop. surviving to last rear in ovele bI extent of I females in total enrollments entering inter-cycle cgugtry gaI a/ Pyg over see. Umner sge. selection I Pigarh recondary 1 Bangladesh 100 24 48 80 8 40 28 19 2 Bhutan 54 17 83 88 13 34 18 17 4 China 90 68 76 81 54 45 40 30 5 India 83 37 72 65 15 40 34 29 C Indonesia 100 60 92 96 46 48 43 32 7 Rarea 100 97 98 95 87 49 47 30 8 Laos 100 40 65 68 21 45 41 36 9 Malaysia 100 97 90 96 79 49 49 45 10 Nepal 75 33 89 81 5 29 23 20 11 Papua new Guinea 74 67 63 95 57 44 36 23 12 Philippines 100 66 74 - 18 49 50 54 13 Singapore 100 100 100 97 99 47 50 42 14 Sri Lanka 100 85 75 100 58 48 53 40 15 Thailand 100 80 91 87 72 48 48 46 Regional average 91 62 80 87 45 44 40 33 Sourcest see appendix 3. a/ Note possible overestimation of entry rate for Bangladesh discussed in appendix table B2.2. b/ The numerator in each column is the number of entrants to the corresponding cycle. See appendix table B1.1 for data on number of years in each cycle or suboycle. See appendix 12.2 for note an possible recent rise in primary cohort survival rates in Indonesia. of See table 4.6 for the definition of this iadex. Note that the larger it is, the more efficient is the selection process in the education system. In secondary education, cohort survival rates are generally higher than at the primary level. However, in the lower secondary cycle, the incidence of dropping out remains significant in Bangladesh, Laos, Papua New Guinea, and to a lesser extent, also in China, India, the Philippines, and Sri Lanka. In upper secondary education, the problem is largely absent in all Asian countries. Taking the primary and secondary cycles together, it is useful to assess the intensity of between-cycle selection relative to that which takes place through dropping out within cycles of education. The former is a more desirable and efficient process, since it maximizes a student's exposure to the full curricula in a particular cycle, thereby improving the chances that he acquires the intended skills. This consideration is particularly relevant in primary education, since the pedagogical objectives at this level concern fundamental building blocks of learning -- basic literacy and numeracy. The index in colmn. (5) of table 1.3 provides a measure of the relative importance of intercycle selection up to the end of secondary education. It ranges from zero, indicating that all selection takes place through the process of dropping out, to 100, indicating that all selection is concentrated at the transition between cycles of education, dropping out within cycles being absent throughout the system. Thus, the larger the index is, the more efficient the selection process in an education system. For the region as a whole, the index has a value of only 45, confirming that dropping - 16 - out is a significant regional issue. It is particularly serious in Bangladesh, Bhutan, India, Laos, Nepal and the Philippines. In these countries, dropping out is the principal instrument, albeit an implicit one, that regulates the flow of students in the education system. Finally, consider an aspect of the system's operation that has received attention in recent years -- its ability to attract and retain females. For the region as a whole, the share of females in total enrollments declines with rising levels of education, from 44 percent in primary education, to 40 percent in secondary education, to 33 percent in higher education. The last figure suggests that, on average, an Asian male's chances of entering higher education are over twice as high as that of his sister's. Countries vary widely, however, both in the overall share of females in the system, as well as in the pattern of access across levels of education. At all levels, females tend to fare badly in low-income countries, including Bangladesh, Bhutan, Nepal, Papua New Guinea, and to a lesser extent, also India. In such countries as Indonesia, Korea, and Sri Lanka, females are relatively well-represented at the primary and secondary levels, but their share of enrollments in higher education is significantly smaller than males'. This pattern suggests the presence of a strong sex bias in the access to higher education. In Malaysia, the Philippines, and Thailand, sex differences are limited or absent throughout the education system. (iv) Education costs and their sources of differences. Expressed as a percentage of per capita GNP, the unit costs of public education in Asian countries are, on average, comparable to those in other world regions except Africa (table 1.4). 1J/ In primary education, they are estimated to be 10 percent of per capita GNP, compared to 12 percent for the Middle East, and 9 percent for Latin America. In secondary education, they are 19 percent for Asia, which compares favorably with the average of 28 percent for the Middle East, and 26 percent for Latin America. In higher education, Asia's figure of 153 percent is comparable to the 150 percent for the Middle East, but is substantially larger than Latin America's average of 88 percent. Within Asia, however, the range in unit costs is very wide, even if the extreme case of Papua New Guinea were disregarded. 11/ The pattern is as follows: 19/ For an explanation for expressing costs relative to the per capita GNP, see footnote 4 in chapter 3. Note also that since Papua New Guinea is an outlier in terms of costs, it is excluded in calculating the regional averages for cost and other related indicators, unless otherwise indicated. lI/ Papua New Guinea's unit costs are three times as high as the regional average in primary and secondary education, and over six times as high in higher education. -17 In primary education, unit costs are lowest in Bangladesh, India, the Philippines and Sri Lanka, and highest in Korea, Malaysia, and Thailand. The difference between these two groups is substantial, unit costs in the second group being, on average, more than twice those in the first group. In secondary education, the differences across countries are even wider. Unit costs remain relatively low in the Philippines and Sri Lanka, but the ranking among the remaining countries is quite different from that at the primary level. For example, Thailand's unit costs are 20 percent below the regional average, while those of Bangladesh exceed it by over 40 percent. In higher education, the overall level of unit costs depends on the distribution of enrollments across delivery systems, particularly berveen conventional and distance education. Since costs are dramatically lower in the latter system, it follows that the greater the reliance on distance education is, the smaller the overall unit costs of public higher education. WZ/ The data indicate that costs are relatively high in Bangladesh, China, India, Nepal, and to a lesser extent, also in Malaysia. On the other hand, they are relatively low in Indonesia, Korea, the Philippines, Sri Lanka, and Thailand, being on average, only about one-third as high as those in the former group. Table 1.4: Patterns of educational costs and their sources of differences, Asia. mid-1980s Unit operating costs of Avg. teachers' pay public education (2 of Unit costs as a percentage (ratio to per ner Canita GNP) La of the rezional averane bi Capita GNP) pupils per teacher country PmaEr Secondary a18itw PEmrvE Secondall Lihe Oerall PriarE Secondary Prlahz Secondary 1 Bangladesh 6 30 285 63 158 186 136 2.2 - 47 26 4 China 9 28 244 91 146 159 132 2.0 3.4 25 17 5 India 6 17 231 59 91 151 100 2.9 3.1 58 20 6 Indonesia 13 23 91 125 123 59 102 2.5 3.2 25 15 7 Korea 17 23 71 163 123 46 111 5.0 5.5 38 34 9 Malaysia 14 21 190 139 112 124 125 2.4 3.1 24 22 10 Nepal 9 14 249 89 71 162 107 2.8 5.0 36 28 11 Papua new Guinea 29 65 1,050 287 343 685 438 6.8 10.0 31 25 12 Philippines 6 9 50 57 45 33 AS 1.6 1.7 31 32 14 Sri Lanka 6 9 83 60 49 54 55 1.6 2.1 32 26 15 Thailand 16 15 40 153 81 26 87 2.5 2.9 19 20 Regional average Escluding 11 10 19 153 100 100 100 100 2.6 3.8 33 24 Including 11 12 23 235 - - - - 2.9 4.0 33 24 Sources: see appendix B. al The unit costs of higher education reflect the average for public regular institutions and distance education, weighted by their respective share of enrollments. bi The figures under the *overall' column are the simple averages of the previous three columns. 12./ Other factors also influence unit costs, as will be discussed later. Here the discussion simply reflects the components that account for the differences. - 18 - To a large extent, variation in unit costs cin be traced to differences in policy choices regarding the organization of the education system, and the use of schooling inputs. In primary and secondary education, the main determinants of costs are teachers' pay on one hand, and pupil- teacher ratios on the other. In the Philippines and Sri Lanka, the low costs are due mainly to the relatively low rate of teacher remuneration, since the other variable, pupil-teacher ratio, is close to the regional average. In India, the main factor is the relatively high pupil-teacher ratio (58 compared to the regional average of 33), since teachers' pay are comparable to the average for the region (2.9 times per capita GNP compared to 2.6 times). It is instructive to compare Korea and Thailand to illustrate differences in choices in the use of schooling inputs. Both countries have relatively high unit cost in primary education, but they arise for very different reasons. The Thai system is more teacher-intensive than the Korean one, its average pupil-teacher ratio being only half as high as Korea's. On the other hand, Korean teachers are paid at about twice the rate for Thai teachers (whose pay are in fact close to the regional average). These effects cancel out, and the result is costs that aro comparably high in both countries. This example indicates that even within a similar resource constraint, a range of options exists in the use of those resources. The data for secondary education suggests a similar conclusion. jj/ The above comparisons show differences across countries separately at each level of education. It is also of interest to assess the costliness of one level of education relative to other levels in a country (which reveals the pattern of resource-intensity per pupil across levels of education); and the overall costliness of a country's system in relation to that in other countries. A useful indicator for this purpose is an index constructed by dividing unit costs at each level, expressed as a percentage of per capita GNP, by the regional average (columns 4-7 in table 1.4). Consider first the overall costliness of education across countries. This assessment can be made using the average of the cost indices at the three levels of education (column 7). Leaving aside Papua New Guinea, which is an extreme outlier, the index for the other countries vary markedly, from 45 percent of the regional average, to 136 percent. Public education in the Philippines, Sri Lanka, and Thailand, is relatively low-cost, while the opposite is true in Bangladesh, China, and Malaysia. In India, Indonesia, Korea, and Nepal, the index is close to the regional average. Wide variation among countries also characterizes the pattern of costliness across levels of education. In one group, comprising Bangladesh, China, India, and Papua New Guinea, deviations from the mean increase rapidly with rising levels of education, indicating a definite bias in costliness toward the higher levels. Nepal shares this pattern, although the bias is more moderate since secondary education is relatively less resource-intensive IV A similar comparison for higher education is not presented for lack of data. - 19 - than primary education. In Sri Lanka, there is relative balance in the structure of costliness across the three levels of education. In the Philippines and Malaysia, the pattern suggests a moderate bias in favor of primary education, while in Indonesia, Korea, and Thailand, this tendency is stronger. (v) Financing and institutional arrangements. A selection of the most pertinent indicators appear in table 1.5. In most Asian countries, primary education is largely a public undertaking, both in terms of provision and financing. For this reason, data on the private share of enrollments and the rate of cost recovery at this level are not presented here. Table 1.5: Indicators of financing and institutional arrangements in education, Asia, mid-1980s Distribution of public spending X of allments by tre of institution at pees in Index of on education d public education private Risher education financing Share of Secondary Righer in higher Gini- 10 X best Country (Z ortigatel f&,E . .R m&. 2SOegas San. Sauga a eduatten of jgA educated I Bangladesh 93 58 41 1 1 4 0 17 85 76 2 Bhutan - 0 96 0 4 - - - - - 3 Burma - 0 54 45 0 - - - - - 4 China 0 0 69 30 1 3 0 0 41 29 5 India 67 b/ 57 37 5 1 12 5 7 66 61 6 Indonesia 50 58 33 9 1 27 19 49 27 21 7 Korea 40 65 21 12 2 34 46 77 16 13 9 Malaysia 2 8 61 1 31 4 6 15 38 32 10 Nepal 10 23 73 1 2 41 10 32 57 52 11 Papua New Guinea - 6 83 9 3 40 0 6 62 55 12 Philippines 42 83 17 0 0 9 15 86 19 14 14 Sri Lanka 2 0 62 29 10 3 3 21 32 27 15 Thailand 20 6 15 78 1 18 5 27 33 23 Regional average Including I & 5 26 28 51 17 4 18 10 31 37 31 Excluding I & 5 21 23 53 19 5 - - - - - Sources: see appendix B. al Data for primary education not shown as most enrollments are in the public sector in all countries. The rate of cost recovery for public education is also very limited. b/ Figure refers to 1980 and is excluded from calculating the average in this column. oI Index reflects the rate of cost recovery across all institutional types weighted by their share of enrollments. For Malaysia, it reflects the extent of cost recovery only for domestic higher education. di See Chapter 5 for details en the calculation of these indices. At the secondary level, private schools enroll, on average, 26 percent of all students, comparable to the share of 29 percent in Latin American countries. The range of variation is very wide, however: countries - 20 - such as Bhutan, Burma, China, Malaysia, Nepal, and Sri Lanka rely little on the private sector, while others, such as Indonesia, Korea, the Philippines do the opposite. The private share of enrollments in Bangladesh and India is exceptionally high, but most private schools are heavily subsidized by the government, and are thus private mainly in terms of management. The institutional composition is more diverse in higher education, with at least four alternatives: private institutions, conventional public institutions, distance education, and overseas studies. In general, conventional public colleges and universities enroll the single largest share of students, but even then that share amounts, on average, to only 51 percent. Private institutions account for 28 percent of enrollments, distance education for 17 percent and overseas studies for 4 percent. These averages mask significant differences among countries, however. In the group of countries that rely mostly on public sector provision (China, Malaysia, Nepal, Papua New Guinea, and Sri Lanka), two have developed a sizable distance system (China and Sri Lanka), while the others have adopted other strategies for satisfying the excess demand for higher education. Nepal relies to some extent on private education, while Malaysia opts to let foreign institutions satisfy it. Countries that place limited reliance on regular public institutions include Korea, the Philippines, and Thailand. The first two has a sizable private sector, while the third has a particularly large system of distance education consisting of two open universities. Arrangements in the financing of education also vary significantly across countries. The rate of cost recovery (fees as a share of unit operating costs) for public higher education is minimal in Bangladesh, India, Malaysia, Papua New Guinea, Sri Lanka, and Thailand. In contrast, it is substantial in Korea, and moderately high in Indonesia, and the Philippines. A noteworthy feature is that in many countries, the rate of cost recovery in secondary education exceeds that in higher education, a pattern that contradicts what might be expected on grounds of equity. The extent of private financing in the sector as a whole depends on the rate of cost recovery across the various types of institutions. The required data are scanty, however, and have been pieced together only for higher education. 2A/ An index of private financing is constructed by weighting the rate of cost recovery in public higher education, private higher education, and distance education by the corresponding enrollment shares; the result appears in column 8 of table 1.5. According to this indicator, four groups of countries emerge. China, India, and Papua New Guinea form one group, where private financing is extremely limited. Indonesia, Korea, and the Philippines form a second group with the opposite characteristic. In Nepal, Sri Lanka, and Thailand the extent of private financing is moderately J./ Note that on efficiency and equity grounds, the case for private financing is probably strongest at the higher education level. - 21 - high. The fourth group consists of Bangladesh and Malaysia, with a somewhat below average rate of private financing. 15 A final aspect worth presenting here is the impact of current financing arrangements on equity in the distribution of public spending on education. Among the various approaches for making an assessment, the one adopted in this study is motivated by the observation that peeple benefit from public spending on education as long as they remain in the education system, and that those benefits accumulate over time, and are largest for people with the longest schooling careers. The focus therefore is on the distribution of cumulative benefits, rather than on single period benefits at a given level of education. To simplify the evaluations, two indices -- the Gini-coefficient and the share of cumulative spending received by the 10 percent best educated people in a generation (columns 9 and 10 of table 1.5) -- were constructed using data on enrollment ratios, unit costs, and rate of private financing across the three levels of education. The ranking of countries is similar whichever index is used. For Asia as a whole, the top 10 percent by education in a cohort receive 37 percent of cumulative government spending on education. In Indonesia, Korea, the Philippines, and Thailand the distribution is much more equitable than is average for the region. In contrast, public spending on education in Bangladesh, India, Nepal, and Papua New Guinea, is highly concentrated in the hands of the lucky few who survive longest in the system. In Bangladesh, for example, the top 10 percent receive an astonishing 76 percent of the cumulative government spending on education. It is of interest to note in passing that public spending appears to be more equitably distributed in countries with significant levels of private financing in higher education, such as Indonesia, Korea, and the Philippines. This finding thus casts some doubt on the commonly cited argument that private financing in education affects equity adversely, and suggests that its application at the higher level may in fact yield the opposite result. This issue will be taken up again in the discussion below. (b) Relationships Between Variables and Their Poligy Implications Beyond the simple statistical description summarized above, some general lessons about educational policies also emerge from the study, stemming from the relationships (or lack thereof) among indicators of 11/ Note that for Malaysia, the index would have been significantly larger had overseas education been included in its computation: the outflow of students abroad is large, and most of the students overseas finance their studies privately. However, for the comparisons here, the lower figure is appropriate as it describes more accurately the extent to which private resources are tapped for the financing of domestic institutions. - 22 - educational outcomes and policy regimes across countries. 1k/ Four main ones are recalled here. (i) Correlates of overall educational development. Regression analysis indicates that while demographic pressures exert a negative impact on educational development, as proxied by the average grade attainment of the current school-age population, a country's wealth and the overall level of public spending on education exert only a weak influence. This result, although seemingly counter intuitive, actually has a sensible explanation. It arises because countries vary widely in the way education is organized and financed, with corresponding differences in the education system's efficiency. 1Z/ Thus, while countries with higher per capita GNP tend to succeed better -- possibly due to the existence of a stronger administrative and institutional infrastructure -- their achievement can sometimes be matched by countries with a smaller per capita GNP. This result signifies that while favorable external conditions may facilitate human capital formation, they do not substitute for appropriate policy choices. A similar explanation applies with regard to the impact of aggregate government spending. Note that its weak influence in this cross-sectional analysis does not contradict the fact that in any single country, the more the government spends, while keeping constant the education system's operational characteristics, the higher the grade attainment or volume of human capital formed. For illustration, the relationship (or lack thereof) between average grade attainment and aggregate public spending on education appears in figure 1.2 (2.2). Countries like Bhutan (2), China (4), India (5), Indonesia (6), Korea (7), Sri Lanka (14) and Thailand (15) have comparable levels of spending, but achieve dramatically different results in terms of human capital formation. Conversely, countries such as Bangladesh (1), China (4), India (5), Nepal (10), and Papua New Guinea (11) have widely different levels of aggregate spending, but comparable outcomes in terms of average grade attainment. IV Individual countries are treated as single units of observations in the analyses. 12/ Differences in those choices are reflected by such indicators as the profile of teacher qualification and pay, pupil-teacher ratios, distribution of class sizes, use of distance education, extent of cost recovery and private financing, and so on. - 23 - Aierage grade attainment * 7 10 13 314 39 8 6. 66 6 -4 4 4 * 2- 0 0 1 2 3 4 5 6 7 Pablic spending on education 1% in GIPI soaes see AlpeIt I. Piure 1.2 (2.2)s Relationship between aggregate public spending on education and average grade attaiament, Asia, mid-1980s. To summarize, the results suggest that educational development clearly depends on policy choices within the sector, even though demographic constraints remain a serious impediment. Of importance is the way the education system is organized and financed. These attributes affect the education system's efficiency, and therefore its costliness, which in turn determines the volume of human capital formation achievable for a given constraint in the public budget. (ii) Correlates of educational costs. The unit costs of public education at all levels vary widely across countries; when expressed in per capita GNP terms, they are only mildly linked to the per capita GNP itself. Not surprisingly, the overall costliness of education, reflecting the average - 24 - of costs across the three levels, also show no systematic relationship to per capita GNP (figure 1.3 (3.2)). JI/ liaex of ovetall costiliess 1.6 1.4 . 1 7 1.25 1 - 10 5. * a 0.8 15i 0.6 -12 0.4 0.2 - 100 1000 10000 Per capita GNP (US$. log scalel PIG. &a estiI, is lot shovs. Soirce: see Aplelit 8. Fisure 1.3.(3.2s Ralationship between overall costliness of education and per capita GHP, Asia, circa 1985. The lack of a relationship arises from the diversity of choices concerning the components of educational costs. In primary and secondary education, the main elements are teachers' salaries and pupil-teacher ratios. In absolute terms, the former variable usually rises with per capita GNP, as market forces cause it to move in step with the rising pay of other productive labor. However, expressed as a share of the per capita GNP, teachers, pay are JI/ The lack of a link between costs expressed in per capita GNP terms to per capita GNP itself contradicts the popular belief that comparisons based on the indicator almost always identify low-income countries as high-cost countries, owing to the impact of the denominator. - 25 - unrelated to per capita GNP, indicating that a range of options exists concerning the pay of teachers relative to that of other productive labor. 12/ Similarly, no relationship is discernible between per capita GNP and pupil- teacher ratios, pointing to the variation among countries in the intensity with which teachers' time is used. To illustrate the diversity in these two components of costs, the data relating to secondary education are plotted in figures 1.4 (3.6) and 1.5 (3.5) below; those for primary education show a similar pattern, and are not presented here. 12 Avq. teachet pay (ratio to GNP/capita) 1211 10 - * 8 - 6 .o 7 4 -46 5. * , i5 *.9 212 2it 0 ' a ma ' ' ' ' a a 100 1000 10000 Per capita GIP (US4. tog scale) Souts: see appeath U. Pilate 1.4 11,_6) a Rlationship between average secondary school teachers' reasaneration and per capita GHP, Asia, circa 1985. 12/ In view of the Senerally positive link between wages and a person's qualification, policies about teachers' pay are obviously related to decisions, implicit or otherwise, concerning the qualification of the teaching force. - 26 - PapIl-teacher ratio 40 35 - .72 30 - 10 3 25 - 1 " 5 *9 13 20 -* .15 15 - 6 10 - 5 - 0 ' ' ' ' ' ii ' I I ' I ' IJ 100 1000 10000 Per capita GIP (USC. log scale) SouS: $seoeppealls 0. Figure 1.5 (3.S)i Relationship between pupil-teacher ratio In secondary education and per capita GNP, Asia, circa 1985. In higher education, costs depend on more factors, including the distribution of enrollments across types of institutions (particularly distance versus conventional universities and colleges) and across fields of study, student-faculty ratios, and the relative pay of faculty. The available data relate only to the first three indicators. As in the case of primary and secondary education, the link between these component variables of costs and per capita GNP is weak, suggesting that they reflect choices rather than inescapable outcomes arising from particular country conditions. For illustration, figure 1.6 (3.7) shows the pattern for data on student-faculty ratios in conventional public institutions. * 27 s Sludeat-Iaculty ratio 7 40 - 30 - 20 - o5 6 100 1000 10000 Per capita GI (US$, log scale) SoUcus: ase table 38.1. Fiate 1.6 (3,s7: Relationship between student-faculty ratio in conventional public higher education and per capita GOP, Asia, circa 1985. The above results suggest that at all levels of education, costs are determined to a significant extent by policy choices about the way education is organized. The lack of a strong relationship between the variables that determine costs, such as the relative pay of teachers and pupil-teacher ratios, implies that the full range of options in these variables is, A priori, available to all countries, regardless of their level of economic development. This finding calls for a greater willingness to explore and adopt, at least in the context of long term planning, desirable shifts in policies that are sometimes never even considered, the quick assumption being that domestic conditions do not support those shifts. - 28 - (iii) The isoact of private financing in higher education Private financing in education is generally achieved through the imposition of user charges for public education, and/or by permitting fee- charging private schools to exist. 22/ Several theoretical considerations suggest that such a policy is most relevant to consider at the higher level: it is probably more feasible to implement in administrative terms; its potentially adverse impact on equity within the subsector is blunted by the fact that most students at this level come from relatively advantaged social backgrounds; and most of the benefits of higher education, particularly for undergraduate studies, are privately captured by students. Beyond these arguments, the practical usefulness of the policy depends on its actual impact, and the nature of the tradeoffs, if any, between its positive and negative effects. Z1/ The country-level data assembled in this study permit a partial assessment of these issues. The first noteworthy feature is that although the degree of private financing tends to rise with per capita GNP, the correlation is quite weak (figure 1.7 (3.8)). Thus, countries with comparable per capita GNPs, such as Indonesia, Papua New Guinea, the Philippines, and Sri Lanka, vary widely in the level of private financing, ranging from 6 percent in Papua New Guinea to 86 percent in the Philippines. On the other hand, Bangladesh and Malaysia share similar rates of private financing, but differ significantly in per capita GNP. The wide variation suggests that while a country's level of economic development affects the administrative and social feasibility of tapping private resources for education, there remains substantial scope for such a policy in all country settings. As indicated before, the options for implementing it involve charging fees in public institutions, and/or allowing the private sector to develop. 2&/ Private schools may of course be partly subsidized by the government. As long as the extent of subsidization is limited, such schools help to mobilize private resources for education. 21/ On the positive side, private financing in higher education is likely to boost efficiency -- therefore reducing costs -- by encouraging greater accountability from school managers, and increasing cost-consciousness among the various actors involved -- students, parents, teachers, school authorities, and so on. An added benefit is that it helps to mobilize resources for educational investments, thus easing constraints in the public budget, and permitting the supply of school places to expand. On the negative side, such a policy may exacerbate biases against low- income groups, for by adding to the cost of forgone earnings or family production, fees could create an insurmountable barrier to higher education for such groups. - 29 - t ate1 of private financing 100- a12 80 - 7 60 - 6. 40 - 10 14 15 20 01 * 100 1000 10000 Per capita GNI (IS$. log scale) Figure 1.7 (3.8): Relationship between per capita GNP and degree of private fimancing in higher education, Asia, mid-1980s. The second finding concerns the impact of private financing on the efficiency of public higher education. The argument is that such a policy creates incentives for cost containment by promoting competition with the private sector, and sharpening cost-consciousness among students and school managers. The data indeed reveal a pattern of declining costliness of regular public higher education with rising rates of private financing in the subsector as a whole (figure 1.8 (4.5)). Note, however, that the relationship flattens out as the rate of private financing goes beyond 40 percent, suggesting that the gains in terms of cost containment are minimal beyond this point. Thus, while private financing is clearly a potentially useful instrument for promoting efficiency, it is not essential to advocate full cost recovery, nor indeed very high rates of cost recovery, to maximize its impact in this respect. - 30 - Costlifess of public hi9er education 21 0 L 0 20 40 60 80 100 Index of private finlancing (S) Seeghters 3 & 4 foi details ona seas;rao of ladices on both aies. Fiaure 1.8 (4.5): Relationship between the costliness of conventional publitc higher education and overall extent of private financing in the subsector, Asia, cirea 1985. With regard to the policy's impact on equity, the results are generally favorable. The data indicate a positive correlation between private financing and coverage in higher education, and with the share of primary education in total government spending on the sector (controlling for the size of overall spending). For illustration, the first relationship is plotted in figure 1.9 (5.7) . The results are intuitively appealing: by supplementing public funds, private financing in higher education enables an expansion of coverage if some of the resources so mobilized are retained in the subsector, particularly when there is excess demand for education at this level (which happens to be the case in the Asian context) . The availability of those resources also permits a partial diversion of govertment spending to the lower levels of education. - 31 - Gross enrollment ratio [%) 40- . 12 30 -7 20 - * 10.14 .10 all 0 1 I 0 20 40 60 80 100 Index of private financing Fiaure 1,9 (5.71t Relationship between overall extent of private financing in higher education and enrollmnt ratios in higher education, Asia, circa 1985. The positive impact on equity of private financing in higher education is further supported by its negative link to the share of cumulative public spending received by the 10 percent best educated people in a generation (figure 1.10 (5.6)). Part of this pattern is due to the fact that (a) cost recovery is more extensive in the wealthier countries in the sample; and (b) their education systems are generally more well-developed at all levels. However, the relationship remains after these considerations have been taken into account. As before, note frou the figure that diminishing returns set in with rising rates of private financing. Thus, a similar conclusion emerges concerning the actual design of policies, namely that very high levels of cost recovery are not necessary to maximize the impact on global equity in the system. - 32 - 80 Top IO's shate of cumalative resources 60 - 40 1 1 5" . 20 - 6 2. 7 0 I I I I I 0 20 40 60 80 100 Index of pilvate financing ouises see Appesali C. Fizure 1.10 (5.6)s Relationship between overall extent of private f£asacing in higher education and share of cumulative public spending on education received by top 10 1 by education, Asia, circa 1985. (iv) Cohort survival patterns in primary education Low cohort survival rates in primary schooling is a major policy issue in many Asian countries. In general, the rates improve with rising per capita GNP, but beyond US$800 (in 1985 prices), the relationship levels off at a reasonably high rate of survival (figure 1.11 (4.1)). There remains substantial variation around the average trend, however. Thus, countries with similar per capita GNP achieve widely different rates of cohort survival; for example, the rate is 85 percent in Sri Lanka, compared to 40 percent in Laos; and 68 percent in China compared to 37 percent in Ind-*a. These comparisons suggest that the wealth of a country is not the sole factor responsible for - 33 - low retention rates, and that policies in primary education can probably also make a difference. Salvival rate () £3 80 4, 5... ...- - . 4 ,1 60 / 6* 40 - 5 910 a 8 20 Al 100 1000 10000 Pet capita GIP (IS$. log scale) Souce: see Apperldx C. Etnre 1.11 (4.11 : Relationship between survival rates in primary education and per capita GNP, Asia, circa 1985. One reason for concern about low retention rates is that people who drop out prematurely are less likely to acquire permanent literacy. The implication is a low rate of literacy among future adults, with adverse effects on economic development. Regression analysis based on a large sample of developing countries indeed suggest a statistically significant and negative relationship between adult literacy rates at one point in time, and the rate of retention in primary education 10 to 15 years earlier. This relationship is fairly strong: a 25 percent increase in cohort survival rates (a magnitude of improvement that most countries in the region would probably need to consider) results in a 7.5 percentage point increase in literacy rates among future adults. Thus, if achieving universal adult literacy is envisioned, strengthening the holding power of primary education is clearly an indispensable component of public policy. - 34 - Low cohort survival rates also raise concern because of their adverse impact on social selectivity. 22/ The reason is that dropping out occurs more often among pupils from disadvantaged groups. The data in this study permits an evaluation of the link between cohort survival rates and females' participation in education. In general, their share of enrollments at both the primary and secondary levels is negatively related to survival rates in primary education; for reasons of space, only the relationship for primary education is shown in figure 1.12 (5.11). The pattern indicates that when an education system's capacity to retain students is weak, it takes a greater toll on females than males. Note that when survival rates are below 50 percent, the sex bias in participation rates is particularly pronounced. % females In primary education 50 - 12 7.9 *013 45 - 40 Di 35 2a 30 1 25 20 ' 0 10 20 30 40 50 60 70 80 90 100 Survival lale In piiary cycle 1%) * f oners to S st year pilnary eatrants sUrTvist9 to eia t fte cycle. SOtes: see Appeix C. Fian= 1.12 M1)la Relationship between female share of enrollents in primary education and cohort survival rates in primary education, Asia, oea 1985. 22/ Note that this adverse result tends to be perpetuated at subsequent levels of education; in fact it often worsens because dropout rates tend to be higher among students from low-income groups. 35 - These results are relevant in considering policies to promote females' participation in education. A basic question is: when should interventions to improve survival rates be targeted to females, and when should they be aimed more generally at the system as a whole? The former approach is warranted mostly when the bias against females is substantial. However, this situation occurs precisely when the system's retention rate is excessively low, say, less than 50 percent. At such levels, internal efficiency is so poor that it calls in fact for intervention to improve overall survival rates. Females would benefit indirectly in view of the general pattern of rising participation rates as the retention rate improves. Above the 50 percent threshold, and clearly when survival rates reach 75 or 80 percent, improvement remains an important objective of policies in primary education, but targeting interventions toward females is generally not warranted since their representation in system is close to 50 percent under these conditions. With regard to interventions to raise overall cohort survival rates, a basic question is whether or not more money would fetch better outcomes. The data for Asia suggest that countries with higher levels of spending per pupil generally achieve better cohort survival rates (figure 1.13 (4.2)). But wide variation remains around the average pattern, implying that while more resources per pupil is likely, on average, to improve a system's retaining capacity, it does not in itself insure against high dropout rates. 120 Survivalrate (s) £20 80 4 1t .12 60 *6 40 -. to. 20 81 0 L 0 5 10 is 20 25 30 Unit costs (as S of per capita GNP) soureest ses hopsall D. Fiauve 1.13 (4.21s Relationship betwoen survival rates ari unit operatiag costs in primary education, Asia, circa 1985. - 36 - Studies elsewhere show that the way resources are spent, and the system organized are also crucially relevant. The aggregate nature of the available data does not permit specific interventions to be identified. 23/ It is nevertheless noteworthy that survival rates show only a weakly negative link with pupil-teacher ratios (figure 1.14 (4.3)). Cohort survival rate (%) 1.20 13 100 - . 9 * 7 14 80 -* .4 *11 60 -6 40 -** 20 - 0 p p 10 20 30 40 50 60 Pupil-teacher ratio 884tte! sIB Ap6td38 3. Fimure 1.14 (4.3): Relationship between cohort survival rates and pupil-teacher ratios in primary fchoolLng, Asia, circa 1985. It appears that raising per-pupil spending in the form of reduced pupil-teacher ratios is probably relevant mainly in such countries as Bangladesh and India where retention rates are extremely low and pupil-teacher 22/ Note that the cost-effectiveness of alternative interventions is likely to depend closely on local conditions. There is therefore no reason to expect that low cohort survival rates can be addressed via the same interventions across countries, or indeed even across different regions within a single country. - 37 - ratios exceptionally large. A/ On the other hand, in countries where the average pupil-teacher ratio is not exceedingly large, say, not over 40, the indications are that such a strategy may not be the most appropriate one. These conclusions would clearly require further refinement in light of country-specific conditions, particularly because the data reflect averages that may conceal highly skewed distributions across localities. B.3 Implications for Policy Dialogue and Further Analysis The study's findings provide a useful beginning for policy dialogue concerning educational development in Asia. Data on the basic indicators facilitate assessment of individual countries in a regional perspective, thereby alerting policy makers to the relative strengths and weaknesses of their education systems. This information is especially relevant in shaping the broad objectives of future policies for the sector, and in assessing the range of feasible options for accomplishing them. The analysis also suggests several general assumptions that should underlie policy discussions in the sector: Substantial scope exists for interventions to improve efficiency and equity in Asian education. The conditions for such policy shifts are likely to be particularly favorable in the coming decade since the prospects for expanding coverage, and/or upgrading services are generally bright for most countries in the region owing to the projected slowdown in population growth and the relatively high rates of economic expansion envisaged. The choice of appropriate policies can make an appreciable difference to a country's performance in education, even though external constraints, especially demographic pressures, remain an important impediment. Poorer countries do not necessarily face a narrower range of potentially feasible policy options, despite weaker administrative structures and possibly more fragile social conditions. 2/ In Bangladesh, a system of double shift teaching is in effect whereby a teacher teaches two shifts of students. At the average student-staff ratio of 46 (a more recent estimate puts it as high as 59), the average class size is relatively low (and could arguably be raised). However, this arrangement also implies that students receive very few hours of actual teaching time, given that almost all schools operate a 10 period day (for about 5 hours), and teachers teach a maximum of four hours daily, a workload advocated by the Primary Teachers' Association. In this context, raising the pupil-teacher ratio would effectively mean increasing the contact teaching time that students receive. It does not mean a reduction in class size. - 38 - . Increasing aggregate public spending on education is an intuit"vely appealing but limited policy option for promoting educational development in most countries. One reason is that while appreciable results tend to materialize only with large increases in spending, such increases are nearly always difficult to provide -- except possibly in countries where current levels of spending are relatively small -- because of the keen intersectoral competition for resources. A more important reason is that increased resources cannot overcome the effects of inappropriate policies within the sector, particularly those affecting the efficiency with which services are provided and financed. . To the extent that poverty alleviation through education is a social objective, two interrelated components of educational policies are essential: increased focus on primary education to improve the rate of cohort survival; and a concomitant reduction in the public financing of higher education (through user charges in regular public institutions, the promotion of a largely self-financing private sector driven by excess demand, and/or use of low-cost distance systems). . Increasing the extent of private financing for higher education, up to a moderately high level, is likely to promote greater efficiency in public institutions (by reducing its relative costliness), and to improve equity, in terms of both the distribution of aggregate public spending on education, and expanded coverage in higher education. Beyond these broad conclusions, based largely on aggregate country- level data, the study also suggests two issues that deserve priority on the agenda for further analytical work. D/ The first is to identify factors that account for the low cohort survival rates in many Asian education systems, taking care to distinguish between those that are open to direct policy intervention (such as conditions of schooling) and those that must be taken as given in the short and medium term (such as poverty-related factors). Further analysis is also needed concerning higher education because of its vital importance in overall sectoral policy. Given the various alternatives for providing services, a central issue is the optimal mix of institutional arrangements across fields of study. Addressing this issue would require assessment of the labor market performance of people who have followed different careers in higher education. Both topics for further study call for the collection and analysis of individual-level survey data. This investment in further analytical work is probably worthwhile as it would strengthen the factual basis for policy dialogue on issues where existing knowledge is inadequate or non-existent. 2$/ There are of course other worthwhile research topics, including in particular, the external efficiency of education by level and type, the appropriate regulatory framework for private education and decentralized control of schools, the cost and benefits of overseas higher education, and local centers of excellence, and so on. - 39 - 2. Educational Development in Asia: Some Basic Characteristics The educational development of countries can be compared along various facets. Consider here such indicators as the schooling attainment of the adult population -- which encapsulates past investments in education --; the present structure of enrollments; the level of government spending on education; and the extent of private involvement in the sector. In what respects does Asia differ from other world regions? How do individual Asian countries compare with one another? What factors account for the differences among countries? 2.1 International comparisons Education systems throughout the world grew rapidly during the sixties and seventies. The fruits of that expansion are revealed by the extent of literacy among today's adults, and in their educational attainment (table 2.1). In Asia the adult literacy rate in 1985 was about 65 percent, slightly higher than the average of about 59 percent for developing countries as a whole, but significantly below the average of 80 percent for Latin America. In terms of the second indicator, the average Asian adult has 5.3 years of schooling,1/ compared to 5.6 years for Latin Americans. The difference is smaller than one might have expected, given the sizable gap in literacy rates. The reason might be that literacy is defined using stricter criteria for Asia; more likely, however, is that the average-years-of- schooling statistic refers mostly to Asian countries with better developed education systems.2/ Note that even in such countries, women receive significantly less education than men, 4.7 years compared to 5.9 years. The gap is much wider than in Latin America, but smaller than in the EMENA region. 1/ In fact, it is more accurate to refer to grades rather than years of schooling, since allowances for repetition are made in reckoning the educational attainment of the adult population. 2/ The nine Asian countries in the sample are Bangladesh, China, Hong Kong, Indonesia, Korea, Malaysia, Philippines, Thailand and Sri Lanka. - 40 - TabeS.it Educational attainment of adults, world regions, 1980s Averase ears of schoolina 1980s Soth X literate. 1985 Males iaeA AseA Asia 64.6 (14) 5.9 4,7 5.3 (9) Africa 47.8 (20) 3.5 2 6 3.1 (12) EMA 48.2 (11) 5.1 3.7 4.4 (12) Latin America 80.2 (12) 5.7 5.4 5.6 (16) Developing countries 58.6 (50) 5.2 4.3 4.8 (56) Developed countries - - 10.3 9.7 10.0 (12) Sourcess Data on percent literate are from UNICEP (1987)1 data on the average years of schooling of adults are from Norn and Arriagada (1986). Among the current population of children and youths, participation in schooling is quite widespread throughout the developing world (table 2.2). In 1985, the average gross enrollment ratio was 90 percent, 37 percent and 10 percent respectively in primary, secondary and higher education for developing countries as a group.,/ Asian countries have, on average, somewhat lower enrollment ratios than those in Latin America and EMENA, but if differences in level of economic development, as indicated by per capita GNPs, are taken into account, they actually achieve better-than-expected coverage (figure 2.1). 3_1 The gross enrollment ratio is defined as the ratio of the number of students enrolled at a given level of education to the population in the corresponding age group. This statistic can sometimes exceed 100 percent because of the presence of overage (or more rarely, underage) students. When the numerator includes only students in the right age range, the result is known as the not enrollment ratio. Data on this statistic are generally more scanty, and are therefore not used here. * 41 - Table 2.2s Enrollment ratios, world regions, 1983 0 countries gpIM SM ar isher mtina Asia 92.0 42.9 11.1 16 ifrioa 77.8 19.5 1.5 30 SEBA 92.9 46.9 14.2 16 Latin America 101.8 468.5 16.5 21 Developtns countries 89.5 36.7 9.5 83 Developed countries 102.7 85.9 29.6 16 Source World Bank DISD database. Gross eaollueat ratio 10 10- Priary 80 - seconfary / - - HIgliet 60 x Asia 0 Africa 40 x . * Latin America 20 - 0 100 1000 10000 Logarithi of ler capita GIP Pet capita GI Is is US$. Sol1ce: see appendix A. IMajIALs Relationship between the gross enrollment ratios and per capita GNP, major world regions, circa 1985. - 42 - The achievement of Asian education becomes all the more remarkable when levels of government spending on education are compared (table 2.3). Expressed as a percentage of the GNP, it is lowest in Asia among all world regions. This apparent paradox -- high coverage despite relatively little "fiscal effort" - - provides a first indication that as a determinant of educational development, public policies in the sector are at least as important as the size of public spending. This idea is in fact quite intuitive, since government policies can affect the extent of private involvement in providing and financing education, as well as the efficiency with which public funds are spent. Both outcomes influence the effective amount of resources available to the sector. Table 2.3: Public spendias on education, world regions, 1985 * countries As I.. o. . re"nrt.M Asia a/ 3.1 16 Africa 4.1 32 meNA 5.3 18 Latin America 3.5 21 Developing countries 4.0 91 Developed countries 5.7 21 Soures UNESCO (1987). al Figure differs slightly from data in tables 1.1 and 2.8 due to differences in the sources of data. Data on the share of private enrollments appear in table 2.4. In Asia, that share is not as large as in Latin America, averaging 20 percent across the three levels of education, compared to 27 percent. Note, however, that it rises from primary to higher education much more rapidly in Asia than in Latin America. In other words, in providing education directly, Asian governments tend to concentrate more on the lower levels of education. On the extent of private financing, it is stressed that the data here serve only as a rough guide, since no distinction is made between government-aided schools, and those that rely mainly on private contributions. - 43 - Table 2.4s Share of enrollments in private sector, world regions, 1985 Share of orivate enrollment ) # countries rmnorting ftaem seaonw hLIft EIasE .&VSABM iaiO Asia a1 3.9 26.0 28.6 10 9 13 Africa 15.4 26.4 - 32 31 - ENNA 8.8 8.1 3.3 15 17 15 Latin America 17.7 29.1 33.6 19 19 7 Developing countries 13.1 21.8 17.0 77 77 31 Developed countries 17.3 19.3 - 20 20 - Sources data supplied by UNESCO, supplemented for Asia, by sources cited in appendix table B1.4; ENA (higher education) by Za'rour (1988); and for Latin America (higher education), by Vinkler (1988). a/ The regional average for higher education differs from that in table 1.5 due to ommission of Laos in that table. See also table 2.9. 2.2 Variation across Asian countries Asia encompasses a great diversity of countries, including some of the more advanced developing economies in the world, as well as some of the poorest. This diversity is reflected in the educational characteristics of individual countries in the region. Data on the attainment of adults appear in table 2.5. In 1985, literacy rates ranged from only 15 percent in Bhutan to over 90 percent in Korea and Thailand. Over time, this indicator improved rapidly for countries with low initial rates of adult literacy, such as Bangladesh, and Nepal. However, the improvement has also been dramatic in countries starting from a higher base, such as India, Indonesia, Malaysia, Papua New Guinea, and Singapore. Adults in Korea and the Philippines have significantly more years of schooling than their peers elsewhere in Asia, 8.4 and 6.9 years respectively. In all Asian countries except Indonesia, the Philippines and Sri Lanka, men receive at least one more year of schooling than women. Since 1970, there have been significant gains in the coverage of education systems in Asia (table 2.6), so that by 1985, most countries in the region have gross enrollment ratios in primary education that exceed, or are approaching 100 percent. The exceptions are Bangladesh, Bhutan, Nepal and Papua New Guinea, where despite past growth, coverage still lags behind the rest of Asia. Bhutan belongs in its own category, however, with a gross enrollment ratio of only 25 percent in 1985, compared to the range of 60-82 percent in the other three countries. The expansion of coverage between 1970 and 1985 has been particularly remarkable in post-primary education; on average, the gross enrollment ratio for secondary education jumped from 26 percent in 1970 to 42 percent in 1985; in higher education, it grew from 5 to 10 percent in the same period. - 44 - Uble 2,3s Bducational attaitamt of adits in Asia, 1970-1980s 1 adulto literate Averase vears of schoolina. 1980s LIM I=A ingrea** maesA YGM2S Ovral I Bangladesh 23 33 43 2.6 1.0 1.8 2 Shutan - 15 - - - - 3 uma 71 - * - - - 4 China- 69 - 6.1 3.8 5.0 S ladia 34 43 26 - - - 6 Iadonesia 54 74 37 4.9 4.3 4.6 7 Korea 88 92 5 9.5 7.3 8.4 8 Laos fI ss 84 154 - - - 9 Malaysia 60 74 23 6.5 4.9 5.7 10 Nepal 14 26 86 - - - 11 Papua New Guinea 32 45 41 - - - 12 Philippines 82 86 S 7.0 6.9 6.9 13 Singapore 69 86 25 6.0 4.9 5.5 14 Sri Lanka 77 87 13 6.1 5.5 5.8 15 Thailand 79 91 15 4.6 3.8 4.2 Sources: As in table 2.1. a/ The exceptionally large ancrease between the two years suggests probable inaccuracy in the data. No country in Asia spends more than 20 percent of total government expenditure on education (table 2.7). However, the pattern of allocation varies widely, ranging from a low of 7.3 percent in Bhutan to 19.4 percent in Thailand. Over time, the share of education has, on average, dropped slightly, from 13.4 percent in 1975 to 12.5 percent in 1985. In most countries, the trend is relatively stable; the exceptions are China, where the share of education increased significantly, and Sri Lanka, where the opposite is true. In both countries, the changes have occurred against a backdrop of unique circumstances.4 In general, therefore, the education sector's claim on government spending neither expands nor diminishes dramatically as a result of the intersectoral competition for public resources within each country. Across countries, however, there are striking differences as to sectoral priorities. In China, the education system was being rehabilitated after the Cultural revolution, while in Sri Lanka, civil unrest probably caused a diversion of public spending for military purposes. - 45 - Tb1l 2.61 Gross enrollent rattos (Z) by level of ed~Cation, A91a, 1970-85 Pr i iar meodr i =e d U9212 22t 1922 1251 J2Z1 ami dl 1 8angladesh 54 60 - 18 - 5.2 2 Bhutan 6 25 1 4 - 0.1 3 Bu 83 107 21 23 2.1 5.4 4 rhia 89 118 24 39 0.6 1.7 5 India 73 92 26 41 8.6 9.0 6 Indonesia 80 118 16 42 2.4 6.5 7 Korea at 103 96 42 75 10.3 31.6 8 Laos 53 94 3 19 - 1.5 9 alaysia 87 99 34 53 2.8 6.0 (8.6) 10 epal bl 22 82 10 25 2.3 4.6 11 Papua Nev Guinea 52 70 8 13 2.5 2.0 12 Philippines 108 106 46 65 18.4 38.0 13 Singapore 105 115 46 71 9.0 11.8 14 Sri Lanka 99 103 47 63 1.3 4.6 (5.1) 15 Thailand 83 97 17 30 3.4 19.6 Trend average .4 76 94 26 42 4.9 10.1 Sources: see appna table B1.2. al The data for secondary education in 1985 is lover than that reported in 1MLICO (1987) since it is the average over both subayoleo of secondary education. The URESCO data refer only to the firat suboyele. bl The statiatic for primary education for 1985 may be overestimated due to inaccuracies in official estiates of the relevant school-age populations see Smith <1988) for furtber details. cl Excludes Bangladesh, Bhutan, and Laos sine data for these countries are incomplete. As a result of this exclusion, the averages for the region may differ slightly from the figures for Asia in table 2.2. dl Pigures in parenthewes refer to the estibated enroll~ent ratio if student# broad are inoluded. They are shown only for Malaysia and Sri -anka, countries with sisable student populations abroadt see also table 3.7 belov. - 46 - Table 2.7; Public spending on education as a percentage of total goverrment spending Latest dat_ I2m2 1.1 AlEf 129.1 ... Iss 1 Bangladesh - 11.8 8.5 10.3 11.3 1988 2 Bhutan - - - 7.3 8.6 1987 3 Burma 11.3 14.1 10.1 10.9 - 4 China 2.9 4.2 6.1 7.8 - 5 India - 14.5 14.5 13.7 13.8 1988 6 Indonesia - - - 15.0 - 7 Rorea - 13.1 14.6 16.6 16.7 1987 9 Malaysia 19.4 13.2 16.0 18.5 1987 10 Nepal - 12.0 9.5 9.6 10.4 1988 11 Papua New Guinea - - - 17.9 15.4 1988 12 Philippines - - 11.1 11.5 14 Sri Lanka 14.4 11.0 7.8 8.1 7.3 1988 15 Thailand - 20.1 19.8 19.4 19.1 1986 Trend average a/ - 13.4 11.6 12.5 Sourcess See appendix table 33.9. at The trend average reflects the average level of spending of the nine countries for which data are available for 1970, 1960 and 1985. To judge the real level of public spending on education, one can express it in relation to a country's GNP (table 2.8). Since total government spending has been rising as a proportion of GNP, public spending on education as a share of the GNP has also been rising, even though its share in total government spending has declined slightly. Bangladesh and Philippines are at the low end of the spectrum in this statistic, while Malaysia and Papua New Guinea occupy the high end. It is interesting to note that very different outcomes in educational development are possible with comparable levels of government spending. Thus, enrollment ratios in the Philippines far exceed those in Bangladesh, and enrollment ratios in Malaysia exceed those in Papua New Guinea. This result signals the strong influence of policy choices within an overall public resource constraint. - 47 - aklA,: Public spending on education as a percentage of the GNP Latest data 12a 192Z 1t1 191 ..L XISU 1 Bangladesh - 1.1 1.3 1.5 1.9 1988 2 Bhutan - - - 3.8 4.0 1986 3 BuSma - 1.7 1.3 1.8 - 4 Chin 1.5 2.1 2.9 3.3 5 India - 2.7 2.7 3.0 3.3 1988 6 Indonesia .- - - 3.7 - 7 Korsea 2.9 2.2 3.1 3.4 3.1 1987 9 Malaysia - 6.1 5.4 6.0 7.3 1987 10 Nepal - 1.4 1.4 1.8 2.1 1987 11 Papua Ne Guinea - * * 6.9 - 12 Philippines - - 1.7 1.8 2.8 1988 14 Sri Lanka 4.2 2.7 2.8 2.8 2.2 1988 15 Thailand 3.5 3.9 3.7 3.6 - Trend average a/ - 2.7 2.7 3.0 - Souroess see appendix table 33.11. al the trend average reflects the average level of spending in the nine countries for which data are available for 1975, 1980 and 1985. The regional average for all 13 countries in 1985 was 3.3 percent. One such policy choice concerns the role of the private sector in providing and financing education. Data on the share of private enrollments appear in table 2.9. As indicated earlier, it is meaningful to distinguish between aided and unaided private schools. In Bangladesh, for example, the government provides subsidies for 70 percent of the cost of teachers' salaries in private secondary schools (World Bank-Bangladesh, 1988b; King, 1988); similarly, private higher education also enjoys very large subsidies (Patwari, 1987). As a result, that sector is "private" mostly in the sense of being privately-managed. In contrast, private schools in the Philippines receive little or no public subsidies, and operate with a minimum of government intervention. This difference is a possible explanation why, despite comparable levels of government spending on education, coverage is so much wider in the Philippines than in Bangladesh. - 148- Table 2.9s The peretage abase of ealsmats i pstvte eduestles. Asia, 1970-1985 Primrs Seemedaww ._ Risher 1aI 1n1. nu au 198 A 1 AIR ASM 1 1 I Bangladesh * 4.1 14.6 11.0 * * 93.0 * * 58.7 2 Bhutan - - * * * * 0 StBurma - -** * ***-***0 4 China * - - 0.0 * * 0.0 * - 0 5 Indiaal * - 15.9 * 08.2 * * * 57.0 6 Indonetia - 15.0 10.0 8.0 49.1 49.S *8.$ 7 lorea 1.1 1.2. I.S. 1.5. 45.4 46.4 59.9 - - * 0.1 8 Laos 11.3 - 0.0 0.0 0.0 0.0 0.0 - - - * 9 Malaysia - - 0.1 * 1.7 - * * 1.0 10 Nepal - * * 5.3 - * 10.4 * * * 23.8 11 Papua New Guinea 63.0 - 2.0 0.5 - * 6.3 12 Philippines 4.9 5.3 5.2 6.0 54.7. 40.2 62.4, 69.6 6.2 64.0 63.2 14 Sri Lanka 7.3 6.0 1.3 1.4 * 2.3 2.4 - a * 0 15 Thailand 14.2 11.1 8.4 9.0 51.7 18.9 20.0 - 0.0 of $.1 6.4 Regional average b/ With 1& - - - 5.9 - 20.0 2 * 2.6 Vithout I & 5 - 3.2 * 18.5 28.2 Sourcess see appendix table 51.4. a/ Data for primary and secondary education refer to the abase in adi A aIded parvate schools as a poreetage of all schoolss figure for higher education refers to the abase of ps.atel-aMaged stitu"eas. b/ No overall trend average is caeculated here due to the small ae of oute for htch the seevent data are available. Recall that private education In Bangladesh. Iadia sad ZademsLts are subsidised by the goverment, the subsidies being particularly large In the tist two eesutles. of Figure refers to the share in 1977. In India, the financing arrangements with regard to private schools are similar to those of Bangladesh's. The share of private institutions is 16 percent in primary, 67 percent in secondary education (Government of India, 1980), and about 58 percent in higher education (Association of Commonwealth Universities, 1987). However, nearly all the schools receive public subsidies. Consequently, the extent of private financing is limited, even though many schools may be privately-managed. Among the remaining Asian countries, private education is well-developed in Korea and Indonesia.,/ In both countries, public subsidization of private institutions exists but is not as extensive as in India and Bangladesh. In Indonesia, for example, about 30 percent of the teachers in private higher education are seconded from government service. (World Bank, Indonesia-1988); and at the lower levels of education, private schools receive some subsidies (Thomas, 1987). In Korea, private institutions are mostly self-financing, as in the Philippines. 1/ In Thailand, the private sector also enrolls a significant proportion of students in regular institutions. Overall, however, its share is dwarfed by size of enrollments in the open universities. In Malaysia, government subsidies benefit only one private college, Tungku Abdul Rahman College. * 49 - 2.3 Sources of differences t4 educational development across countries Educational outcoames vary widely across countries, but so do constraints and policies. A question of interest is therefore the relative importance of these factors in determining outcomes. If they are primarily influenced by the system 'kconstraints, then little room exists for policy intervention, but if policy choices are also important, it becomes imperative to identify and promote those that lead to better outcomes. The constraints on educational development are determined largely by demographic and macroeconomic conditions./ Their impact relative to that of policies is not easily ascertained, but the data presented so far present a first opportunity to examine it. In subsequent chapters, this issue will be reexamined as new data are presented. (a) Demoraphic and economic constraints A useful indicator of demographic pressure is the dependency ratio, defined here simply ab the rati6 of 'the school-age population (aged 5-14) to the working population (males and females aged 15-65). The smaller this ratio is, the lighter the fiscal burdn on taxpayers, and therefore the less binding the constraint on educational development. The historical evolution and future projection of this variable appears in table 2.10. Demographic and macroeconomic conditions are assumed to be constraints rather than variables since the first is not easily changed in the short run and the second cannot be affected in a predictable fashion. - 50 - Table 2,101 Levels and trends in the dependency ratio, Asia, 1970-2000 Population aged 5-14 as percent of Percent change in nonulation aaed 15-65 al donendenow ratio i 1g Agg Agf Lg 1970-1985 1I9-2000 1 Bangladesh 55 56 46 49 42 -11 -14 2 Bhutan 46 47 44 44 47 -4 7 3 Burma 44 45 47 47 35 7 -26 4 China 44 39 44 33 26 -25 -21 5 India 49 48 45 44 35 -10 -20 6 Indonesia 50 50 48 46 35 -8 -24 7 Korea 52 43 37 31 24 -40 -23 8 Laos 45 46 60 53 49 16 -8 9 Malaysia 56 53 44 41 33 -27 -20 10 Nepal 46 48 49 49 51 7 4 11 Papua New Guinea 47 48 53 50 40 6 -20 12 Philippines 53 55 48 47 37 -11 -21 13 Singapore 48 37 30 26 20 -46 -23 14 Sri Lanka 49 47 42 36 31 -27 -14 15 Thailand 56 55 47 41 29 -27 -29 Regional average 49 48 46 42 36 -13 -17 Sources: Computed from BESD databaseg and Zechariah and Vu (1988). a/ For the present purposes, this statistic is the dependency ratio. On average, the dependency ratio in Asian countries declined from 49 in 1970 to 42 in 1985, and is projected to drop to 36 by the end of the century. The rate of decline has been sharpest since the 1980s. Across countries, differences exist in the variable's size and trend over time. In 1970, the school-age population in most Asian countries was close to half that of adults. By 1985, however, a more differentiated pattern had emerge: in one group of countries comprising China, Korea, Singapore and Sri Lanka, the ratio was at or below 36; in a second group comprising Bhutan, India, Malaysia, and Thailand, it was moderately high in the 36 to 44 range; in a third group comprising Bangladesh, Burma, Indonesia, Laos, Nepal, Papua New Guinea, and the Philippines, the ratio exceeded 44. These patterns imply sizable differences in the tax burden of financing education. For example, if China had Bangladesh's dependency ratio, the tax contribution of each Chinese adult would have had to rise by nearly 50 percent in order to finance the current level of coverage in the education system, assuming that all else in the system remains unchanged. With regard to the future, the fiscal burden on taxpayers will probably lighten in most Asian countries as the dependency ratio continues to drop. However, in a few countries, it remains high (Bangladesh, Laos, and Papua New Guinea, and may even rise (Bhutan and Nepal), implying that demographic pressures will remain a significant obstacle to educational development. - 51 - A second aspect of constraints on the education sector is the macroeconomic environment. In the last decade or so, most Asian economies have expanded in real terms at moderate to high rates. In every country in the region except Papua New Guinea, the rate of economic growth exceeded the growth rate of the school-age population (table 2.11). Thus, even with a constant share of education spending in GNP, resources were available for expansion (either in terms of coverage or additional inputs per pupil) beyond what was needed to maintain current enrollment ratios. In countries where education's share of GNP also rose at the same time, the pace of educational development was naturally even faster.2/ Table 2.11: Population and real economic growth rates (percent p.a.), Asia, 1975-2000 PoMMation aaed 5-14 Real economic Arowth L"5,*,5 1985-000 a1 197-8 1990-2000 a 1 Bangladesh 1.8 2.0 4.4 4.9 2 Bhutan 1.6 3.0 6.1 - 3 Buzma 2.4 1.0 5.8 3.5 4 China 0.2 0.4 7.8 6.6 5 India 1.6 1.3 4.4 4.8 6 Indonesia 1.8 1.0 6.1 3.9 7 oresa -0.8 0.3 7.4 6.8 9 Malaysia 0.8 1.4 6.3 5.0 10 Nepal 2.7 3.1 3.1 3.8 11 Papua New Guinea 2.6 2.0 1.5 5.1 12 Philippines 2.1 1.5 2.5 5.3 14 Sri Lanka 0.1 1.1 4.9 4.8 15 Thailand 0.6 0.5 5.8 6.0 Average 1.4 1.4 5.1 5.0 Sources: calculated from population and CNP data In MSD database (UNESCOMD and SOCIND) projected economic growth rates are from AMDREX database. a/ Projected. 2/ Note that this assessment is only a rough one, since no account is taken of the evolution of educational costs over time. The comparison between the economic and population growth rates nevertheless provides a sense of the budgetary constraints on educational development. * 52 - Most Asian economies are projected to grow at moderate to high rates in real terms, averaging about 5 percent a year between 1990 and 2000. In contrast, the school-age population is forecast to grow more slowlye.by an average of 1.4 percent a year. The macroeconomic constraint is therefore likely to ease considerably. But the outlook.is not uniformly bright in all countries. Nepal, in particular, faces much tighter constraints than other countries in the region because its economy is projected to grow only slightly faster than the school-age population; the scope for expanding education (in terms of coverage or increased inputs per pupil) under these conditions is therefore likely to be limited. Note, however, that even in countries with a favorable outlook, the size of the "extra" resources that will in fact materialize for expansion depends on how educational costs evolve with time and rising levels of per capita GNP. In most Asian countries, some of those resources are likely to be absorbed by the "natural" rise .in costs as a relatively young teaching force ages and advances up the wage ladder. Moreover, since costs will also depend on sectoral policies, the favorable macroeconomic outlook does not guarantee future progress in education, but only makes it more probable. (b) The imnact of constraints and sectoral Rolicies A preliminary assessment via regression analysis is possible here using the cross-sectional country-level data presented so far. For the present purposes, the dependent variable is educational outcomes proxied by the average grade attainment of the current school-age population. This statistic summarizes the structure and coverage of the education pyramid, as well as the length of study at each level.J/ As defined, it is a rough measure of the volume of human capital formation among the current generation of children and youths.2/ g/ It is calculated as follows. Suppose the enrollment ratio in primary, secondary and higher education is P, 8, and H respectively, and that these cycles of education last p, a and h years. Among a cohort of people of the same generation, some will have more education than others. In particular, if P is less than 100 percent, then (100-P) will have no education. At the end of (p+s+h) years, H will have attained higher education, (S-H) will have attained secondary education, and (P- S) will have attained primary education. Weighting this distribution by h, s, and p respectively yields the average years of educational attainment in that generation. In some countries, the primary enrollment ratio exceeds 100 percent, in which case the calculation assumes that P is 100 percent. Since repetition is not taken into account here, the result shows the average grade attainment, not the average years of schooling of the population. 2/ The average grade attainment is only a rough proxy since its relationship with educational costs is non-linear. However, this defect probably does not invalidate the results in table 2.12. . 53 - The independent variables in the regression include two proxies for the constraints on educational development: demographic pressure, simply represented by the dependency ratio, as defined above; and a country's wealth. A grgri, the larger the dependency ratio, the heavier the fiscal burden education places.on taxpayers, and the more difficult it is to achieve a high level of human capital formation. A possible proxy for the second constraint is a countryOs per capita GNP,Ig/ but because of the variable's strong correlation with the dependency ratio, it is not used directly. Used instead are the residuals (RGNP) from a regression linking the per capita GNP with the dependency ratio.11/ A qrioi, RGNP is likely to have a positive effect on the dependent variable. The reason is that, in relative terms, the prices of educational inputs tend to be more expensive in poor countries: some books and equipment are imported at international prices, and teachers are in shorter supply (and therefore earn larger salaries relative to the per capita GNP). Thus, it is as if poor countries face stricter economic constraints than rich countries,j2/ and so achieve lower levels of human capital formation. A third independent variable is the overall level of government spending on education, expressed as a percentage of the GNP. It is less rigid a constraint than the demogrphic pressures facing a country, or its level of wealth, since the appropriations to education are the result of policy decisions about intersectoral allocations of public funds. Moreover, those appropriations are not entizely independent of choices about the financing arrangements within the education sector. Thus, while short-run shifts in spending may be difficult to achieve, they are possible within a longer term perspective. As such, the level of public spending on education imposes a *fuzzy" rather than a strict constraint. Finally, another independent variable used in the regressions is the extent of private involvement in the education sector, proxied here simply as the average share of private enrollments across primary, secondary and higner education.1j/ 10/ The rate of economic growth relative to ti at of the school-age population is not a suitable proxy since all the other variables in the regression are single period indicators. 1],/ This procedure follows standard econometric method for addressing multicollinearity in regression analysis. 12/ Note, however, the prices of educational inputs are not entirely exogenous factors; teachers, for example, often are part of the civil service, and their pay is determined as part of government wage policies. As a result, this constraint may not be as binding as at first sight. J3/ Since the cost of education rises with level of education, it may appear appropriate to weight the shares of private enrollments accordingly. However, the independent variable proxies the overall extent of private involvement, which depends equally on the volume of enrollments. The latter is much large at the lower levels of education, and therefore . 54 - The regression results appear in table 2.12. The first equation is based on international data, and the second, on a subset of Asian and Latin American countries, for which data on private enrollment shares are available for all three levels of education. The coefficients on the first two variables -- dependency ratio and wealth (RGNP) -- have the expected signs, and are statistically significant (at the 5 percent level of confidence) in both regressions, signalling their importance as determinants of human capital formation in a country. Table 2.12s OLS regression results using grade attainment of current population as dependent variable Dependency ratio - 11.67 * -13.85 * (6.SS) (2.01) ROg 1.16 * 1.86 * (4.10) (2.49) Education share in GUP 0.23 * -0.17 (1.86) (0.37) Average private share of enrollments - 0.01 (0.30) Asia 1.47 * 0.34 (2.21) (0.37) a8RA 0.42 - (0.57) Latin Anerlca 1.83 * * (2.97) Developed countries 1.78 * - (1.76) Intercept 11.47 - 15.05 ' (7.52) (3.65) N 82 15 R-squared 0.66 0.48 Notes * denotes statistically significant at 5 1s and **, at 10 1 confidence levels. Variables definitions are as follows: the av# rage grade attainment is calculated by weighting the proportion of a cohort exiting the education system with primary, secondary and higher education, by the length of schooling at each levels the dependency ratio is the ratio of the population aged 5-14 to the population aged 15-64 OM is the residual from regressing the logarithm of the per capita GP against the dependeany ratios education share in GNP is public spending en education in 1985 expressed as a percentage of the GNPs average private share of enrollments is the simple average across primary, secondary and higher education. The remaining are regional duomy variables which take on the value of one if a country is in the region, and sero otherwise. The emitted region is Africa in regression Is and Latin America in regression II. partly offsets the effect of the rising cost pattern. - 55 - With regard to the overall level of government spending on education, its coefficient is positive and statistically significant (at the 10 percent level of confidence) in the first regression. To appreciate the magnitude of its effect, consider the following: an increase of one standard deviation in this variable, from the sample mean of 4.35 percent to 6.26 percent (representing an increase of 44 percent), raises the average grade attainment by only 0.44 years, up from a sample mean of 8.47 years. The increase represents merely 13.9 percent of the dependent variable's standard deviation. This calculation suggests that across countries, large differences in government spending on education account for relatively small disparities in the volume of human capital formation. The result arises partly because of wide variation in the way countries organize the provision of education (involving such choices as teacher qualifications and pay, pupil-teacher ratios, distribution of class size, and so on) and its financing. What countries achieve depends as much on policy choices on these matters as on the overall amount of public resources appropriated to the sector. Clearly, this result does not contradict the fact that in any single country, the more the government spends (while keeping its organizational setup the same), the higher the grade attainment. In the second regression, government spending on education is not a statistically significant determinant of grade attainment, which reinforces the findings discussed above. For illustration, figure 2.2 depicts the relationship (or lack thereof) between these two variables for the sample of Asian countries in this study. The regression also indicates that the average private share of enrollments exerts no influence on the level of human capital formation, contrary to initial expectations. Note, however, that countries differ widely in the extent of public subsidization of private schools, and the presence of cost recovery in public education. As a result, the variable used in the regression is, at best, a poor proxy for the overall extent of private financing of education. The size of the private sector is, in itself, not as important for human capital formation, as the policies -- such as public subsidies for private education, and fees for public education -- that affect the contribution of private financing ' r education, and therefore, the extent to which constraints in the public budget are softened. - 56 Averale grade attalument . 7 10 12" 14 8 6. 15 6 4 - . 10:'1 22 0 0 2 3 4 5 6 ? Pablic spendiag on educatioa t% ta GIP) Fales: s21 elatpas~ g. Flaue 2.2 Relatinna4ie bet.een aa6zgqate public spending on datLi and average grad. attainnent, Asia, .1*.a mid-1980s. . 57 - 3. The Costs of Education and its Financing: Further Comparative Statistics Variance in educational achievements stem in part from differences in policy choices which affect the costs and financing of education. Data are presented in this chapter to document the diversity among Asian countries in three aspects of those choices: the allocation of spending among levels of education, the pattern of unit costs, and the extent of private financing. 3.1 The intrasectoral allocation of public spending on education The pattern of allocation captures the combined effect of a range of polices in the education sector, including the structure of the enrollment pyramid, the public-private division in the financing and provision of education, the structure of unit costs, and so on. As such, it provides an early, if rough, description of a government's priorities in the sector. On average, countries in Asia allocate 48 percent of public spending to primary education, 31 percent to secondary education, and 19 percent to higher education (table 3.1). This declining pattern is similar to that in Latin America, where the corresponding shares of public spending are 51, 26 and 24 percent (World Bank, 1986). Table 3.1s Level and distribution of public spending on education, 1985 Distribution of public spendins Speft las by level of education overall govt. bv level of ednation (X) el as a moment of R (Z) spendlue on education l A= S aM& gA prmaEE second&" Higie I Bangladesh 1.5 49 34 15 41 2 0.7 0.5 0.2 4 China 3.3 41 42 16 0 1.3 1.4 0.6 5 India 3.0 27 47 19 6 0.8 1.4 0.6 6 Indonesia 3.7 62 27 9 2 2.5 1.0 0.3 7 Rerea 3.4 57 34 9 0 1.9 1.1 0.5 9 Malaysia 6.0 36 34 26 4 2.2 2.1 1.5 10 Nepal 1.8 41 21 5 S 0.7 0.4 0.6 11 Papua nov Guinea 6.9 45 18 28 10 5.1 1.2 1.9 12 Philippines 1.8 64 16 20 0 1.2 0.3 0.4 14 Sri Lanka 2.0 43 41 16 0 1.2 1.2 0.5 15 Thailand 3.6 58 24 12 6 2.1 0.8 0.4 Average 3.0 48 31 19 3 1.6 1.0 0.7 Sourcest see appendix table 35.8 and BS.11. al Expressed as a percentage of the GUP. b/ For India, the data en the distribution of spending refer to 1960. of Figures may not add up to 100 percent due to roundin errors. See also footnote AI In text. d/ Figure includes expenditure en universities, colleges polyteobates and teebutcl Institutes. - 58 - It is instructive to compare the pattern of allocation among countries in the region.j/ Those showing a definite emphasis on primary education include Indonesia, Korea, the Philippines, and Thailand, where the subsector's share of the budget for education ranges from 57 to 64 percent compared to the regional average of 48 percent. Except for the Philippines, the other three countries' overall level of education spending also exceeds the regional average. As a result, their fiscal "effort" in favor of primary education turns out to be exceptionally strong. As indicated before, the pattern of allocation reflects the combined impact of sectoral policies. For example, in the Philippines, the existence of a mostly self-financing private sector in higher education makes it possible for the government to allocate a large share of its limAted budget for primary education. This contrasts sharply with the pattern in Bangladesh where private institutions at the secondary and higher education receive substantial government subsidies. As a result, public spending on primary education, expressed as a percentage of GNP, in the Philippines is over 1.7 times as high as in Bangladesh. Differences in the structure of enrollments, pattern of unit costs and financing arrangements are other relevant factors to consider. For example, the share of secondary education is significantly lower in Papua New Guinea than in Malaysia (countries with comparable levels of public spending on education) since the system's coverage at this level is much smaller in the former country. On the other hand, in such countries as Indonesia, Korea and Thailand, the intrasectoral distribution of public spending on education is similar despite differences in enrollment structures. This result is partly due to differences in unit costs and extent of cost recovery in the sector. 3.2 Variation in the unit costs of public education 2/ Since the purpose is to compare costliness across levels of education and across countries, the appropriate statistic is the operating 1/ For completeness, table 3.1 show the distribution of spending across all three levels of education. Note, however, that in some countries, the classification of enrollments and expenditures at the secondary and higher levels is not always made on a consistent basis. Care is therefore is needed when interpreting the data for indications of the relative emphasis on secondary and higher education. In most countries, however, the definition of primary education is unambiguous, so its share of total spending provides a good idea of the overall emphasis it receives in terms of government spending on education. 2/ The focus here is on public education, partly because little data exist on private education. . 59 - costs of education, regardless of the source of finance.J/ Following common practice, unit costs are expressed as a percentage of the per capita GNP.&/ To render the comparison among countries even more transparent, an additional cost index is defined as follows: for each level of education, unit costs are expressed as a ratio to the regional average unit cost.5/ Thus, an index exceeding unity implies that unit costs in the country lie above the regional average, and an index below unity implies the opposite. To complete the picture, an overall cost index is also defined, being the average of the cost index for the three levels of education. It is a measure of the global costliness of a country's education system in relation to its Asian neighbors. (a) Comaring countries by level of education The relevant data appear in table 3.2. Unit costs in Papua New Guinea are exceptionally high at all levels of education, comparable to the costs of education in African countries. Since the country is an outlier in this respect, it is not taken into account in the comparisons discussed here. For the remaining Asian countries, the unit costs of primary education are on average 10 percent of the per capita GNP, comparable to the level in Latin America. They fall into three groups: Bangladesh, India, the Philippines and ,/ In deriving unit operating costs from basic statistics, care was taken to add fees to the public unit cost if the latter reflects only what the government spends from general revenue. This adjustment is required, for example, where fees are retained by the school (normally when such collections are small), as in Malaysia. _/ This treatment avoids the problems associated with currency conversion, and helps to control for differences in the price of educational inputs. It is especially relevant in primary and secondary education where most inputs are non-tradeables. In higher education, the case is stronger for conversion to a single currency, since some inputs --such as books, equipment and even some categories of teachers -- are tradeable, and their prices tend to concentrate in a narrower band, irrespective of the conditions within a single country. If the costs of inputs primarily reflect the outcome of international market forces, then expressing the unit costs of education in terms of a country's per capita GNP would distort comparisons, since they would always tend to be high in a poor country, and low in a rich country, largely because of the size of the denominator. However, analysis in a later chapter shows that in fact little relationship exists between unit costs expressed as a percentage of the per capita GNP, and the per capita GNP itself. So, standardizing costs in terms of the per capita GNP remains a valid basis for comparison. 5/ Papua New Guinea is not included in reckoning the regional average unit cost, since the costs of education in this country are exceptionally high. Its inclusion would have caused most countries' cost index to fall below the average. - 60 - Sri Lanka appear to be low cost countries, with average unit costs around 6 percent of the per capita GNP; a second group consists of Indonesia, Korea, Malaysia and Thailand, with unit costs which are two or three times as high; China and Nepal form the in-between group, with moderate levels of costs. In secondary education, Asia's average unit costs are 19 percent of the per capita GNP. This figure is somewhat less than the average of 26 percent for Latin America. The Philippines and Sri Lanka have the lowest costs, less than 0.5 times the regional average; Bangladesh and China, on the other hand, have the highest costs, about 1.5 times the regional average. Among the remaining countries, India, Nepal and Thailand have slightly less than average levels of unit costs, while Indonesia, Korea, and Malaysia have above average costs. Iable 3,2s Unit operating costs of publio education in Asia, mid-1980s As I of aer eamita CAP Ratio of unit cost to reaional averase * 2imEy Secondar ax.h Risher bScondaCr gLag I Bangladesh 6.4 30.0 284.6 0.63 1.58 1.86 4 China 9.2 27.6 243.8 0.91 1.46 1.59 5 India 6.0 17.3 231.1 0.59 0.91 1.51 6 Indonesia 12.6 23.3 91.1 1.25 1.23 0.59 7 gorea 16.5 23.4 70.6 1.63 1.23 0.46 9 Nalayeta 14.1 21.3 190.3 1.39 1.12 1.24 10 Nepal 9.0 13.5 249.0 0.89 0.71 1.62 11 Papua Nev Guinea 29.0 65.0 1050.0 2.87 3.43 6.85 12 Philippines $.8 8.6 50.0 0.57 0.45 0.33 14 Sri Leaka 6.1 9.3 83.3 0.60 0.49 0.54 15 Thailand 15.5 15.3 39.9 1.53 0.81 0.26 Average at Without 11 10.1 19.0 153.4 - - - Vith 11 11.8 23.1 234.9 - - - Sourcest see appendis table D4.1. At The Unit Costs of education In Papua New Gainea (P80) are exceptionally high, thus lifting up the average for Asia considerably. In comparng individual countries to an Asian average (Last throo coluss of this table), the relevant denominastor is the average that that exludes data for MU8. bl In countries with distance systems, the Unit costs of hIgher education is the average for public regular sad distance education, weighted by their respective enrollment shares. of Pigure refers to unit costs in universities. . 61 - The unit costs of Asian higher education are, on average, 153 percent of the per capita GNP, compared to an average of.88 percent in Latin America. As before, the variance within the region is very wide: the Philippines and Thailand have unit coats which do not exceed 50 percent of the per capita GNP, while Bangladesh and China have costs which are nearly 6 times as large. It is sometimes argued that poor countries tend to have higher costs when these are expressed in relation to the per capita GNP: at this level of education, some educational inputs, being tradeable goods, command international prices, regardless of a country's domestic conditions. If this argument holds, then it would be more valid to express costs in absolute terms, using a common currency. Further, if the international prices of inputs are of primary importance, then one should expect a rather flat relationship between absolute unit costs and the per capita GNP. But this expectation is not entirely borne out in figure 3.1. I l costs (1St. iog scalel 1fi0000 6 18 1oo 1000 t100 Patrapita GNP (IS$.Ilog scale) Fiaure 3.1t Relationship between the unit costs of xesular pubic hiiher education and per capits, WH, Asia, oirca 1985. The international-prices argument Is further weakened when it is noted that countries with similar levels of per capita GNP have vastly different unit costs: US$900 in China compared to US$599 in India; US$453 in Bangladesh compared to US$354 in Nepal; US$2,132 in Korea compared to 3,540 in Malaysia; and so on. What these comparisons suggest is that the costliness of - 62 - a country's higher education is influenced at least as much by the internal policies pursued in the sector, as by external factors. The data on costs in table 3.2 reflect the average across all types of institutions. In higher education, it is important to distinguish between regular and open universities,L/ since their cost structures are so different (table 3.3). On average, the unit costs of open universities are only 18 percent as large as those of regular public institutions. The gap between the two types of instititions is widest in Korea and Thailand: open universities cost no more than 10 percent as much as the regular institutions. Table 3.3: Unit operating costs of public higher education, mid-1980s Regular Inatitutionm Unit costs of open/distance ed. As X of ratio to those of DC GNP ,,, ,u,, resular institutions 1 Bangladesh 284.6 453 4 China 329.8 900 5 India 231.1 $99 - 6 Indonesia 105.7 497 0.35 7 lorea 104.5 2132 0.10 9 Malaysia 190.3 3540 - 10 Nepal 249.0 354 11 Papua New Guinea 1050.0 6521 12 Philippines 50.0 291 - 14 Sri Lanka 111.2 416 0.20 15 Thailand 177.9 1267 0.08 Average Excluding 11 183.4 1045 0.18 Including 11 262.2 1543 Sources: see appendix table B4.1. Unit costs converted to US$ using exchange rates for 1985 reported in UNESCO (1987). (b) Comaring the global costliness of education across countries The global costliness of an education system is summarized by the overall cost index, calculated as the average of the cost index at the three levels of education (table 3.4). Papua New Guinea's figure is an astounding 4.38, which means that its system is nearly four and a half times as costly as that of its Asian neighbors. Among the remaining countries, three groups can be distinguished: Bangladesh, China, Korea and Malaysia have high to ,/ For the purposes of this paper, education via correspondence and other similar arrangements are included whenever open universities are referred to. - 63 - moderately high unit costs; India, Indonesia, Nepal and Thailand have costs that are close to the average for the sample; the Philippine and Sri Lanka are well below the average. In the last two countries, unit costs have in fact risen sharply since 1985, due to substantial increases in teacher salaries, but even so, they remain below the Asian average.2 Table 3.4: Index of overall costliness of public education, Asia, aid-1980s 1 Bangladesh 1.36 4 China 1.32 5 India 1.00 6 Indonesia 1.02 7 torea 1.11 9 Malaysia 1.25 10 Nepal 1.07 11 Papua New GULea 4.38 12 Philippines 0.45 14 Sri Lanka 0.55 15 Thailand 0.87 Sourcet Data refloot the arithmetic mean of the last three colums of table 3.2 The costliness of education in a country appears to depend little on its level of per capita GNP (figure 3.2). For example, Indonesia's overall unit cost index is 85 percent higher than Sri Lanka's, even though both countries have comparable levels of per capita GNP. The wide disparities suggest that there is indeed substantial room for choosing policies with vastly different cost implications. Note, however, that when a country's educational costs differ from the regional mean, it does not necessarily imply that costs should be raised or reduced. The regional mean facilitates comparative analysis, but is clearly not a normative target in policy making, since the costs of education have to be balanced against the educational outcomes. Having said that, the existence of high costs nevertheless flags a strong need to examine the efficiency of the system, focussing, in particular, on the policies that have resulted in high costs. Z/ For the data on unit costs in 1988, see appendix table B4.1. - 64 - 1tol of CVetail Costilless 1.6 0.2 1 10 s. *6 0.e 15 0.6 12 0.4 0.2 100 1000 10000 Pet capita 01P loS$, 10 2cale) PIG. as atilot, is st o.. Sauce: see lpeadia . Figure 3.21 telationship beteen the overall costliness of public education and per capita GNP, Asia, circa 1985. 3.3 Sources of variation In unit costs Most of the variation in costs stems from differences in the way education is provided. Of relevance are such aspects as the grouping of students into schools and classes, the pre- and in-service training of teachers, their pay scale and workloads, the use of pedagogical methods and materials, the nature and extent of the evaluation and supervision of teachers and students, the criteria used in student selection between levels of education, and so on./ Because teachers are typically the most important item of cost, and because of data constraints, we focus here mainly on teachers' salaries and pupil-teacher ratios (which may be taken as a rough indicator of the intensity in the use of teachers' time). In higher education, an added dimension is the distribution of enrollments in selective and open universities, two types of institutions with vastly different cost structures. For ease of presentation, the data for primary and secondary education are discussed separately from those for higher education. L/ Apart from their impact on costs, these aspects in the organization of schooling are, to some extent, also likely to affect student achievement. This issue will be touched on in a later chapter, but the treatment is necessarily brief due to the scarcity of data and resource materials. - 65 - (a) Primary and secondary education The salaries of primary school teachers in Asia in the mid-1980s averaged about 2.9 times the per capita GNP (table 3.5), comparable to the mean of 2.4 in Latin America (Mingat and Psacharopoulos, 1985). 2/ They ranged from a high of 6.8 times in Papua New Guinea, to a low of 1.6 times in the Philippines and Sri Lanka. As a result of recent policy changes, teachers, salaries in the latter two countries have, respectively, risen to 2.2 and 2.0 times the per capita GNP by 1988, thus, bringing them nearer the norm in Asia. Excluding the sample's outliers (Korea and Papua New Guinea),. the average salaries of primary school teachers concentrate in a narrow band, between 2 and 3 times the per capita GNP. In all countries, secondary school teachers earn higher salaries than their primary school colleagues, averaging 4 times the per capita GNP. In some countries, however, the gap in pay is nevertheless quite small. As before, secondary school teachers in Korea and Papua New Guinea have exceptionally high salaries compared to their counterparts in other Asian countries. Table 3a.s Anmsal teacher reameration and pupi-teacher ratio: Io public primary and secondary sbools, aid-1980s Teacher remuneration as ratio to noriett.0 .a No. of *anits got teacher PCgg 8aofner PARMg Scnd 1 Bangladesh 2.2 - 47.0 26.2 2 hutan - * 38.5 10.1 3 Bums - - 46.4 28.5 4 Chias 2.0 3.4 24.9 17.2 bamg Kong - - 27.S 25.1 3 India b/ 2.9 3.1 57.6 20.2 6 Indonesia 2.5 3.2 25.3 15.3 7 Korea b/ 5.0 5.5 38.3 34.3 8 Laos - - 24.9 11.2 9 Malaysia 2.4 3.1 24.1 22.1 10 Nepal 2.8 5.0 35.5 27.5 11 Papua NOW Guinea 6.8 10.0 31.0 25.4 12 Philippines 1.6 1.7 30.9 32.2 13 Sinaspoer - - 27.1 20.4 14 Sri Lanka 1.6 2.1 31.7 26.1 15 Thailand 2.5 2.9 19.3 19.6 Average Excluding 11 2.6 3.3 - - Including 11 2.9 4.0 33.1 22.6 Sourcess see appendix tables B2.4, 32.5, And 34.2. a/ Teacher remmeration includes basic pay and allowances. bI Data on remaneration are estimates based on unit costs and pupil-teacher ratios. 2/ The salaries of other comparable professionals might have been a better numeraire than per capita GNP for comparing teacher's salaries; unfortunately, this alternative could not be implement for lack of data. - 66 - As for the pattern of pupil-teacher ratios, the regional average is 33 in primary education, and 23 in secondary education. These averages contrast with the corresponding figures of 29 and 18 in Latin America, and 19 and 14 in developed countries. Overall, it appears that Asian teachers are used more intensively as an educational input than teachers elsewhere.IQ/ However, the variation among Asian countries is very pronounced. The ratio ranges in primary education, from 58 in India to 19 in Thailand; and in secondary education, from 34 in Korea to 10 in Bhutan. It is interesting to contrast the sources of cost differences among countries. For example, in both Bangladesh and Sri Lanka, the unit costs of primary education are only about 60 percent of the regional average, but in the former country, the result is due mostly to very high pupil-teacher ratios, while in the latter, it stems largely from the relatively low teacher salaries. In India and Thailand, primary school teachers' salaries are comparable, but owing to the vast differences in pupil-teacher ratios, the result is a wide gap in unit costs, 6 percent of the per capita GNP compared to 15.5 percent respectively. Because of their impact on costs, teacher salaries and pupil-teacher ratios are potential objects of policy intervention. The diversity among countries in these characteristics, coupled with the observation that neither is strongly linked to levels of economic development, as proxied by the per capita GNP (figures 3.3 to 3.6), suggests that there is probably scope for policy maneuver in this regard. Levels of teacher salaries are affected, for example, by such instruments as public wage policies, and regulations on teacher qualification; and pupil-teacher ratios depend on such choices as the grouping of students, the use of multiple-grade teaching and of specialized teachers and other academic staff, and so on. J1/ A high pupil-teacher ratio reflects a higher number of classes per teacher and/or large class sizes. Both arrangements signify that teachers are used more intensively. - 67 - Avg. teacher pay itallo to GNPIcapita) 7 6 5 4 3 g• isi 3 - 0 100 1000 10000 Per capita Gli? (US*. log scale) Sosme: see appesti I. Fia= 3.3: Reattona~hp betwen aveae priaay schooi oø:a re~oneration and per capita GP, Asia, oiro 198$. 60 Pplil-keacler ra1o se 50 1u .3 40 *. 10 14. 11 30 18°° ;i . 6 .9 20 -15 10 100 1000 10000 Per capita Gi? (95. log scale) 809its: s ppsisI 9 i 5_,4: V*lattcanb betvee pupil-taa~r ratio in priay eduation and per capita GM, Asia, cIrca 1985. - 68 - 40apI-teacher ratio 35 12 .7 30 - 10 .3 25 - 14 s 9 13 20 - .15 15 , 6 5 - 100 1000 10000 Per capita GI? (Us*. lag scalt) segne, se appetis a. FIL 3.2: Relatiahip betwne pupil-te~aber ratio Ln secondary education and per ~apita GM, Asia, oiroa 19e$. Av. teacher pay (ratio to GUP/caplta) 12 is 10 8- 60 10 .? 4 - 6 &" • .15 .9 2 - ;< 1 100 1000 10000 Per capita GI? (98*. tog scale) gey": se appeatia I. FSaIKe.6:J' Relat~anip between pupiL-tesobr ratio ta sooday oduatjom and per capita OP, Asia, etrea l9e~. - 69 - Higher education Data are available on pupil-teacher ratios, but not or the salaries of teaching staff (table 3.6). On average, the ratio is 14 in regular institutions in the public sector, comparable to the average for Latin America (13.5) and that for developed countries (12). As before, great diversity exists among countries in the region: the ratio is relatively low in China, Papua New Guinea, and Thailand (less than 9), and remarkably high in Korea (nearly 40); the remaining countries have ratios in the 10 to 20 band. In distance education, the variation in pupil-teacher ratio is even wider, stretching from a low of 36 in China to a high of over 770 in India. These differences illustrate the tremendous range of possible choices in pupil- teacher ratios, both in regular institutions, and in the open universities. Given the absence of a systematic link between student-faculty ratios and per capita GNP (see, for example, figure 3.7), these choices appear to be, a prJori, feasible in all country settings. Table 3.6: Higher education student-faculty ratios, 1985 zeaflar institutions Open/distance pblic PrEvate education 1 Bangladesh 15.9 - * 2 Bhutan 10.9 - - 3 Sums a/ 30.3 - - 4 China 5.2 - 36.0 5 India 15.7 - 776.5 of 6 Indonesia 14.0 46.1 689.7 7 Korea 42.2 41.1 414.7 8 Laos 10.1 - - 9 MalaysIa 11.4 - - 10 Nepal 13.2 - - 11 Papua new Guinea 7.7 - - 12 Philippines 16.0 48.0 - 14 Sri Lanka 10.7 - 84.9 15 Thailand 8.3 17.6 618.8 Averae b/ 13.9 - - Sources: see appendix table 32.6. a/ Data includes pupils in correspondence courses. bt Data for Suzma excluded in calculating the average for the reason given in footnote al above. of Refers only to Andbra Pradesh Open University. - 70 - 50 Studeat-laculty ratio 7 40 30 20 . 5" .6 . 12 2 it 114.15 100 1000 10000 Pat capita GNP (1S, log scale) sMVes see tests 33.6. PaiAtre 3, Relationship between student-faclty ratio in conventional public higher education and per capita 01P, Asia, circa 1985. In four countries -- Indonesia, Korea, and the Philippines -- data are also available for private higher education. A striking feature is that only in Korea is the ratio comparable between the public and private sectors. The gap is very wide in the other two countries, the ratio in private higher institutions being two to three times as high as in the public sector.,/ A possible reason for Korea's uniqueness in this respect is that public higher education is subsidized to a lesser extent than in the other two countries, so that from a financing perspective, public institutions are rather like private ones, and therefore face similar pressures to use teacher time efficiently.4/ The distribution of enrollments among types of institutions also affects the overall costliness of public higher education. For completeness, the data in table 3.7 show the distribution in local institutions (including U]/ In most countries, the private sector tends to concentrate more on the provision of non-science courses. Probably because there are fewer logistical constraints on the use of large classes in such courses, pupil-teacher ratios tend to be higher in the private sector. 2/ The impact of private financing on unit costs will be explored in more detail in a later chapter in this report. - 71 - private ones) as well as those overseas. The institutional composition of the sector reveals the alternative ways that countries have taken to satisfy the demand for higher education. In all of them, the government is involved in the direct provision of higher education, but the extent and nature of its involvement varies widely. Table s7; Distribution of higher education enrollments by type of institution, mid-1980s Percent of total enrolled in Total enrolled Local institutions Overseas (6000L PMuli Distanc Priyate a/ Institutions 1 Bangladesh 811.4 40.5 0.6 58.4 bi 0.5 2 Shutan 1.1 95.8 0.0 0.0 4.2 3 Burma 185.1 54.4 45.4 0.0 0.2 4 China 3470.6 68.6 30.2 0.0 1.2 5 India 3314.5 37.4 4.8 57.3 b/ 0.5 6 Indonesia 1295.6 32.5 8.8 57.7 1.0 7 lorea 1478.2 21.4 12.0 65.1 1.5 9 Malaysia 131.2 60.5 1.0 7.6 30.9 10 Nepal 74.3 73.3 1.3 23.4 2.0 11 Papua New Guinea 11.4 82.6 8.8 6.1 2.5 12 Philippines 1549.6 16.7 0.0 83.0 0.3 14 Sri Lanka 34.5 62.0 28.5 0.0 9.5 15 Thailand 724.0 14.5 77.9 6.3 1.3 Sources* see appendix table 51.6. a/ Data in this column differ slightly from those in the last column of table 2.9 since overseas enrollments are included in the denominator in this table, but not in table 2.9. The discrepancy in wide only for Malaysia owing to the large number Malaysian students abroad. b/ Data include enrollments in privately-managed institutions. The public sector is clearly the predominant sector in all Asian countries except Indonesia, Korea and the Philippines. Bangladesh and India also belong to this group despite the apparently large share of private enrollments since most private institutions are in fact heavily subsidized by the government. A distinction should be made, however, between countries that rely heavily on relatively cheap distance education or open universities (Burma, China, India, Sri Lanka, and Thailand), and those that rely on the more costly regular institutions. In Indonesia, Korea, and the Philippines the public sector's share of enrollments lies below 50 percent. As in the previous group of countries, differences exist in the strategy adopted to address excess demand. In Indonesia and Korea, the government has relied on private institutions as well as open universities, both of which are cheaper alternatives to regular public institutions. In Indonesia, however, private institutions are subsidized, since close to 30 percent of their staff are seconded from government service. - 72 - In the Philippines, distance education does not really exist, so that the demand for higher education is satisfied almost entirely by the private sector. Finally, a note on overseas studies. Among the Asian countries, overseas enrollments are substantial only for Malaysia: over 30 percent of higher education students are abroad, most of them self-financing. Sri Lanka has the next highest share of enrollments abroad, but at 9.5 percent, the outflow is hardly of the same order of magnitude as Malaysia's. In a sense, Malaysia's large overseas student population represents the country's private sector, albeit one that exists largely in foreign countries.1j/ It is instructive to recall that Thailand faced a similar situation in the 1960's (Watson, 1981). At that time, the government embarked on a dual-pronged strategy to stem the outflow: it lifted the ban on private education and invested heavily in open universities. Judging from the institutional makeup of higher education in the country today, those policies seem to have succeeded remarkably well. 12/ The outflow of students abroad is a familiar phenomenon wherever the demand for education outstrips the capacity of local institutions. In Greece, for example, a government ban on private universities, coupled with limited provision of heavily subsidized public higher education, has led to a large outflow of students to institutions abroad. Psacharopoulos (1988) has described this outflow as representing the country's de facto private sector. - 73 - 3.4 The private financing of education Just as countries vary in the costs of education, they also differ in the way those costs are financed. In public education, there are two main sources of financing for public education: government appropriations, and private contributions in the form of tuition and other fees. IV In some countries, additional resources are generated from income-earning activities (such as consultancies, school production, and fund-raising events) and through endowments and other donations from parents' associations, local communities, philanthropists, and so on. In aggregate, however, these sources yield relatively little income for public education, and are not as amenable to policy intervention as direct charges on students. For these reasons, the data in table 3.8 simply show the share of unit costs financed through school fees; its complement is assumed to be the share of government funding. Table 3.8l Pees for public education as a percent of unit operating costs., mid-1980 bohe ed&oatian Prile saur &manu 2M 1 Bangladesh 7.4 4.0 0.1 - 4 China 4.8 $.1 0.3 - 5 India j 11.6 4.5 59.0 al 6 Indonesia 7.1 27.4 18.9 - 7 Korea 0 34.2 45.9 32.0 9 Malaysia 3.7 4.0 5.8 - 10 Nepal 0 40.7 10.4 - 1 Papua no Oune.a 8.7 39.8 0.0 - 12 Philippines 0 9.3 15.3 - 14 Sri Lanka 3.1 3.1 3.4 57.7 15 Thailand 0.1 18.3 5.0 27.s Average 3.2 17.8 10.0 - Sources: see appendix table B4.3. a! Rafers only to Andhra Pradesh Open University. To te more precise, private contributions should also include school- related private expenses other than fees, such as those covering books, materials, and transport. In some countries, these items are provided by the school, in which case, the amount of fees paid relative to schools' operating costs would, by and large, capture the private share in the total direct costs of education. In other countries, they are not provided by the school, but are purchased separately by students. In this case, fees alone would understate the share of private contributions. Although some micro-level data are available on the magnitude of private spending on education other than for fees (e.g. the 1982 Household and School Matching Survey in the lhilippines), they are patchy, and usually dated; and are therefore not considered here. - 74 - On average, fees finance a larger proportion of the costs of public education in Asia than in other developing regions: 3.2, 17.8, and 10.0 percent respectively in primary, secondary and higher education, compared to the corresponding figures in Latin America of 0.9, 1.7 and 6.6 percent (World Bank, 1986). There is, however, substantial deviation from the mean: some Asian countries rely to a significant extent on private financing (e.g. Indonesia and Korea), while others rely little, if at all, on it (e.g. Bangladesh, China, Malaysia and Sri Lanka). Moreover, countries differ in the structure of cost recovery across levels of education. In Korea and the Philippines, the share of costs recovered via fees show a definite and consistent rise with the level of education. The opposite trend is present in Bangladesh, and China. Yet a third pattern exists, in which the rate of cost recovery rises steeply from primary to secondary education, and then drops, sometimes very sharply, at the third level (Indonesia, Nepal, Papua New Guinea, and Thailand). Data on the financing of private education are scanty, and have been pieced together to provide rough estimates for higher education (column 1, table 3.9). Three models appear to exist among Asian countries. In Korea, Malaysia, Nepal, the Philippines and Thailand, the private sector is largely self-financing, depending on non-government funding, usually student fees, for the bulk of operating incomes. At the other end of the spectrum belong such countries as Bangladesh and India, where the private sector receives the major share of funds through government subventions. Indonesia is alone among the Asian countries in which private institutions receive a moderately high level of public subsidies, covering about 30 percent of their costs. Table 3.9: Estimated rate of private financing in higher education, Asia, circea 1985 Inde of Cost recovery overall in private private anctor W tCaLat () .aL I Bangladesh 28 16.5 2 Bhutan 3 Burma 4 China - 0.3 5 India 5 7.1 6 Indonesia 70 48.7 7 Korea 95 76.6 9 Malaysia 90 15.1 (33.1) 10 Nepal 100 31.8 11 Papua New Guinea 100 6.3 12 Philippines 100 85.8 14 Sri Lanka 20.5 1Z Thailand 100 26.9 Sources: authors' estimates based on tables 3.8, 31.6 and 34.3 and data, discussion, and estimates in aing (1988) for Bangladeshi ACU (1988) for Indiai World Bank (Indonesia-19886b) Government of Korea (1987)t Government of Malaysia (1988)o Tiailsina (1988) for Nepali World Bank (Papua New Ouinea-1987)i and Mingat and Tan (1988) and James (1988) for the Philippines. a/ Reflects the rate of cost recovery across institution types, veighted by t6heir share of total enrollments. Figure in parentheses for Malaysia denote the rate of private financing if privately-financed overseas education wre included. - 75 - The overall extent of private financing in higher education can be summarized by constructing an index based on the rate of cost recovery across types of institutions, weighted by the corresponding shares of total enrollments. The results appear in column 2 of table 3.9. According to this indicator, four groups of countries emerge. China, India, and Papua New Guinea form one group, where private financing is extremely limited. Indonesia, Korea, and the Philippines form a second group with the opposite characteristic. In Nepal, Sri Lanka, and Thailand, the extent of private financing is moderately high. The fourth group consists of Bangladesh and Malaysia, with a somewhat below average rate of private financing. Note, that Malaysia's index would have been significantly higher had overseas education been included. However, for comparison across countries, the lower figure is relevant since it reflects the extent private resources are mobilized to finance the development of local higher education. To what extent does the extent of private financing depend on the wealth of a country? A.2rior, a positive link might be expected for at least two reasons: the demand for education is probably stronger owing to the better job prospects in such settings, and therefore people are more willing and able to pay for their education; and the institutional infrastructure is better developed to support cost recovery policies. Figure 3.8 shows that although the index of private financing is positively correlated with per capita GNP, the link is in fact quite weak. Thus, countries with comparable per capita GNPs, such as Indonesia, Papua New Guinea, the Philippines, and Sri Lanka, vary widely in the level of private financing, ranging from 6 percent in Papua New Guinea, to 86 percent in the Philippines. On the other hand, Nepal and Malaysia share similar rates of private financing, but differ significantly in per capita GNP. The pattern of variation suggests that while a country's level of economic development affects the administrative and social feasibility of tapping private resources for higher education, there remains substantial scope for such a policy in all country settings. The options for implementation include charging fees for public education, and/or allowing a largely self-financing private sector to develop. - 76 - 100 adel of pliat flaaacka 80 - 60 - 6. 40 10 to £0- i 208 .5 ,l1 100 1000 10000 Per capita el? js18. og scala) PI~ae.: 3 l,asis Mp betueen per capita OMP en degree of private finaning in higher education, Asia, MU4-19809. . 77 - 4. Efficiency i the Provision of Education The data presented in the preceding two chapters provide a regional perspective on the costs and financing of education in individual Asian countries. For assessing alternative policies, however, they are inadequate in themselves, since the choice of policies depends, not so much on a country's ranking in relation to its neighbors (at least in the aspects considered so far), as on its educational and social objectives. In general, these objectives include enhancing efficiency and equity in the use of educational resources. L/ Issues relating to efficiency in the sector are discussed in this chapter; those relating to equity, in the next. The term efficiency describes the relationship between inputs and outputs. When output refers to broad societal goals -- such as the production of educated manpower for the labor market, better health, lower fertility rates, and so on --, the analysis focusses on the external efficiency of education, and involves assessment of such issues as the economic returns to investing in education in general, and the allocation of spending across levels and types of education. When output refers to goals internal to the education system -- such as students' achievement of the curriculum objectives, dropout and retention rates, and so on --, the focus is on the system's internal efficiency. Both aspects of output are important to consider, since they complement each other in determining the global efficiency of the education system. Because the subject is so broad in scope, it is clearly not possible to cover all facets of it in the discussion here. Touched upon will be topics that are normally considered important, and for which the relevant data are available. Where an issue has already been covered in the extant literature, the presentation will be kept brief so as to leave more space for the discussion of new data and findings.2/ 1/ Political objectives clearly also affect policy choices, but they are beyond the scope of this paper to address. Such objectives depend strongly on country-specific political conditions, and are difficult to analyze in the economic framework adopted here. 2/ The length of discussion on a topic should therefore not be taken to indicate our judgement of its relative importance as a policy issue. In fact, since its significance depends on the circumstances of individual countries, and since those circumstances vary widely, there is no basis to establish a general ranking of issues by their importance for policy purposes. * 78 - 4.1 External efficiency The accumulated evidence on the impact of education on economic growth and social welfare has been summarized succinctly in an excellent paper by Psacharopoulos (1984). In general, the data reveal that countries with a better educated population had higher economic growth rates and more equitable distributions of income; and at the micro level, individuals with more education tind to enjoy higher incomes, more geographical mobility, better health and so on. Weighing such evidence against the recent attacks on the role of education on productivity growth and income distribution,J/ Psacharopoulos concluded that the argument for investing in schools and training remains unassailable, particularly in developing countries where human capital tends to be relatively scarce. Mincerian regressions show, for example, that education yields an average economic return exceeding 12 percent in developing comntries worldwide.A/ The results suggest that education is in general at least as profitable an investment as physical capital. If more is to be invested in education, where should resources be allocated in the sector? A common approach is to compare the rate of return to the different levels and types of education.5-/ What lessons for Asian countries may be drawn from international evidence in this respect? Do the available data for countries in the region yield similar conclusions about investment priorities? 2./ These attacks attempt to cast doubt on the social role of schooling by offering alternative explanations for the observed earnings advantage of the more educated. Among them, Psacharopoulos (1984) listed screening for ability, job competition, labor market segmentation, nonclearing wages, nonprofit maximizing public sector pay scales, social class, and youth unemployment. W/ In Mincerian regressions, the dependent variable is income, and the independent variables, schooling and experience. When income is expressed in logarithmic terms, the coefficient on the schooling variable (measured in years) is the estimated rate of return to an additional year of schooling at the sample mean. w/ Rates of return are derived by comparing the productivity of educated people to those with less education, and weighing the difference against the costs of their extra schooling. Although the statistic suffers from various well-known flaws, it provides perhaps the most concise basis for comparison since its method of calculation reflects both the costs and benefits of an investment. Although externalities and other non- quantifiable aspects are neglected by the statistic, the results remain valid as a first indication of the pattern of returns to different investments in the sector. - 79 - (a) The economic returns to investing in education The overall pattern across world regions and country type is summarized by the data in table 4.1. L/ In the typical developing country, the social returns to education show two characteristics: they are high at all levels, and they decline with ascending levels of education. A further finding is that returns drop with rising levels of economic development. The decline is gradual, however, when compared to the vast expansion of education as countries develop, a pattern revealed by cross-section and time-series data (Psacharopoulos, 1981). These international patterns have two direct implications for Asian countries: for the less developed countries in the region, primary education deserves top priority in the intrasectoral allocation of educational resources; for the more developed countries where education is better developed, the lower levels of education continue to warrant the highest priority. This bias in the allocation of public spending on education is reinforced when considerations of external efficiency and equity are taken into account. Table 4.1: International patteras of social returns to education Social rate of .retura (U. Restonlomtry tre Primary Eon Hish, Africa 26 17 13 Asia 27 1 13 Latin America 26 18 16 LDC 27 16 13 Intesmeadiate 16 14 10 Advanced - 10 9 Sources: Psacharopoulos (1981, 1985). How do these general conclusions fit in with the data on individual Asian countries? Recent estimates of rates of return to education exist for only seven of them (table 4.2). India, Indonesia and Papua New Guinea have the classic pattern of returns across levels of education: highest in primary education, and lowest in higher education. In Papua New Guinea, however, the returns to higher education, at an estimated 2.8 percent, are so low as to render investments at this level "wasteful" (Gannicott, 1987). One reason for hi The rates of return data may be biased if the estimates do not make allowance for quality as an output. Few studies on the returns to investments in quality exists, however (an exception is Behrman and Birdsall, 1983). - 80 - this result is the country's astronomical unit costs at this level of education. Likewise, the somewhat below-average returns to higher education in Malaysia is partly attributable to its higher-than-average unit costs. Table 4.2t Rates of return to education in Asia, latest available year soCIAL PRIUn IMSS Piary Iscondary RLhe PrUarE Sesnay igher India 1978 29.3 13.7 10.8 33.4 19.8 13.2 Indonesia 1982 18.0 15.0 10.0 - - - (14.5) - - - * - Korea 1982 - 10.9 13.0 - - - Valaysia 1983 * - 7.6 - - 12.2 Papua New Guinea 1982 19.9 12.0 2.8 29.4 14.7 8.1 Philippines 1985 11.9 12.9 13.3 18.2 13.8 14.0 (4.4) (9.3) (11.6) (7.2) (10.2) (12.5) Thailand 1975 12 24 12.8 - - - 1985 - - 13.3 - * 17.4 Sources: Psacharopoulos (1985) for India USAID (1986) for Indonesial Nebmet and Tip (1986) for Malaysia# Korea Educational Development Institute (1983) for Koreas Gannatt (1987) for Papua New Quineas Tan and Paqueo (1988) for Philippines; and Govt. of Thailand (1987) and Suppachai (1976) for Thailand. Note: The figures in parentheses denote the rates of return for incomplete education. The pattern of returns in the other Asian countries is also interesting. In the Philippines, the returns to education at all levels are currently still moderately high, despite the rapid expansion of education during the last two to three decades. The time series data for Thai higher education show the same result. A second noteworthy feature in the Philippines is that the returns are comparable among the three levels of education. Both outcomes stem in part from the country's relatively low unit costs, particularly in higher education. The low costs are in themselves a result of the system's relatively heavy reliance on private financing.2/ A final comment is that at all levels of education, the returns to complete cycles of education exceed those to incomplete cycles. The gap in returns emphasizes the social profitability of reducing the incidence of dropping out, particularly in primary education where that gap is widest. The data for 2/ Fcr various reasons discussed later in this chapter, there appears to be an inverse relationship between the costliness of a system and its degree of dependence on private financing. - 81 - primary education in Indonesia point to a similar conclusion. In Thailand, the highest returns are associated with secondary education, in contrast to the normal pattern for other developing countries. Although the data are somewhat old, they jibe with the current perception that this level of education is relatively underdeveloped compared to the other levels (World Bank, Thailand-1988). So far, no distinction is made concerning the returns to the different types of educational investments. Such differentiation is particularly relevant in secondary and higher education, where wide variation in unit costs exist between general and vocational programs, and among alternative specializations. The relevant international data appear in table 4.3. They point to the obvious: expensive programs tend to have lower returns than cheaper ones. However, this generalization should be interpreted cautiously: the same program labels may describe quite different curricula across countries, with correspondingly different unit costs. Moreover, labor market conditions are likely to change from country to country, affecting the pattern of earnings across specializations. Thus, for example, in Thailand -- the only country for which we found recent data on rates of return by university programs, - - courses in agriculture currently yield a respectable 15 percent in economic returns to society, compared to only 10.4 percent for medicine (see appendix table B5.3); this pattern contrasts sharply with that reported in Psacharopoulos (1985). In science, engineering programs fetch a higher return than courses in the pure sciences; and in the arts, law courses earn much better social returns than those in education. Table 4.3: International evidence on the social returns to selected secondary and university programs Rate of return (M) General, academic 16 Vocational, technical 12 UniversLtv sconomics 13 Law 12 Social Sciences 11 Nedloine 12 amineering 12 Sciences, math, physics 8 Agriculture 8 Sources Psacharopoulos (1985) - 82 - From the existing data, it is clear that specific conclusions about investment priorities in the sector depend on each country's unique conditions. However, in broad terms, worldwide evidence supports placing priority on the lower levels of education. As indicated earlier, the argument for this ranking is reinforced if considerations of social equity are also taken into account. If it is accepted that the lover levels warrant priority, there is the further question of how resources should be allocated within the subsector to maximize efficiency. Since the acquisition of literacy is a primary reason why returns are high, an insight into this issue may be obtained by analyzing the determinants of literacy, a topic discussed in the next section. (b) Investing in basic literacy Many studies on literacy and its retention exist (see Hartley and Swanson (1984) for a survey of the literature). Most rely on micro level analyses, using individuals as units of observation. Here we )resent some fresh findings based on data at the country level, which help to shed light on macro aspects of the policy choices involved in enhancing literacy in a country. What factors account for the variance in literacy rates across nations? A Rrior, one would expect that past investments in primary education are of central importance. Also crucial are such factors as the length of the primary cycle, and the proportion of pupils who actually persist to the en1 of the cycle. To test the relationship, a regression equation was estimated, using as the dependent variable, adult literacy rates in 1985. The independent variables zeflect primary schooling conditions in the past, and include the enrollment ratio, the cohort survival rate, and the length of the cycle. For these variables, data for 1975 were used. The regression results, based on a sample of 46 developing countries worldwide for which data are available, appear in table 4.4. - 83 - Table 4.4s OLS regression of the determinants of adult literacy rates, 1985 RearessLon Sample Standard Coefficient I-atAatic m.aB. deviatioan Prifiy enrollment ratio, 1975 0.70 * 8.90 79.7 27.3 X surviving to end of primary education, 1975 0.30 * 2.74 68.3 20.6 Length of primary cyalt 1975 9.43 * 3.19 6.2 0.8 Constant -75.64 * 3.37 - - Dependent variables adult literacy rate, 1985 59.3 25.5 Number of observations 46 R-squared 0.7 Sourcess data on literacy rates are item UNICEF (1987)s data on primary enrollment ratios in 1975, length of primary cycle in 1975 are from UNESCO (1987) 1 data on per capita GP to 1985 are from World Bank (1986)1 and data en X surviving to end of primary cycle in 1975 are from World Bank (1986). Note: ** denotes that the coefficient is statistically significant at the 5 percent confidence level. Overall, the model explains a remarkably high proportion (70 percent) of the variance in literacy rates across countries. L/ All the independent variables have the expected sign and are statistically significant at the 5 percent confidence level. An increase of one standard deviation in the primary enrollment ratio (from the sample mean of 79.7 percent to 107 percent) lifts the adult literacy rate by 19.1 percentage points, from the sample mean of 59.3 percent to 78.4 percent. A comparable one-standard- deviation increase in the cohort survival rate, or in the length of the primary school cycle would raise the adult literacy rate by 6.2 and 7.5 percentage points respectively. The results confirm an intuitive idea: simply expanding coverage does not maximize the achievement of literazy in a population. Also important is that pupils persist through the system once they are enrolled.2/ Most j./ The regression results are suggestive only, however, owing to questions about the source and reliability of data on literacy rates. 2/ Persistence through the system is important for two reasons: people may not learn sufficiently to be literate if their education is interrupted too early, and even if they did, they may lapse into illiteracy (Lestage, 1981). - 84 - Asian countries have achieved, or are currently close to achieving, universal coverage in primary education. Improving persistence through the system therefore becomes important if literacy rates among future adults are to rise. To this end, two types of policy options exist: enhancing cohort survival rates, while keeping unchanged the length of the primary cycle; or lengthening the cycle, on the assumption that current rates of cohort survival remain constant. Both strategies raise the average years of schooling in the population, but the first is probably less elitist in its impact. Another argument for choosing it is that primary schooling in all Asian countries already exceed the minimum four years generally considered necessary for achieving permanent literacy (see appendix table 11.1). The scope for improving cohort survival rates is addressed below as an issue on internal efficiency. 4.2 The Internal efficiency of the education system An education system is judged to be internally efficient if it produces the desired output at minimum cost; or equivalently, if for a given input of resources, it maximizes the desired output. The latter is often measured by such indicators as cohort survival or retention rates (that is, the proportion of pupils persisting to the end of the cycle), student's cognitive and technical skills, scientific know-how, conformity to social norms of behavior, and so on. A wide range of policy interventions exist for enhancing a system's internal efficiency. They include such choices as the selection processes that affect students' progression through the system; the mix of school inputs and the use of alternative pedagogical methods for maximizing student achievement; the curriculum design; the incentives to encourage students, teachers, and other actors in the system to perform their best; the spatial distribution of schools; the grouping of students in classes; and so on. The following sections will touch upon selected topics in this regard. * 85 - (a) Efficiency of selection arocesses in orimary and secondary education Selection in the education system occurs via various mechanisms, the result of which is reflected in the pattern of cohort survival IQ/. It is useful to distinguish between selection that takes place within and between cycles of education since this helps in gauging the internal efficiency of the system. A high rate of survival (or retention) within cycles of education, particularly in primary and secondary education, is a necessary, though insufficient, mark of efficient systems; conversely, systems that exhibit low intra-cycle retention rates are invariably inefficient ones. The reason is that the curriculum for a given cycle of study is normally designed to impart and reinforce certain cognitive skills, so that students who exit before the end of the cycle will acquire these skills not only partially, but probably also temporarily. To the extent that this outcome holds, the resources invested in their education would be wasted, leading to inefficiency in the system. (i) Patterns of cohort survival in Asia. The relevant data appear in table 4.5. 11/. Also presented for completeness are data on the gross enrollment ratio in primary education, and the estimated proportion of school- age children entering grade one. 10/ From this pattern we can calculate the cohort survival rate (CSR) (or equivalently, the cohort retention rate), defined as the proportion of grade one entrants who survive to a given grade, say, the end of the primary cycle. The complement of CSR is the cohort dropout rate (CDR), defined as the proportion of grade one entrants who dropout before reaching that grade. By definition, CSR - 1 - CDR. Thus, the two indicators can be used interchangeably. A common mistake is to confuse the CDR with the rate obtained by dividing the number of pupils who dropout in any one year by the enrolled population. This calculation yields the period dropout rate (PDR) since it refers to the incidence of dropping out at a specific calendar time. The PDR is often much smaller than the CDR because its denominator includes people from several cohorts (the number of cohorts depending on the length of the cycle of education in question). The two statistics have very different uses: the CDR is relevant for assessing a system's internal efficiency and its pedagogical performance; the PDR is more relevant for space planning purposes. 11/ These profiles were derived from aggregate time-series enrollment data for the mid-1980s, and were adjusted for the incidence of repetition. As a result, they reveal patterns of survival, not in terms of years of schooling, but of grade attained. For more information on the year to year survival pattern, see the appendix table on which table 4.5 is based. That table also retains information on the length of primary and secondary education in each country, the end of each cycle or sub-cycle being denoted accordingly. - 86 - Table-4.5s Patterns of cohort survival in primary and secondary education, mid-1980s I grade 1 entrants suriYIna to Primary gross X population enrollment entering End of Lover secondary Ver secondar ratio, 1985 grade 1 gXIMg l jegg Lasar Ist year Last year I Bangladesh 60 100 24 22 11 5 4 2 Shutan 25 54 17 8 7 3 3 4 China 118 90 68 41 31 7 6 5 India 92 83 37 31 22 17 11 a/ 6 Indonesia 118 100 60 37 34 18 18 7 Korea 96 100 97 95 93 46 44 8 Laos 94 100 40 30 20 11 7 9 Malaysia 99 100 97 78 70 42 40 10 Nepal 79 75 33 32 28 25 21 It Papua Nev Guinea 69 74 61 25 16 2 2 12 Philippines 106 100 66 56 41 na na 13 Singapore 115 100 100 75 75 20 20 14 Sri Lanka 103 100 85 76 57 19 19 I Thailand 97 100 80 32 29 15 13 Regional average 91 91 62 46 38 18 16 Source: see appendix table 32.2. at Some enrollments in tertiary education in fact correspond to the last two years of the four-year upper secondary cycle. For this reason, data from schools show a much smaller number reaching the end of upper secondary. The data from UNESCO have been adjusted accordingly. Access to primary schooling is universal in all but 5 of the 14 Asian countries in the sample (note, however, possible overestimation of the entry rate for Bangladesh, as indicated in appendix table B2.2). China and India are only a short way from that goal; Nepal and Papua New Guinea are somewhat further away; but Bhutan is very far away from it. Perhaps more dramatic, however, are the differences across countries in the pattern of cohort survival through the education system. In the primary cycle, for example, dropping out is negligible in only four countries - - Korea, Malaysia, Papua New Guinea, and Singapore. At the other extreme are countries like Bangladesh and Bhutan where dropping out occurs so frequently that only 24 and 17 percent, respectively, of all grade one entrants reach the last grade of primary schooling. It is of interest to note that countries with comparable gross enrollment ratios can have very different student flow patterns. An example is the case of Bangladesh and Papua New Guinea (whose gross enrollment ratios are, respectively, 60 and 69 percent). Bangladesh's figure reflects a high entry rate, combined with a high drop out rate, while that of Papua New Guinea reflects a somewhat lower entry rate, combined with a much smaller dropout rate. These characteristics have very different implications for designing appropriate strategies for expanding coverage: in Bangladesh, the main focus should be on improving the system's capacity to retain pupils, but in Papua New Guinea, it should be on attracting more school-age children to enter the system. - 87 - Another noteworthy point is that a high gross enrollment ratio does not necessarily imply universal or close-to-universal coverage. India's figure, for example, is probably inaflated by the presence of overage pupils,12/ since the entry and survival rates ia primary education suggest that a much smaller proportion of the of-age pupils are in fact enrolled.1,/ Similarly, caution is advisable in reading the data f-,r such countries as China, Indonesia, Laos, and the Philippines. With regard to the pattern of survival for the system as a whole, big differences exist in the proportion of grade one entrants reaching the end of secondary schooling. Korea and Malaysia belong in one group, with a survival rate of about 40 percent; Indonesia, Nepal, the Philippines,IA/ Singapore, and Sri Lanka comprise another group with a moderately high survival rate of about 20 percent; India and Thailand form a third group with a survival rate around 12 percent; the remaining five countries have very low rates of survival, averaging no more than 5 percent. (ii) Comparing intra- and inter-cycle selection. The relative importance of the various loci of selection is revealed by the data in table 4.6. The first three columns show an expected pattern, with the proportion dropping out within each cycle or subcycle diminishing with rising levels of education.2/ The exception is Papua New Guinea, where the survival rate in lower secondary education is actually slightly smaller than that in primary education. Across countries, the survival rate in lower secondary education is low only in Bangladesh (less than 50 percent); at the upper secondary level, India's statistic lags significantly behind that of other countries. I/ The presence of underage pupils can also inflate the gross enrollment ratio, but the number of such students is probably limited. I/ If dropping out occurs uniformly throughout the system, the net enrollment ratio (defined as the ratio between the number of pupils in the school-age range to the total population in that age range) can roughly be estimated by taking the average of the entry and survival rates. Note, however, that in some countries (such as Bangladesh, Bhutan, and Nepal), dropping out actually occurs much more frequently in the early years of the primary cycle, so that this method of estimation would be invalid. The relative size of the overage population is indirectly revealed by the difference between the gross and net enrollment ratios. IA/ Although the Philippines does not have a formal system of upper secondary educatizn, many students in fact enroll in the equivalent of such education in non-degree courses at the tertiary level. The proportion entering such courses is probably between 15 to 25 percent of the cohort. L/ This pattern is to be expected because those who have reached higher levels are precisely those that have a lower propensity to drop out. - 88 - Table 4.6s Intra- and Luter-cycle selection in primary and secondary education, Asia, mid-1980s Percent of first year entrants suryvianto last "ear in ovTeal I Index of extent of inter-cycle pia Lammasea. RIyaSLx aslet b/ 1 Bangladesh 24 48 80 8 2 Shutan 17 83 88 13 4 China 68 76 81 54 5 India 37 72 65 15 6 Indonesia 60 92 96 46 7 Rorea 97 98 95 87 8 Laos 40 65 68 21 9 Malaysia 97 90 96 79 10 Nepal 33 89 81 5 11 Papua New Guinea 67 63 95 57 12 Philippines 66 74 - 18 13 Singapore 100 100 97 99 14 Sri Lanka 85 75 100 58 15 Thailand 80 91 87 72 Average 62 80 87 45 Sourcess as in table 4.5. al The denominator for each column to the number of entrants to the corresponding cycle. b/ The Index is defined as the ratio between the proportion of grade 1 entrants eliminated at the transition between cycles of education to the total proportion eliminated from the system by the end of secondary schooling. It ranges from 0 to 100s the larger it is, the move efficient is the selection process in the education system. The last column of table 4.6 is an index defined to reflect the relative weight of intra- and inter-cycle selection (see footnote b/ in table 4.6 for the precise definition). The greater this index is, the more intense the selection between cycles is; or equivalently, the less students drop out within cycles of education. Thus, systems with a higher index have a more efficient selection process than those with a lower index. Note, however, that the value of the index depends partly on the size of the group that drops out between grade one and the end of secondary schooling. As a result, comparison is probably valid largely among countries that are similar in this respect. Of the countries with the lowest survival rates to the end of secondary education (Bangladesh, Bhutan, China, Laos, and Papua New Guinea) China's and Papua New Guinea's systems achieve the best results (even though dropping out is nevertheless still high within primary and lower secondary education), while Bangladesh's lags the furthest behind. In the second group of comparable countries (India and Thailand), the performance of India's system lags considerably behind that of Thailand. In the third group (Indonesia, Nepal, the Philippines, Singapore, and Sri Lanka), Nepal's index (and to some extent, also the Philippines') is noticeably worse than that of the other countries. Finally, both countries in the fourth group (Korea and Malaysia) appear to be comparably efficient in the selection process. * 89 - (iii) Linkages between cohort survival patterns and country characteristics. To what extent are the cohort survival patterns discussed above linked to variation in economic conditions across countries; and to what extent to differences in policy choices? First, consider the pattern of cohort survival in primary education. Figure 4.1 indicates that in general, the poorer a country, the lower tends to be the proportion of grade 1 entrants reaching the end of the primary cycle. However, there remains considerable variation around this overall pattern; for example, China and India share similar levels of per capita GNP, but the cohort survival rate in the former is 68 percent, compared to only 37 percent in the latter. This result suggests that low cohort survival rates are not inevitable outcomes in poor countries, and that policy choices within primary education probably also play a role. Countries in which cohort survival rates are currently below the level suggested by the per capita GNP include Bhutan, India, Indonesia, Laos, Papua New Guinea and the Philippines. Servial late (X) 13 80 60 -* 40 * 10/ * 8 20 - 0 l i A t I pppI I I I B l 100 1000 10000 Pet capita GIP (IS$. lot seate) Sarce: see Alpees1 C. Logge 441 Relationship betwee survival rates in primary education and per capita W1P, Asia, aira 1985. The specific design of effective interventions to enhance retention rates in primary education depends on country conditions, and is beyond the scope of this study to examine in detail. However, it is of interest to note - 90 - that cohort survival rates are low in poor countries partly because of inadequate levels of resources per-pupil in primary education (measured as a percentage of the per capita GNP).16/ This relationship is suggested by figure 4.2 which reveals an overall pattern of rising cohort survival rates with rising levels of per-pupil spending.2/ The argument for more resources should not be pushed too far, however, since wide variation exists around the average pattern depicted in the figure. For example, Bangladesh and Sri Lanka share the same level of unit operating costs (as a percentage of the per capita GNP), but the retention rate in primary schooling is 85 percent in the latter country, compared to only 24 percent in the former. There is probably potential for improving survival rates through a better use of existing resources iii Bangladesh, Indonesia, Nepal, Papua New Guinea, and Thailand. Some increase in the level of per-pupil spending is probably also called for Bangladesh, India, and Nepal. The aggregate nature of the data in this study precludes detailed examination of how increased funding should be allocated. Indeed, since country conditions are likely to vary widely, allocations that are suited to one setting may be inappropriate in others with different initial characteristics. It is nevertheless noteworthy that survival rates show only a weakly negative link with pupil-teacher ratios (figure 4.3). Thus, it appears that raising per-pupil spending in the form of reduced class sizes is probably relevant mainly in such countries as Bangladesh and India, where retention rates are extremely low and pupil-teacher ratios exceptionally large. On the other hand, in countries where the average pupil-teacher ratio is not large, say not over 40, the indications are that such a strategy may not be the most appropriate one. These cot, lusions obviously require further confirmation in light of country-specific conditions, particularly because the data reflect averages that may conceal highly skewed distributions across localities. lj/ As will be shown in a later chapter, the level of per-pupil spending in primary education (measured as a percentage of the per capita GNP) is more a reflection of priorities embodied in public policies in the sector, than of inevitable links between the level of spending and a country's per capita GNP. 12/ A simple straight-line regression linking the cohort survival rate to the unit operating cost shows an R-squared statistic of 0.141 if the data for Papua New Guinea (which is an outlier) were included; and 0.399 if it were excluded. The regression coefficient on the unit cost variable is +0.375 and +0.632 respectively. - 91 - 120 sUtvival gate (m) 100 -,..? 0 £4 80 4 it 60 -* 40 - a Ina 20 *e 0 0 5 10 15 20 25 30 Vai costs (as % of per capita GNP) Beotems: see appeasts 3. nIe 42: Relatlon~hip between cohort surval ratae and uiat operating costa ta primary educatton, As~a, circa 1985. 120 Cokott srvival sate (1) 100 -, 190 - as 60 60 -69 40 - 8 20 0 10 20 30 40 50 60 Pipti-teacher ratlio 8asreesa ses aspeess: 3. FL==r 4.3: RelatlnaMp betwen co~ort survival rates and pupl-teacher ratios ta prlm~y school , Aala, circa 1985. - 92 - Turn now to consider the relative importance of inter- and intra- cycle selection in an education system (up to the end of secondary schooling). Figure 4.4 suggests that the poorer a country, the more does selection take place within cycles of education.M/ As before, however, wide variation exists around the average pattern. For example, although the Philippines and Thailand are at similar levels of economic development (as indicated by the per capita GNP), the gap between them in the efficiency of the selection process is very wide: in Thailand, 80 percent of the selection in the system takes place between cycles of education, whereas in the Philippines, it is only 18 percent. A similar comparison can be made between China and India; and between Sri Lanka and Laos. It appears, therefore, that the efficiency of the selection process depends not only on a country's level of economic development, but also on policy choices in the sector. Index of ltercycle selection 100 - 7 60 - 6 40 - 8 20 - a s 10. 100 1000 10000 Pet capita GNP (IS$, 1of scale) Setsm Appeals 0. ZIMM4.A: Relationship between the relative Importance of intercycle selection In primary and secondary education and per capita Q1, Asia, circsa 1985. Al/ Recall that pedagogical and economic considerations suggest that intracycle selection is probably less desirable than intercycle selection. - 93 - (b) Student achievement in basic education Cognitive skills are desired outputs of the education process. Thus, an education system or school that maximizes student achievement for a given input of resources can be considered internally efficient. Its efficiency derives from using school inputs that have the greatest impact on student learning per unit of resources spent. This aspect of internal efficiency is not easily assessed in the context of studies, such as the present one, in which countries (rather than schools or individual students) are the basic units of observation. A first problem is that internationally comparable data on achievement are scanty, particularly for developing countries. However, even if abundant, such data provide a limited basis for identifying inefficient systems. For example, not all low-scoring countries have education systems that are internally inefficient, since their poor performance could well be the result of inadequate levels of funding, and not of inefficient operations. A further limitation is that clearcut conclusions about efficient mixes of school inputs are hard to derive: countries differ widely in their initial conditions (including current provision of the different school inputs, and their relative prices), so that even if an input were shown to be generally effective in lifting achievement scores, it does not follow that increasing its provision would be the most efficient intervention in all settings.12/ With these caveats in mind, the following discussion will be limited to a simple presentation of the latest available achievement data. Under the study conducted by the International Association for the Evaluation of Educational Achievement (IEA), separate science tests were administered to 10- and 14- year old students in a number of countries.29/ The results appear in table 4.7. China and Papua New Guinea also participated in the study, but their data are as yet unavailable. Among the five developing Asian countries in the study, only Korea's achievement scores exceed the sample mean for both the 10- and 14- year old populations. On the other hand, the performance of Philippine students lies well below the average. The Roh statistic also provides some interesting insights into the operation of each country's education system. A value of 0.16, for example, I2/ It is generally known, for example, that textbooks have an important impact on student achievement. However, Paderanga (1987) and Jimenez, Paqueo and de Vera (1988a) found in separate studies that their impact is limited in the Philippine context. One plausible explanation is that textbook provision has already reached more or less adequate levels in most Philippine schools. As a result, this input has a smaller marginal impact on student achievement than other inputs, such as the subject knowledge of teachers. 2g/ The test for each population was identical for all the 24 countries in the study. Data are currently available for only 17 of these countries. In some countries, only one of the two populations was tested. * 94 - indicates that the variance in achievement between schools is 16 percent of the total variance in achievement in the population. Thus, the larger this statistic is, the more schools in the country differ in the academic performance of their students. This result may reflect a "deliberate [government] policy of differentiation among schools in terms of resources and curriculum," (IEA, 1988, pg.8), as is probably the case in Singapore. In the absence of such policy, a high Roh statistic suggests that there may be inefficiencies in the allocation of resources across schools, and in the ure of resources within schools. This situation probably exists in the Philippines. Table 4,7s Solence achievement in Asia and other countries, mid-1980e 1Q-wear-old Ronlation 14-year-old Dogulation NOVID. dev. Norm. dev. a h b from jean ofo c b fo M of Asian countrs long Kong 11.2 - -1.1 16.4 0.29 -0.6 Korea 15.4 0.16 1.4 18.1 0.15 0.2 Philippines 9.5 0.56 -2.1 11.5 0.48 -2.9 Sinsapore 11.2 0.39 -1.1 16.5 0.56 -0.5 Thailand - - - 16.5 0.24 -0.5 Selected developed countrkes Japan 15.4 0.04 1.4 20.2 0.04 1.2 Finland 15.3 0.07 1.3 18.5 0.05 0.4 Sweden 14.7 0.03 0.9 18.4 0.08 0.4 United States 13.2 0.14 0.1 16.5 0.29 -0.5 Mean score for all countries 13.1 - - 17.6 - - in the study at Std. deviation of mean score 1.7 - - 2.1 - - Maximan socre 24.0 - - 30.0 - * Source: IEA (1986) a/ In addition to the above countries, the others were Australia, Canada, England, Sungary, Italy, Netherlands, Norway, and Polands for the 10-year old population, data for 15 countries (excluding Netherlands and Thailand) were availables while for the 14-year old population, data for 17 countries were available. b/ The Roh statistic indicates the proportion of the total variance in achievement score that is accounted for by between school differences in achievement. *I Calculated by dividing by the sample standard deviation, the difference between a country's mean score and the mean score for all countries in the sample. * 95 * (c) Private financin as an incentive mechanism The internal efficiency of an education syrtem or school depends not only on its administrative characteristics, but also on the incentives that motivate the behavior of students, teachers, and school administrators. These incentives may exist in several forms. The testing and evaluation of achievement constitute an incentive mechanism, since they can stimulate students and teachers to better performance by clarifying expectations, and providing the basis for making comparisons.21/ Merit pay is another arrangement that is oometimes thought to promote effective teaching by tying a teacherOs pay to students' academic achievement. At the institutional level, financing arrangements are yet another incentive mechanism. A 2riori, the expectation is that the more a school depends on private financing, through fees collected from students, and/or contributions from the local community, the more it is likely to use resources efficiently, that is, to provide the services in demand at least cost. The reason is that when people share directly in the cost of a service, they are likely to monitor costs more closely, and to guard against waste.2/ Such behavior on the part of students and their families promotes greater efficiency by sharpening cost-consciousness among school managers, and creating pressures for greater accountability from them. Note that even when public institutions charge no fees, the incentive for them to achieve greater efficiency can be generated simply by allowing fee-charging private institutions to emerge and survive. The presence of such institutions helps to promote a competitive climate in the system, and provides information (on such things as costs and student achievement) for judging the performance of public institutions. Even if no explicit comparison is made, outright wastefulness and inefficiency in the public institutions are less easily concealed from public scrutiny. The argument for private financing is of course strongest in higher education, since these efficiency considerations are reinforced by considerations of equity. Evidence on the impact of private financing is scanty, due partly to the lack of data, and partly to researchers' inattention to the issue. Jimenez, Paqueo, and de Vera (1987b) appear to have been the first to provide 2/ The IEA study cited above is an example of how testing can provide a useful incentive for low-scoring countries to examine the operation of their education systems more closely. 22/ When part of the cost of education is borne by a student and his family, the student is also more likely to avoid waste in his own behavior. For example, in education systems with no tuition charges, and generous stipends (which are untied to academic achievement), students tend to take longer than the normal time to complete their course. On the other hand, students who pay for their education are more likely complete their course on time, and sometimes even before the normal time. - 96 - some evidence in this regard. Their analysis (based on data from the Philippines) indicates that unit costs across schools varied inversely with the degree of dependence on local sources of funds, a finding that persists when differences in quality are taken into account. The authors argue that it suggests that administrators in schools financed by local funds face greater incentives to minimize costs while maintaining quality. A second piece of evidence is based on data collected for the present study. The analysis focuses on the impact of private financing in higher education, using individual Asian countries as units of observation. The hypothesis is that the more a country relies on private financing, either by charging fees in public institutions or by encouraging self-financing private institutions, the less costly are its gublic institutions likely to be. The estimated regression equation is as follows (t-statistics in parentheses): UC - 2.22 - 0.028 x PF + 0.0096 x PCGNP (3.01) (1.59)* (0.13) R-squared - 0.27; N - 11 * statistically significant at 15 percent confidence level. UC represents the unit costs of public higher education. It is important to emphasize that UC refers to overall costs, not just the part financed by the government. It is expressed in terms of a country's per capita GNP, and then divided by the average figure for the sample. Measured in this way, UC yields a direct sense of the costliness of a country's public higher education relative to that of its neighbors. PF is an index of the extent of private financing in the system. It is a composite variable derived by weighting the distribution of enrollments between public and private institutions by the respective rate of cost recovery in each sector. Thus, it takes into account not only the existence and level of fees in public institutions, but also the extent of public subsidization of private institutions. PCGNP is simply the country's per capita GNP in 1985, expressed in units of US$1,000. The coefficient of the PF variable is negative and statistically significant at the 15 percent level of confidence. Interestingly, that of the per capita GNP variable is statistically insignificant, even though it has the expected positive sign. The results are tentative at this stage, given the small sample size; however, it cannot be ruled out that the more a system of education relies on private financing, the lower tends to be the overall unit operating cost of its public institutions.2/ This relationship is depicted graphically in figure 4.5. 21/ This result implies that in systems that rely on private financing, the government bears a lighter fiscal burden because it finances a smaller share of a lower overall unit cost. . 97 - It is noteworthy that the curve flattens out as the rate of private financing approaches about 40 percent. 2/ The marginal gain in efficiency thus starts to diminish significantly beyond this point. This result has an important policy implication, for although it argues in favor of a moderately high rate of cost recovery, it also cautions against extremely high rates.2,,/ The reason is that the tradeoff between efficiency and adverse social selectivity in the system tends to deteriorate with rising levels of private financing. Costiliess of public hig bet education 2- 44 5. \ -. * t 1 *6 * ** 0.5 14'~*------ * 12 0 0 20 40 60 80 100 Index of pilvate financing (%I) See chapters 3 & 4 fat doUlls oa to1t01ttles of iathes at blol ate. Fisure 4. 3 Relationaship between the costliness of public higher education and the overall extent of private fiaancag in the subsector, Asia, cirea 1985. 2A/ This pattern is of course based only on data from Asian countries. The results nevertheless provide a benchmark for judging whether or not the extent of private financing in a system is high enough to exploit its potential impact on efficiency in public higher education. 2j/ Recall that increased private financing in higher education can be accomplished in at least two ways: imposing/raising user charges in public institutions, and/or promoting the expansion of largely self- financing private institutions. * 98 - (d) Efficiency in the phsical organization of schooling The provision of education requires organizing students into discrete groups in schools and in classrooms.2&/ This aspect of schooling affects the efficiency (and hence the operating unit costs) of the education system largely through its impact on the utilization of staff time.22/ In all schools, a number of non-teaching staff are employed whose time is generally more fully occupied in large schools. Further, if the curriculum calls for the use of specialized teachers, as it often does in secondary and higher education, the time of teachers also tends to be more fully utilized. As a result, costs per student are often found to be lower in larger institutions than in smaller ones. 21/ How costs actually vary with school size depends on unique country characteristics with regard to the curriculum, and staffing norms and practices. The economic benefits of larger schools will therefore differ from place to place. In terms of practical feasibility, local conditions are perhaps even more crucial to consider. For example, in rugged terrain with sparse population and weak transportation networks, the consolidation of small schools into (a few) larger ones would greatly increase the average distance between pupils' homes and their school. As a result, transportation costs will rise (perhaps sharply), thus reducing the net economic benefit of larger schools; and some students may even be forced to quit. In such settings, it would be desirable, for equity reasons, to continue with small schools. Note, however, that the argument for siting schools close to pupils' homes is probably most persuasive for primary schooling. In secondary and higher education, it is less so partly because students are older, and can travel longer distances from home; and partly because the difference in unit costs between large and small institutions tends to be much wider.22/ Since local conditions are an important determinant of the behavior of education costs, no single optimal size can be specified for all types of institutions in all countries, nor indeed even for all regions within one 25/ In some situations, it also involves organizing pupils into smaller groups within classrooms, as in the case of multi-grade teaching (one teacher in a classroom of pupils at different grade levels). 22/ The grouping of students also affects efficiency in the utilization of physical facilities, but the costs affected are capital costs rather than recurrent costs. 21/ The issue of economies of scope (i.e. having more than one level of education under a single administration) is not addressed here. See, however, Jimenez (1986) for a discussion. 22/ At these levels of education, the provision of boarding is feasible as an option for enhancing the access to education for students from remote areas. - 99 - country. The following discussion is therefore intended only to illustrate the existence of economies of scale, and the potential benefit of improving the size distribution of institutions. The data relate to secondary schools in the Philippines, and higher education in China. (i) Economies of scale in Philippine secondary schools.3/ Evidence from a sample of 497 national secondary schools suggests that unit costs drop with rising enrollments. The decline tapers off, however, beyond enrollments of about 1,200 students (figure 4.6). This pattern exists in both comprehensive and vocational secondary schools, although unit costs are higher in the latter type of institution at all enrollment levels. Opetlag costs Per stidel (Pesos 3000 - 2500 - 2 00 0 * *. --. Compreheisive 1500 -Ariculture 10.... All sclool types 1000 - 500 - 0 " p p p p 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Etouseolat of school IThosands) Sots: lajal aed lea (1086]. Figute 4.61 Relationship between unit operating costs and school saie, secondary education in the Philippines, 1986. 3_Q/ Up until 1986, the public sector comprised national schools (funded directly by the central goverment) and local schools (funded by local governments, and through fees paid by students). In that year, however, the goverment nationalized local schools; and in the following year, it abolished fees for public secondary education. - 100 - The scope for exploiting economies of scale is substantial due to the predominance of small schools in the system at present (table 4.8). Nearly 30 percent of the national schools enroll less than 500 students. Among the newly nationalized local schools, the corresponding figure is nearly 85 percent for schools outside the national capital area (which comprise the major share of all local schools). As these local schools acquire the staffing standards of national schools, and as the use of specialized teachers become more widespread, unit costs are likely to rise sharply, by as much as 72 percent in the long run, according to one estimate (Mingat and Tan, 1988). The indications are that it would be increasingly inefficient to maintain small secondary schools. Table .8s Sise distribution of Philippine public secondary schools, 1984 Local schools National National stae ofaschool shoels. MLAea AftOther LMo < 100 0.2 0.0 6.9 100 - 200 3.6 1.1 26.8 200 - 300 9.8 3.3 27.2 300 - 500 16.1 8.8 24.0 500 - 1,000 32.7 13.2 12.4 1,000 - 2,500 23.9 33.0 3.1 2,500 - 5,000 11.4 34.1 1.2 > 5,000 2.4 6.7 0.4 All schools 1002 100 1002 Total enrollments 716 196 946 (000) Average school aie 1145 2150 373 Sources Mingat and Tan (1988) - 101 - (ii) Economies of scale in Chinese higher education. The existence of economies of scale is suggested by a World Bank (1986) study based on data from the 1980s for 136 institutions of higher education. The results, summarized in table 4.9, show that unit costs in all types of institutions fall sharply with rising enrollments. Among those offering science-based courses, for example, institutions with 4,000 students have unit costs which are only 59 percent that of institutions with 1,000 students. The decline in unit costs is most dramatic in colleges that specialize in the social sciences. At present, less than 10 percent of the country's 1,063 institutions enroll more than 4,000 students each; and nearly 60 percent enroll fewer than 1,500 students (table 4.10). So there appears to be substantial scope for rationalizing their size distribution. However, shifting toward bigger sizes would probably require a concomitant move away from excessive institutional specialization by subject area. In this regard, it is noteworthy that only among the comprehensive universities are large institutions common. All other types of institutions are more specialized, and very few of them enroll over 3,000 students. Converting such institutions into comprehensive universities therefore appears to be a necessary condition for improving the internal efficiency of the system. As a policy issue, it is probably of some importance, since over 95 percent of all institutions currently belong to the specialized group. Table 4.9: Economies of scale in Chinese higher education, 1980s Predicted unit cost with Reference enrollment ver Lastitution of bi unit cost a/ Tree of Institution (Ya) L 200 AAM 6.00 Agriculture 1,789 100 73 59 55 Science & technology .,751 100 72 59 54 Social sciences 1,321 100 63 45 39 Comprehensive 1,490 100 68 51 46 Teacher education 1,498 100 68 52 46 Sources derived from table 3.3, World Bank (China-1986). al Data refer to the unit coast predicted froe regression analysis for institutions with 1,000 students. b/ Data are presented as percentages of the reference unit cost. - 102 - Table.4.1 Sise distribution of tertiary Institutions Lt China by type, 1987 Size of Agriculture Science & Teacher Social halttain All M2es & forest technolgoty a/ education saesE Comsoehensive Othere bL < 300 7.2 1.4 3.3 1.5 8.8 2.1 27.0 301 - 300 8.1 5.6 3.8 4.2 12.4 0.0 23.6 501 - 1000 23.6 21.1 18.5 26.5 30.1 6.4 32.0 1001 - 1500 19.7 22.5 20.3 29.2 19.5 2.1 7.9 1501 - 2000 13.7 11.3 16.5 18.1 12.4 6.4 5.1 2001 - 3000 12.6 28.2 18.0 7.3 10.6 12.8 3.4 3001 - 4000 5.6 9.9 8.4 2.7 4.4 10.6 1.1 4001 - 5000 3.2 0.0 3.8 4.6 1.8 10.6 0.0 > 5000 6.3 0.0 7.4 5.8 0.0 48.9 0.0 All institutions 100 £ 100 Z 100 I 100 1 100 X 100 1 1001 Number of institutions 1063 71 394 260 113 47 178 Sources Governmnt of China (1987a). al Includes edical colleges. b/ Includes colleges of physiel education and fine art. (e) The efficiency of alternative trpes of hiaer education In all countries, higher education is the most costly part of the education system. For this reason, cost-saving interventions are especially relevant at this level of education. Exploiting economies of scale within conventional institutions is one option, as discussed above, but a more radical solution is distance education. In some countries, this alternative consists of correspondence courses offered by regular universities; in others, it involves a substantially larger operation in the form of open universities. In all settings, distance education is dramatically cheaper than conventional higher education. Of course the fact that unit costs are lower does not necessarily imply greater efficiency since completion rates are often also lower in open universities. This outcome occurs for various reasons: open universities tend to attract weaker students since entry is usually non-competitive; perhaps as important is the tenuous link that a student has with the institution, his teachers, and fellow students. Under such circumstances, his personal - 103 - motivation and discipline to complete a course lack the reinforcement that a more supportive environment might provide. Adjusting for differences in completion rates between open universities and regular institutions, would the former still produce graduates more cheaply? As before, the answer is likely to vary according to country conditions. Here, d.ta from Thailand illustrate the comparison (table 4.11). Table 4.11s Graduate output from various types of hiher education in Thailand, 1980s Public institutions Private SelectivA gM istutions Student-staff ratio 8.1 744.5 17.6 Ratio of graduates to enrolled population 0.258 0.050 0.172 Graduatea-staff ratio 2.1 37.3 3.0 Source: Gove=nment of Thailand (1985b). Thailand's system of higher education comprises three main types of institutions: selective regular institutions, and open universities in the public sector; and private colleges. Unit costs are lowest in the open universities because of the exceptionally high pupil-teacher ratios: the average is 745 compared to only 8 in the selective public institutions, and 18 in the private sector. However, the selective public universities achieve better graduation rates. On average, graduates in any one year represent 26 percent of the enrolled population, compared to only 5 percent in the open universities, and 17 percent in private institutions. The lower rates in the latter settings reflect one of two possibilities: students either take longer to complete their course, or they drop out. But despite the small completion rates, the open universities still produce more graduates per teacher than the regular institutions, owing to their high pupil-teacher ratios. Conventional universities achieve much less, but the performance of the public selective institutions appears to be particularly weak in this respect. It is to be noted that this comparison provides only a rough (and incomplete) assessment of the internal efficiency of the various types of institutions. The measure of efficiency used here -- graduates per teacher -- relates to only one aspect of the problem, since it reflects the aggregate production of graduates without differentiating by field of study. A more - 104 - general shortcoming is that the comparison does not deal with issues of external efficiency. Clearly, even though open universities produce graduates efficiently, the output would be wasted unless graduates are productive in the labor market. A more global comparison would therefore require assessment of the labor market performance of people who have followed different university careers. However, this type of analysis is beyond the scope of the present study since it requires survey data. * 105 - 5. Equity Considerations in Education The term equity has a wide meaning. In the broadest sense, it refers to the distribution of income, wealth, position, and power in society. Public policies are an important determinant of that distribution because they define the conditions and rules of interpersonal competition. In this regard, policies in the education sector are particularly relevant since they affect access to schooling, and subsequently, to jobs and income. For this reason, the impact of educational policies on equity is often a major issue in the sector. To place this chapter's discussion in perspective, it is useful briefly to recall the various analytical approaches for assessing equity in education. From the vast literature, three complementary types of analyses may be distinguished: those that (a) evaluate differences in the access to specific levels or types of education; 1/ (b) compare the distribution of benefits among people with different education;V and (c) assess who pays for and who benefits from the provision of education. The third approach is clearly the most comprehensive, but it usually requires survey data, thus making it unfeasible in the context of this study. The data and analysis presented below fall largely under the rubric of the first two approaches. The discussion has two parts. In the first, the .ducation systems of Asian countries are compared on the basis of their aggregate features. The focus is therefore on global aspects of equity in education. Building on this assessment, the second part addresses the issue of selectivity in the education system. J/ Studies in this tradition assume that education is beneficial without specifying the nature or value of the benefits. 2/ These benefits may materialize while a person is being educated or after graduation. In the first case, the benefits consist of the public resources that accrue to people who are enrolled in the system; in the second, those benefits appear in the form of higher earnings and upward social mobility. - 106 - 5.1 Global aspects of equity What aggregate features of an education system have important implications for equity? At least two features are relevant in this respect: the degree of emphasis on higher education in the system; and more generally, the distribution of cumulative public spending on education implied by the structure of enrollments and government subsidization across levels of education. These two features describe slightly different but complementary aspects of an education system: the first focuses mainly on relative emphasis Ithin the education system, whereas the second views the education system as a whole, and incorporates in the assessment the extent of non-enrollment. Beyond characterizing the degree of equity in the education system, the following discussion will also assess the potential impact of increased private financing as an policy option for enhancing overall equity in the sector. (a) Dearee of bias toward higher education The composite elements of a bias toward higher education include relatively high costs per student in public institutions; excessively wide coverage at this level relative to other levels; and extensive government involvement in the subsector as a whole, particularly in financial terms. The differences among Asian countries in each of these respects are discussed separately, and then summarized to compare the net degree of emphasis. (i) Pattern of unit operating costs across levels of education.3/ Unit costs expressed in terms of a country's per capita GNP indicate the pattern of resource-intensity across levels of education /. For a meaningful comparison of the structure of costs across countries and levels of education, unit costs are expressed as deviations from the regional mean at each level of education. 5/ The relevant data appear in table 5.1. 2/ These costs refer to total operating costs per pupil in public education, irrespective of their source of financing. 4/ Recall that this statistic is valid for the comparisons here since it is not related to per capita GNP. An alternative index for comparison is the ratio between a country's unit costs and the corresponding regional average at each level of education. But the resulting indices would be asymetric in the sense that for costs below the average the indices' potential range is zero to one, while for those above the average, the range is one to infinity. This problem is avoided when using deviations from the mean as a basis for comparison. - 107 - Table 5.11 Unit operating costs of public education, deviation fram regional mean, and Index of bias tovard higher education, Asia, mid 1980s Unit operating costs Percent unit costa deviate as I of ar canita GP_ from retional mean Index of bias toward higher Zriaxg Secondarw Las E.aimr Second= Hie education.a/ I Bangladesh 6.4 30.0 284.6 -37 58 86 61 4 China 9.2 27.6 243.8 -9 46 59 34 5 India 6.0 17.3 231.1 -41 -9 51 46 6 Indonesia 12.6 23.3 91.1 2S 23 -41 -3S 7 Kores 16.5 23.4 70.6 63 23 -54 -58 9 Malaysia 14.1 21.3 190.3 39 12 24 -8 10 Nepal 9.0 13.5 249.0 -11 -29 62 37 11 Papua New Guinea 29.0 65.0 1050.0 187 243 585 199 12 Philippines 5.8 8.6 50.0 -43 -55 -67 -12 14 Sri Lanka 6.1 9.3 83.3 -40 -51 -46 -S 15 Thailand 15.5 15.3 39.9 53 -19 -74 -64 Regional average Excluding 11 10.1 19.0 153.4 - - - - Including 11 11.8 23.1 234.9 - - - - Sourcess as in table 3.2. a/ The index is calculated by taking the average of the lines Joining the index of deviation for the three levels of education. For Chins, for example, the figure of 34 - [(59+46)12 - (46+(-9))12)]. See also figure 5.1. The deviations from the mean at each level of education are plotted in figure 5.1, each country being represented by a set of three bars. Taken together, the size of the bars and their location provide an indication of overall costliness. In Papua New Guinea, for example, all three bars lie to the right by substantial margins, reflecting exceptionally high costs relative to other countries in the region. The opposite is true of the Philippines and Sri Lanka. Perhaps of greater interest here is the relative size and location of the three bars within each country, for they reveal the pattern of relative resource-intensity at the three levels of education. In countries where deviations increase with rising levels of education, there exists a bias in resource-intensity toward higher education (e.g. Bangladesh). Countries with the opposite pattern exhibit a bias in favor of the lower levels in this - 108 - respect (e.g Korea). For ease of comparison, the pattern of bias is summarized by calculating for each country the average rate of increase in deviations from the mean as the level of education rises from primary to higher education. 6/ The result is the index reported in the last column of table 5.1. EPrimary 0Secondary r7ligier Bangladesh - China - Ind ia - Indonesia - o1ea - lalaysla - lepal - Papua lev Guinea - Philippines - Sti Lanka - Thailand - -100 -75 -50 -25 0 25 50 75 100 % Bill costs deviate 1101 legional Real $uies: see tabia 1.1 lola: devisshas a 1ss1 sta4 at 103 ja tte . Flaie 5.11 Deviatien of the unit operatiag costs of public education from the regional average, Asia. oirea 1985. See footnote to table 5.1 for an example of the calculation. - 109 - Countries differ significantly in the bias in resource-intensity toward the higher levels of public education. In one group comprising Bangladesh, China, India, and Papua New Guinea, the deviations increase rapidly with rising levels of education, indicating a definite bias toward the higher levels. This pattern also exists in Nepal, but the bias is more moderate since secondary education is relatively less resource-intensive than primary education. In Sri Lanka, the pattern is generally balanced across the three levels of education. In the Philippines, and Malaysia there is a moderate bias in favor of the lower levels of education, while in Indonesia, Korea and Thailand, this tendency is stronger. In the last two countries, for example, the relative unit cost in public primary and secondary education is two to three times that of public higher education, a pattern denoting a clear emphasis in resource-intensity toward basic education. (ii) The quantitative develoMent of higher education. It appears attractive at first sight to develop an index of bias similar to the one above by comparing a country's enrollment ratio at the different levels to the average for the sample. However, this approach is flawed because enrollment ratios are linked, even if roughly, to a country's level of economic development, and so poor countries will always tend to achieve below average coverage. A better approach is to compare the actual enrollment ratios in a country to that predicted on the basis of its per capita GNP (table 5.2). Table 3,.: Actual and predicted esollment ratios in Asia, aid-1980s Aatual.enrellment.ratios (X) . rediated enrollment ratios fl) Per capita (us ) P Secondary $Lahor primar Seondary Righl 1 sangladesh 159 60 18 5.2 82 24 5.0 4 China 273 118 39 1.7 63 27 6.1 5 India 259 92 35 9.0 83 27 5.9 6 Indonesia 503 118 42 6.5 86 35 8.2 7 Norea 2,040 99 74 31.6 103 72 18.8 9 Malaysia 1,860 99 53 6.0 101 69 17.9 10 sepal 142 79 25 4.6 82 23 4.8 11 Papua New Guinea 621 64 14 1.7 88 38 9.2 12 Philippines 561 106 65 38.0 87 37 8.9 14 Sri Lanka 374 103 63 4.6 85 31 7.0 15 Thailand 712 97 30 19.6 89 41 10.0 Sourcess As in table D1.2 for data on actual enolleAt ratios and per capita 0UPs predicted enrollment ratios are derived from the rearessions for Asia in table C.1. - 110 - For primary education, this comparison does not work as well as at the other levels because most Asian countries are approaching universal coverage (as far as official statistics are concer;ed). As a result, the differences in coverage among countries tend to be compressed. More importantly, the gross enrollment ratio is flawed as an indicator of coverage particularly at the primary level, since i.' is contaminated by the pattern cohort survival, as the analysis in the previous chapter shows.2/ In view of these methodological and data problems, it is not possible to derive a quantitative indicator of bias in the coverage of the education system. However, a qualitative sense of this bias emerges by assessing the size and direction of the deviation of actual enrollment ratios from those predicted from the per capita GNP (figure 5.2). 2/ At the other levels, the cohort survival rate is generally much higher, and there is greater similarity among countries. - 111 - Primary B Secondary C- Higher Bangladesh - Bhutan.... Burma , INepla - Papia Nev Guiea - Philippines - Si Lana - Thailand - -80 -60 -40 -20 0 20 40 60 80 % actual tatio deviates from predicted loti see table 5.3. ate: 6108viats ta et Us ata at pius $ad alads 80 5, Tinure 5.2 Deviation of actual enrollment ratios frets those predicted on the basis of per capita GIP, Asia, cirea 1985. As an example, consider the data for Bangladesh and Malaysia. The former country's gross enrollment ratio for primary education is 60 percent compared to a predicted value of 82 percent. In higher education, the actual ratio is slightly larger than the predicted figure. These comparisons suggest a relative emphasis on the quantitative development of higher education. In Malaysia, on the other hand, the enrollment ratio for higher education is only 6 percent, compared to the predicted value of 18 percent. Since primary education is universal in this country, this result suggests a relative emphasis on the coverage of primary rather than higher education. 112 - Following this procedure for all countries in the sample, a ranking can be made of their emphasis on the quantitative development of higher education. Bangladesh, India, and to a lesser extent, the Philippines, aty be characterized as countries with a strong bias in favor of higher education. In Korea and Thailand, this bias is less pronounced, since basic education is generally well-developed and does not suffer from high dropout rates. The coverage of Nepal's system is broadly balanced across the three levels of education. In Indonesia, Malaysia, Papua New Guinea, and Sri Lanka coverage is moderately skewed in favor of the lower levels of education. In China, this bias appears to be stronger. (III) Extent and nature of government involvement. The government's involvement in higher education is reflected not only by the share of enrollments in public institutions, but also by the extent of public financing. The latter is mainly an outcome of the strategy used to accommodate the demand for places. Vo minimize the costliness of higher education to the public purse, countries may use three (not mutually exclusive) options: low-cost.distance education; (largely) self-financing private education; and user charges for public higher education. The degree of reliance on these instruments provides a qualitative basis for assessing the costliness to the government of higher education as a whole. Countries vary widely in this respect. Korea, for example, exploits all three options, while the Philippines and Thailand rely heavily on only one of them (self-financing private education in the former, and low-cost distance edication in the latter). In all three countries, the result is less government involvement in the financing higher education than in other countries. On the other hand, countries like Bangladesh, India, Malaysia, and Papua New Guinea fall at the high end of the spectrum, since they rely little on all three instruments.§/ Between these extremes are the other Asian countries. In Indonesia, the extent of public subsidization is not ranked as low as in, say, Korea, even though the country has a large private sector and an open university. The reason is that its private sector is heavily subsidized by the government through the secondment of civil servants to teaching posts in private institutions. China's ranking is moderately high, despite the presence of distance education, because the system fails to exploit the substantial economies of scale associated with it. Nepal is ranked in the middle of the spectrum, since self-financing private education is moderately well-developed. Sri Lanka is also ranked in this position because the country has a reasonably large system of low-cost distance education. I/ In the former two countries, the share of enrollments in the private sector is significant, but private institutions are heavily subsidized by the government. - 113 - (iv) Overall assessment. An education system's performance along each of the three dimensions discussed above can be scored using pluses and minuses.2/ Adding these qualitative scores provides a consolidated index of the overall bias toward higher education. The results appear in table 5.3. This composite index is clearly only a rough indicator of the government's implicit priorities across the various levels of education, revealing broad rather than fine differences across countries. It is nevertheless a useful index, since large gaps among countries can be interpreted as signifying the existence of real differences in educational priorities among them. Table 5.3: Relative degree of emphasis on higher education in Asia, mid-198)s a/ Structure of Costiness to Overall degree pubie unit costs geggge the severnment of empbasis bl I Bangladesh + + + + + + + 6 4 China ++ - + + 1 5 India + + + + 5 6 Indonesia - * * * 3 7 Korea *- + -- - 3 9 Malaysia - * ++ 0 10 Nepal + 0 0 + 1 11 Papua New Guinea .+ + + + + 4 12 Philippines - + * - - 2 14 Sri Lanka 0 * 0 - 1 15 Thailand + - - 2 Sources authors' assessment based on data on unit costs, coverage, and the extent of public subsidisation of higher education. See discussion in text for additional Information. al Positive signs signify a bias toward higher education in the country compared to its Asian neighborss a negative sign, the opposites while sero's denote the absence of a bias in either direction. The larger the number positive or negative signs, the stronger is the bias toward higher education. b/ Derived simply by sumaing up the positive and negative signs in the first three columns, on the implicit assumption that each aspect of the bias toward higher education carries equal weight. 2/ The stronger the bias toward higher education is, the larger the number of pluses; the absence of a bias is scored with a zero; while a bias away from higher education is denoted by negative signs. - 114 - In general, the Asian countries in the sample fall into three groups. The bias toward higher education appears to be particularly strong in Bangladesh, India, and Papua New Guinea. In China, Malaysia, Nepal and Sri Lanka, this bias is largely absent. In the remaining countries -- Indonesia, Korea, the Philippines and Thailand - - the emphasis is on the lower levels of education. It is interesting to note that China's ranking reflects a strong tradeoff between costs and coverage. This tradeoff is missing in Bangladesh and India; and only mildly present in Papua New Guinea. To what extent is a country' s propensity to emphasize higher education linked to its level of economic development? A weak inverse relationship appears to exist, as indicated by figure 5.3 in which the index of overall bias is plotted against per capita GNP. This result implies that the poorer a country, the stronger tends to be the overall emphasis on higher education. However, there exists wide variation around this general pattern. Papua New Guinea and Thailand, for example, have similar levels of per capita GNP, but the bias toward higher education is much stronger in the former country. A similar comparison can be made between China and India. The implication of these comparisons is that poor countries are not necessarily condemned to systems that are skewed toward higher education: the degree of bias depends as much on policy choices as on conditions in the country. Iadez of emphasis on hitker education 8- 6 5. 4 oil 2 0 .44 9 -2 ~ .1.5 6" 7" -4 I i 1 8 a I a I 100 1000 10000 Pet Capita GNP (US$. 1og scale) See text xes ils e 4ad of FlaUse_5 Relationship between emphasis on higher education and per capita GNP, Asia, circa 1985. -115 - (b) Eauity jin the distribution of cumulative Rublic sending on education To analyze this issue, begin with a definition: cumulative public spending on education refers to the total amount of public resources appropriated to a generation of people passing through their schooling years. The concept of benefits is therefore a longitudinal one; it differs from that normally used in other analyses in which benefits refer to public spending on education in a given period (which could be a year or longer). The focus is on the distribution of cumulative benefits among people with different schooling careers, rather than on the distribution of single seriod benefits at a given level of education.j&/ Equity in the distribution of those benefits depends on two aggregate features of the education system: the structures of enrollments and public subsidization across the three levels of education. For this reason, the subsequent discussion will sometimes refer to a system's structural equity. The outcome depends more on broad policy choices in education than on specific interventions aimed at altering the social composition of individuals in the system. Thus, if a government allocates most of its spending on education to the higher levels to benefit a few people, leaving few resources for primary school pupils, it achieves a lower level of structural equity in the education system than a government that pursues the opposite policies. To quantify the degree of structural equity, the first task 4S to estimate the amount of public subsidies accumulated by people with different schooling careers. Data on the structure of recurrent public subsidies by level of education appear in the first three columns of table 5.4. They reflect the weighted average of subsidies in the public and private sectors.fl/ The amount accumulated by a person exiting the system with primary education is the product of this average and the number of years in the cycle.12/ A person exiting with higher education accumulates the sum of subsidies in all three cycles of education. The last three columns summarizes this calculation, showing the cumulative benefits per person according to the level at which he exits the system. IQ/ See Kingat and Tan (1985) for a fuller discussion of this approach; and Mingat and Tan (1986) for its application to an analysis of the distribution of cumulative public spending by socioeconomic groups. 11/ As such, they differ from the data on overall operating unit costs in table 3.2. See footnote b/ of table 5.4 for the details on the calculation of public subsidies per student in the system. The share of public subsidies received by people with the same terminal level of education will differ according to the type of institutions they attended. However, in the context of the global analysis here, this distinction is not taken into account. 12/ See appendix table B1.1 for data on the length of educational cycles in Asian countries. - 116 - Table 5.1 Average public subsidies per pupil, Asia mid-1980s a/ Subsidies hy level of education bi Cumulative subsidies of blamaT s***ndarr ishaba Priaxyg $00onds" igher 1 Bangladesh 5.3 28.0 250 27 223 1223 4 China 8.8 26.7 248 53 213 1205 5 India 3.2 15.0 160 26 116 836 6 Indonesia 10.8 16.3 30 65 163 283 7 torea 16.6 8.3 12 100 149 197 9 Malaysia 13.6 20.5 170 82 225 905 10 Nepal 8.5 7.2 170 43 79 759 11 Papua new Qaina 26.4 39.0 978 158 392 4306 12 Philippines 5.5 4.5 6 33 51 77 14 Sri Lanks 5.8 8.8 77 35 96 404 15 Thailand 11.6 13.3 30 70 149 269 Average 10.7 17.1 196 63 169 951 Sources authors' estimates based on method discussed in test. a/ Spending per pupil is expressed as a percentage of the per capita GP. b/ The data refer to the average over the public and private sectors at each level. They reflect overall operating costs per pupil mins the amount financed through fees, and are adjusted for the share of private education, and the extent of government subsidisation of that sector. of The data reflect the annual recurrent subsidies multiplied by the number of years in the cycles the figures for secondary ad higher education include the subsidies received in the previous cycles of education. The next step in the analysis involves estimating the distribution of educational attainment among the current generation of school-age children as they reach adulthood. The results, based on data on current enrollment ratios, appear in table 5.5 (first 3 columns). To illustrate the method of derivation, consider the data for Bangladesh. The proportion with higher education is the same as the current enrollment ratio (5 percent);/ V the percentage without schooling is the residual of the primary enrollment ratio (40 percent - 100 - 60). The proportion with primary education is the difference between the enrollment ratios in primary and secondary education (42 percent - 60 - 18); the share of those with secondary education is derived likewise (13 percent - 18 - 5). LV/ This calculation and subsequent ones implicitly assumes that dropping out does not affect the comparisons across countries, even though the actual profile of future adults' educational attainment in each country will in fact be influenced by it. - 117 - Zbek 5.3A Projected distrtbution of the educational attainment and cumative public subsidies among members of current school-age population, Asia aid-1980s (percent) a/ eet a of tamiatin with Share of cumulative subsidies (1) No No Scholan pugaim secondarr Hig SchoiA g SAua Hishe 1 Bangladesh 40 42 13 5 0 6 29 65 4 China 0 61 37 2 0 24 60 16 5 India 8 51 32 9 0 11 30 60 6 Indonesia 0 58 36 7 0 33 51 16 7 Korea 4 21 43 32 0 14 43 43 9 Malaysia 1 46 47 6 0 19 54 28 10 Nepal 18 57 20 5 0 32 21 46 11 Papua New Guinea 30 57 11 2 0 41 20 39 12 Philippine* 0 35 27 38 0 21 25 54 14 Sri Lanka 0 37 58 5 0 15 64 21 15 Thailand 3 67 10 20 0 40 13 47 Average 9 48 30 12 0 23 37 40 Scurcess authors' estimates based on data on enrollment ratios from Unesco (1987) and data on public subsidies per pupil presented in earlier chapters. In all but three Asian countries (Korea, the Philippines, and to a lesser extent, also Thailand), the enrollment pyramid is steep, since no more than 10 percent of future adults will exit the education system with higher education. In countries like Bangladesh, Nepal and Papua New Guinea, it is particularly steep, since a sizable group in the cohort does not even enter the system. The data in tables 5.4 and 5.5 (first 3 columns) provide a basis for estimating the distribution of cumulative public subsidies among people belonging to the same generation. / The results are reported in the last 3 columns of table 5.5. They show, for example, that in Bangladesh, the 42 percent in a cohort who exit the system with primary education accumulate only 6 percent of the total cumulative public subsidies appropriated to the entire cohort. On the other hand, those who exit with higher education comprise 5 percent of the cohort, but their share is an astonishing 65 percent. In contrast, the distribution in Korea, for example, is much less skewed. The 21 percent with only primary education obtain 14 percent of the cumulative subsidies; while the 32 percent with higher education get a share of 43 percent. I/ The calculation involves two steps. First, multiply the share of people with the various levels of terminal education (first 3 columns of table 5.5) by the corresponding amount of cumulative subsidies per person in the group (table 5.4) to obtain the group's total cumulative subsidies. Then divide the result by the cumulative subsidies aggregated over all groups to get each group's share of those subsidies. - 118 - To render the comparisons across countries more transparent, two summary statistics, based on the data in table 5.5, are presented in table 5.6. The first is the Gini-coefficient, an index of the concentration of cumulative public spending on education.,./ It has a value ranging from 0 to 100; the closer is the index to 100, the more is public spending on education concentrated among people who reach the higher levels in the system, and therefore, the more unequal is the distribution of those resources. The Gini- coefficient varies widely across Asian countries, ranging from less than 20 in Korea and the Philippines, to over 60 in India and Papua New Guinea, and over 85 in Bangladesh. jble 5.6: Distribution of public spending on education, Asia mid-1980s I of eumulative Gint- spending received by coefficin al best educated 10 X 1 Bangladesh 85.2 75.8 4 China 41.4 28.9 5 India 65.7 60.8 6 Indonesia 27.3 21.1 7 Korea 15.8 13.3 9 Malaysia 37.9 32.1 10 Nepal 57.1 52.1 11 Papua new Guiaea 62.2 55.3 12 Philippines 18.6 14.1 14 Sri Lanka 31.7 27.0 15 Thailand 32.9 23.4 Sourcess authors' estimates based on data in tables 5.2 and 5.31 see text discussion an method of derivation. a/ This statistic has a range of 0 to 100. The closer it is to 100, the more unequal is the distribution of public spending on education in a generation of school-age population. The second summary statistic is the share of cumulative public spending appropriated to the 10 percent best educated members in a generation. The cutoff share of 10 percent is an arbitrary choice, but it does provide a reasonably good idea of the concentration of public spending on education. Three groups of countries ire distinguishable: Korea and the Philippines form one group where the 10 percent best educated people in a generation obtain no more than 15 percent of the aggregate cumulative resources received by the entire cohort. A second group comprises China, Indonesia, Malaysia, Sri Lanka and Thailand where this group's share of resources ranges from about 20 to 30 percent. In the third group of countries -- Bangladesh, India, Nepal and 1/ See Mingat and Tan (1985) for the method of calculation. - 119 - Papua New Guinea - - the corresponding share of resources ranges from 50 to nearly 80 percent. There appears to be an inverse relationship between equity in the distribution of cumulative public spending and a country's level of economic development, as figure 5.4 shows. As before, however, the relationship is only a weak one, as there is wide variance around the average pattern. The implication is that policy choices can affect the degree of equity in the distribution of public spending on education. Top 10%'s share of cuadative resouces 80 2 0 6.... 40- 214 0 j 100 1000 t0000 Pet capita GIP JUS$. lot scale) StMtes, set 10Sat 1S ta1 lot 1t. Piweae 5.4: Relationship between share of cumulative public spending on education received by the 10 X beat educated and per capita GNP, Asia, circa 1985. The potential for improvement is especially strong in countries with two characteristics: cumulative public spending is highly skewed toward the best educated; and this group's share of resources exceeds what might be expectw on the basis of the country's per capita GNP. Figure 5.5 aids in the identification of such countries. The vertical axis shows the share of cumulative resources received by the 10 percent best educated in a cohort; the vertical axis, the deviation of this share from that associated with the country's per capita GNP (estimated roughly from figure 5.4). The figure is divided into four quadrants, using lines through the zero deviation point on the horizontal axis, and through an arbitrary cut-off level of 40 percent on - 120 - the vertical axis. Countries in the top right quadrant are those in which the current distribution of public spending is highly inequitable, and where the scope for improvement is large. These countries include Bangladesh, India and Papua New Guinea. Although Malaysia's location is in the bottom right quadrant, there is also room for improving equity in the distribution of public spending on education, mainly by reducing the high cost and heavy public subsidization of higher education. 80 hare received by 10 X best edacated 801 60 - it a sIo' 40 20 - 0 * s e I I I a -30 -25 -20 -15 -10 -5 0 5 10 5 20 25 30 Deviation tie tread lize il (igate 5. Sorces: see teit 8lscissil. Figue 5.5t Share of cumulative public spending received by the 102 best educated, in absolute terms (y- axis) and relative to the trend line in £igure 5.4 (a-ais), Asia, 1985. (c) The imactof grivate financing on-equity As noted elsewhere, the argument for increasing private financing is most persuasive when applied to higher education. Evidence from simple graphs and regressions is examined below regarding the likely impact of such a policy on equity in the system as a whole, and within higher education. (i) Overall eauity in the system. A useful index in this regard is the share of cumulative public spending on education accruing to the best educated 10 percent in a generation. Figure 5.6 shows that it declines as private financing in higher education expands. Part of this pattern is due to - 121 - the fact that (a) cost recovery is more extensive in the wealthier countries in the sample; and (b) their education systems are generally better developed at all levels. But an earlier discussion suggests, the link between per capita GNP and extent of private financing is a weak one. Thus, part of the variance in overall equity stems from the extent to which private sources of finance are tapped in higher education. Note from the figure that diminishing returns set in with rising rates of private financing. This pattern has an important implication regarding the actual design of cost recovery policies in higher education, namely that very high rates, say, beyond 40 percent, are not necessary to maximize their impact on global equity. 80 'bp 101's shate of ciaulative resoutces 80 - 51 . 60 1911CS* Mzinc O .. 0 206 0 08 0 ladex of private floaadagq aIEUMe,A: Relationship between overall extent of private £inancins Lh hisher education and share of cumative public spending on education received by top lO by education, Asia, circa 1985. As another indicator of overall equity, consider the share of primary education in total public spending on education (Pshare). It is a simple measure of overall equity with an intuitively appealing meaning, since systems in which primary education has a large share of total public spending are likely to be more equitable than those with the opposite characteristic. The impact of private financing in higher education on Pshare is examined through the simple regression below, based on country-level data for around 1985: - 122 - Pshare - 36.2 + 0.47 x Eshare + 0.32 x PF (4.62) (0.28) (3.19)* R-squared - 0.58; N - 11 * statistically significant at 5 percent confidence level. Eshare is the share of total public spending on education as a share of GNP; 1/ and PF is the index of private financing in higher education defined in chapter 4. 1/ The result, although tentative at tli stage in view of the limited sample size, is nevertheless suggestive. The PF variable has the predicted positive sign and is statistically significant at the 5 percent level. A 10 percent increase in its value (corresponding to a shift of 28 percent of its standard deviation from the sample mean of 32.7 percent) raises the share of primary education by 3.2 percentage points, from a sample mean of 47.6 percent, corresponding to an average rise of about 6.7 percent. The positive link revealed by the regression is due to two reinforcing effects: private financing in higher education reduces unit cost in public institutions (as demonstrated in the previous chapter), and diminishes the government's share in the financing of that smaller unit cost. As a result, a larger share of government resources can be channelled to primary education. (ii) Eauity within higher education. It is sometimes argued that private financing reduces the access to education. The counter argument is that it helps to mobilize resources, thus augmenting public funds for expanding coverage. The relationship depicted in figure 5.7 -- between the gross enrollment ratio in higher education (ERH) and extent of private financing -- provides preliminary support that the latter outcome is more probable. IV It is important to control for this variable in the regressions to avoid a spurious link between the extent of private financing and the share of primary education in total public spending. With private financing in higher education, government spending on this level of education is likely to be smaller, leading to a lower level of overall spending on education. The spurious link arises because the relationship reflects mostly the impact of a smaller denominator. 1/ To recall, it is a composite variable derived by weighting the distribution of enrollments between public and private institutions by the corresponding rates of cost recovery in each sector. - 123 - Gross erolIlmet ratio (%) 40 30 - 20 . 10 -to 9 8 *4 *jf0 4 1 *1t 0 20 40 60 80 100 Wadel of private finanago iran 3.7: felationship between iades of private £inoaa s in higher education ad eros en=ollaemt ratios in higher education, Asia, circa 1985. Regression analysis, based as before on country-level data for around 1985, also yields tentative support for this conclusion: ERH - 2.20 + 3.8 x PCGNP . 0.92 x Eshare + 0.33 x PF (0.30) (0.77) (0.48) (3.04)* R-squared - 0.80; N - 11 * statistically significant at 5 percent confidence level. PCGNP is the per capita GNP expressed in units of US$1,000. Eshare and PF have the same definitions as in the previous regression. The coefficient on the PF variable has the expected sign, and is statistically significant at the 5 percent level. An 10 percent increase in this variable leads to a 3.3 percentage point rise in the enrollment ratio of - 124 - higher education, which corresponds to a 28 percent increase from the sample mean. Note, however, that few countries in Asia have well-developed student loan schemes, and it is possible that the overall expansion of educational opportunities resulting from an increase in private financing would also be accompanied greater social selectivity in the system. In the absence of loan schemes, only students with access to private funds can enter the system, since most commercial banks do usually accept future earnings as collateral for loans. Evidence on this potential adverse outcome is scarce, however, so an indepth assessment of the problem is not attempted here. 5.2 Social selectivity in gAucation Equity in an education system is affected not only by its aggregate structure, but also by differences among population subgroups in the access to education within that structure. Thus, two systems might have identical distributions of cumulative public spending on education, but one would be less equitable than the other if its best educated graduates come largely from the more advantaged groups in society. In assessing social selectivity in the education system, subgroups in the population can be defined according to such criteria as sex, income and occupation of students' parents, geographic origin and so on. Since each definition yields insights into different facets of selectivity, the analysis should ideally segregate the population in as many ways as feasible. In practice, however, the requisite data are often lacking. An added problem in the present study is the scarcity of data that are comparable across countries. In light of the data gaps, the following discussion will address only some aspects of social selectivity in Asian education. In particular, it compares Asia as a region to other Yorld regions in terms of the distribution of cumulative public spending on education by socioeconomic group; evaluates differences in females' access to schooling among Asian countries; and assesses selectivity in the education systems of India, the Philippines and Thailand, as reflected by differences in access among urban and rural populations, and among people of different social backgrounds. (a) Distribution of public spending on education by socioeconomic group As before, the benefits refer to the cumulative public spending that is appropriated to a generation of people through publicly-financed education. The distribution of these benefits among socioeconomic groups depends on differences in the access to education. The data in table 5.7 permits a comparison between Asia and other world regions in this regard. These aggregate data clearly hide wide variation among countries in each region, but they nevertheless provide a broad picture of regional conditions. - 125 - table .71 88 composition of school end refersence populatM62, major world reions () at School population as ratio of Sohool nontlatin bz level refereace poodation oL Reference rimE Secondar_ &BS pentaton bt Pggigr $econds= Rsher Asia 1001 1002 1001 1001 Fasmers 53 25 19 58 0.91 0.43 0.33 Blue collar 34 43 38 32 1.06 1.34 1.19 White collar 13 32 43 10 1.30 3.20 4.30 Anglophone Africa 1002 1001 1002 10D Famers 74 36 39 76 0.97 0.47 0.51 Blue collar 18 29 21 18a 1.00 1.61 1.17 White collar 8 35 40 6 1.33 5.83 6.67 Francophone Africa 1001 1001 1001 1002 Farrs 61 36 39 76 0.80 0.47 0.51 Blue collar 26 27 21 18 1.44 1.50 1.17 White collar 13 32 43 6 2.17 5.33 7.17 Latin Amea 1002 1002 1001 1001 Farmers 31 12 10 36 0.86 0.33 0.28 Blue collar 32 54 45 49 1.06 1.10 0.92 White collar 17 34 45 15 1.13 2.27 3.00 Middle East and North Africa 1001 1001 100 1001 Parcers 33 15 22 42 0.79 0.36 0.S2 alue collar 43 57 31 48 0.90 1.19 0.65 White collar 12 28 47 10 1.20 2.80 4.70 010E countries 1001 1001 1001 1002 Fasmers 12 11 11 12 1.00 0.92 0.92 Blue collar 53 45 32 53 1.00 0.85 0.60 White collar 35 44 57 35 1.00 1.26 1.63 Source: Mingat and Tan (1986). al 8E8 groups in the school and reference populations are defined according to the occupation of students' fathers. For each reSion, the coln figures add up to 100 percent. b/ The reference population refers to the population of parents with school-age children. al Derived by dividing the first three columns respectively by the fourth colum. The first three columns show the socioeconomic composition of the school population in primary, secondary and higher education; the fourth column, the composition of the population of parents with school-age children. A bias exists in the access to a given level of education if a group's share in the school population exceeds its share in the reference population. This comparison is simplified by using an index of bias defined as the ratio between the school and reference population shares at each level of education (last three columns of the table); a value below 1 signifies that a group is under-represented in the school population, and a value above 1, the opposite. In Asia, selectivity in the access to education echoes the characteristic pattern in all developing countries: it is biased toward the * 126 - children of white collar parents at all levels, but especially in higher education. There are also some subtle regional similarities. Compare, for example, Asia and Latin America. In both regions, the access to secondary education improves substantially for children from blue and white collar backgrounds at the expense of farmers' children; the probability of entry in the former two groups is respectively 3 and 7 times as high as that for the farmer group.11/ In higher education, that figure remains around 3-4 times for the blue collar group, but climbs to over 10 times for the white collar group.12/ Given the bias in access, farmers' children will clearly be less well-educated than their peers from other social backgrounds. As a result, they benefit less from public spending on education. Their share of cumulative resources can be estimated using the same method discussed earlier in this chapter. The results appear in table 5.8 (first three columns). Comparing this distribution to that of the reference population reveals the extent of bias against farmers in the allocation of cumulative public spending on education (columns 4-6). In Asia, farmers' share of these resources is much less than their population share, while the opposite is true for the white collar group, the ratio between the resource and population shares being respectively 0.59 and 2.79 for the two groups. This pattern in fact is typical of most developing countries. However, if the outcome in developed countries is any guide, there is probably some scope for reducing the degree of social bias in the distribution of public spending on education. 1L/ For example, in Asia, the chances of entering secondary education for the blue collar group is 3.1 times (-1.34/.43) that for the farmer group. 12/ Taken as a whole, these results have important implications for the design of policies to improve equity in the system. If the objective is to expand access to higher education for the farmer group (usually the most disadvantaged), the intervention must begin at an earlier cycle of education. On the other hand, if the target group are those from blue collar backgrounds, it could focus simply on higher education itself. - 127 - Tble .8t Distribution of cumulative public spending on education by SBS group, major world regions Share 00 cumlative spending galative to Index of shaeof wMA&tion bias to r I Blue White Blue White Eament sMa sMM sMa" Dolaa Asia 0.59 1.19 2.79 2.0 4.8 Anglophone Africa 0.74 1.19 3.78 1.6 5.2 Francophone Africa 0.58 1.15 5.93 2.0 10.3 Latin Amersa 0.49 1.04 2.03 2.1 4.3 Middle East and North Africa 0.60 0.35 2.87 1.6 4.8 OECD countries 0.95 0.87 1.20 0.9 1.3 Sources Mingat and Tan (1986) a/ This index is calculated by dividing the figures in the preceding two colums by the corresponding figures for the farmer group. The larger this index public spending on education is concentrated in that group relative to the share by the farmer group. (b) Sex differences in the access to education Data on the share of females in total enrollments at the three levels of education appear in table 5.9. In all Asian countries, that share has improved significantly in the fifteen years to 1985. However, substantial variation remains across countries, particularly in higher education. Table 5.9: Percent females enrolled by level of education, Asia, 1970-1985 Primary S Ay lis 12ZO 1985 1970 1985 19 1285 1 Bangladesh 32 40 - 28 10 19 2 Shutan 5 34 3 18 - 17 3 Burm 47 - 39 - 38 - 4 China - 45 - 40 - 30 5 lndia 37 40 28 34 22 29 6 Indonesia 46 48 34 43 25 32 7 Korea 48 49 38 47 24 30 8 Laos 37 45 27 41 19 36 9 Malaysia 47 49 41 49 - 45 10 Nepal 15 29 17 23 - 20 11 Papua er Guinea 37 44 27 36 - 23 12 Philippines - 49 - 50 56 54 13 Singapore 47 47 48 50 30 42 14 Sri Lanks 47 48 51 53 43 40 15 Thailand 47 48 42 48 42 46 Average - 44 - 40 - 33 Sourcess see appendix table 81.3. - 128 - In general, females' share in the education system appears to be positively linked to a country's level of economic development (figures 5.8 - 5.10). In primary education, this relationship levels off above a per capita GNP of US$500, signifying that male-female gap in enrollment shares vanishes at about this point. In secondary education, the graph levels off as the per capita GNP rises to around US$700. In higher education, it rises continuously over the range of per capita GNP represented in the sample. Thae of females (%J 50 14It IS1 45 -4. It. 40 35 - 30 - le 25 20 ' ' ' S ' ' I ' I t eS too 1000 10000 Per capita GOP (05. log scale) Soing: see Appoilt 0. YLire 5.8: Relationship betven femle share of enrollents In primary education and per capita GI, Asia, circa 1965. The differences among countries nevertheless remain after controlling for the variation in per capita GNP, signifying that educational policies can affect females' access to education. For example, the share of females in primary education is higher in Bangladesh (40 percent) than in Nepal (29 percent), even though both countries share comparable levels of per capita GNP. A similar comparison can be made between China and India, and between Papua New Guinea and the Philippines. At the secondary level, there is probably scope for improving females' access to schooling in Bhutan and Papua New Guinea, two countries that achieve much less than others at similar levels of economic development. The diversity among countries is especially wide in higher education. The below-average performers include Bhutan, Korea, Papua New Guinea, and Singapore. Korea's weakness in this respect (30 percent females) is brought out strikingly by comparing it to Malaysia (45 percent), a - 129 - Sts at eMales IS) 50 - IS* 4 *13 40 / 6 30 - 20 - to 10 100 1000 10000 Pei capita GNP (S. log scale) 8oette: see Appeads 4. FLAe 5.9: Relationship between female share of enrollments In secondary education and per capita GP, Asia, oIe& 1985. country with a somewhat lower per capita GNP. Singapore's inclusion in this group is also quite surprising. What factors explain sex differences in the access to schooling? Cultural and economic considerations clearly account for part of the diversity among countries. But schooling conditions are probably also important. Consider first primary education. Figure 5.11 shows that the lower are cohort survival rates, the smaller tends to be females' share of enrollments. This relationship levels off as survival rates rise above 50 percent. The implication is that when an education system's holding power is extremely weak (due, for example, to exceptionally poor quality), girls tend to suffer more than boys. Thus, aside from being a desirable pedagogical objective in itself, strengthening the retention capacity of primary education is also a strategy for widening female participation at this level.2&/ 29/ Note from figure 5.8 that in Nepal, the share of females in primary enrollments is smaller than that in systems with a similar level of cohort survival rate. One possible interpretation is that there exists an especially strong cultural or economic bias against female education. • 130 Share et lemales 1%1 50 , sj0 - .6 30 - 10 - 10 100 1000 10000 Per tapita GiR [IS$, log stale) ae,teet see Agpesti I. Piaure $.10: Relationship between føale share of enrollments in higher eduation and per capita GNP, Asia, circa 1985. 1 females In prinary education 50 - 27.-9 45 - • -'" • 40 -• 5 35 2 30 - 1 25 - 20 I iI I 0 10 20 30 40 50 60 70 80 90 100 Survival tate in primary cycle 1%) * Reiere to Ist yesar paliaary estrans setlivitt to ead si tie cycle. sostce: see AppeadI C. Fiaure 5.11: Relationship beteen female share of enrollmnta in primary edueation and cohort survival rate in prma~ education, Asia, circa 1985. - 131 - In secondary education, the picture is more complicated. A plot of the cohort survival rate against the share of females reveals a random relationship (figure 5.12).21/ To unravel this puzzle, recall that selection at this level of education occurs mostly at the transition between subcycles of secondary education, comtuconly on the basis of examination results (see table B2.2). Therefore, one interpretation of the random pattern is that girls survive as well as boys iii the competition for places within the secondary cycle. It is not a surprising result, since girls who reach secondary education represent a more select group of pupils than boys, the weakest of their cohort having exited the system at an earlier stage. X Iualses is secoadaty ea0caMle0 se4* 50 a 40 -* e all 5. 30 - 20 - 10 - 0 j I I 0 10 20 30 40 50 60 70 Survival rate it secondary cycle (K1 * Maume 5.12: Relationship between female share of enrollments in secondary education and cohort survival rate In secondary education, Asia, crca 1985. The bias against female enrollments in secondary education in fact originates mainly in primary education. A plot of the cohort survival rate in primary education against the female share of enrollments in secondary education reveals an upward-sloping relationship which levels off as survival rates rise to about 70-80 percent (figure 5.13). This result is intuitively 21/ The survival rate in this figure refers to the proportion of first-year entrants to secondary education who reach the end of the entire secondary cycle. The data are based on appendix table B2.2. * 132 - appealing since access to secondary education is contingent upon the completion of primary schooling. A further source of bias is that not all females who complete primary schooling continue to the next level. Note from figures 5.11 and 5.13, that when the primary cohort survival rates is 30 percent, for example, the share of females is 38 percent in primary education, but only 26 percent in secondary education. The gap between the latter two figures suggests that females experience a smaller rate of transition to secondary education. However, as the primary cohort survival rate rises, this gap diminishes, disappearing altogether as it reaches 80 percent. Beyond this point, sex differences in the transition to secondary education are largely absent. In general, the analysis suggests that an effective way to boost females' share of secondary enrollments is through a broad policy of enhancing cohort survival rates in primary education. Direct interventions within secondary education are likely to be less effective, even though they may appear at first sight to be more focussed and therefore more promising. The former strategy is stronger largely because it addresses the problem at its point of origin in the system. X te3ales in seceaduy e12cation /It 20 /' 30 20 0 10 30 30 40 50 60 70 80 90 100 Suivl tale I prizary cycle (9) Fla 3.13 1 Relationship between female shae of easonts in secondary education and cohort survival rate in primaY education. Asia, oirca 1985. - 133 - (c) Patterns of social selectivity in India. the Philionines and Thailand The following discussion illustrates the extent of this problem in three countries at different levels of economic development.22/ (i) Inia. Table 5.10 shows the patterns of cohort survival rates by sex and urban-rural residence based on the Fourth All-India Educational Survey.2/ This source actually provides information up to grade 12, but since a significant share of students at this level are enrolled in colleges rather than in the school system, they are not captured in the Survey. For this reason, this analysis concentrates on the first 10 grades of the system. Table 5.10: Proportion of grade 1 entrants surviving first 10 erades of schooling, India, 1980 al Percent of trade 1 entrants survivin to 9.as.. S r-d tade.. Ak10 Overall population Boys 39.5 25.0 15.9 Girls 32.1 16.9 9.7 Both sexes 36.5 21.7 13.4 Urban population Boys 60.7 54.0 40.3 Girls 55.7 42.2 27.4 Both sexes 58.4 48.6 34.4 Rural population Boys 34.9 18.8 10.1 Girls 25.5 9.9 4.8 Both sexes 31.2 15.5 8.3 Sources authors' estimates based on Goveramnt of India (1985). 22/ The choice of these countries is largely determined by the availability of readily accessible data at the time of writing this report. 21/ A fifth survey has in fact been completed recently, but no published data are as yet available. It would be interesting to examine changes in survival rates between the two surveys. Note, however, that selectivity in an education system tends to evolve only slowly over time. So it is unlikely that the more recent survey will reveal dramatic changes, particularly because the gross enrollment ratio in primary education rose only slightly between 1980 and 1985, from 81 to 92 percent. - 134 - On average, about a third of grade 1 entrants reach the end of the primary cycle (grade 5).2A/ Girls lag slightly behind boys at this stage. By the end of middle school (grade 8) however, only 17 percent of the entering class of female first-graders remain, compared to 25 percent among males. This divergence arises because the transition rate to middle school is lower for girls than for boys, and the dropout rate within this level of schooling is higher among them. Beyond middle school, girls' survival rate is comparable to that of boys', so that the male-female gap in survival rates remains stable up to grade 10. The bias against females is strongest in rural areas. Only 25 percent of female first-graders in rural areas reach grade 5, compared to 35 percent among rural males. By grade 10, rural girls' survival rate is only half as high as that of rural males. Further calculations based on the Survey reveal the pattern of enrollment ratios by grade presented in figure 5.14. 21/ To illustrate the dramatic gap between the most and least advantaged groups, only boys and rural girls are represented in the figure. It is beyond the scope of this report to analyze the reasons for these social differences in education. However, it is of interest to note the impact of demand patterns reported by Caldwell et al.(1985) based on a survey of schooling in south India (Karnataka). Parents in the sample cite economic reasons for terminating a child's schooling 44 percene of the time, regardless of the child's sex. Weak academic performance has a weight of 55 percent among boys, but only 35 percent among girls. The onset of menarche is the other main reason for terminating a girl's education, with a weight of 20 percent. The importance of the last factor is an indication of the strength of cultural norms in affecting the demand for education. However, it is not as important as the other two factors -- economic conditions and academic performance. This result suggests that policy interventions, aimed possibly at improving girls' academic performance and reducing the economic cost of schooling (including that of forgone child labor), would probably be effective interventions to enhance the representation of females in the education system. 2A/ The figures in table 5.11 differ slightly from those in table 4.5 because they come from different sources. L/ The grade 1 enrollment ratios for urban boys and rural girls are estimated on the basis of data on enrollments by sex and grade, assuming that all boys enter grade 1. The enrollment ratios at subsequent grades are derived simply by multiplying the result by the corresponding cohort survival rate. - 135 - 90otte m* tells .10. t et entllmeal lato (%I) 100 80 ai boys 10 - .06 60 - '-. 5 4 j,041Sae 0 I 2 3 4 5 6 7 8 9 10 li Grade Glades t-6 iota the plualsy scle; 6-0, tie ati le cycle, aid 9*t0. past of the iII Uplet SIOCdary S714. Fiure l.s et enrollment ratios by srade in overall population, and among urban boys and rural girls, India, circa 1980. (ii) The Philionines. The general pattern of cohort survival in Philippine education appears in figure 5.15. The entry rate into primary schooling is nearly universal, but only 66.4 percent of the entrants reach the end of the cycle. A few of them fail to graduate, and some do not enter the next cycle. As a result, only 56.9 percent of the cohort begin secondary education. The incidence of dropping out at this level shrinks the proportion reaching the end of the cycle to 41.4 percent. Of the secondary school graduates, half (corresponding to 20.7 percent of the original cohort from grade 1) enter college.2&/ Those completing college education comprise 16.7 percent of the cohort of grade 1 entrants. To what extent does this pattern result in social selectivity in the education system? Given the relatively low rate of cohort survival in primary education, social differentiation in the system probably originates at this level of education. The data in table 5.11 indicate that wide disparities 26/ The entrance to college is governed by the National College Entrance Examination. The passing mark is generally set at the 50th percentile of the distribution of scores. - 136 - Senlival rate (1 100 80 60 - 40 - 20 - 20 1 2 3 4 5 6 7 8 9 10 it 12 13 14 15 Glade Gades 1id. *10. a Ilotsra. resp.. the ItIall. secoataty aet h at tes Soue: Ia1at tal Tas (1988). igaure 5.15t Cohort survival rates In the Philipptaes education system, Circa 1985. indeed exist in the cohort survival rates among children from different socioeconomic backgrounds. For example, 43 percent of the first-graders from families with income below P10,000 in 1982 drop out by the end of primary schooling, compared to only 11 percent among children from families with an income exceeding P30,000. For illustration, figure 5.16 show the differences among groups defined according to the educational attainment of pupils' fathers. - 137 - Table 5,11a Cohort survival rates in elementaxy education by sooioeconomie groups, Philippines, 1982 Survival rate () al Family income (pesos a year) < 10,000 57 > 30,000 89 Geographc origin Rural 57 Urban so Pather's education No schooling 19 Elemntary 51 Secondary 69 College 92 Father's occupation tasmer, fishermen 5S Manual worker 75 White collar worker 91 Sources Mingat and Tan (1968) a The cohort survival rate is defined as the proportion of entrants in grade 1 who survive to the end of a given cycle of education (here eleasmtary education). It is worth mentioning in this context that the goverrment recently implemented policies under which locally-financed secondary schools were nationalized, and fees for all public secondary education abolished. These policies address inequities that are widely-perceived to exist in secondary education.22/ To be stressed, however, is that they are likely to have only a limited impact on reducing overall social selectivity in the system. The reason is that free secondary education benefits only the 63.2 percent of the cohort who in fact reach the end of the primary cycle, thus automatically bypassing those who dropped out in earlier grades, the majority of whom are from the poorer social classes. To enhance educational opportunities for these groups therefore calls for fresh policies aimed specifically at improving cohort survival rates in primary education.2i/ 22/ See Kingat and Tan (1988) for a more detailed discussion of these inequities in secondary education. 2&/ This concern will partly be addressed through a World Bank loan currently under preparation. However, the design of more detailed interventions that are applicable on a country-wide scale would require analysis of the causes of dropping out using survey data. - 138 - Savival late (M 100 1 0 91 0 '*i- All gtops 60 - College 60 -alitbo '(1 ***** 11gh school *0* Blementary 40 + an education 20 - 0 0 1 2 3 4 5 6 7 8 9 10 it 12 13 14 15 Glade Gades 1-6, 10, ad 1.14 !ota, resp., the Iiaaty SICOMAT sal hlht CycleS Seace: 1agat ai Tal T988). Fire 5.161 Coboart survival rates In primary education according to educational attainment of pupil's father, Philippins, 1982. (iii) Thailand. The relevant data on selectivity in the system appear in table 5.12. The degree of social bias can be assessed by comparing a group's share of enrollments to its share in the population. In upper secondary education, for example, the professional group's share of enrollments is 25 percent compared to their population share of only 3 percent, the ratio between these shares being 8.1. On the other hand, farmers' share of enrollments is 20 percent compared to their population share of 69 percent, the corresponding ratio being only 0.2. The other social groups' access (business and laborer classes) to upper secondary education also exceeds that of the farmer group, although by a smaller margin. - 139 , 2abl 5,12t Access to secondauy and haber education An Thailand, 1980s at Share of enrollments Enrollment relative to share Selectivity index share (M) of Voculation tlasErs 1.0) Lo Population Upper Upper Upper .ae..LE go.r UY o.S$ U s AU Socioconomlo group b/ Profesetonal 3.1 25.1 36.5 8.1 11.8 27.0 59.0 Business 9.0 19.1 28.0 2.1 3.1 7.0 15.5 Laborer 19.4 35.6 23.9 1.8 1.2 6.0 6.0 Farmer 68.5 20.2 11.6 0.3 0.2 1.0 1.0 All groupe 100 100 100 - - - - Source: authors' estimates based on Goverment of Thailand (1987). al Data for open uaniversities are not included under the *univ." columns. bi Defined according to the occupation of students' fathers. of Calculated frm the preceding two columns by dividing the corresponding figure by the figure for the farmer group. To render the comparison more transparent, an index of selectivity can be computed by dividing the ratio of the enrollment and population shares for each group by that for the farmer group (columns 6-7). It can be interpreted as the probability of entering upper secondary education for a child from a given social group compared to one from a farming background. Thus, in upper secondary education, professionals' children have a relative entry probability that is 27 times as high as that of farmers' children. The corresponding index for the children of businessmen and laborers is 7 and 6 respectively. With regard to selective universities,2/ the index for the children of professionals rises to almost 60. That for businessmen's children reach nearly 16, while that for laborers' children remains the same as at the upper secondary level. 22/ These universities (which include both public and private institutions) differ in kind from the unselective open universities because they apply stricter entry criteria. The competition for places is usually very keen because they are more prestigious; in addition, the public selective universities are heavily subsidized by the government. * 140 - Since primary education is nearly universal, differences among socioeconomic groups occur mostly in the access to subsequent levels of education. In fact, it is at the transition between cycles of education that these differences arise, since the extent of intracycle selection is limited in Thailand (see table 4.5 in previous chapter). Thus, the patterns revealed by table 5.12 suggest that the professional and business groups continue to strengthen their advantage in the transition from upper secondary to higher education. The bias in favor of the laborer group occurs mainly at the transition from lower to upper secondary education, and does not widen subsequently. Some of the selectivity against farmers' children probably occurs as early as primary schooling, since the overall cohort survival rate at this level is only 80 percent. However, the bulk of the bias against them is traceable to the transition from primary to lower secondary education, since only 40 percent of the primary school graduates systemwide are selected to continue. These patterns of selection have important ramifications for policies to improve educational opportunities for the disadvantaged. If laborers' children, for example, are the target, then policy interventions should focus on widening their access to selective universities. On the other hand, if the target group are farmers' children, those interventions should occur much earlier in the system. Taken together, the foregoing results for the three countries-- India, the Philippines and Thailand--hold a significant implication for policies to ameliorate social selectivity in an education system. In particular, it highlights the necessity first to distinguish between intra- and inter-cycle selection in locating the origin of adverse selectively. The result has very different requirements with regard to the design of potential interventions. In situations where the rate of completion within a cycle, particularly at the primary level, is low, reducing adverse social selectivity in the sector as a whole invariably calls for interventions within primary education. On the other hand, in systems where selection takes place mostly in between cycles of education, the issue of biases in the selection mechanism becomes much more prominent. In addition, since the pattern of survival and transition can differ significantly among population groups, clarifying the intended target group is a necessary ingredient in formulating appropriate interventions. - 141 - 6. Conclusions Several general lessons about policy choices in education emerge from the preceding chapters. Their implications for policy dialogue concerning educational development in the region are summarized below, while more specific assessments of individual Asian countries are discussed in Appendix A. Priorities for future analytical work are briefly touched on in the concluding section. ,/ 6.1 Overview of findings and assessment of policy issues Perhaps more than any other world region, Asia is characterized by striking diversity in educational development across countries. The range stretches from Bhutan with an adult literacy rate of only 15 percent and a primary enrollment ratio of 25 percent, to such countries as Korea and Thailand where over 90 percent of the adults are literate, and primary education is largely universal. Over time, most Asian countries have made substantial progress in expanding coverage at all levels of education, but in some countries significant numbers of primary school-age children remain outside the education system, because they either have never been enrolled, or have dropped out from it prematurely. (a) Causes of differences in performance across countries The analysis in this paper suggests that two broad factors account for the sharp contrast in educational outcomes among countries: exogenous constraints and the cumulative effect of policy choices in the sector. By most measures of educational development, low-income countries in the region invariably perform worse than richer countries, reflecting the adverse impact of rapid population growth combined with relatively slow economic expansion. In the former countries, these problems are often compounded by the high cost of educational inputs, notably imported pedagogical materials and supplies, and sometimes even expatriate teachers; the weak demand for schooling due to unfavorable cultural and economic factors, including the costliness of forgone child labor; and the sparseness of institutional infrastructure. However, a consistent finding is that among countries facing comparable external constraints, some have performed better than others owing to differences in the choice of sectoral policies. This result conveys a positive and optimistic message, for it means that appropriate educational policies can 1/ It is useful to recall here that the analysis relies largely on aggregated data, and therefore shed light mainly on broad policy issues regarding structural characteristics of the sector. Countries clearly also face concerns in education that are unique to the local environment, but these are not addressed here since they belong more properly to country-specific sector analysis. - 142 - make an appreciable difference to a country's educational achievement, even though external constraints remain an important impediment. (b) Generallv bright outlook for the future The two factors that explain cross-sectional differences among countries will obviously also affect the prospects for future progress. With regard to demographic and macroeconomic conditions, the outlook appears favorable in most countries. On average, the tax burden of financing education is likely to lighten considerably as the dependency ratio (ratio of school-age population to working adults) drops from around 0.42 in 1985 to 0.36 by the end of the century; the ratio remains high, however, in Bangladesh, Laos, and Papua New Guinea, and is even expected to rise somewhat in Bhutan and Nepal. A second reason for the favorable prospects in Asia is that the region's economies are expected to grow at moderate to high rates, averaging about 5 percent a year between 1990 and 2000, while the school-age population is forecast to grow much more slowly, by about 1.3 percent a year. Thus, if government spending on education as a share of the GNP remains unchanged, public resources would be available not only to maintain current enrollment ratios, but also for expansion (in terms of coverage and/or additional inputs per pupil). As before, however, the prospects are less bright in some countries, notably in Nepal, and to a lesser extent, also in such countries as Bangladesh and Papua New Guinea, where the gap between the projected economic and population growth rates is relatively slim. Clearly the amount of "extra" resources that actually materializes for expansion depends on the evolution of educational costs over time. Little can be predicted in this regard, since the available cross-sectional evidence points to the absence of a clear link between the costliness of education and a country's per capita GNP. At one extreme, it is possible that costs remain unchanged in real terms as the economy grows; at the other extreme, they may grow at the same pace as the GNP. The future is likely to lie somewhere between these extremes, its location being largely a function of future government policies in the sector. If it is closer to the first extreme, significant resources for expansion would become available beyond what is needed to maintain enrollment ratios at current levels; but if nearer to the second, then no scope for real expansion exists since all the "extra" resources will have been absorbed by rising costs. (c) Policy challenges These broad characteristics of the external environment define the future policy challenges in the sector. In the handful of Asian countries where demographic and macroeconomic conditions will probably remain difficult (Bangladesh, Bhutan, Laos, Nepal and Papua New Guinea), appropriate educational policies are crucial, since the wrong choices are likely to exacerbate the inequities that characterize the sector at present, and may even reverse past gains. For example, policies that perpetuate high costs in higher education or raise them will reduce the effective amount of resources available for the lower levels, a particularly worrisome prospect since - 143 - primary education in these countries currently still suffers from incomplete coverage, high dropout rates, and inadequate resources. In the other Asian countries where the exogenous constraints on education are likely to ease considerably, the challenge is to avoid the temptation of complacency. The goal of consolidating and further strengthening past progress can hardly be ignored since the education sector in many of these countries continue to be marked by striking inequities and inefficiencies in the use and distribution of resources. Just as countries differ in the current status of and future prospects for educational development, they also differ in their policy objectives for the sector. Political considerations clearly matter a great deal in defining those objectives, but they almost always depend on unique domestic circumstances, and therefore do not lend themselves readily to intercountry comparisons. However, few governments would disagree with the general goal of alleviating poverty. Since education is a chief avenue of social and economic advancement, this objective may be taken as a primary aim of policies in the sector. In some situations, the promotion of equity may imply sacrificing efficiency, but the analysis in this report points to a fortuitous absente of conflict, at least at the aggregate level, between equity and efficiency goals. In fact, improving efficiency in the sector is inescapable if equity is to be enhanced since inefficiency results in high costs, which in turn would shrink the amount of "extra" resources potentially available for interventions to alleviate poverty. (d) Policy option 1: Increased public spending on education in selected anuntries Several results emerge with regard to policies to promote educational development. An intuitively appealing intervention is to increase the aggregate level of public spending on education, but this obvious candidate is in fact quite limited as a policy lever. First, the extent to which it can realistically be manipulated is circumscribed by the keen intersectoral competition for resources; this factor probably explains the relatively stable trend in most countries' spending on education as a share of the GNP. Second, and perhaps more importantly, regression analysis based on cross-sectional country-level data suggests that appreciable variation in educational outcomes emerges only with relatively large differences in aggregate levels of public spending. In other words, countries which spend more do not necessarily achieve better outcomes than others which spend somewhat less. This result is actually less paradoxical than it appears because educational services are produced in different organizational setups, with corresponding differences in efficiency. It implies that for most countries in the region, the potential for improvement resides largely in the choice of policies within the education sector. In such countries as Bangladesh, Nepal, Sri Lanka, and until recently, also the Philippines where current levels of public spending are much below the Asian mean, the case for increased public funds for education is stronger, but even here policies within the sector remain important. - 144 - (e) Polic oRtion 2: Imoroving primary education The precise design of educational policies will clearly vary from country to country according to local conditions. However, the findings suggest that as a broad strategy, the central focus should be on improving primary education. Expanding coverage and minimizing dropout rates yield external benefits by producing literate future adults, with implications for the prospects for economic development. Equally important is that such interventions promote equity insofar as the poor and such disadvantaged groups as females are the ones most adversely affected by incomplete coverage and low retention rates. In a few Asian countries (Bhutan, China, India, Nepal, and Papua New Guinea), a two-pronged approach is required since not all school-age children enter grade one, and many of those who enter drop out before the end of the cycle. In the remaining countries where entry into primary education is more or less universal, the main focus should be to reduce the incidence of non-completion. Korea, Kalaysia, and Singapore are the only countries in which the problems of incomplete coverage and high dropout rates in primary education have largely disappeared. (Hong Kong and Taiwan are probably also in this group, but no data for them were collected in this study). The analysis suggests that increased levels of spending per pupil in primary education may increase the attraction of schooling and its holding power by promoting educational quality. However, "throwing more money" at schools is not the be all and end all of a strategy for improvement since countries with similar levels of resource-intensity per student achieve strikingly different results in terms of entry and completion rates; conversely, countries achieve comparable results under widely different schooling conditions. Thus, beyond a general policy of augmenting resources for primary education, a further, and probably more important, issue is to identify the most effective ways of spending those resources. This question is relevant not only in countries which currently spend little on primary education, but also in those where seemingly adequate levels of spending still lead to poor results. No specific answer on this issue emerges from the present study, however. A fundamental reason is that non-entry and high dropout rates are caused by factors that vary across countries and probably also across regions within countries. Thus, it is unlikely that a single approach or intervention will be effective in all situations. Unfortunately, little can be added beyond this obvious recognition owing to the scarcity of information on the conditions under which alternative policies are likely to succeed. It is nevertheless clear that if governments are to overcome the weak performance of primary education, further analysis of this topic must be accorded priority in future work. (f) Policy option 3: Freeing up resources for primary education As noted earlier, the orientation of public policies in favor of primary education requires concomitant changes in other subsectors. Increasing efficiency, especially in secondary and higher education where - 145 - operating costs are often much larger than in primary education, is crucial since it would increase the amount of resources effectively available for expansion. Many options are of course relevant here, including the design of administrative rules and regulations regarding teachers' qualifications, their pay and the criteria for promotion; allocation of teaching duties within schools and scheduling of classes; arrangements for multigrade teaching; use of evaluation instruments and definition of management and teaching styles; choice of the mix of school inputs, and so on. However, data are lacking on most of these options, not least due to their diverse outcomes in different settings. The available evidence relates only to the physical organization of providing education services, as reflected in the size distribution of institutions. In some countries, the predominance of small establishments suggests that economies of scale, which are often present in secondary and higher education, are probably not fully exploited. Although the potential savings in costs, and the practical feasibility of consolidation are likely to vary across countries, this source of efficiency gain should probably not be neglected. At the aggregate level, lessening the bias toward higher education in public policy is perhaps the most clearcut and important change called for. In all countries, government spending on education tends disproportionately to benefit the most advantaged group -- here defined as the best educated people in a generation. However, the bias is less strong in some countries than in others, owing to differences in policies. None of the countries showing only a modest bias toward higher education relies on administrative regulation to ration places at this level. Instead, they have sought ways to reduce the public cost of financing higher education. Three strategies, not mutually exclusive, have been used with success: increased private financing in public institutions; promotion of largely self-financing private institutions; and low-cost distance education. Some countries rely on all three strategies (Korea), while others choose only one or two of them (Philippines with private education, and Thailand with open universities and private institutions). Although the practical relevance of these options clearly differs among countries, the experience to date clearly suggests that they lead to more equitable outcomes compared to situations in which higher education consists largely of conventional institutions that are operated and financed by the government. The positive outcome of increased private financing in higher education -- whether resulting from charging fees in public institutions or encouraging the growth of private institutions -- arises partly because it helps to promote efficiency in public institutions. As such, it permits the government to do two things simultaneously: lower overall operating costs, as well as its share in financing those costs. The savings in turn permit increased allocation of public spending toward the lower levels, a pattern revealed by the analysis of this study. With regard to distance education, the cost advantage is well-known and highly attractive. However, little is known about the fields in which such an arrangement provides an effective alternative to conventional teaching. To pursue this promising option would require additional assessment, particularly to compare the labor market performance of people who have followed different careers in higher education. - 146 - To summarize, the overall findings suggest that substantial scope for policy intervention exists to promote equitable and efficient outcomes in education, even though exogenous conditions will continue to hamper progress. The diversity in Asia, in terms of both the strategies that governments have chosen and their corresponding outcomes, provides a particularly rich basis for identifying potential policy options. To the extent that poverty alleviation through education is a policy objective, the analysis indicates that an effective policy package would comprise two essential components: increased focus on primary education, and reduced public financing of higher education. Its precise design will obviously need to respond to unique country conditions, but the general thrust is probably relevant in all settings. 6.2 Priorities for future research on education in Asia Asian countries share many common problems in education. The most important of these is the education system's generally weak capacity to retain students, particularly at the primary level. A second issue relates to the choice of a delivery system in higher education that satisfies demand without diverting resources away from basic education, while at the same time achieving equitable and efficient patterns of investment within the subsector. Further research on both issues is worthwhile as the results would strengthen the analytical basis for policy dialogue. (a) Study 1: Factors affecting retention rates in primary education The rate of retention (or completion) in primary education varies widely among countries around the regional mean of about 60 percent. This pattern signifies that substantial scope for improvement exists, particularly to promote efficiency and equity in the education system. Knowledge about potentially effective interventions is limited, however. What is known from the available data is that low retention rates can occur under very different conditions of schooling. The objective of a regional study would be to discover the main factors, in-school and out-of-school, that affect a pupil's probability of dropping out prematurely. Bearing in mind the design of policy interventions, it is crucial to distinguish between factors that can be manipulated, and those that must be taken as exogenous. Among the former class of factors, a further issue is to identify the most cost-effective interventions in different settings. This differentiation is important since interventions that work among rural schools, for example, may not be as cost-effective in suburban settings; and it is likely that within countries, regional differences are also likely to call for varied interventions attuned to local conditions. Possible country settings for such a study include Bangladesh, India, Indonesia, Nepal, Papua New Guinea, and the Philippines. - 147 - (b) Study 2: Alternative ways of groviding higher education Higher education can be provided through various arrangements: conventional universities and colleges in the public and private sectors, correspondence courses, and open universities. Asian countries have followed markedly divergent strategies in this respect. In a significant number of them distance education is an important mode of providing services, with the distinct advantage of affordability and wide accessibility. In other countries, the strategy has been to rely on private education. The various types of institutions differ in many ways, particularly in terms of costs and entry criteria. Although some knowledge exists on their relative internal efficiency and the social characteristics of their students, the available information is generally tentative and incomplete. A more serious deficit is the paucity of knowledge on the labor market performance of people who have followed different careers in higher education. Such outcomes generally depend on a student's academic and social background, as well as the characteristics of the institutions. Sorting out the nature of the relationships involved is desirable since it would (a) help to clarify the role of the various types of institution; and (b) strengthen the factual basis for policy dialogue regarding the development of higher education. Possible country settings for the study include Burma, China, Korea, Sri Lanka, Thailand, and possibly also India and Indonesia. - 148 - APPENDIX A: INDIVIDUAL COUNTRY PROFILES IN A REGIONAL PERSPECTIVE The comparative analysis in foregoing chapters provides a benchmark for evaluating the status of educational development in individual Asian countries. Below is a succint assessment for 13 countries (in alphabetical order) for which the relevant data exist. For reasons of comparability, we use data for 1985, since it is the year for which the data are most complete for the largest number of countries. Where available, more recent data are reported in the basic appendix tables. A standard format with five sections is adopted: overall assessment; development of the education system; constraints and prospects; the system's operational characteristics; and patterns of cost and financing. The meaning of all the indicators have been defined before, but some of the more specific ones are repeated here to facilitate the discussion: grade attainment refers to the average grade that will be attained by the current school-age population given the present structure of the enrollment pyramid; dependency ratio refers to the ratio of the school-age population (aged 5-14) to the adult population (both sexes aged 15-65); comletion rate (used interchangeably with survival rate and retention rate) refers to the percentage of first-year entrants in a cycle of education surviving to the end of the cycle; pupil-teacher ratio refers to public institutions. In countries where distance education is sizable in higher education, the ratio refers only to regular conventional institutions, this being denoted by the word "regular" at the appropriate place in the table. index of costliness refers to unit costs (expressed as a percentage of per capita GNP) relative to the unit costs averaged across all Asian countries in the sample, except Papua New Guinea; % of cost recovery to the share of operating unot costs financed privately. It usually refers to public institutions. However, where the word "global" appears, it is the weighted average rate of cost recovery across all institutional types. This statistic is presented in situations where the contribution of financing through private education (and overseas education in the case of Malaysia) is substantial. distribution of resources refers to the distribution of cumulative public spending on education appropriated to an entire generation as it passes through its schooling years. A more detailed discussion of the longitudinal concept involved can be found in chapter 5. - 149 - Note that the larger the Gini-coefficient, the more inequitable is the distribution of those resources. For quick reference, the data for all 13 countries are summarized in a single table at the outset (table A.1) before presenting the individual country profiles. For completeness, the table also contains some additional data on the education system's operational characteristics and unit costs. A qualitative evaluation is provided in table A.2: positive signs signify that a country's performance on the variable under consideration is better than is average among Asian countries, negative signs denote the opposite, and a zero signifies that it is about average. 1/ The larger the number of positive (negative) signs, the better (worse) is its performance. Where data are missing a question mark is used. For reasons discussed in an earlier chapter, the data on enrollment ratios are not compared to the Asian average in the qualitative evaluation, but to the that predicted on the basis of the country's per capita GNP. For all but two variables, positive signs are assigned for positive deviations from the regional mean, and negative ones for negative deviations. In the case of pupil-teacher ratio and distribution of resources, positive signs are assigned for negative deviations from the mean since the smaller are the statistics, the better are outcomes in terms of quality and equity respectively. This procedure ensures consistency throughout the table in that positive signs always mean better outcomes than negative signs. Tabte A Sumy of comparative data on edu tional developant in AsLa, mid-1980s ffimtan B 9~g lak li~ &&Mg Nlay.r N~ ~9lE. Sri L. Tha, SY Develoa~t of the education avste. ~mrolimant ratio (X) Priry 60 25 107 118 92 118 96 99 82 70 106 103 97 90.2 •eøou*a~ 18 4 23 39 41 42 75 53 25 13 65 63 30 37.8 Rlgher 5.2 0.1 5.4 1.7 9.0 6.5 31.6 6.0 4.6 2.0 28.0 4.6 19.6 9.6 Grade attatemn 3.9 1.4 - 5.1 4.8 7.3 11.4 9.2 3.6 4.3 10.2 9.5 7.0 6.5 Adultliter y rate (Z) 33 15 - 69 43 74 92 74 26 45 86 87 91 61.3 Contraint and *rofneota Grøvth rata (x p.a. to 2000) Shool-age pop. 2.0 3.0 1.0 0.4 1.3 1.0 0.3 1.4 3.1 2.0 1.5 1.1 0.5 1.4 Sean~m 4.9 - 3.5 6.6 4.8 3.9 6.8 5.0 3.8 5.1 5.3 4.8 6.0 5.0 C Depanjna~ ratio 1985 49 44 47 33 44 46 31 41 49 50 47 36 41 42.9 2000 42 47 35 26 35 35 24 33 51 40 37 31 29 35.8 The system8' operational characteristios Completion rate (X) Prtmary 24 17 - 68 37 60 97 97 33 67 66 85 80 60.9 Loer seo. 48 83 - 76 44 92 98 90 89 63 74 75 91 76.9 Upper Seo. 84 88 - 81 77 94 95 96 81 95 - 99 87 89.0 Table A.13 (eåialisation al 19 1219 1imi 1 angladesh 91~nties 73908 147835 30.8 33.9 Social science 93978 165481 39.1 37.9 medicin 8347 14687 3.5 3.4 Science & technology 62988 105830 26.2 24.2 Other 951 2782 0.4 0.6 Total 240172 436615 100.0 100.0 4 Ohin- 8u 1anities 406128 572817 35.0 32.2 Social science 43551 190442 3.7 10.7 Nedicin 142737 166008 12.3 9.3 Science & technology 559419 835373 48.2 47.0 Other 9605 13968 0.8 0.8 Total 1161440 1778608 100.0 100.0 5 India Diunties 2832302 - 53.0 - Social science 1033385 - 19.3 - Nedicine 146472 - 2.7 - Science & technoloay 1317880 - 24.7 - Other 15541 - 0.3 - Total 5345580 - 100.0 - 6 Indonesia 9u~ 49ties 169738 268574 30.0 27.4 Social science 244059 474065 43.2 48.4 Nedicin. 22194 24855 3.9 2.5 Science & technolosy 127173 208545 22.5 21.3 Other 2337 4123 0.4 0.4 Total 565501 980162 100.0 100.0 7 Korea Nu~nities 159841 438178 26.0 29.6 Social science 111685 447733 18.1 30.2 edicin 41420 80651 6.7 5.4 Scionce t techology 301874 488738 49.0 33.0 Other 632 26011 0.1 1.8 Total 615452 1481311 100.0 100.0 9 alaysia REnantties 18894 26418 32.8 28.3 Social science 15986 28689 27.7 30.8 Medicin 1639 2920 2.8 3.1 Science & tectnology 21052 27561 36.5 29.6 Other 79 7661 0.1 8.2 Total 57650 93249 100.0 100.0 - 204 - Table 81.7: 32.0 27.2 19.6 17.0 b 8.2 7.4 6.8 el 3.2 2.8 0.0 0.0 3 Durma 102 - - - - - - - - - - - - - 4 hmaa 124 90.0 100.0 84.0 79.0 75.0 67.5 b/ 40.5 37.3 30.9 c 7.4 7.1 6.0 0.0 0.0 5 1a 92 83.0 100.0 70.0 56.0 44.8 36.7 b/ 30.5 25.6 21.8 17.0 13.4 cl 4.7 2.7 0.0 6 Ianncia el 118 100.0 100.0 89.0 82.8 75.3 69.3 59.6 bl 36.9 35.1 34.0 c 18.4 17.7 17.7 0.0 7 Forea 96 100.0 100.0 99.0 99.0 97.0 97.0 97.0 bl 95.1 94.1 93.2 cl 46.6 45.2 44.3 0.0 8 Laos 94 100.0 100.0 100.0 71.0 50.4 40.3 bl 30.2 23.3 19.6 cl 10.6 8.2 7.2 0.0 0.0 9 talaysia 99 100.0 100.0 100.0 99.0 99.0 98.0 97.0 b/ 77.6 69.9 69.9 c/ 41.9 40.2 dl 7.2 6.9 10 Nepal 79 75.0 100.0 48.0 41.8 37.6 33.1 bl 31.8 28.3 cl 25.4 23.7 20.6 0.0 0.0 0.0 0 11 Papus wG 69 74.0 100.0 91.0 84.6 79.6 73.2 67.3 bl 24.9 23.2 16.7 15.7 el 1.9 1.8 0.0 1 12 Philippine 106 100.0 100.0 86.0 80.0 76.0 71.4 66.4 b/ 55.8 49.7 44.7 41.1 el - - - 13 Singapore 115 100.0 100.0 100.0 100.0 100.0 100.0 100.0 bl 75.0 75.0 75.0 75.0 el 20.3 19.6 d/ 5.7 14 Sri Lanha 103 100.0 100.0 100.0 100.0 99.0 92.1 84.7 bl 76.2 69.4 63.1 57.4 57.4 el 18.4 18.4 15 Thailand 97 100.0 100.0 89.0 88.1 86.3 84.6 80.4 bl 32.2 30.5 29.3 cl 15.0 13.2 13.0 0.0 Source: Computed from data on the SD (UNUCOED database. al The rate of entry to primary education in Sangladeah (and possibly also Laos) may be overestimated du to the presence of "baby*classes hich contain =ndrage children. ln fact, when estiating the entry ~*te from enrollment and population data, this factor produced a rate ece«ing 100 percent. We have therefore attributed an entry rate of 100 percent here, thus diluting the ~ayU clasa factor to som extent. The result may of course still be overestimated, although to a lesser extent than at firat sight. b/ E of priary eyele. el End of lover secondary eyele. In the Philippines, the entire oyele is only four years. di An additinal melection point vithin upper secondary cducation. e/ Data frm (World ank - Indonesia 1989) suggeast that the proportion surviving to the end of primry education may have Improved significantly, rising to as much as 75 percent ln the latter half of the 1980s. - 207 - Tabe 2.3u E~oess demand for higher education, Asia, mid-1980s # of entranta al * in last gra o of Local Local Distance Be*oda~ ewa~tokera gEMIg vrjvate oducatio 1 ianl1deth 132580 194764 156389 -----> 3288 2 iuten 298 - -- 3 Burma 133133 - 25151 - 21013 4 chna 2145000 - 809960 0 452344 5 India 1136075 - 343354 -----> 4456 6 Indonesia 519176 983263 97416 142714 40698 7 Korea 371507 790874 313640 -----> 39351 9 Nalaysia 146388 - 24499 -----> 713 10 epa1 54611 - 13619 4339 - 11 Papua Nw u~a 828 - 895 ----- - 12 Phllppnea 660126 595575 64621 321510 - 14 Sri Lmna 67714 150000 5318 0 3096 15 Thailand 290409 - 15194 11425 142467 Sourcos: Data on the uber in the last grade of secondary education are from the ESD (UNESCED) database. Data on the number of eem-takers are fr=m the folloVinS sourcess Govt. of angladesh, (1986) Govt. of bn~onea-USAID <1986) (the data refer to the nuber of applicant for university entrace): Govt. of the Philippinea, National Educational Teating and lesearh Center (1987 personal CmmatIon to Jee-Peng Tan)§ World ank (sri Lanka-1986)8 and Korean Council for University ducation (1988). Data on the intake to higher education are from the followng souroes: ADN (1987) for India and Papua New Guineax Govt. of Bangladeshs (1986>; covt. of Nurma (1986): Govt. of China (1986)1 Govt. of Malaysia (1986)8 Govt. of szt Lanka <1988>3 (Govt. of Thailand, 1985, 1987b): Korean Council for University Eduatton (1988)s Mingat and Tan (1988) for the Philippines and World Bank (Indonesia-1988b). Except for China, India, Indonesia, Sri Lanka and Thailand (regular institutions), the data on fntake are estimated by spreadins total h~nhe «edation enrollmeont over en assumed four- year Cyele. at Arrowa indicate that the numbers to the left refer to data for entrants to boch Local public and private institutions. Nota: Given the difficulty of measuring eoess demand, these data are only suggestive at best, and should therefore be used with extrem caution. J一，、11二。二。。，:1- l‘一‘.&3&&,,&．馮？&o I忽－幼J&..”遺”J -，一‘.&.&&,31&.&&& -‘一1,&,&&o&&.&&, ―綱騙，,&o,1&SO&1忽，&, 〕，-&1·，‘露·1&‘·，,？·‘ 61州緘闖個皂為乞個，參念9。S，忽。為念5。3 1寫必驚e•，頂．。喊開．，馳二心7。3歸。S •L論•忽6.0鷓．翁鈴．0皺。9 ，袖縫啊悶認，1 .0忽1,，念7。忽念鳥．1 功魚零曉觀。7幼．7，必．鳥語．3 11州戶．魚”O暱細∥．幼·7 51.6,1．為51.0 雄網緘11神細”•幼。．2，•O幼。轟論。， 1忽S認開開”討．7論．3閱．•27 01 1為Srl訕臘由．.－一，1。7 絮騙馱閱．矗1。1,1。7，忽。專51。7 幼個驢U劇鑪S矗。7鴆．0顫．7 19．忽 劇口■口•口•■•．•■•■■• ‘跚劇滷•，嬰竺j纖中中•》戶蚌山甲，＿興神坤國曉“州戶“&_f戲．切紹以沁 〔甚，•，J審U煙唱唱．0蠶寫綱蔔勵總《1，螂•】露孕州儿．。蠶Ch鰓騙（1個細夠，1個目6）審 。戲吃。緘取d么由【1，挪挪）．劉憑實dl蘊ei98自》劍必劍bllld扯ti，口7、•”d ,r1t認b觔驪。U《19以）邊“細d認。以”化。緘蠶b“•（19開》， •d黝緘d為閱自《，用陣曉細由O“紅滷觔1開才》． - 209 - PWII-te=Mr ratio In ««~ edmation, Asla, 1970-1~ 23.8 26.2 10.1 9~ 32.1 37.3 34.1 28.5 21.0 17.9 17.2 21.9 24.9 29.3 25.1 20.9 20.3 19.4 20.2 6 Imom sta 13.1 14.4 14.9 15.3 I torga, la. of 38.5 37.1 39.1 34.3 8 La09 18.9 19.2 11.2 9 målayola 25.6 27.3 22.8 22.1 10 »~ - 24.3 31.3 27.5 11 ~ Nav Guinea - 20.3 22.0 23.4 12 ~ Ipplms 33.1 31.5 34.1 32.3 13 sin~ re 19.9 23.1 19.4 20.4 14 Sxl 11-mär- - - - 26.1 Utveft - - - - 15 Thall~ 13.3 27.2 19.6 S~ s: Ag la Table 32.4. rör ~ ~ data for 1985 am fr« Omorment o£ TballaM (1987*). - 210 - Ta M2.6 Pupil-teacher ratio in tertiary edgoation, Asia, 1970-1985 al 1985 Private åp22 ågå M R~SSa Distange ~nsitutkon i angled8sh bi 16.3 11.7 21.6 15.9 - - (21.6) (23.9) 2 ~hutan - - 8.7 10.9 - - 3 Burma 13.2 - - 30.3 0/ - 4 Chia 0.4 3.2 4.7 5.2 36.0 - an Kong 15.2 14.6 12.5 - - 5 India 18.8 19.6 19.3 15.7 776.5 di 6 Indonesia 12.4 6.0 9.3 14.0 689.7 46.1 7 Korea, Re. of 19.3 20.8 29.1 42.4 414.7 41.1 8 Laos - - 10.1 10.1 - - 9 Nalsysla - - 10.4 11.4 - 10 Nepal - 15.5 13.2 - 11 Papua Nev Guinea - - 7.9 7.7 - 12 Philippines el 22.5 24.2 29.2 16.0 - 48.0 (31.7) 13 Singapore 11.9 15.6 10.2 - - - 14 Sti Lanka 8.3 7.7 8.9 10.7 84.9 - 15 Thailand 7.4 14.4 18.4 8.3 618.8 17.6 SourBes: =S8 (1NESCOED) for 1970-1980 for all coutries and for 1985 aupplanted by estimates based on Goveranmnt of Bangladesh (1986): Government of China (1987a) iACU (1987) £or India# Jaiwa (1988> for the Philippiness Government of Sri Lnka (1988) and Governmant of Thailand (1985). at Data refr to regular public institutionz unless otherwise indioated. bi Data refer only to universities; figur. in paranthests reflect the averages across all types of higher eduoation, el Figure reflecto veighted average for regular and correspondence higher education. 4/ Vigure reflects data for Anda~ Pradesh Open University. Data £rom the other open university are not available. e/l figures for 1970-80 overestiate the true pupil-teacher ratio for tertiary education sinee many state universities enrolled a large nubr of primary and secondary level students. Such student& vere reamoved fram the numerator in James (1988) estibate for 1985s £or comparison the unadjusted f~igur appears in parentheses. - 211 - Ta 3.-11 Dependemy rattos, Asa, 1970-2000 Population aged 0-14 a* ratto of Populatton ased 0-14 ag ratio of Donulation ed over 15 to nonulation a&ed 15-64 Iffi Iff u~ua m im Iff i= m 1 Bangladesh 0.82 0.81 0.71 0.78 0.62 0.55 0.56 0.46 0.49 0.42 2 hutan 0.68 0.68 0.64 0.66 0.71 0.46 0.47 0.44 0.44 0.47 3 suma- 0.64 0.66 0.67 0.70 0.50 0.44 0.45 0.47 0.47 0.35 4 China 0.61 0.52 0.53 0.40 0.36 0.44 *.39 0.44 0.33 0.26 5 India 0.71 0.69 0.63 0.61 0.49 0.49 0.48 0.45 0.44 0.35 6 Indonesta 0.75 0.73 0.68 0.63 0.50 0.50 0.53 0.48 0.46 0.35 7 Korea 0.69 0.58 0.48 0.42 0.34 0.52 0.43 0.37 0.31 0.24 8 Laos 0.68 0.69 0.90 0.81 0.77 0.45 0.46 0.60 0.53 0.49 9 Nalaysia 0.79 0.72 0.61 0.55 0.45 0.56 0.53 0.44 0.41 0.33 10 Npal 0.69 0.70 0.71 0.71 0.60 0.46 0.48 0.49 0.49 0.51 11 Papua NevGunea 0.69 0.69 0.70 0.62 0.76 0.47 0.48 0.53 0.50 0.40 12 PhilIpplnes 0.79 0.78 0.10 0.63 0.53 0.53 0.55 0.48 0.47 0.37 13 Sinapore 0.60 0.47 0.40 0.36 0.27 0.48 0.37 0.30 0.26 0.20 14 Sri Lanka 0.68 0.61 0.56 0.50 0.42 0.49 0.47 0.42 0.36 0.31 15 Thailand 0.81 0.77 0.64 0.55 0.40 0.56 0.55 0.47 0.41 0.29 Sources Cputed frm data on the ME8D (SOCID) database for 1970-85s and from Zechari*h and Vu (1988> for 2000. ble33.2i: Population and eon~omio grouth rates, Asia, 1975-2000 ANNUJA. RATE OP POPULATION GRONTH REAL EO0NMIC GROTH Overall Poulation Ponulation 5-14) {t x.a.I 1975-85 1985-2000 1975-5 198-2000 al 1975-85 1I 19 200 al 1 an&1ash 2.4 2.2 1.8 2.0 4.4 1.8 4.9 2 »huten 2.0 2.3 1.6 3.0 6.1 - - 3 aua 2.0 1.9 2.4 1.0 5.8 5.1 3.5 4 China 1.3 1.4 0.2 0.4 7.8 7.6 6.6 5 India 2.2 1.8 1.6 1.3 4.4 7.2 4.8 6 Indonesta 2.2 1.8 1.8 1.0 6.1 3.6 3.9 7 Korea 1.4 1.3 -0.8 0.3 7.4 7.8 6.8 9 Nalaysia 2.4 1.9 0.8 1.4 6.3 4.1 5.0 10 Nepal 2.5 2.6 2.7 3.1 3.1 5.7 3.8 11 Papua New Gu~nea 2.6 2.2 2.6 2.0 1.5 2.0 5.1 12 PhilippLAS 2.6 2.1 2.1 1.5 2.5 5.0 5.3 14 Sri 1.nkA 1.8 1.5 0.1 1.1 4.9 5.5 4.8 15 Thailand 2.1 1.7 0.6 0.5 5.8 5.8 6.0 Sources: Cenpued from populatton and GNP data on the BED (SOCIND and UNESCOED) databaset data on projected population and ea~c" Sroth ar. repectively frm ZebarLah and Vu (1988) and the World Bank's ANDEE database. al Projeoted - 212 - Tab -33: Pfrten adulta iterate, dMia, 1970-1985 1 Bangladeh 23 26 29 33 2 heonn - - 5 15 3Sua 71 - 66 - 4 China - - 66 69 5 Indla 34 - 45 43 6 Indonska 54 - 67 74 7 Korea 88 - - 92 8 Laos 33 - - 84 9 Mlaysia 60 - - 74 10 Npal 14 19 24 26 11 Papua wv ou~nea 32 32 32 45 12 Philipplnea 82 - - 86 13 Sngapore 69 83 - 86 14 gr£ Lanka 77 - 86 87 15 Thailand 79 - 88 91 Soures: UNICir (1987) for 1970 and 1985. por 1975 and 1980, 8E8D (0C~UD), supplemented by World Bank (Bangladesh-1987) for Bangladeås World Bank (åhutan- 1988~): World Bank (China-1983) World Bank (China- 1988)* World Bank (Indla-1988)& World Bank (Papua Nw GU~a-1986): World Bank (Papa New Gunea-1988)l World Bank and World ~ (INU) s World ~ (Ps~ Mm Gu~-1987)* World ~ 1.4 4.3 6.3 0.6 9 ralaysi 14.1 21.3 190.3 190.3 - 1.5 13.5 13.5 - 10 Nepa 9.0 13.5 249.0 249.0 - 1.5 27.7 27.7 - 11 Papua New oin a 29.0 65.0 1050.0 1030.0 - 2.2 36.2 36.2 - 12 ~bilippine 1985 5.8 8.6 50.0 50.0 - 1.5 8.6 8.6 - 1988 8.5 11.8 - - - 1.4 - - - 14 Sri L-an. 1985 6.1 9.3 83.3 111.2 22.7 1.5 13.7 18.2 3.7 1988 6.6 9.6 - - - 1.5 - - - 15 Tha1and 15.5 15.3 39.9 177.9 14.2 1.0 2.6 11.5 0.9 Souroes: KIng (1988) for Banldeaaht Govt. of Chna (1987, 1986 and 1984) Govt. of inAI (1983) and Borda (1988) for Indias Govt. of nadnasia-USAID (1986) Govt. of India (1987); World Bank (Koraa-1985), Govt. of Korsa (1987) and Rorean Comncil for Univergity ueoation (1988) Govt. of Kalaysta (1988, 1986) and Laya (1988) for MalaysLas Govt. of Nepal-USAD (1988> World Bank (Papua New Gunea-1987) Xingat and Tan (1988) for the PhIlIppiness Govt. of Sri -anka (1986 and 1987) and S~ith (1988) for Sri L.aas World Bank (Thalland-1985) and Govt. of ThaLland (1988). Data on unit ooat in open 4 ther edueation relot estimtes based on Brisa-n (1987). ai eighted by dL&triUtion of enroll~anta in publie seator aoross regular institution and distane education. - 221 - l 4.21 eahev rewma~tio as ratto to per ofptta MM, daa, ~td-19809 gjingn Becondar gagghe 1 aladesh 2.2 - - 2 hutan - - - 3 Sua - - 4 ChIva 2.0 3.4 6.9 5 India al 2.9 3.1 - 6 Indonesta 2.5 5.2 5.7 bi 7 Korea ai 5.0 5.5 - 9 malaygia 2.4 3.1 - 10 mpal 2.8 5.0 di - 11 Papua New Guinea 6.8 10.0 - 12 PhtiLPpine 1985 1.6 1.7 - 1988 *1 2.2 2.4 - 14 Srti Lnka 1986 1.6 2.1 - 1988 of 2.0 2.3 - 15 Thailand 2.5 2.9 - Sou~es: Govt. of ba~1sah (1984)s Govt. of ChiMa (1987, 1986) Govt. of Tlwd na£a-U~AID (1986) Laya (1988) for malaystsa Govt. of Nepal-USAID (1988): Norld ~ank (Papua New aunea-1987)8 Ninat and Tan (1988) for the Philippuness Govt. of Sri Lanka (1986, 1987) nd Govt. of Thailand, (19870). al Furs raeloot eatimates based on data on nit Gosta ad pupil- te~ire ratio*. bligure reflaota salaries after the 69 per=ent luores bet~een 1984 and 1985. of Data rafleot «etbmat&s based on the new salary ~es for tashrs. di F1ure refleats the weishted aver~ae salaries for lover and uper seoo~uy teachers. - 222 - Table 24.3: Fe* for public education an Z of uait operattnø costa, Asia, mid-1980s aiher Oducation rgi~lg 5e8odar Rgg~ar ag 1 Bangladagh 7.4 4.0 0.1 - 2 btan - - - - 3 Burma - - - - 4 China 4.8 3.2 0.3 - 5 India al - 11.6 4.98a - 6 Indonesia 7.1 27.4 18.9 - 7 Korea 0.0 34.2 45.9 32.0 9 Malaysia 3.7 4.0 5.8 - 10 Noel 0.0 40.7 10.4 - 11 Papua Nov Guinea 8.7 39.8 0.0 - 12 Philippinøs 0.0 9.3 bi 15.3 - 14 Sri Lnka el 3.1 3.1 3.4 57.7 15 Thailand dl 0.1 18.3 5.0 27.5 Source: King (1988) and world Bank (Banladesh-1988-b) for Banladeabø world Bank (Chlna-1986b)j Kolbatkar (1988) and assoiation of Cmmaoualth ULivritis (1987)8 for hiser edueation in Indiat and Govt. of India (1987) for primary and seflnay education Gavt. of Indonesia-USAID (1986); Govt. of Korea (1987)8 Noook (1985> for Nalaysia,i Govt. of Nepal-USAID (1988> and ililsina (1988> for Nepal; World Bank (Papa New Guinea-1987)1 Ningat and Tan (1988) for the Philippieal Govt. of Sri Lanka (1986, 1988); aud Govt. of Thailand (1987). Data on the level of cost reovery la open higher edueation aro from Sriaa-an (1987). al For primary and øe«ondary eduoation, data ar. estiated from e schedules lo highet ation, the figur. ig uh =maller than the 20% figur. o£ted la Kol hatkar because hig cost data did not Inalude the east of faculty salaries. bi Yees for public 9e~adary eduation, inaluding In local schoola, vere abolished in 1988. - 223 - Tabl 5.11 Rato* of return to investnt in education, Aala, latest available Year &patal Private IaM uE soEE 8on sjahoE Prma &eod Eidm 5 India 1978 29.3 13.7 10.8 33.4 19.8 13.2 6 Indoneaa 1982 18.0 15.0 10.0 - - - (14.5) - - - - - 7 tor.a 1982 - 10.9 13.0 - - - 9 malayala 1983 - - 7.6 - - 12.2 11 Papua New Guinea 1982 19.9 12.0 2.8 29.4 14.7 8.1 12 Philippines 1985 11.9 12.9 13.3 18.2 13.8 14.0 (4.4) (9.3) (11.6) (7.2) <10.2) (12.5) 15 Thailand 1975 12.0 24.0 12.8 - - - 1985 - - 13.3 - - 17.4 Souoeas Peaoharoposlos (1985) for Indial USAID (1986) tor Indonestaf EEDI (1983) Lor Rorea~ Nehmet and Tip (1986) for kalaysia Gannioott (1987) Lor Papua New Guinea; Tan and Paquo (1988) Lor the Philippuest and oet. of Thailand (1987b)$ and Suppachal (1976) Lor Thailand. Nete: P8ures in parenthses denote rates of return tor in~mplete education. - 224 - Tabl I. Dates of establb sam of open nIverattlea, mua 1 he ~s1adoh R 2 ~an nå 3 Ruma al 1970's 4 China bl 1960's 5 India el 1982, 198 6 rnanla 1984 Japan 1985 7 Korea di 1972, 1982 9 Malayeia Rh 10 ~epa m 11 Papua No~ inea da 12 Philippineå R 14 eri Lanka 1980 15 Thailand 1970, 1978 8ourcess for 3um, fram Go . of Burma, budt dotmants for Undla from oläatkat, M.R. (1988). *Coutry Poper: India* Paper presented at tho UNESCO resional smar on mobilisation of addtional fundia for blahb education, san~k, Auust 22-27, 1988 for r sdonnia, from World ank (1988) SA# for Zorea, from Govt. of Korea (1988> Eannaion ~0 Korea 1987-1988, for Gri Le~, from World (198M) Ori ~-an. gnation and Tra~in Sontor ~em~randuaa for Thailand, from Chantavanich, 8. and Q.N. ry (1988) hailand la T. Neville Postlevatte (ed), The .nyolopedia of comparattve oducation and national nystems of edu~ation, Peram~n Press: New York. al University eorresponao~ cours offered. bl U~opened in 1979. 0d First date denotes opnin8 of the ndhra Pradesh Open University and the oenna the opning of Indira Gamdhi National Open Univeraity. di Firat date denotes opening of of Korea Air and corrspndonc University. The second date donotes the ostabli~heent of the first of a open colleges, three of uhich ame privat,. * 225 - TableJ.s Returne by field of study in higher edneation, Thailand, 1985 Pure science 13.6 19.5 Agrioulture 15.0 19.0 Fine Artalarmbitecture 10.0 17.9 Busanitieslsooial so. 14.5 15.9 Medicioe 10.4 13.8 Medical technology 7.8 17.5 Accountancy 18.0 17.5 Law 18.0 15.4 Education '10.0 15.9 Engneeriag 17.9 22.0 AVERAGE 13.5 17.4 Sources Govt. of Thailand (19876) pp. 6-35. Mote that the author(s) report estimates by type of firm in which a graduate is isorkiag. For brevity, the estimates selected for reporting here as followas the social rates of retur are the returns calculated from the earninge profile of graduates working in the private sector where wages are more likely to reflect marginal produatIvitys the private returns are these calculated from the earnings profile of graduates otkingin all sectors. * 226 - APPENDIX C: NISCEIUANEOUS TABLES REFENCED IN TEXT AND FIGURES - 227 - le .L: LS regreason of th relatonshtp betee Gross enrollma ratto* ad the per ofpita om, m1d-1980s ASIA WORLD Ea Sø scondazv fluffer Pri~ar SØGGndarv fisiher Per capita GP x 101-2 1.44 3.47 1.03 0.57 1.19 0.40 (0.9) (2.0) (1.8) (3.2) (8.3) (5.3) (Per capita GP)**2 x 10E-6 -1.40 -0.40 -1.33 -0.32 -0.53 -0.16 (0.6) (2.3) (1.6) (2.5) (5.1) (3.0) Interept 79.4 18.2 3.38 82.74 24.31 5.06 Nuber of observationa 16 16 16 100 99 91 R-squared 0.46 0.55 0.24 11.5 58.5 43.9 Sourcs: Data on enrollment ratios from UMES00 (1987) data on per ofpita OM frm orld ank BSD database. lable C.2t Average per capita om and gross enrollmant ratios, orld rogios, 1980. Gress enrollment ratio$ (2) No. of Per capita Coutiriea P £UM) praE sonar Iah Asia al 14 585 90.5 41.0 11.0 Africa 35 517 77.8 19.5 1.5 Latin America 20 1401 101.8 48.6 16.5 1mena bl 14 2902 92.9 46.9 14.2 at Excludea HK, Sngapor, Japan. bi Exludes UAS, Ku~ait Sources: A* n Table C.1. Table C.: Data for te~t f£aur~s al Tstar of Index of prIvate Per capita U.C. (US$) Index of X reaobJnn Pr~ry ed, extent of Uat cost £U•nal? 0Cm (195) of re~ular overall end of n±It Costa ter-yelc of pub. reg. la bh~r USM hie ed. postIMs na (l p.0 l ) selection habeh Lu gdaon 1 Bas1 159 453 1.36 24 6.4 8 1.57 16.5 2 phutan 151 17 13 2 mm 184 4 China 273 900 1.32 68 9.2 54 1.82 0.3 5 Zndia 259 515 1.00 37 6.0 15 1.27 7.1 6 1~9adana 470 497 1.02 60 12.6 46 0.58 48.7 Sorea 2040 2132 1.11 97 13.5 87 0.57 76.6 8 Laos 332 40 21 9 alaysia 1860 3540 1.25 97 14.1 79 1.05 15.1 10 NpaL 142 354 1.07 33 9.0 5 1.37 31.8 11 Papua Ne, Guinea 621 6521 4.38 67 29.0 57 5.78 6.3 12 hIlippInes 581 291 0.45 66 5.8 18 0.28 85.8 13 Singapore 7093 100 99 14 gri Lanka 374 416 0.55 83 6.1 58 0.61 20.5 15 Tha~ln 712 1267 0.87 80 15.å 72 0.89 26.9 0 al All fIur.. refer to early to m~d-1980s. bl Elative to regional averaff. Table C.3: ct.) Share of Index of oaumlativo Devaton Per capita Ratio of niLt costs to overall resoures frm GM" (198$) reg8onal ave«ae mphtass on e~tved by cu=ve ta CG~zvE Uva . r EMg sconda~ 2~ag h~ae ed. to» 10 1 ft~x 5.31 1 S aladesh 159 0.62 1.60 1.86 6 75.8 0.10 2 Bhutan 151 3 Sum 184 4 Chia 273 0.88 1.47 1.59 1 28.9 -0.40 5 Inda 259 0.58 0.92 1.51 5 60.8 0.20 6 lna~la 470 1.21 1.25 0.59 -3 21.1 -0.25 7 Norea 2040 1.59 1.25 0.46 -3 13.3 0.00 8 Laos 332 9 alayia 1860 1.36 1.14 1.24 0 32.1 0.35 10 Nepal 142 0.87 0.72 1.62 1 52.1 -0.40 11 Papua New Gunaa 621 2.79 3.47 6.85 4 55.3 0.50 12 hLippnes 581 0.56 9.46 0.33 -2 14.1 -0.30 13 Singapore 7093 14 Sxl ~enka 374 0.59 0.50 0.54 -1 27.0 -0.25 15 Thaland 712 1.49 0.82 0.26 -2 23.4 0.00 TabLe C.3 (contd.) 9 of Ist year mtrants X dov:Latim of it costa ftm survivng regona~ avermo Por capita n emales in total enrollments to end of ei_le GNP (1985) g29gM US£~ g gg scondp~ m~B pgi~ag SeOndarr prie~ scoo~ gigh B Eangladoch 159 40 28 19 24 18 -0.38 0.6 0.86 2 ama 151 34 18 17 17 38 3 Bua 184 4 C~Ina 273 45 40 30 68 15 -0.12 0.47 0.59 5 Ta6a 259 40 34 29 37 36 -0.42 -0.08 0.51 6 TndntsLa 470 48 43 32 60 49 0.21 0.25 -0.41 7 Korea 2040 49 47 30 97 46 0.59 0.25 -0.54 8 Laos 332 45 41 36 40 23 9 alaysLa 1860 49 49 45 97 51 0.36 0.14 0.24 10 E.pal 142 29 23 20 33 66 -0.13 -0.28 0.62 11 Papua Ne Guinea 621 44 36 23 67 8 1.79 2.47 5.85 12 hi ~pins 581 49 50 54 66 36 -0.44 8.46 -0.67 13 Slagapor 7093 47 50 42 100 27 14 Sri Tanka 374 48 53 40 s 25 -0.41 -0.5 -0.46 15 ha~land 712 48 48 46 80 41 0.49 -0.18 -0.74 Souroes: see Appadi 8. !aЫ® C.s: (o0atd.) АУ8. tl8СЪО1 рСу а3 1'at10 КtlФЪtt Of jWj/S1i te евк eaeita 6ЯР �!! tвiCЪR! аьвг c198s) � � � �ик � � 1 в�� и9 2.2 - а7.о 2в.2 2 вь�еоn и1 - - 9s.s 1о,1 9 т,� 1еь - - ав.ь 2s.s ь СЫаа 279 2.0 9.4 ?4.9 17.2 S Iadia 2S9 2.9 9.1 ST.6 20.2 б 2nдOnesis 470 2.S 9.2 ц.3 13.9 7 Rosma ZОь0 S.0 S.3 98.9 94.9 8 LoOS 392 - - 24.9 11.2 9 ltвlayaia 18б0 2.ь 3.1 24.1 22.1 10 Наря1 142 2.8 S.0 3S.S 27.3 11 Papus 8a�r О�а б21 в.8 10.0 31.0 25.4 12 Philipplaea S81 1.б 1.7 50.9 92.9 13 8ingaposa 7099 - - 27.1 20.б 1ь 8rl Lваlев 374 1.б 2.1 51.7 2б.1 ! и lheцa:sd 712 2.S 2.9 19.9 19.б �у W и✓ 80us0es: 8ае Арралд3s В. � - 232 - I Bibliography Association of Commonwealth Universities (ACU) (1987) Commonwealth Univergities Yearbook 198Z. 3 Volumes. London: ACU. Asian Development Bank (ADB) (1987) DistMe Educati . Proceedings of the Regional Seminar on Distance Education, 26 November - 3 December 1986. Manila: Asian Development Bank. Asian Development Bank (ADB) (1986) ftgal. Education Sector StUft. Manila: Infrastructure Department. Bank of Korea (1988) EconomLc Statistics Yearbook 1988. Seoul: Bank of Korea. Behrman, Jere R. and Nancy Birdsall (1983) " The quality of schooling: quantity alone is misleading," Ameriem Economic Lai , 73(5): 928-946. Bordia, A. (1988) "India", in Poatlethwaite, T. Neville (ed) Ma gn=lgaeadia of C2=arati-9 ad National fintAns of Educati . Now York: Pergamon Press. British Council (1984) Education Prglile, India. London: Editorial Section, Central Information Service. Caldwell, J.C.; P.H. Reddy; and P. Caldwell (1985) "Educational Transition in rural south India," Pogulation and DaYA120= Revi , Vol 11 (1): 29-51). Gannicott, Ken (1987) The ftalMtion of Human CAULW in E#M Now Guine Canberra: Islands/Australia Working Paper, National Centre for Development Studies, Australian National University. Government of Bangladesh (1984) New 1Afe in Education. Dhaka: Ministry of Education. Goverment of Bangladesh (1986). Educational Statistics 1986. Dhaka: BANBEIS, Ministry of Education. Government of Burma (1986) Roort ta lbo Philbu Hluttaw on the FinancLal. Economic and Social Gondltig= of the Socialist L92Mblic of Burma for 1985/6. Rangoon: Ministry of Finance. Goverment of China (1984, 1986) Achievement of Education--in China. Beijing: People's Education Press. - 233 - Government of China (1985) 1982 Poulation Census of China. Beijing: Statistical Publishing House. Government of China (1987) Yearbook of Chinese Education Statistics (in Chinese). Beijing: Ministry of Education. Government of China (1987) Statistical Yearbook of China 1987. BeiJi-ag: China Statistical Information and Consultancy Service Centre. Government of India (1980) Fourth All-India Educational Survey. New Delhi: National Council of Educational Research and Training. Government of India (1983) Analysis of BudAeted Expenditure on Education 1980/81 to 1982/83. New Delhi: Planning and Monitoring Unit, Ministry of Education. Government of India (1985) The Seventh Five-Year Plan. New Delhi: Ministry of Finance. Government of India (1987) A Handbook of Educational and Allied Statistics. New Delhi: Ministry of Human Resource Development. Government of India (1987) Selected Information on School Education in India. New Delhi: Planning, Monitoring and Statistical Division, Dept. of Education. Government of India (1988) Selected Educational Statistics 1986-87. New Delhi: Ministry of Human Resource Development. Government of Indonesia - USAID (1986) Indonesia: Education and Human Resources Sector Review. Washington, D.C.: Ministry of Education and Culture with the United States Agency for International Development (USAID). Government of Korea (1987) Statistical Yearbook of Education 1977-87. Seoul: Ministry of Education. Government of Korea (1988) Education in Korea 1887-1988. Seoul: Ministry of Education. Government of Malaysia (1986) Fifth Malavsia Plan 1986-1990. Kuala Lumpur: Office of the Prime Minister. Government of Malaysia (1988) Belaiawan Persekutuan. Kuala Lumpur: Ministry of Finance. Government of Nepal - USAID (1988) Nepal: Education and Human Resources Sector Assessment. Virginia: Ministry of Education and Culture with the United States Agency for International Development (USAID). - 234 - Government of Sri Lanka (1986) Some Indicators and Projections Relevant to the First and Second Level General Education. Colombo: Ministry of Education. Government of Sri Lanka (1988) Statistical Handbook 1986. Statistics on Hipher Education in Sri Lanka. Colombo: University Grants Commission. Government of Sri Lanka (1987) 1988 New Consolidated Salaries Imlementation of Recommendations of Administrative Reforms Committee - Salaries and Cgdes, Colombo: Department of Government Printing. Government of Thailand (1985a) Statistics Report on Education_in Thailand. Bangkok: National Commission on Education. Government of Thailand (1985b) General Information, Bangkok: Ministry of University Affairs. Government of Thailand (1986) Statistical Yearbook of Thailand 1985-86. Bangkok: National Statistical Office. Government of Thailand (1987a) The Sixth National Education Development Plan. Bangkok: Ministry of Education. Government of Thailand (1987b) Costs and Contribution of Higher Education. A Case Study of Thailand. Bangkok: Educational Research Division, Office of the National Education Commission. Government of Thailand (1987c) Basic Educational Statistics for Pre-primary and Primary Education. Bangkok: National Education Commission. Government of Thailand (1988a) Thailand's Budet in Brief. Bangkok: Bureau of the Budget. Government of Thailand (1988b). A Glance at Educational Statistics during the Fifth Plan (1982-1986). Bangkok: Office of the National Education Commission. Hartley, Michael J.; and Eric V. Swanson (1984) Achievement and Wastage: An Analysis of the Retention of Basic Skills in Primarv_Education. Final report of the international study of the retention of literacy and numeracy: An Egyptian case study. Washington D.C.: The World Bank (mimeo). Husen, Torsten; and Postlewaite, T. Naively (1985) The Encycloedia of Educatin. New York: Pergamon Press. International Association for the Evaluation of Education Achievement (IEA) (1988) Science Achievement in Seventeen Countries. A Preliminar Report. U.K.: Pergamon Press. - 235 - James, Estelle (1988) Philioines Education Sector Review. Higher Education. Washington, D.C.: Asia Country Department II, The World Bank, mimeo. Jimenez, Emmanuel (1986) "The structure of educational costs: multiproduct cost functions for primary and secondary schools in Latin America," Economics of Education Review, 5(1): 25-29. Jimenez, Emmanuel; Vicente Paqueo; and Ma. Lourdes de Vera (1987a) "The relative efficiency of private and public high schools in the Philippines," Washington D.C.: Population and Human Resources Department, The World Bank (mimeo). Jimenez, Emmanuel; Vicente Paqueo; and Ma. Lourdes de Vera (1987b) "Does local financing make public primary schools more efficient? The Philippine case," Washington D.C.: Population and Human Resources Department, The World Bank (mimeo). King, Elizabeth (1988) ReoOrt on Education Sector for Bangladesh Public Expenditure Rview. Washington, D.C: World Bank Population and Human Resources Department, mimeo. Kolhatkar, X.R. (1988) "Country Paper - India", paper presented at the Regional Seminar on Mobilization of Additional Funding for Higher Education, 22-27 August 1988, Bangkok: UNESCO Principal Regional Office for Asia end the Pacific. Korea Educational Development Institute (KEDI) (1983). Education and Economic Development, Seoul: KEDI report by Kong En-Bae et. al, in Korean. Korean Council for University (1988). Equity. Onality. and Cost in Higher Eduatio. Bangkok: UNESCO Principal Regional Office for Asia and the Pacific. Kumar, Krishna (1985) "Reproduction or Change? Education and Elites in India," Economic and Political Weekly, Vol XX, No. 30: 1280-1284. Laya, Jaime (1988) Malavaa financing dmar and secondary education. Washington, D.C.: Asia Country Department II, mimeo. Lescage, Andre (1981) Literacy and Illiteracy. Paris: UNESCO, Educational Studies and Documents #42. ebmet, Ozay; and Yat Hoong Yip (1986) Human Capital Formation in Malaysian Universities. Kuala Lumpur: Institute of Advanced Studies Occasional Papers and Reports KLB #2. Mingat, Alain and George Psacharopoulos (1985) Education Costs and Financing in Africa: Some Facts and Possible Lines of Action. Washington D.C.: Education and Training Department, The World Bank, report No. EDT13. - 236 - Mingat, Alain; and Jee-Peng Tan (1985) "On equity in education again: an international comparison,'0 Journal of Hwman Resources, 20 (2): 2498- 308. Mingat, Mlain; and Jee-Peng Tan (1986) "Public Funding of Education: A Comparison,' Comgarati3e Education !eview, Vol. 30, No. 2. 260-270. Mingat, Alain; and Jee-Peng Tan (1988) The cost and financLug of education i the Philippines: A documentation and assessment of goligX issues. Washington D.C.: Asia Technical Deatmen, mimeo. Moock, Peter R. (1985) Education in WaWaWia: A Review of lapeditures and Discussion of Issues.. Washington, D.C.: World Bank Education and Training Department, Report OEDT11. Mullick, S.P. (1987) 'Distance Education in India," in Proceedings of the Regional Seminar on Distance Education, 26 November - 3 December 1986, Bangkok: Asian Development Bank. Murray, Thomas (1988) "Indonesia,' in Naively T. Postlevaite (ed.) MM En=1cloedia of Qonaratikn Education and National SXstems of Educat , New York: Pergamon Press. Paderanga, Cayetano, (1987) 'Exploring the PRODED results: preliminary report on the HSMS Project,* Manila: Government of Philippines, Department of Education, Culture, and Sports, (mimeo). Patwari, A.M. (1987) 'The universities of Bangladesh," inComrat Universities Yearbook 1987. London: Association of Commonwealth Universities. Postlethvaite, T. Neville (1988) The ln=y1lo0dia of Coanarati3m Educatio and National System of Education. New York: Pergamon Press. Psacharopoulos, George (1981) 'Returns to education: An updated international comparison," in CqaaieEuai vol. 17, no. 3, pp 321-41. Psacharopoulos, George (1984) 'The contribution of Education to Economic Growth. International comparisons,' in John W. Kendrick (ed) International CogUarisons of Productivity and Causes of the Sl 4. Cambridge: American Enterprise Institute/Ballinger Publishing Company. Psacharopulos, George (1985) 'Returns to Education: A further international update and implicatiovs," Journal of Human Resources, vol. xx, no. 4 (Fall) pp. 584-604. Psacharopoulos, George (1988) 'Efficiency and Equity in Greek Higher Education,' Miera 26(2): 119-137. - 237 - Smith, W.J. (1988) "Chapter III Education and Training," contribution to 1988 Country Economic Report for Sri Lanka. Washington, D.c.: Asia Country Department I, mimeo. Srisa-an, Wichit (1987) "Financing and Cost-Effectiveness of Distance Education," in Proceedings of the Regional Seminar on Distance Educatio, 26 November - 3 December 1986, Bangkok: Asian Development Bank. Srisa-an, Wichit (1988) "Country Paper - Thailand," paper presented at the Regional Seminar on Mobilization of Additional Funding for Higher Educatio, 22-27 August 1988, Bangkok: UNESCO Principal Regional Office for Asia and the Pacific. Suppachai, Panitchapakdi (1976) "Changes of the rates of return to investment on education," in Rak Mueng Thai (Love of Thailand), Bangkok: Sciences Society of Thailand. Tan, Jee-Peng; and Vicente Paqueo (forthcoming) "The economic returns to education in the Philippines," in International Journal of Educational Development. Tilak, Jandhyala B.G. (1985) Discriminatory Pricing in Education. NIEPA Occasional Paper, New Delhi: National Institute of Educational Planning and Administration. Tilak, Jandhyala B.G. (1987) Educational Finances in India. NIEPA Occasional Paper, New Delhi: National Institute of Educational Planning and Administration. Tilak, Jandhyala B.G. (1988) Educational Finances in South Asia. Nagoya: United Nations Centre for Regional Development Working Paper #88-1. Timilsina, Parthibeshwar (1988) "Mobilizing of additional funding for higher education in Nepal," paper presented at the Regional Seminar on Mobilization of Additional Funding for Higher Education 22-27 August 1988, Bangkok: UNESCO Principal Regional Office for Asia and the Pacific. UNESCO (1984) "Evolution of Wastage in Primary Education in the World Between 1970 and 1980," paper presented at the International Conference on Education, International Conference Center, Geneva, 16-25, October, 1984. UNESCO (1985) Development of Education in Asia and the Pacific: A Statistical Review, Fifth Regional Conference of Ministers of Education and Those Responsible for Economic Planning in Asia and the Pacific, organized by UNESCO with the cooperation of ESCAP, Bangkok, 4-11 March. UNESCO (1987a) Statistical Yearbook 1987. Paris: UNESCO - 238 - UNESCO (19876) Statistical Digest 1987. Paris: UNESCO UNESCO (1988) Education Statistics -- Latest Year Available. Paris: UNESCO. United Nations (1985) Selected Statistics and Indicators on the Status of Wmeo. World Conference to Review and Appraise the Achievements of the United Nations Decade for Women: Equality, Development and Peace, July 15-20, 1985. Nairobi: Report to the Secretary General. Watson, Keith (1981) "The higher education dilemma in developing countries: Thailand's two decades of reform," in Higher Education, 10: 297-314. World Bank (1986) Financing Education in Developing Countries. An Exploration of Policy ORtions. Washington D.C.: The World Bank. World Bank (1988) World Development Report. Washington D.C.: Oxford University Press for the World Bank. World Bank (Bangladesh - 1978) Bangladesh. Education and Training Sector Memorandm. Washington D.C.: East Asia and Pacific Regional Office, report #2037a-BD. World Bank (Bangladesh - 1981) Bangladesh. Education Sector Memorandum. Washington D.C.: South Asia Projects Department, Report #3548-BD. World Bank (Bangladesh - 1987) Bangladesh. Promoting Higher Growth and Human Deelopment. Washington D.C.: World Bank Country Studies Series. World Bank (Bangladesh - 1988a) Bangladesh. Adiustment in the Eighties and Short Torm Prospects. Washington D.C.: Asia Country Department I, Report #7105-BD. World Bank (Bangladesh - 1988b) Bangladesh, A Review of Selected Issues in Education. Washington D.C.: Asia Country Department I, Report #6770-BD. World Bank (Bangladesh-1988b) Bangladesh. A Review of Selected Issues in Educatio. Washington D.C.: Asia Country Department I, report # 6770-BD. World Bank (Bhutan-1988a) Bhutan Primary Education Project. Staff Appraisal Report. Washington D.C.: Asian Country Department I, Report #7010-BHU. World Bank (Bhutan-1988b) Bhutan. Development Planning in a Unique Envmznt. Washington D.C.: Asia Country Department I, Report #7189-BHU. World Bank (Burma - 1985) Burma Policies and-Prospects for Economic Adjustment and Growth. Washington D.C.: South Asia Regional Office, Report #4814-BA. - 239 - World Bank (China - 1983) China Socialist Economic Development. Vol. III The Social Sectors. Washington D.C.: World Bank Country Study series. World Bank (China - 1985) China. Issues and Prosects in Education. Annex to China. Lon-Te=a Development Issues and Options. Washington D.C.: World Bank Country Study series. World Bank (China - 1986a) China. Growth and Development in Gansu Province. Washington D.C.: East Asia and Pacific Regional Office, Report #6064-CHA. World Bank (China - 1986b) China. Management and Finance of Higher EducAtio. Washington D.C.: World Bank Country Study series. World Bank (China - 1988) China Teacher Training Proiect Washington D.C.: World Bank Staff Appraisal Report #7187-CHA. World Bank (India - 1987) India. An Industrializing Econoun. In Transition. Washington D.C.: South Asia Regional Office, Report #6633-IN. World Bank (India - 1988a) India. Recent Developments and Medium-Term Issues. Washington D.C.:' Asia Country Department IV, Report #7185- IN. World Bank (India - 1988b) State Finances in India The Case of Gujarat. Washington D.C.: Asia Country Department IV, mimeo. World Bank (Indonesia - 1982) Indonesia. Financial Resources and Human Development in the Eighties. Washington D.C.: East Asia and Pacific Regional Office, Report #3795-IND. World Bank (Indonesia - 1984) General Secondary Education in Indonesia: Issues and Programs for Action. East Asia and Pacific Regional Office, Report #4954-IND. World Bank (Indonesia - 1988a) Indonesia. Adjustment. Growth and Sustainable Developmnt. Washington D.C.: Asia Country Department II, Report #7222-IND. World Bank (Indonesia - 1988b) Indonesia. Higher Education Development Project. Washington D.C.: Asia Country Department II, Staff Appraisal Report #7085-IND. World Bank (Indonesia - 1988c) Indonesia. Selected Issues of Public Mana9emet =Washington D.C.: Asia Country Department II, Report #7007-IND. World Bank (Indonesia-1988b) Indonesia. Higher Education Development Project. Washington D.C.: Asia Country Department II, Staff Appraisal Report #7085-IND. - 240 - World Bank (Korea - 1985) Impact of World Bank Lending for Educational Development in Korea: A Review. Operations Evaluation Department, Report #5950. World Bank (Malaysia - 1988) Malaysia. Matching Risks and Rewards in a Mixed Econo. Washington D.C.: Country Department II, Report #7208-MA. World Bank (Nepal - 1988a) An Assessment of Education in Neal. Washington D.C.: Asia Country Department I, mimeo. World Bank (Nepal - 1988b) Policies for Improving Growth and Alleviating Poverty. Washington D.C.: Asia Country Department I, Report #7418-NEP. World Bank (Papua New Guinea - 1986) Papua New Guinea. Country Economic Memor , Washington D.C.: East Asia and Pacific Regional Office, Report #5967-PNG. World Bank (Papua New Guinea - 1987) Paua New Guinea. The Costs and Financing of Education. Washington D.C.: East Asia and the Pacific Regional Office, Report #6767-PNG. World Bank (Papua New Guinea - 1988) PaDua New Guinea. Policies and ProsRects for Sustained and Broad-Based Growth. Washington D.C.: Asia Country Department V, Report #7121-PNG. World Bank (Philippines - 1988) The Philippines. Country Economic Memorandum. Asia Country Department II, Report #7438-PH. World Bank (Sri Lanka - 1988) Sri Lanka. A Break with Past: The 1987-90 Program of Economic Reforms and Adjustment. Washington D.C.: Asia Country Department I, Report #7220-CE. World Bank (Sri Lanka - 1986) Sri Lanka. Education and Training Sector Memoranum, Washington D.C.: South Asia Regional Office, Report #5696-CE. World Bank (Thailand - 1985) Completion Report. Thailand Fourth Education Proiect. East Asia and the Pacific Regional Office: Report #1271- TH. World Bank (Thailand - 1986) Thailand. Growth with Stability. A Challenge for the Sixth Plan Period. Washington D.C.: East Asia and the Pacific Regional Office, Report #6036-TH. Zachariah, K.C.; and Vu, My T. (1988) World Population Proiections 1987-88 Edition: Short and Long-Term Estimates. Baltimore: Johns Hopkins Press for the World Bank. ASIA REGION DISCUSSION PAPER SERIES I1 Til uhrDate Orisinag IDP2 The Labor Force Participation of women in the Republic of Korea: Evolution and Policy Issues C. Grootaert may 1987 F. Iqbal IDP15 The Role of Exchange Rate Policy in Sans-Woo Nam May 1988 D. Leipaiger Four East Asian Countries 78841 IDP28 The Small-Scale Enterprise Credit Program (S.S.E.P.) Under the Second and Third Calcutta Urban Development Projects P. Kahnert March 1988 F. Kahnert (CUDP iI and CUDP III) - An Assessment 76376 IDP35 $1 Improving Tax Policy Advice: Lessons and U. Fleisig June 1989 H. Fleisig Unresolved Issues from Asia Experience 76375 IDP36 Direct Taxes and Fiscal Policy Issues: A. Virmani June 1989 a. Fletsig An Illustration for East Asia 76375 IDP37 Commodity Taxation in Selected Countries 2. Shalisi June 1989 S. Fleisig in South East and East Asia 76375 IDP38 Tax Analysis in Developing Country R. Musgrave June 1989 H. Fleisi Settings 76375 IDP39 Indonesia: External Shocks, Policy Sadiq Ahmed June 1989 Kyle Peters Response and Adjustment Performance 74270 IDP42 An Analysis of the Nature of Unemployment W. T. Dickens July 1989 R. Zagha in Sri Lanka and Kevin Lang 78049 IDP44 Assisting Poor Rural Areas Through Friedrich Kahnert August 1989 C. Chamberli Groundwater Irrigation 75465 IDP51 Educational Development in Asia: A Comparative Study Focussing on Cost Jee-Peng Tan October 1989 Joe-Pens Tan and Financing Issues Alain Mingat 73746 1/ Extra copies may be obtained from the Asia Information Service Center. 1 Not released yet.