CPD8502 Pricing Policy in the Social Sectors Cost Recovery for Education and Health in Developing Countries ErnrnanuelJnnenez Published for the World Bank The Johns Hopkins University Press Baltimore and London © The International Bank for Reconstruction and Development / The World Bank 1818 H Street. N.W., Washington, D.C. 20433. U.S.A. All rights reserved Manufactured in the United States of America The Johns Hopkins University Press Baltimore, Maryland 21211 First printing February 1987 The findings, interpretations, and conclusions expressed in this study are the results of research supported by the World Bank, but they are entirely those of the author and should not be attributed in any manner to the World Bank. to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. Library of Congress Cataloging-in-Publication Data Jimenez. Emmanuel, 1952­ Pricing policy in the social sectors. Bibliography: p. 1. Social service-Developing countries-Finance. 2. Price policy-Developing countries. 3. Education­ Economic aspects-Developing countries. 4. Public health-Economic aspects-Developing countries. I. World Bank. II. Title. HV525.J56 1987 338.4'33621'091724 86-27590 ISBN 0-8018-3501-1 Contents Acknowledgments v 1. Efficiency, Equity, and Cost Recovery: A Summing Up 1 Efficiency and Equity in Current Pricing Policy 1 Pricing for Greater Cost Recovery 4 The Issues 2. Current Pricing Policy 11 Prices and Cost Recovery 11 Prices and the Allocation of Services 22 Justifications for Current Pricing Policy 23 3. The Financial Crisis and Underinvestment in the Social Sectors 27 Social Objectives and Fiscal Constraints 27 Alleviating Underinvestment 35 4. In1/iciencies in the Social Sectors 39 The Mix of Services 39 The Mix of Inputs 43 The Access to Services 49 5. Inequities in the Provision of Services 52 The Distribution of Subsidies 52 Reasons for Inequities 57 Changing Pricing Policy 6. Basic Principles Revisited 65 Pricing Policy for Efficient Provision 65 Pricing Policy for Efficient and Equitable Access 71 Technical Note 74 iv CONTENTS 7. Changing Pridng Policy in Education 77 Behavioral Parameters 77 Prices and Expansion 83 Prices and Efficiency within Schools 92 Prices and Equity 101 Private Schools 102 8. Changing Pridng Policy in Health 106 Behavioral Parameters 108 Prices and Expansion 114 Prices and Efficiency 118 Prices and Equity 121 9. Feasibility of Policy Change 124 Appendixes A. A Review of Alternative Pridng Polides 131 General Pricing Rules 131 Basic Needs 133 Distorted Markets 134 Pure Public Goods 136 Economies of Scale 136 B. Pridng Polides under Budgetary Restraints 139 Fixed Subsidy Allocations 139 Fixed Prices 140 C. The Algebra of User Fees 142 Impact of Changes in Subsidy Allocations on Fees 142 Impact of a Fee Increase under a Fixed Subsidy Allocation with Excess Demand 146 D. Tables 147 References 155 Index 164 Acknowledgments A PRELIMINARY DRAFT ofthis monograph was completed in 1984 while I was a consultant to the Resource Mobilization and Public Management Division in the Country Policy Department of the World Bank. The initial work was commissioned and supervised by Lyn Squire and Nancy Birdsall. I benefited significantly from their general direction. as well as from comments by other staff and consultants. including Martha Ainsworth, Trent Bertrand, and Oey A. Meesook. Punam Chuhan and Rama Seth provided research assistance. In mid-1984 the paper was presented at a conference sponsored by the Country Policy Department on "User Charges and Cost Recovery in the Social Sectors." The manuscript was revised and updated in 1985 after I joined the Bank's Education and Training Department as an economist. The revision benefited substantially from parallel and collaborative policy work I did with George Psacharopoulos and Jee-Peng Tan on the financing of education. My dis­ cussions with them are reflected in many parts of the text, particularly in chapter 7. Any remaining omissions are my responsibility. I would like to acknowledge the considerable editorial assistance of Bar­ bara de Boinville and Richard Kollodge. Many typists labored professionally on the many drafts, including Teresa Hawkins, Setsuko Oiyama, Peggy Pen­ der, and Althea Skeete. v 1 Efficiency, Equity, and Cost Recovery: A Summing Up PRICES HAVE GENERALLY PLAYED a minor role in generating resources for educational and health services in developing countries and have not been used to determine who should have access to them. Instead, the public sector, which is the most important provider of most of these services, has traditionally relied on general revenues for financing. These revenues, in turn, are financed through direct or indirect taxes, inflation, or budgetary deficits supported by domestic and foreign borrowing. Tightened budgets have forced many developing countries to make pain­ ful choices regarding the financing and provision of their major social ser­ vices. Competition for scarce government resources is becoming more in­ tense. Unless they find alternative sources of funding, policymakers must either scale down their expectations regarding how much their economies can devote to these services or dilute quality. In either case, they run the risk of sacrificing long-term development to meet short-term exigencies. This book reviews recent research-much of it supported by the World Bank-on the desirability of increasing the role of prices in financing and allocating educational and health services, usually the largest components of the social sector budget. The book describes current pricing policy and its ef­ fect on efficiency and equity. It then shows how efficiency and equity can be improved by adjusting cost-recovery policies. Efficiency and Equity in Current Pricing Policy Most prices for publicly provided educational and health services are very low or nonexistent. Even if prices are defined to include any charges levied on users of a service, the proportion of a government's cost that is recovered through pricing revenue remains small. Although data are scarce and some 1 2 EFFICIENCY, EQUITY, AND COST RECOVERY may not be strictly comparable, chapter 2 examines this problem and pro­ vides a rough indication of pricing trends. In twenty-eight developing countries, the public cost recovered through prices for higher education is 9 percent; secondary education, 15 percent; and primary education, 5 percent. In health, the recovered cost is 7 percent. These figures must be revised upward somewhat if the percentage of social cost borne by users is calculated. A proportion of the cost may be borne directly by users and not reimbursed by public providers-such as the forgone income of years of schooling or the transport cost of traveling to a clinic. However, even if these figures are taken into account, the proportion of social cost recovered from users is relatively small. Low or nonexistent prices tend to apply across the board for different types of services, as well as for different types ofindividuals. For example, in many countries free education from primary school to university is guaran­ teed for those who are able to obtain access, regardless of their income. The same can be said for many types of health services. Traditional economic justifications for heavily subsidized educational and health services are numerous: benefits flow not only to individuals but also to society at large (externalities); small private providers operate inefficiently (scale economies); fmancial and labor markets are distorted; and the costs of collecting and processing fee revenue are high. These reasons, while valid for some social services, are not valid for all. Moreover, they do not justify the most popular pricing policy-free educational and health facilities for all users. Blanket policies have led to inefficiency and inequity. Efficiency Contrary to intentions, current pricing policies have made the provision of educational and health services less efficient. Signs ofthis inefficiency are un­ derinvestment in these sectors relative to others, misallocation of resources within each sector, and an inability to ration services according to need. Chapter 3 documents the financial crisis confronting the social sectors. For many governments, the real resources spent on educational and health ser­ vices have been declining or rising at a rate slower than the population growth rate. Moreover, expenditures by both the public and private sectors of most developing countries are far less than the amounts needed to meet their stated social goals, such as universal primary education and access to primary health care. One estimate of the cost of financing a so-called basic human development package implies budget shortfalls for the average developing country as high as 17 percent of gross national product (GNP) (Meerman 1980). A SUMMING UP 3 These trends are disturbing because of the evidence that there is underin­ vestment in these sectors: high social rates of return are persistently found. Budgetary constraints have particularly affected recurrent expenditures overall, with drastic repercussions for health and education, where these ex­ penditures represent the bulk of spending. At the same time, there is excess demand for many social services and an observed willingness to pay for them. Current fee policies are not flexible enough to use this demand to mobilize financial resources. Current pricing policy has also contributed to inefficiency within the so­ cial services, as documented in chapter 4. This is partially reflected in a dis­ proportionate allocation of resources for services with low social returns. Uniformly low prices for different types of services imply large subsidies for high-cost services without correspondingly high levels of benefits. For ex­ ample, in the poorest countries in sub-Saharan Africa, public expenditures for education have been concentrated in higher education. At that level of schooling, students receive not only free tuition, but also living allowances that far exceed what is required. Allowances are nearly as high as starting wages in the working world. Under tight budgetary restraints, allocations for higher education come at the expense of other educational services. In these countries, primary school enrollment ratios remain low and so does the quality of education at that level. Another manifestation of inefficiency in current pricing policy is the in­ creased unit cost of providing any given educational or health service. A key to efficiency in provision is accountability. Administrators have to be given incentives to minimize costs. These incentives are less obvious in centralized systems where the few inspectors may be undertrained and inadequately directed or in systems where administrators are given lifetime job security. In general, inefficiency is exacerbated if there are fewer links among those who pay, those who provide, and those who use a service. Finally, inefficiency occurs in the way subsidized services are allocated among the beneficiaries. To ensure efficient allocation in education, those who are most able to augment their productive capacity should be able to secure a place. In health services, at least those who have most to gain from being well or who are the sickest should be treated. Free provision of all public services, however, does not mean that everyone, and particularly those who would benefit the most, will enjoy this access. When places are scarce, access must be rationed. Chapter 4 shows that this rationing has not been fully efficient in education. Because of privately in­ curred cost (mainly in the form of forgone income), free provision alone has not resulted in adequate representation of the brightest students from the lower-income groups. In many countries, these students do not even apply for entrance to institutions of higher education. Those who do are often at a 4 EFFICIENCY, EQUITY, AND COST RECOVERY disadvantage in competing for available places with students from higher­ income families. Although they may be more able, they do not invest in tutoring or in extracurricular materials to assist them in passing entrance ex­ aminations. As a result, despite large subsidies lower-income groups are severely underrepresented at higher levels of education, where privately in­ curred costs are greatest. In some countries, higher education, and its atten­ dant subsidies, is almost an exclusive preserve of the upper classes. In the health sector, the most popular quantity-rationing device appears to be the queue. Although systematic evidence is not available, various field reports cited in chapter 4 are replete with examples of overcrowding in free health centers. It is only for inpatient care, where physicians have an oppor­ tunity to assess needs, that this rationing device may be efficient. For out­ patient care, it is more difficult to ensure that those who need the service the most are given priority. Equity Another major justification for a policy of uniformly low prices is equity. To redistribute resources, social services should be financed from progressive taxes, rather than from price revenues. But this objective is rarely met. As chapter 5 indicates, progressive redistribution has not always taken place. Uniformly low prices mean that the most costly services (for example, higher education or health care in urban hospitals) are subsidized the most heavily. For reasons mentioned in the above discussion on access, higher­ income groups tend to consume more of these services. In the case of educa­ tion, high personal costs make it prohibitive for lower-income groups to compete successfully with higher-income groups for the subsidized school places. In the case of health, many of the subsidized services are in urban areas inaccessible to the rural population. The tendency toward inequity is not reversed even when the sources of financing are considered. Although most countries have progressive personal income taxes, these taxes account for a relatively small proportion of collec­ tions. The incidence ofindirect and foreign trade taxes, which are the bulk of revenues, is not progressive, according to evidence cited in chapter 5. Pricing for Greater Cost Recovery Given the inefficiency and inequity of current pricing policy, what is an appropriate response? The rest of this book examines the role of increased prices in recovering a greater proportion of cost in education and health. Chapter 6 reviews basic pricing principles and the tools of applied welfare A SUMMING UP 5 economics. Chapters 7 and 8 discuss the implications for the education and health sectors. The general analytical framework in chapter 6 is standard partial equilib­ rium analysis. Although a general equilibrium approach is beyond the scope of this book, since it would require a relatively complex extension of the ex­ isting literature, general equilibrium interactions are discussed qualitatively wherever possible. A move away from across-the-board zero pricing for all educational and health services is advocated. Pricing policies across all types of publicly pro­ vided services should not be uniform but should differ depending upon the "public goods" characteristics of the social service. If the externalities and scale economies are great or if exclusivity ofuse cannot be enforced, the role for subsidies in financing a public service is increased. Pricing policies should also depend upon initial market conditions: how much of a service is being demanded relative to supply at current prices? For example, if public budgets have fixed subsidy allocations to a given service and ifthere is excess demand for that service, then prices could be raised. The revenue could be used to finance expansion of the service with the highest so­ cial return. If the source of fee revenue is the same service as its destination, prices could be raised until excess demand for that service is eliminated. For services that require relatively large outlays, such as higher education or prolonged hospital stays, efficiency and equity in pricing policy depend upon the development of related markets. For example, education fees could be accompanied by credit schemes, and medical insurance schemes could be introduced. Efficiency and equity are also improved by differentiating prices, not just by type of service, but also by type of individuaL In this way subsidies could be targeted toward low-income groups rather than distributed indis­ criminately. If the costs of implemeting such schemes are too high, prices could be differentiated by groups of individuals, such as rural as opposed to urban residents. Education The applicability of increased pricing in education is reviewed in chapter 7. The case for greater cost recovery through increased user charges is strongest at the level of higher education. Most of the benefits flow directly to the student, and there is also excess demand at the currently low level of prices. Increased prices are thus unlikely to cause any decline in overall en­ rollment for most countries. The disposition of the revenues would depend upon country conditions. Any extra resources generated through user charges should be devoted to the 6 EFFICIENCY, EQUITY, AND COST RECOVERY activity with the highest social returns. In countries that are far from reaching universal primary enrollment, these resources could be used to expand primary education. For example, if all living expenses for higher education in sub-Saharan Af­ rica were to be covered through increased user charges, the primary educa­ tion budget could expand, on average, by 18 percent. Ifoperating costs were also recovered, the primary education budget could be increased by 41 per­ cent (Mingat and Tan 1985a). Although full cost recovery in higher education is not necessarily recommended, these figures provide a rough idea of the magnitude of change that would be possible. In countries that are already close to achieving universal primary educa­ tion, such as those in East Asia or in Latin America, high social returns may be gained by expanding some disciplines in higher education or by improving the quality of most levels of education. If there are no scale economies, a 10 percent increase in fees at higher levels would allow a 1 percent expansion of enroIlment at those levels. With just moderate scale economies, however, such an increase could accommodate an expansion of 2 percent. If fees are raised at higher levels and used to expand primary levels, equity should improve, given the current distribution ofsubsidies. There would be a countereffect, however, if those who drop out at higher levels are from lower-income groups. This effect could be mitigated if higher fees were in­ troduced along with selective subsidies and, eventually, educational credit. Health Chapter 8 reviews recent studies of health services where increased cost recovery through pricing would be appropriate. Although it is more difficult to categorize health services than educational services because of the many different types, one useful distinction is between curative and primary health care. Most patient-related curative services are not characterized by ex­ ternalities. Compared with education, there is less systematically collected informa­ tion about the market characteristics of curative services. Anecdotal evidence suggests that many countries have excess demand for them. In these coun­ tries, a 10 percent increase in fee revenue from curative services could in­ crease the amount of the health services by about 1 percent if all of the revenue were used to finance service expansion. If there is no excess demand, the impact on utilization of a fee increase (to counteract a decline in subsidy allocations, for example) would depend on the elasticity of demand for health. For the average developing country, a rise in unit fees of about 7.2 percent would be required to counteract a 10 percent A SUMMING UP 7 decline in subsidy allocations without creating excess demand. This would result in a drop in utilization of about 2.5 percent. The efficiency and equity gains from user charges would be enhanced in a system where medical insurance was widely available. Otherwise, it would not be feasible to increase user charges for high-cost services to levels near cost recovery. Feasibility The last chapter of the book, chapter 9, assesses the feasibility of policy change. Information about the costs of administering cost-recovery schemes in educational and health services is relatively scarce. This would be a fruitful subject for further research. But the fact that many developing countries are already instituting policies to recover more of their costs suggests that they can be implemented. The Issues 2 Current Pricing Policy As CHAPTER t MADE CLEAR, prices have been little used to finance edu­ cational and health services in developing countries and to determine who gets access to them. Governments have traditionally been the predominant providers of these services, and they have relied on tax revenues to subsidize users. The percentage of provider-borne costs covered by prices is small. In fact, many governments have strictly regulated the types and amounts of fees that private providers can charge users. This chapter examines current pricing policies in education and health in developing countries and the impact of these policies on cost recovery and the allocation of services. Prices and Cost Recovery When the user of a service incurs significant consumption costs, the relationship between prices and cost recovery is complex. In this book, price is defined as any user payments, per unit of a service, to the provider of the service. Cost recovery per unit of a service provided by the government will then be the proportion ofits unit cost that is covered by price (P).1 Unit cost is the sum (per unit) of the recurrent cost (rc), the annualized capital cost (cc), and any direct transfers (t) to the users, such as educational scholarships or health allowances. The net unit cost for the government (or unit subsidy, s) is (2-1) s = rc + cc + t - p. In symbols, a cost-recovery ratio for the government (rg) is defined as 1. Hereafter, "charges," "fees," and "prices" are terms that are used interchangeably. All lower case symbols signify a magnitude per unit. 11 12 THE ISSUES (2-2) rg p / (rc + cc + t). From the social point of view, however, this definition is not complete. Because of consumption costs, users may have to make payments that are in­ dependent of price. These might include opportunity costs (oc) of time while the service is being used and any direct private costs (dpc), such as transport in getting to the service, net ofany transfers paid to the user (t). In symbols, unit private cost (pc) is (2-3) pc = p + oc + dpc - t. Thus, the unit social cost (sc) is the sum of government and private net unit costs: (2-4) sc =s + pc = rc + cc + dpc + oc. A cost-recovery ratio for society (rs) can be defined as the portion of unit so­ cial cost that is recovered from the user: (2-5) rs = pc / sc. In this book, measures of the governmental cost-recovery ratio (2-2) will be derived for selected countries. Because of the difficulty of obtaining the ap­ propriate data, only a rough estimate of the societal cost-recovery ratio (2-5) will be presented. This estimate, however, is still valuable. Governmental cost-recovery policy is not limited to setting p. Cost recovery can also be af­ fected by setting transfers and choosing how much of direct personal costs (such as books, in the case of education) are provided. The preceding definitions of cost recovery do not measure actual collec­ tions. They indicate only the proportion of cost that is intended to be collec­ ted. Thus, even the relatively low cost-recovery ratios cited in this chapter are underestimated somewhat. The extent of the underestimate varies from country to country, according to various reports cited in Ainsworth (1984), and cannot be accurately determined. Cost-recovery estimates are difficult to make in many countries because social services, such as health and education, are provided publicly and privately. Private-provider cost-recovery ratios are higher than governmental cost-recovery ratios, and they frequently approach unity. Thus, an overall picture of cost recovery in the presence of a significant number of private providers must take account of the private providers' higher cost-recovery ratios. CURRENT PRICING POLICY 13 1kble 2-1. PubUc Provision of Education and Health in Selected Regions, 1975-80 ~rcetltage of students Public expenJiillre as public schools, 1975' percetltage of total health expenditure, Region Primary Secondary 1975-8ri' Mrica 62 East 57 55 West 82 70 Asia 87 71 30 Latin America and Caribbean 87 67 51 Europe, Middle East, and North Africa 94 92 57 a. Tan (1985h) and World Bank estimates. h. de Ferranti (1985). Cost Recovery in Public Provision The public sector has been the most important provider ofeducational ser­ vices in developing countries. The most accurate indicator of this is public school enrollment. In 1975, for example, 57 percent to 87 percent of all primary-level students in Latin America, Asia, and Africa were enrolled in public schools. Fifty-five percent to 71 percent were enrolled at the secon­ dary level (see table 2-1). The proportions at the university level were also substantial according to World Bank data. These figures have significant implications for government expenditures, since fees recover very little of the costs in providing education. Table 2-2 summarizes the available evidence on fees as a percentage of costs per user in the education sector (rg from equation 2-2). Fees in education consist primarily of tuition, books and supplies, registration and entrance fees, ex­ amination fees, boarding charges, and charges to parents and community associations for capital improvements. In the thirty-six developing countries for which data were available, 39 percent, 25 percent, and 30 percent charged no fees for public primary, secondary, and higher education, respectively. Of those that did charge fees, the amount as a percentage of publicly incurred unit costs was very low: 7.9 percent for primary education, 14.7 percent for secondary education, and 8.2 percent for higher education. In the nineteen African countries in the sample, fees tended to be lower for higher education than for education at other levels. Whereas 42 percent and 26 percent ofAf­ 14 THE ISSUES Thble 2-2. Cost Recovery in Education: User Fees as Percentage of Unit Cost in Selected Regions, 1980 Percentage of countrit:s User foes lIS percentage of Number unit cost" with IW fees of Region countries Primary Secondary Higher Primary Secondary Higher Mrica 19 42 26 69h 13.4 19.1 8.3 Asia 8 38 13 13 2.4 16.0 12.0 Latin America 9 33 22 OC 2.4 2.4 5.9 All 36 39 25 30d 7.9 14.7 8.2 a. Calculated as an average of all countries with nonzero fees. b. For thirteen countries. c. For twelve countries. d. For thirty-three countries. Source: Adapted from appendix table D-1. rican countries charged no fees at the primary and secondary levels, respec­ tively, 69 percent charged no fees at the higher levels. Moreover, of those countries where fees were charged at higher levels, the proportion of unit costs recovered was lower than that recovered at primary and secondary levels. In addition to the direct governmental expenses that are not recovered through user fees, public educational costs are boosted by direct payments to students for living and other expenses and by transfers to subsidized private schools. (See appendix table D-2 for a breakdown of government expen­ ditures for selected countries.) Approximately 37 percent of public education budgets for higher education and 21 percent for secondary education in 1980 in Africa were reserved for scholarships and welfare payments (Unesco 1984 Statistical Yearbook). Even though user fees as a percentage of unit cost are low, as indicated in table 2-2, they playa key role in financing certain types of expenditure. 2 This is because of the way that educational subsidies are distributed to various schools. In countries where school fees are retained at the local administrative 2. In many countries listed in table 2-2, government expenditures appear to be directed toward financing one major item of educational expenditure-staff salaries. Other operating expenditures are financed out of fees. For example, in Lesotho, one-half of all fee revenue is estimated to be used for textbook expenditures. one-tenth for school feeding, and the rest for maintenance and other operating expenses (Lesotho, Government of 1983). In Swaziland, a study of a sample of government and subsidized schools indicated that the share of salaries in government expenditures is equivalent to the ratio of unit fees to unit costs (Swaziland, Government of 1981). CURRENT PRICING POLICY 15 level or are earmarked and passed on to the provincial or national levels, much ofthe user-fee revenue is used to finance teaching materials (Ainsworth 1984). Country-specific data indicate that users finance a large portion of several forms ofteaching materials. For example, in Lesotho, all textbook ex­ penses at the primary level and 63 percent at the secondary level are financed by user fees (Lesotho 1983). Similar findings hold for Malawi (Thobani 1983) and many other countries. But no systematic studies have evaluated whether the level of these fees has been appropriate to finance the level of expenses. Educational fee revenue is declining for many countries. Zimbabwe, Lesotho, Kenya, Indonesia, and Botswana have abolished primary-level fees during the past decade. In other countries, user fees have not changed to keep up with inflation (Ainsworth 1984). A similar pattern emerges in health services, although the public sector share of total expenditures for health is smaller than it is for education. The figures vary from country to country for definitional reasons, but gross es­ timates ofpublic expenditures as a percentage oftotal health expenditures are about 62 percent in Africa, 51 percent in Latin America and the Caribbean, and 30 percent in Asia (table 2-1). In general, the percentage of public health expenditures recovered by user fees is small. As table 2-3 shows, in fourteen of the seventeen countries for which data are available, the amount is 8 per­ cent or less. For many countries, this amount has declined in the past decade (Ainsworth 1984). Fee policies vary among service levels, which implies that fees may ac­ count for a small portion of the overall budget, but are more important for certain types of services. For example, in Malaysia only 5 percent of total government health costs are collected by user charges, while 8 percent of hospital in-patient services, which account for over half of all health expen­ ditures, are recovered through user fees (Meerman 1979). These unreimbursed expenditures, or those not recovered through user fees, imply that education and health subsidies (direct payments or fees below unit costs) account for a major share of the public budget. For example, developing countries (not including oil-exporting ones) allotted more than 10 percent of their public budgets to education in 1980. In Africa, that number rises to more than 16 percent (see table 2-4). The percentage share of health expenditures in the government budget was about 4 percent for Asia and the Middle East, and 5 percent for Africa. Social Cost Recovery in Public Provision To calculate social cost recovery, as in equation 2-5, estimates of direct private costs in using a service and of opportunity costs are required. These estimates are available for only a few countries and vary across and within 16 THE ISSUES Table 2-3. Percentage of Recurrent Public Health Expenditures Recovered by User Fees in Selected Countries Percentage Country Year recovered Note Botswana 1978 2.8 Burundi 1982 4.0 Ministry of Public Health expenditures only. Colombia 1980 28.4 Percentage of the amount budgeted, both recurrent and capital. Increased from 11.9 percent in 1975. Ghana 1976-77 3.0 Percentage of recurrent expenditures. De­ clined from 4.9 percent in 1966-67. Indonesia 1980-81 15.5 Expenditures at all levels of government. If standard payroll deductions for insurance are included, 17.1 percent. Jordan 1982 13.2 Lesotho 1980-81 6.0 Increased from 11 percent in 1970-71 to 16 percent in 1974-75. Since then, declined. Malawi 1980-81 3.0 Pakistan 1980-81 2.5 Expenditures at all levels of government. Peru 1981 8.0 Percentage of both capital and recurrent expenditures. Philippines 1981 6.8 Ministry of Health expenditures only. De­ clined from 14 percent in 1978. Rwanda 1977 7.0 Total revenue, including payroll deduc­ tions, was 16.5 percent of expenditures. Sri Lanka 1982 0.7 Sudan 1980-81 1.4 Hospital and other fees, central govern­ ment only (which accounts for only one- fourth of total expenditures on health). Togo 1979 6.0 Excludes fees from and budget of Lome University Hospital Center. Tunisia 1982-83 2.0 Zimbabwe 1980-81 2.2 Ministry of Health expenditures only. Ex­ cludes budget of Parirenyatwa Hospital. Declined from 9.5 percent in 1974-75. Sources: Ainsworth (1984) and de Ferranti (1985). CURRENT PRICING POLICY 17 Table 2-4. Public Expenditures on Education and Health. as Percentage of State Budget in Selected Regions. 1980 Regkll1 EduC4tion" Mrica East 14.1 5.3 West 21.5 4.7 Asia East Asia and Pacific 14.0 5.1 South Asia 8.8 3.7 Latin America 16.4 8.2 Europe, Middle East, and North Africa 14.6 4.2 Developing countries 15.5 5.4 a. Unesco, 1984 StatistiC41 Yearbook, eighty-four countries. b. de Ferranti (1985). countries. Nevertheless, averages can be used to approximate overall social cost recovery. In education, there are other expenses aside from direct monetary expen­ ditures on tuition, books, and supplies. Expenditures for transport to school, uniforms, and school-provided meals, in excess ofwhat would normally have been spent on transport, clothing, and food if the child had not gone to school, must be considered. There are also indirect expenses, such as the op­ portunity cost of having a child in school instead of at home working for the family business. Information on the composition of personal education costs incurred by households in developing countries is limited by the paucity of studies. Nevertheless, it is clear from table 2-5 that fees represent only a small portion of total personal costs to the household. Opportunity costs, however, may be substantial at all education levels, especially in rural areas, and at secondary and higher levels in urban areas. Actual earnings of school-age children were used to calculate the oppor­ tunity costs of schooling among the poorest 70 percent of urban households in E1 Salvador. Earnings of children out of school were compared with children of similar ages in school. The results, shown in table 2-6, indicate that the average household's educational opportunity cost, measured as the current expected income of those members in school, is about 4 percent of household income. The opportunity cost is, on average, 43 percent of total educational costs to the household. Opportunity costs are lower for lower­ income households because the number of schoolchildren, particularly at higher levels of schooling, is lower. Table 2-5. Private Costs of Publicly Provided Educational Services in Selected Countries (current u.s. dollars) Average household expenses per sI.!llienl Per azpit4 Avera:e income in.!Ul1lt')' bursary Indirect Opportunity Country Survey yel1T year T~ of service provided Fees coSls' cosls Kenya 1980 420 Rural primary 0 0 9 0 Malawi 1983 210 Standard 8 0 5 34 Secondary 86 126 - Co Malaysia 1974 680 Primary Secondary Postsecondary 1 5 7 2 30 141 51 123 236 Mali 1981 190 Rural primary 0 2 5 0 Tanzania 1981 280 Secondary 0 0 137 1,115b - Not available. a. Includes transport; auxiliary materials such as books, school supplies. and unifonns; and board and lodging if they are available. Some indirect costs are living households might have had to incur had a member of the housebold not gone to school. b. As perceived by the student. Sources: Bertrand and Griffin (1983) for Kenya; Tan, Lee, and Mingat (1984) for Malawi; Meennan (1979) for Malaysia; Birdsall (1983a) for Mali; and Tan (1985 zania. Per capita income from World Bank data. Table 2-6. Opportunity Costs (OC) of Education by Income Quintile in EI Salvador Urban Households (1980 colones a month) Totalrost Average eJucati AC, the answer is ambiguous. at which the marginal will­ ingness to pay is exactly nil. (2) Merit goods lack of information As in (1), if information As in (1). programs are infeasible. lack of income The optimal (first-best) As in (1). policy is to have income transfers at p" = MC. As in (1), if income transfers are infeasible. (3) Other public aspects Nonexclusivity To circumvent free-rider If benefits flow to other than problem, a benefit taxation users, subsidy is needed. scheme may be better than user charges. Decreasing AC p" = MC. Full cost recovery from users is not warranted. unless a two-part pricing scheme is introduced. (4) Market failures in Depends upon distortion in Depends upon the nature and other sectors the other market. In general, extent of distortion in related policy should attempt to market. correct for the distortionary effects of the related market. The flTst-best policy would be to correct the distortion (for example, by setting up an educational credit scheme). Note: See appendix A for a rigorous explanation of the efficient pricing scheme for each case. MC = marginal cost to the provider; AC average cost; p = price; p. efficient fee level. CURRENT PRICING POLlCY 25 generally weigh only the personal benefit against the personal cost; they should be induced to consider the impact of their consumption on others. A marginal cost pricing scheme that results in just enough consumption to equate personal benefit with cost is suboptimal. Second, potential users may be unaware of all the personal benefits of educational and health services. In addition, even those who are aware of them may have insufficient income to consume the minimum amount con­ sidered socially desirable by public authorities, without unacceptable sac­ rifices in the consumption of other basic commodities, such as food, clothing, and shelter. Thus, educational and health services are said to have the charac­ teristics of "merit goods." In other words, public authorities may have more information and resources concerning what is best for users than the users have themselves. Third, some services are provided to the general public because access to them cannot be physically limited to those who are willing to pay. For exam­ ple, insect spraying to control disease in a neighborhood benefits payers and nonpayers alike. Fourth, marginal cost pricing may be inefficient because of distortions in related markets: markets for inputs (such as teachers, when their salaries are inefficiently subsidized); markets (such as the labor market) for other services for which the output (graduating students) of the social sector (education) would be considered as an input; and markets for financial services (such as credit for education and insurance for health). Access to credit and insurance markets, then, has important implications for efficiency. For example, without access to financing, a brilliant child from a deprived background can­ not invest in higher education, even though the future returns may be very high. Thus, in the absence of a credit market, the social benefit of a unit of higher education may exceed the private benefit. For health, the relevant related market is insurance. Since many households cannot set aside a large amount of money to be used in the event of major illness, they may want to purchase medical insurance. But the lack of an insurance market prevents the efficient sharing of the risk of paying for major illnesses. Pricing subsidies are one way of circumventing these problems. If the lack of credit and insurance markets is viewed as a divergence between private and social benefits, the analysis for services with externalities would apply. The labor market is also likely to be inefficient. The returns on in­ vestments in human capital depend upon the discounted value of future wage payments. If those wages are offered at a socially inefficient level, then marginal cost pricing in education would probably be inefficient. For exam­ ple, if the main employer of university graduates (such as the civil service) of­ fers a wage that is higher than the cost to society of attracting the labor from other productive work, more graduates would be tempted to go to school to 26 THE ISSUES secure those jobs than would be socially efficient. 4 (In technical terms, the marginal social benefit of a unit of education is less than the marginal private benefit.) These negative externalities imply that a price greater than marginal cost should be imposed. Negative externalities tend to offset the positive ex­ ternalities discussed earlier and to reduce the need for price subsidies in education. Equity Much of public intervention in the health and education sectors has been justified on equity grounds. Some argue that public expenditures on social services can be used as a method of income redistribution, particularly if the services are financed by progressively collected revenues. Others claim that health and education should be considered "basic human needs," and therefore the public sector should guarantee access to some minimum threshold amount. Challenging Traditional Justifications The efficiency and equity arguments outlined above and summarized in table 2-8 imply a departure from marginal cost pricing. Appendix A presents the reasoning behind each recommended pricing scheme and gives specific examples. These arguments have traditionally been used to justify heavy public subsidization. The last column of table 2-8 presents the conditions under which efficient pricing schemes imply subsidization. As long as the average cost of providing a service is constant with respect to scale, marginal cost pricing is equivalent to cost-recovery pricing. Thus, prices less than marginal cost imply the provision of subsidies. If there are scale economies, even marginal cost pricing implies the provision of subsidies. In recent years, however, many developing countries have been forced to tighten their budgets, and other conditions have changed substantially since their education- and health-financing policies were first established. Thus, it is time to investigate the record of public investments in meeting goals of equity and efficiency. There is wide room for improvement in both areas. The next three chapters reexamine traditional assumptions behind public in­ tervention in education and health, and their findings lend support for greater private participation. 4. This might be so even if the limited number ofgovernment jobs resulted in unemployed graduates. If the decision to enter school were based on expected wages, there might still be an incentive to overinvest in schooling (Blomqvist 1982). 3 The Financial Crises and Underinvestment in the Social Sectors BECAUSE OF LOW or nonexistent prices, publicly provided educational and health services have had to rely on the budgetary allocations of central governments. The fiscal crises recently experienced by many developing countries have slowed the flow of resources into the social sectors at a time when social objectives have been far from met. Even for those countries that have been able to expand their social services, the gap between available resources and social demands is expected to widen. Social Objectives and Fiscal Constraints Education As table 3-1 shows, the social rates of return or the net returns that societies receive on their educational investments are high for all forms of education, especially primary education. The rates are generally computed by comparing the gain in earnings because of schooling with costs of providing that schooling. 1 The figures show, on average, that these rates are greater than the benchmark return on physical capital. 1. Many rate-of-return analyses in education require empirical refinements because of the unavailability of data. For example, estimates of social rates of return for alternative types of education represent only the monetary benefits of gains in earnings. They account for oppor­ tunity costs plus provider-incurred direct costs. Thus, the social rates of return to primary education are likely to be underestimated because private and social benefits are presumed to be equal, while the opportunity cost is probably overestimated. At alllevcls ofanalysis, tuition payments are ignored. In addition, several methodological problems arise from the use of aggregate data; from inattention to the effects of ability, employment, and the variance of in­ come distribution; and from the underreporting of income. Nevertheless, these figures pro­ vide a generally accepted guide for investment priorities. For a complete discussion of social rates of return on education, see Psacharopoulos and Woodhall (1985). 27 28 THE ISSUES 'Thble 3-1. Social Rates of Return on Investments in Education by Country Group, Region, and Level of Education Level of education Country grOlip ana region Primary SeCtJnJary Higher Developing Africa 28 17 13 Asia 27 15 13 Latin America 26 18 16 Europe and Middle East 13 10 8 Developed - • 11 9 a. - Not available because of the lack of a control group of illitetates. SOllree: Psacharopoulos (1985). Estimates based on sixty countries. Rates of return have been documented in the urban sector, but the rural, agricultural sector also may benefit from social investments in education. For example, annual yields of farmers with four years of primary schooling are, on average, 9 percent higher than those of farmers with no education Oamison and Lau 1982). Education also generates indirect benefits. There is evidence of a com­ plementary relationship between capital and education. Physical capital in­ vestments tend to be more profitable when workers have been educated and are therefore literate and numerate. In other words, educated workers operate production machinery more efficiently. Education provides several other nonmonetary benefits. For example, educated people are more capable of complying with tax collection laws than uneducated people are because they have the skills to fill out tax forms. In­ creased tax compliance means more tax revenue. In addition, a literate pop­ ulation can learn about birth control methods more easily than an illiterate one can-an obvious benefit to countries interested in controlling population growth. Finally, literacy promotes citizenship: literate people can easily learn about laws and social mores (Haveman and Wolfe 1984). Many developing nations, aware of the benefits from educational in­ vestments, have set constitutional goals to achieve universal primary educa­ tion (usually obligatory school attendance for all children between the ages of six or seven and twelve or fourteen), eradicate illiteracy, or provide universal access to all levels of education. Developing and maintaining programs to meet these goals, however, are extremely expensive. Thus, the amounts that governments spend on education do not always meet constitutional obli­ gations. Carnoy and others (1982) call the difference between actual UNOERINVESTMENT IN THE SOCIAL SECTORS 29 educational expenditures and the amount needed to achieve constitutional or stated national goals the "legitimacy gap." Thble 3-2 presents the legitimacy gaps for primary and secondary educa­ tion for six developing countries. These gaps are indicated in two ways: as the actual amount that countries fall short of their goals and as a percentage of the expenditure needed to meet the goals. For example, Brazil spent 33,300 million cruzeiros on education in 1978, but its goal would have required spending 44,000 million. The legitimacy gap, then, is 10,700 million, or 24 percent less than what Brazil should have spent in 1978 if it had met its goals for that year. A negative percentage means that a particular country's educational expenditures exceed the amount necessary to meet national goals. Half of the countries in table 3-2 will have to increase their primary educational expenditures between 1978 and 1988 to meet national goals. Egypt's projected 1988 primary educational expenditures, for example, will be about 32 percent less than what is necessary to meet stated goals. The 1988 figure for Brazil (-78 percent) implies excess educational expenditure, but it is not necessarily correct because projected expenditures were based upon the country's unusually high growth rates during the early 1970s. Brazil's recent growth rate is much smaller than in previous years. From 1970 to 1975, the gross domestic product (GOP) grew at an average annual rate of more than 10 percent, but between 1976 and 1981, that rate fell to about 4 percent (World Bank 1983). Therefore, if Brazil continues to have a falling GOP growth rate, it might actually experience a legitimacy gap even though table 3-2 shows a surplus of primary educational expenditure in 1988. In secondary education, large legitimacy gaps are apparent from the table in every country except Egypt, which plans on spending slightly more on education than is necessary to meet its national goals. All six countries will experience a legitimacy gap in at least one level of education in 1988. Health Although it is more difficult to estimate a summary measure of social returns from health services than from educational services, increased in­ vestments in the health sector are widely believed to contribute to economic development (World Bank 1980b, 30). For example, absenteeism related to illness can reduce the availability of labor. According to a careful study of tuberculosis control in the Republic of Korea, an optimal health program resulting in increased work life and decreased absenteeism would yield a return of $150 for each dollar spent (Feldstein, Piot, and Sunderesan 1973). Table 3-2. Legitimacy Gaps and Actual and Projected Spending in Primary and Secondary Edncation in Selected Countries. 1978 and 1988 (millions of national currency) 1978 1988 Expenditure Legitimacy gt1p Expenditure Legitimacy gt1 tu!eJeJ /0 Project€d tJ;!eded to Level of edualtion Adlull expenditure meet goals Amount" Percent· expenditure meet goals AlIIOII1It' P andcOlIntry (1) (2) (3) (4) (5) (6) (7) Primary Brazil 33,300 44,000 10,700 24 103,000 58,000 -45,000 Egypt 74.2 131.9 57.7 44 120.3 162.8 42.5 India 4,450 7,700 3,250 42 6,500 9,500 3,000 Kenya 1,150 1,300 150 12 2,800 1,850 -950 ..... Q Mexico 25,600 25,900 300 1 46,500 36,500 -10,000 Thailand 6,600 13,300 6,700 50 13,700 17,500 3,800 Secondary Brazil 12,400 47,900 35,500 74 37,800 63,200 25,400 Egypt 175 216 41 19 280 270 -10 India 10,600 23,000 12,400 54 15,500 28,300 12,800 Kenya 315 1,180 865 73 770 1,680 910 Mexico 16,000 22,000 6,000 27 28,900 31,300 2,400 Thailand 2,300 6,500 4,200 65 4,700 8,600 3,900 Note: 1988 currency expressed in 1978 prices. a. Column 2 minus column 1. b. Column 3 divided by column 2, times 100. c. Column 6 minus column 5. d. Column 7 divided by column 6, times 100. UNDERINVESTMENT IN THE SOCIAL SECTORS 31 An unhealthy workforce can result in low productivity. For example, a World Bank study showed that, in Indonesia, the prevalence of hookworm infestation in construction and ruhber plantation workers was 85 percent, and 45 percent ofthe victims suffered from a resulting iron deficiency. Treatment of the anemic workers with elemental iron for sixty days, at a total cost of US$O.13 per laborer, increased productivity by approximately 19 percent. The resultant benefit-cost ratio is 280 to 1 (World Bank 1980b, 31). Health is related not only to productivity but also to education because ill­ ness impairs students' ability to learn (Selowsky and Taylor 1973). The World Health Organization's A1ma-Ata declaration of1978 outlines a "global strategy" for health for all by the year 2000 through primary health care systems (WHO 1981). WHO estimates that the annual per capita cost ofim­ plementing these systems would be an additional US$15 for most developing countries. Since per capita public spending is currently US$2.30, there is an average annual resource gap of US$50 billion for all the developing coun­ tries. Even if developing countries could fund as much as 50 percent of this amount, which would mean a quadrupling of average annual per capita domestic spending, they would have to seek external funding about seven times the present level of international transfers (WHO 1981). The growth of per capita domestic public spending, therefore, is not anywhere near the levels required to meet the goals of the global strategy. The Cumulative Budgetary Impact Even though governments and international organizations have decided that universal education and health care systems are desirable, they have not been able to cope with the financial problems associated with providing them. Table 3-3 relates the combined government costs of basic human develop­ ment packages to the resources that governments can collect. The financing of individual components "will normally range in costs from at least 11 per­ cent to 21 percent of GNP" (Meerman 1980, 122). These figures imply that it is possible for low-income countries to provide comprehensive social service packages, but many will be unable to do so through government finance. The recent worldwide recession has restricted the ability of many governments to mobilize resources to meet the health and educational needs of their growing populations. Per capita GDP growth rates from 1970 to 1980 have fallen in low-income countries (excluding China and India) by more than 50 percent from the previous decade (table 3-4), while population con­ tinues to grow. The gap between income and population growth is widening, which makes it more difficult for governments to meet society's health and educational goals. 32 THE ISSUES Table 3-3. Hypothetical Government C~sts to Provide Human Development Services in Low-Income Countries Percentage Human development services and rosls of GNP Adequate nourishment (to cover average food deficit in caloric requirements) 2-4 Universal primary education 3-5 Health care, hygiene, family planning (Malaysian standard) 1-2 Pure water and sanitation (universal water supply) 1-2 Subtotal 7-13 Associated investment 1-3 Adjustment for underfunding of recurrent costs 1-2 Subtotal 9-18 Incremental costs of covering the most expensive 85th to 95th percentile of the population 2-3 Total government costs 11-21 Total resources mobilized 10-20 Resources available for human development services 4-14 Shortfall (costs less resources available) 0-17' a. Limiting cases of high resource generation and low resource requirements. Source: Meerrnan (1980), pp. 124-29. In addition, high administrative costs prevent many developing countries from reaching their "taxation potential" -the maximum proportion of national income that can be diverted for public services by means of taxation (Meier 1976, 271). Because inflation taxes do not need to be administered, they might be an easy alternative to mobilize resources for the public sectors, but there still may be severe adjustment costs as well as regressive incidence. Thus, many countries are left with the option ofdiverting increased resources to education and health at the expense of other sectors. There is no evidence, however, that the average share ofthe public budget devoted to education and health has been increasing at an adequate pace. In fact, for many countries this share has been declining. Table 3-5, based upon data from the International Monetary Fund and other sources, indicates the extent to which the education and health share of the government pie has been shrinking for some developing countries. Although the share of educa­ tion and health in total government expenditures has risen or remained stable UNDERINVESTMENT IN THE SOCIAL SECTORS 33 'Thble 3-4. Growth Rates of GDP per Capita by Country Group GDP per capita growth rates Number of Population Country group countries (millions) 1960-70 1970-80 Low-income 33 2,161 2.3 2.5 Excluding China and India 511 2.0 0.9 China and India 1,650 2.4 3.0 Middle-income 63 1,139 3.4 3.2 Net oil exporters 497 3.7 2.9 Net oil importers 642 3.4 3.3 Industrial market economies 19 714 4.2 2.4 Source: World Bank (1983), calculated from table 2, p. 150, and table 17, p. 180. for the world, especially for industrialized countries, it has fallen, on average, for the non-ail-exporting developing countries since 1973. The downward trend in education, which was more pronounced between 1977 and 1980 because of the declining share of education in Latin American countries, is different from that in health, which seems to have been most affected be­ tween 1973 and 1975 by the adjustment to oil price shocks. These trends differ somewhat across regions. The steadily declining share of health expenditures is most evident for Africa. Although the share of education expenditures in Africa rebounded some between 1977 and 1980, it has yet to reach 1973 levels. A recent African strategy paper by the World Bank attributes perfonnance in both sectors to the "crisis management of re­ cent years [that] has resulted in widespread neglect of programs dealing with the long-tenn constraints on development. In an environment of overall financial tightness, intersectoral competition for resources has made large so­ cial sectors the inevitable victims of budget cuts" (World Bank 1984b, 6). These figures do not necessarily mean that the real public resources avail­ able for education and health have declined only during those periods when total public resources have declined. Frequently, educational and health ex­ penditures suffer in relative terms, especially when total government spend­ ing increases or remains unchanged. For nineteen countries that experienced an average drop of 11 percent in real government expenditures, the average decrease in social sector spending was only 4 percent (Hicks and Kubisch 34 THE ISSUES 1984). Although this study used a broader definition of social sectors than just publicly provided education and health, data for those two sectors alone ap­ pear to corroborate its qualitative findings. Table 3-6 lists the average growth rates of real per capita government ex­ penditures for education and health. It shows no evidence that the social sec­ tors have been more adversely affected than others when total government expenditures have been reduced. Between 1972 and 1975, when per capita total government expenditures fell an average of 30 percent for seven coun­ tries, average per capita health expenditures fell by approximately the same amount, and educational expenditures fell by only 3 percent. The same trend can be found between 1975 and 1979 for the twelve countries with declines in per capita total expenditures of 45 percent. Health declined propor­ tionately, while education declined less than proportionately. These tentative findings indicate that although social sectors have been "squeezed" in favor ofother government expenditures over the past decade, this has not occurred when governments have had to decrease total per capita spending. Rather, so­ cial sector spending has grown less rapidly during times of government expanslOn. For example, more than half of the twenty-eight countries in table 3-6 ex­ perienced declines in per capita expenditures in education or health (or both) between 1972 and 1975, while eight experienced declines in total govern­ 1kble 3-5. Expenditures 011 Education aDd Health as Percentage of 1btal Government Expenditures by Co_try Group aDd RegiOll EJucation Health Gnlntry group tmJ~ 197; 1975 1977 1980 197; 1975 1977 1980 World 6.31 6.81 6.67 6.39 9.07 9.98 10.16 10.36' Industrial 5.08 5.70 5.54 5.20' 9.94 11.28 11.47 11.73 Oil exporting 11.60 11.49 10.90 13.07' 5.20 4.20 4.07 4.72" Non-oil developing 11.76 11.42 11.45 10.47 5.45 4.49 4.86 4.73' Africa 18.62 15.71 15.68 16.76 6.12 5.35 5.46 5.09 Asia 9.31b 8.45 8.90 8.38 2.89b 3.26 3.27 2.82 Middle East 8.38'> 7.41 8.17 8.72" 2.95b 2.96 3.33 2.73 Western Hemisphere 12.64 12.28 12.21 11.50' 6.03 5.41 6.15 6.10' a. For 1979. h. For 1974. Sqwra: Intematiorual Monetary Fund (1981, 1982). UNDERINVESTMENT IN THE SOCIAL SECTORS 35 ment expenditures between 1972 and 1979. During the latter part of this period (1975-79), sixteen of the twenty-eight experienced real declines in per capita spending in education or health (a fourth experienced declines in both), while twelve experienced declines in total government expenditures. These trends indicate that many countries are (and will continue to be) un­ able to allocate sufficient funds to meet their stated health and educational goals. According to the World Bank's 1984 World Development Report, the resource requirements to meet these goals are expected to increase rapidly because of unprecedented high population growth rates. To meet their goals in the face of troublesome financial constraints, governments often make un­ fortunate tradeoffs between quantity and quality that lead to a severe mis­ allocation of very limited resources. Alleviating Underinvestment Because health and education in many developing countries are considered responsibilities of the state, users are often charged minimal fees or none at all. And since users' fees are limited, so is cost recovery. This means that the health and education sectors can turn only to the central government for help. But, as demonstrated earlier, governments' abilities (or willingness) to allocate more funds to these sectors is limited, even though demand continues to grow. The result has been underinvestment in both of these subsectors. At the same time, the private sector has often been prohibited by law from operating in health and education, so it cannot compensate for inadequacies that arise in the purely public systems. Ironically, laws intended to shield society from the underprovision of a completely private market have pre­ vented the sectors from responding to demand. Simple tools of economic analysis illustrate the crucial role of pricing policy in alleviating underinvestment. Although this technical section can be skipped without losing the flow of the later analysis, it does show how economic analysis has been applied to pricing in the social sectors. In figure 3-1, the private demand for a service is depicted as Dp and the so­ cial demand as D,. These are not equivalent for reasons outlined in the pre­ vious chapter: externalities, failures in related markets, and concerns about equity. The total social marginal cost of providing the service is assumed to be constant at c. This unit cost is the sum of directly incurred private costs, cPI exclusive of prices, and the unit costs borne by the government provider, ccp' The socially optimal amount of the social service that should be provided and consumed is that amount at which the additional gain to society from another unit of consumption is equal to the additional cost. In figure 3-1, this i i I , f , 1: l Table 3-6. Average Growth Rates of Real Per Capita Govemment Expenditures for Education and Health by Income Group and Country I t "'" 0\ Income group and country Lower income Total 1972-75 EtiuCdtWn Health Total 1975-79 EduCdtion Health Total 1972-79 EduCdtion Healt Malawi 1 -41 3 29 29 38 39 -24 33 Nepal -2 4 22 24 48 42 39 108 52 Tanzania 54 11 60 -18 1 -83 66 24 Lower-middle income Bolivia 23 -27 -7 16 49 10 20 10 10 Costa Rica -12 -20 5 50 83 736 32 46 778 EI Salvador 14 27 -14 14 -6 22 31 20 4 Guatemala 2 -15 17 34 4 19 37 -12 10 Honduras 4 -1 31 43 20 -10 49 19 18 Kenya 0 1 1 34 0 21 35 10 23 Korea, Rep. of -10 -10 -23 55 29 64 40 16 26 Morocco 86 45 4 17 38 -1 118 100 39 Paraguay -3 7 -23 23 16 64 19 24 25 Peru 5 -6 -14 -86 91 -83 -85 -91 -86 Philippines 17 -13 50 -12 14 4 3 -1 5 Syria 99 -42 15 -25 2 32 49 -42 5 Thailand 0 -1 43 47 78 44 48 7 Tunisia 75 22 49 32 10 37 130 34 10 Turkey 8 36 -2 -36 -49 -39 -31 -30 -4 Zambia 39 2 9 -52 -49 -43 -33 -49 -3 Upper-middle income Argentina -72 -67 -77 -99 -99 -98 -99 -10 Brazil -5 -4 -4 -65 -72 -59 -65 -73 -6 Chile 1 -1 -1 -80 -85 -99 -99 -1 Iran, Islamic Rep. of 85 39 63 17 21 -12 17 68 4 Malaysia 32 28 33 54 13 10 54 44 4 ...., Mexico 35 49 13 -6 -28 -36 -6 6 -2 '-l Uruguay -76 71 -42 -92 -74 -63 -92 -92 -8 Venezuela 26 18 -2 21 4 -9 21 23 -1 Yugoslavia 16 -44 16 -1 -44 -9 - Not available. Source: International Monetary Fund (1982). 38 THE ISSUES Figure 3-1. Pricing a Social Service p ~~--------~~-------P~--------~~- is q*, which is the amount of the service where the marginal gain, measured by the social demand curve D,. is equal to the social marginal cost, e. Although for some countries q* may be equivalent to declared social objec­ tives, such as universal primary education or universal access to primary health care, these objectives may imply magnitudes beyond q*. They are fre­ quendy formulated without regard to e-the cost to society of providing them. Suppose that prices are nil. (According to the previous chapter, this is true for many public providers in developing countries; almost all charge prices close to nil.) To ensure that the socially optimal amount ofthe service is pro­ vided, the public sector must offer a unit subsidy (51) equal to ccr At this level of subsidy, it must also restrict access to q*, and there will be excess demand q*q1' The total subsidy allocation (S) required to finance optimal consump­ tion would equal the area eFEep' which is S1 = (51)(q*). The problem arises because governments allocate subsidies of an amount So less than S1- Suppose this subsidy is equal to eBGep less than eFEe,. Given 510 the total quantity that can be provided is qo less than q* , and there will be so­ ciallosses of an amount ABF. 4 Inefficiencies in the Social Sectors CURRENT PRICING POLICIES have also contributed to inefficiency within each of the education and health sectors, as well as to underinvestment in relation to other sectors. In particular, prices are low, which implies that ser­ vices are provided with heavy public subsidies. Moreover, prices are uni­ formly low regardless of the type ofservice or consumers' willingness to pay. As a result, governments have tended to allocate less to services with high rates of return, a lack of accountability has decreased internal efficiency, and spending has not been successfully targeted to those who could benefit most from the services. The Mix of Services Within the education and health sectors, various types of services are pro­ vided. Their different characteristics are important factors to consider in resource allocation. These characteristics depend upon whether the services exhibit externalities, whether users are knowledgeable about their benefits, and whether markets that affect them are distorted (see chapter 2). Using these criteria, analysts have grouped educational services into primary, secon­ dary, and tertiary levels, and health into preventive and curative services. According to the efficiency criterion for investing in social services, the most resources should flow to those services with the highest social rates of return. Although estimates of social rates of return are, at best, imprecise, es­ pecially for education and health, some quantitative and qualitative evidence suggests that the current pricing structure in many countries contributes to an inefficient mix ofsocial services. Educational and health services with low so­ cial rates of return are generally more attractive to consumers and are pro­ vided more readily than are services with high rates of return. 39 40 THE ISSUES Education The pattern of subsidies leads to a great stimulation of demand for higher education. Households make judgments about sending children to school on the basis of the private rate of return, which differs from the social rate of return because of externalities on the benefit side or because of subsidies on the cost side. If households faced the true social costs of obtaining more education, they would confront a lower rate of return; instead, they are in­ duced to obtain more schooling. Table 4-1 reveals that for Africa the difference between private and social rates of return is greater for higher education relative to primary education. Much of this difference reflects direct payments to students for living allowances, which amount to 35 percent (East Africa) and 67 percent (West Africa) of the recurrent budget for higher education (World Bank 1986). Yet, the social rates of return on investment in education decline as the level of education for all country groups increases (table 3-1). According to those rates, it would be most efficient to give priority to investments in primary education, particularly in countries in sub-Saharan Africa. For any level of education, the social rates of return reported in table 3-1 were based on the difference between average earnings of graduates at that level and earnings at the previous level. These earnings were not adjusted for the possible effects of externalities or distortions in related markets, which may differ by schooling level. If these factors influence earnings, differences between social and private rates may be justified on the basis of economic ef­ ficiency. But even if externalities and labor market distortions are con­ sidered, the larger difference between private and social rates cannot be fully Table 4-1. Ratio of Private to Social Rates of Return in Education Level of education Region and country group Primary Secondary Higher Mrica 1.42 1.34 1.77 Asia 1.57 1.10 1.33 Latin America 1.28 1.15 1.17 Average developing country 1.32 1.19 1.73 Intennediate' 1.38 1.15 1.55 b Advanced 1.55 1.32 a. Cyprus, Greece, the Islamic Republic of Iran, Israel, Spain, Turkey, and Yugoslavia. b. - Not available because of the lack of a control group with no education. Source: Psacharopoulos (1985), tables 1 and 2. INEFFICIENCIES IN THE SOCIAL SECTORS 41 explained. Positive externalities, such as good citizenship and the benefits to social transactions from a literate population, are apparent at primary levels of education but would likely diminish at higher levels. In contrast, the benefits ofa greater amount ofhigher education are realized primarily by the individuals who consume the education and who are paid according to the value of their additional contribution to production. Therefore, externality arguments for subsidization of higher education are weak. In many countries, reallocation of resources from higher education to ex­ pand primary and basic education would lead to gains in efficiency. In fact, estimates indicate that ifpublic resources were reallocated to equalize the so­ cial rates ofreturn at all educational levels, the efficiency gains would amount to 2.6 percent of GDP in certain African countries and more than 3 percent in some Latin American countries (Dougherty and Psacharopoulos 1977). These gains are comparable to a doubling of the public budget for education. Although this analysis does not advocate an immediate and drastic reduction in resources allocated to secondary and higher education, it does show that the present pattern of subsidies favoring secondary and higher education at the expense of primary and basic education is inefficient. In other words, current subsidies are not distributed in relation to expected social returns. There is a need, therefore, to increase the relative proportion of public resources devoted to primary education over other levels of education. Health In the health sector, it is important to distinguish between curative and preventive services. Curative health care involves the treatment of the sick. Preventive health care might include all services that ensure good health, such as immunizations and nutritional advice. Some preventive health services have attributes that inhibit the role of prices in promoting efficiency: high levels of consumption externalities (im­ munization and disease control programs), unidentifiable individual bene­ ficiaries, and a lack of information among users regarding benefits. As de Ferranti (1985) points out, it is probably impossible to implement national user fees for many types of preventive health services. For example, when a government agency sprays an area to combat insect infestation, it is imposs­ ible to identify who benefits and who is to pay for this service. User fees may be appropriate for certain kinds of preventive and curative health care, such as in- and outpatient treatments, drug sales, and water sup­ ply; in these cases positive externalities are negligible since most benefits ac­ crue to the individual. This is especially true for curative treatment of nonin­ 42 THE ISSUES fectious illnesses. Society as a whole obtains no additional benefits from the curative treatment of ailments such as stomach disorders. And when patients return to work after this type of illness, there would be no additional social benefits if they were paid the competitive wage-the value of their marginal product (de Ferranti 1982; Blomqvist 1979). Yet many countries charge uniformly low unit prices or give uniformly high subsidies regardless of the type of service. The greatest proportion of government health expenditures in developing countries goes for curative care, probably owing "in part to the professional bias of physicians and the mystique and popular appeal of hospital-based health care" (World Bank 1980b, 40). In Senegal and Brazil, for example, 72 percent and 85 percent of the health budgets of the central governments are devoted to curative care (de Ferranti 1983, 68-69). These figures indicate that governments may be investing the majority of their health care resources in services that promise only negligible positive externalities. This does not automatically mean, however, that all curative care should be abandoned or charged at full cost. A comprehensive curative health care system can be very expensive for both consumers and providers. Consumers must incur private costs, such as transportation and opportunity costs, even when curative health facilities charge no fees. Also, curative services, such as hospitals, require huge initial investments, even though they are likely to experience decreasing costs­ their average costs would decline as the quantity of services provided in­ creased. Thus, within curative health care there appears to be little effort to differentiate prices according to the private costs of services and the in­ dividuals who must incur them. In particular, the rates charged for services in rural and urban areas tend to be uniform. A Graphic Presentation The differential welfare effect of a uniform pricing scheme on any two types of services can be illustrated with standard economic techniques. For this example, assume that prices are nil for all types of service. As before, Dp and Ds depict the respective private and social demand curves; c and cp are constant total marginal costs and private costs, respectively. In figure 4-1, zero prices result in an inefficiently low use of the service, since the presence of private costs inhibits demand too much. Thus, qo is less than q*. One can argue that this figure depicts the case for primary education or primary health care, particularly in rural areas. Transport and, in the case of primary education, opportunity costs would be the principal components of private costs. The shaded area represents the social loss. INEFFICIENCIES IN THE SOCIAL SECTORS 43 Figure 4-1. Underutilization of a Free Social Service p 7~----------~------~~~------------ q In figure 4-2, zero prices result in overutilization of the service but for dif­ ferent reasons. Some services, such as secondary education, may have sub­ stantial externalities, but private costs are a small proportion of total costs because of scholarships. In figure 4-2A, this results in overutilization of ql greater than q*. Other services, such as curative outpatient care or university education, may imply high private costs but with little externalities. In figure 4-2B, this results in overutilization of q2 greater than q*. In both cases, the shaded area depicts the service losses if the service is not rationed. The Mix of Inputs A high level of unit subsidy has contributed to inefficiencies in the ways that various educational and health services are provided. Reliance on sub­ sidies from central governments has led to inappropriate input mixes. In par­ ticular, this method of financing has caused underutilization of variable fac­ tors of production, such as nonlabor inputs, relative to other factors. There is a growing body of evidence that recurrent expenditures are being crowded out, even in cases where total health or educational expenditures are not being eroded. In addition. resources tend to be misallocated among variable factors, with labor inputs being favored over material inputs. 44 THE ISSUES Figure 4-2. Overutilization of a Free Social Service A. Low Private Costs p ~r------------------+----~~--~-- q B. Low Level of Externalities p ~~--------------------~-ir-~---- INEFFICIENCIES IN THE SOCIAL SECTORS 45 Frequendy, the root of the problem is a centralized and rigid system of allocating expenditures. When the bulk of the funding comes from subsidies, field administrators are not accountable to the families who use the services they dispense. Rather, they are accountable to the funding source, which relies necessarily on preset formulas to allocate inputs. Then formulas often are inflexible to meet local needs. Recurrent Costs It is difficult for the analyst to determine a priori whether the mix of capi­ tal and recurrent spending is optimal. Depending upon technology and prices, it may be efficient not to operate and maintain existing capital stock at capacity. Recent studies of the so-called recurrent cost problem suggest, however, that at present there is a misallocation. A program should not be established unless the resources that are required to set it up and run it year after year are considered to have higher returns in that program than elsewhere in the economy. Ifgovernments are incapable of drawing, to the extent that it is economically worthwhile, all the resources they require to run and maintain the program, a recurrent cost problem exists (Heller 1979; USAID 1982). There are many reasons why developing countries experience considerable difficulty in mobilizing the resources they need to finance sound economic projects. Table 4-2 summarizes the chief causes that are considered most relevant for certain countries. Most of the reasons are related to underfunding. Heller (1979, 40) notes that "the problem pardy derives from the administrative structure common to the financial and sec­ toral ministries of developing countries which separates the investment and current budgeting functions." This problem is inherent in centrally con­ trolled systems. It is difficult to plan, invest in, and administer programs in large systems when all financing comes from a common pot. Unforeseen budgetary crises create short-term expediencies that can be addressed by reducing the size of the pot. The relatively high recurrent cost content in the social sectors makes them more vulnerable to budgetary restrictions than other sectors are. The amount of recurrent costs that need to be incurred to operate a unit of capital expen­ diture is higher for health and education than for other publicly provided ser­ vices. Table 4-3 provides a comparison of the ratios of recurrent to capital costs. Some recent aggregate studies of recurrent expenditures corroborate these conclusions. Between 1965 and 1973, although there was no serial decline in the overall share of current government expenditure in total government ex­ penditure on average for a group of developing countries, "the strongest 46 THE ISSUES Thble 4-2. Causes of Recurrent Cost Problems in SpeciSc Countries Causes Countries Inadequate revenues Inadequate resource mobilization Haiti. Malawi. Pakistan, Zaire Sudden change in economic situation of country Liberia. Mexico, Niger. Peru. Rwanda. Tanzania Foreign exchange shortages Burma. Liberia. Panama Effects of poorly set producer prices Mauritania, sahel region Project-specific problems Longer than expected gestation; poorly specified project technologies Sahel region Excessive investment activities in past Domestically induced Guyana. Honduras. Madagascar. Malawi. Philippines, Sri Lanka, Tanzania, Thailand, Zaire Externally induced Haiti, Madagascar, People's Republic of the Congo, Philippines. Rwanda. Sahel region. Sri Lanka, Thailand. Togo, Zaire Excessive growth of other nondevelopment expenditure Madagascar. Mauritania, Niger. Pakistan, 'Thnzania, Zambia Pressure to generate budgetary savings for investment Honduras. Sri Lanka, Thailand Budgeting difficulties Weak budget planning and forecasting; Cote d'ivoire, Honduras, Madagascar, poor coordination of finance and plan­ Malawi, Nepal, Pakistan. People's ning ministries; lack of information; Republic of the Congo, Philippines, dichotomy of recurrent and develop­ Rwanda, Sri Lanka. Tanzania, Togo, ment budget; management difficulties Zaire associated with multiple externally financed projects Earmarking Costa Rica, Haiti Note: The fact a cause was not listed by no means denies its relevance to a particular country. Source: Heller and Aghevli (1985). Reprinted with permission. evidence of a secular decline was found in the health sector" (Lim 1983a. 378). A regression ofthe share ofrecurrent health expenditure of total health expenditure with respect to time yielded a negative and significant coeffi­ cient (at 5 percent) for 40 percent of the countries. More than 60 percent of INEFFICIENCIES IN THE SOCIAL SECTORS 47 Thble 4-3. Recurrent Expenditures as a Proportion of Investment Ratio of recurrent to Sector capital costs Agriculture Fisheries 0.08 Forestry 0.04 General agriculture 0.10 Livestock 0.14 Rural development 0.08-0.43 Veterinary services 0.07 Buildings om Education' Agricultural colleges 0.17 Polytechnic schools 0.17 Primary schools 0.06-7.00 Secondary schools 0.08-7.20 Universities 0.02-0.22 Health District hospitals 0.11-0.30 General hospitals 0.18 Medical auxiliary training school 0.14 Nurses college 0.20 Nutrition rehabilitation unit 0.34 Rural health centers 0.27-0.71 Urban health centers 0.17 Roads Feeder 0.06-0.14 Paved 0.03-0.07 Social and rural development 0.04 Tourism 0.05 a. The relatively high figures for education may be due to an underestimate of capital costs, some of which are borne privately, particularly in several Mrican countries. This underestimate is not expected to be severe. Source: Heller (1979), p. 39. Latin American countries exhibited this result. Moreover, recurrent expendi­ ture is more unstable than capital expenditure in the health and education sectors, the reverse of the trend in "hard" sectors, such as agriculture, transport, and communications. In another study, Lim (1983b, 450) speculates that "this might have reflected the unfortunate widespread tendency to treat 48 THE ISSUES recurrent spending as consumption and development spending as invest­ ment." These findings could be expected to be worse in the post-1973 period as a result of generally tighter budgetary restraints. Observable declines in the quality of educational and health services reflect this pressure on the recurrent budgets. The operational effects of the problem have been bad maintenance and underutilization of the capital stock. This is often manifested in undertrained staff, or none at all. A case in point is Kenya's primary education system. In 1974, school fees were abolished for the first four years of primary education, and, in 1979, all other primary school fees were abolished. Enrollment in 1974 was 253 percent more than in 1973; enrollment in 1979 was 163 percent more than in 1978. Since the government could not mobilize the necessary resources to replace this financ­ ing quickly enough, the effects on quality were immediate. In 1974, the per­ centage of untrained teachers increased from 22 to 33 percent. In 1979, student-teacher ratios were allowed to increase (table 4-4). Given resource constraints, this reflects the tradeoff between increased coverage and lower quality. In both sectors, however, the problem is most severe in underspending for nonlabor inputs, such as school materials, maintenance, and pharmaceuticals. This points to another difficulty: when a recurrent cost problem exists, the nonlabor component usually suffers the most. Thble 4-4. Primary Education in Kenya Gross Percentage Percentage of enrol/ment growth of untrained Student-teacher Year rati" teachers teachers ratio 1972 0.66 22 1973 0.69 6 22 32 1974 1.03 39 33 35 1975 1.02 10 36 33 1976 0.99 3 37 32 1977 0.99 1 34 33 1978 0.96 3 31 33 1979 1.14 26 40 Not available. a. Enrollment in primary schools divided by population of primary school age. Source: Bertrand and Griffin (1983). INEFFICIENCIES IN THE SOCIAL SECTORS 49 The Choice of Recurrent Inputs There are several reasons why, in the face of tightening constraints, the pressure to forgo nonlabor recurrent expenditures is the greatest. One has to do with the strong labor unions; teachers in particular are a powerful political force to be reckoned with in many countries. Another simply concerns the way central authorities distribute subsidies. In many instances, staff are ap­ pointed from central ministries and are given field positions. Field super­ visors, when told to cut back, have no leeway to do so through stafflay-offs; they must cut back nonwage expenditures. For example, a recent study concluded that the increasing fiscal pressures on education over the past decade have affected the way the twenty-eight developing countries in the study "resolved the conflict between the budgetary pressures confronting all elements of the public sector and the political pressures for expanding enrollments and the attendant fiscal im­ plications of such expansion" (Heller and Cheasty 1984,23). The public sec­ tor response was to increase enrollments by squeezing recurrent expenditure per student, particularly at the higher levels of education. This was done primarily by allowing a clear deterioration in the share and level of real ex­ penditure per student on nonwage recurrent inputs (table 4-5). In many countries, teachers' wages were allowed to increase more rapidly than were expenditures per student. This serious deficiency is reflected in a lack of books and other school materials. The Access to Services When resources are scarce, it is efficient to give priority to users who will benefit the most from a service. Because of externalities or failures in related markets, it is generally not efficient to rely solely on willingness to pay in determining this priority. But charging nothing or close to nothing also will not lead to efficient allocations. In education, it would be efficient to provide priority access to those who are most able to augment their productive capacity. If there is some com­ plementary relationship between schooling and innate ability to improve productivity, the most able should be ensured access. Because the distribution of innate ability is not likely to be correlated with willingness to pay, an allocation system that relied on the income criterion alone to distribute a small number of school places would result in an underrepresentation of the most able students. Zero pricing schemes, on the surface, would seem to be efficient in this respect. Upon closer inspection, however, they also base allocations on 50 THE ISSUES 'Th.ble 4-5. Potential Indicators of Quality Change in Education by Level of Education, 1965-78 (percentage of countries) W~ rates increased more rapidly than expenditure Nonwage expenditure per per student student decreased Latin Other middle- Latin Other middle- Level of education America income countries America income countries Primary education 1965-70 53 (67)· 33 n.a. n.a. 1970-75 27 (55)' n.a. n.a. n.a. 1975-78 29 (71)· 20 n.a. n.a. 1970-78 27 25 n.a. n.a. Secondary education 1965-70 43 40 17 40 1970-75 70 71 55 71 1975-78 60 20 80 20 1970-78 78 25 83 25 Tertiary education 1965-70 69 40 50 50 1970-75 50 50 50 50 1975-78 60 50 75 66 1970-78 80 100 100 100 Note: The percentages refer to the number of "unfavorable" observations divided by the number of total available observations in each category; thus. where observations are few, the description of the state of education may not be accurate. n.a. Assumed not to be applicable for primary education. a. Figures in parentheses refer to cases in which wages increased at least at the pace of ex­ penditures. Source: Heller and Cheasty (1984). Reprinted with permission. willingness to pay. First, even with zero prices, private costs are a significant factor in determining household choice. This is particularly true for educa­ tion, a sector in which opportunity costs make privately borne costs as high as publicly incurred costs of free schooling. Free provision does not mean free consumption. These nonfee costs would cause poor households to demand less education than rich ones. Moreover, the poorest households may choose not to participate at all. Second, by rationing, many countries have limited the number of people who are allowed access. The impact on efficiency depends crucially upon the INEFFICIENCIES IN THE SOCIAL SECTORS 51 rationing scheme's success in detecting those who would benefit the most from the service and upon what sort of behavior the rationing scheme induces. The most popular method of allocating scarce public school places is by performance on examinations. Although this may appear to select students solely on the basis of ability, in practice it leads to sorting on the basis of willingness to pay. Rich households have more resources than do poor households to invest in tutoring or in materials that prepare their children for these examinations. In African countries such as Mauritius, students are even kept longer in school (through repetition) solely to enhance progression to higher levels of ~ducation (Woodhall 1983). The result is an unwarranted overrepresentation of rich households in the rationed schoolleve1, which in tum will result in excess investment in schooling inputs that are socially costly and only marginally beneficial in raising productivity. The efficiency loss, then, is substantial. In general, as will be discussed in the "prescriptions" part of this book, a combination of user charges with a loan scheme would ensure efficient allocation. Piiiera and Se10wsky (1981) estimate that if the allocation of school places were based more on ability and all the most able students were given the op­ portunity to benefit from secondary and higher education, the efficiency gains over the current allocation systems would amount to 5 percent of GNP in Latin America, Africa, and the less developed countries of Asia. A more re­ cent study in Colombia concluded that 54 percent of first-year students in tertiary institutions obtained lower scores in aptitude tests than did the cor­ responding cohort of secondary school graduates who did not attend such in­ stitutions (Jimenez 1985). In curative health, it is not particularly useful to speculate on which type of individual should get priority. Presumably, it is not the individual that mat­ ters, but the type of malady; the most serious and life-threatening malady should be given priority treatment. In health services as in education, it is inefficient to rely solely on prices to determine priority. Diseases or accidents strike randomly and, as will be dis­ cussed later, efficient allocation must be based on an insurance scheme that will allow low-income and high-income individuals equal access for similar treatments. Even with insurance schemes, however, it would be efficient to have some fee for payment to deter those who are not in dire need of a ser­ vice from dogging up the system. Some anecdotal evidence confirms that short-term excess demand in health has been met by declines in the level of service quality through over­ crowding and long waiting lines. World Bank staff have documented instan­ ces in which heavily subsidized district hospitals and urban hospital out­ patient departments "are swamped with patients" (Mills 1984, 17). 5 Inequities in the Provision of Services THE SUBSIDIES INHERENT in publicly provided educational and health ser­ vices have not been equitably distributed. Only a small proportion of the population is able to obtain access to these subsidies. Moreover, the poorest socioeconomic groups are not given priority. Indeed, at least for education subsidies, the richest income groups obtain a disproportionately large share. The Distribution of Subsidies Education A great proportion of government subsidies for education is directed at the highest level. Table 5-1 compares the unit public subsidies for primary, secondary, and higher education. On average over all developing countries, unit subsidies at the higher level are twenty-six times greater than at the primary level and nine times greater than at the secondary level. The dis­ crepancy is largest for the two African regions. At the same time, the propor­ tion of the world regional populations with no education or only primary schooling compared with the proportion with higher education is very small-72 percent versus 6 percent, respectively-as shown in the last four columns of table 5-1. These figures can be combined to derive a picture of the shares ofsubsidies acquired by different proportions of the populations (table 5-2). The figures account for the subsidies received by those in higher education throughout their school career, including subsidies at the primary and secondary levels. In the developing countries, 71 percent of the population in each generation receive only primary schooling or less and obtain only 22 percent of the resources devoted to education. The proportion receiving higher education is 52 INEQUITIES IN THE PROVISION OF SERVICES 53 Table 5-1. Per Student Public Subsidy on Education and Educational Attainment of the Population in Major World Regions, ca. 1980 Per student public subsidy as percentage of per capita GNP Enrol/ment ratio Primary Secondary Higher No Primary Secondary Higher Region education education education sclwoling education education education Mrica Anglophone 18 50 920 23 60 15.8 1.2 Francophone 29 143 804 54 32 11.6 2.4 South Asia 8 18 119 29 52 14.6 4.4 East Asia and Pacific 11 20 118 13 44 33.9 9.1 Latin America 9 26 88 10 46 32.0 12.0 Middle East and North Africa 12 28 150 18 46 26.6 9.4 Developing countries 14 41 370 25 47 22.0 6.0 Developed countries 22 24 49 0 20 59.0 21.0 Source: Based on Mingat and Tan (1985a). Reprinted with permission; copyright 1985 the University of Wisconsin Press. only 6 percent, but it obtains 39 percent of total resources. In francophone Africa, 2 percent of each cohort in the population attain higher education and receive 40 percent of the public resources devoted to education. In anglophone Africa, 1 percent of the cohort receives more than one-quarter of the public educational resources allotted to it. The main reason for this disproportionate allocation of resources is the large public reimbursements for out-of-pocket personal costs, such as transport and books (Eicher 1985; Hinchliffe 1985; Mingat and Psacha­ ropoulos 1985). For example, in anglophone Africa, "student subsidies repre­ sent 14 percent [of total subsidies] at secondary and higher education levels, while in francophone Africa the figures are much higher, 23 and 43 percent respectively" (Mingat and Psacharopoulos 1985, 36). Also, per student scholarships as a percentage of per capita GNP are 120 percent in the Cote d'ivoire, 160 percent in Senegal, 700 percent in Mali, and 800 percent in Niger and Burkina Faso. The rich get a larger proportion of the subsidy because they have more children in school, particularly at those educational levels where subsidies are 54 THE ISSUES 'Illble 5-2. Population and Total Educational Resources by Terminal Level of Schooling in Major World Regions, ca. 1980 Primary education or less Higher education Region Population Resources Population Resources Africa Anglophone 83 39 26 Francophone 86 16 2 40 South Asia 81 23 4 39 East Asia and Pacific 57 19 9 40 Latin America 56 16 12 42 Middle East and North Africa 64 19 9 45 Developing countries 71 22 6 39 Developed countries 20 8 21 37 Source: Mingat and Tan (1985a). Reprinted with permission; copyright 1985 the University ofWisconsin Press. highest. The total monetary effect of this on various world regions has re­ cently been estimated by Mingat and Tan (1986a). Table 5-3 shows the pro­ portion of total public resources appropriated for education by different socioeconomic groups. Because of the paucity of income data in developing countries, the distribution figures are categorized on the basis of occupation rather than income. The figures take into account the cumulative effects of subsidies obtained at previous levels of education. A comparison of the share of resources received by three socioeconomic groups with their share of the total population of school-age children provides a measure of the benefit that each socioeconomic group derives from education subsidies. This com­ parison, which is termed the "subsidy-benefit ratio," is much higher for the white-collar group, indicating that it enjoys a disproportionate share of all education subsidies. In most developing regions, the children of white-collar workers gain nearly six times more benefit from public education subsidies than do the children of farmers. In francophone Africa, the contrast is even more marked-ten to one. Thus, the provision of free or heavily subsidized education does not ensure equity in the distribution of public resources. Health Information on the distribution of public health subsidies by income class is even more limited than for education subsidies. Table 5-4 summarizes the 'DIble 5-3. Public: Resoa.rc:es Appropriated for Education by Different Socioeconomic Groa.ps in Major World Regions, ca. 198 Ratio between proportion of rescurce PerCetll4ge in the population Percenl4ge of public sdwol rescurces and poptlLJtion (1) (2) (2)/(1) Manual White- Manual UJhite- Manual UJ workers and collar workers and colLJr workers and co Region Farmers ~ workers Farmers tratkrs workers Farmers trdders wor Africa ~ Anglophone 76 18 6 56 21 23 0.73 1.19 3. Francophone 76 18 6 44 21 36 0.58 1.15 5. Asia 58 32 10 34 38 28 0.59 1.19 2. Latin America 36 49 15 18 51 31 0.49 1.04 2. Middle East and North Africa 42 48 10 25 46 29 0.60 0.35 2. OileD" 12 53 35 11 46 42 0.95 0.87 1. a. Organization fOr Economic Cooperation and Development. Source: Mingat and Thn (19800). Reprinted with pennission. Table 5-4. Public Health Subsidies by Income Group in Selected Countries Percentage share by income group' Country Survey year Type of health subsidy Poorest 2CJ% 20-4CJ% 40-6CJ% 60-8CJ% 80-10CJ% Chile 1969 Public health 31h 35 35 C Colombia 1974 National health service 30 23 20 IB 12 Social security system hospital 8 15 29 24 23 Health center 25 29 23 15 8 Overall public 20 21 20 20 20 Indonesia 1980 Overall public 1 qo- In appendix C, a more general model is developed that can be used to determine (1) the marginal impact of a drop in subsidy allocations (S) on utilization (q); (2) the price increase needed to offset a marginal drop in sub­ sidy allocations; and (3) the impact of a marginal price increase on utilization and overall household payments. For example, under the assumption of ex­ cess demand, the magnitude of (1) is (6-3) dq/dS = l/[c(e + 1) - p] where e is the unit cost elasticity with respect to the amount of the services provided. The magnitudes for the other expressions are derived in ap­ pendix C. 4. Differentiating equation 6-1 with respect to p and multiplying the resulting expression by (P/t{) yields (p/t{)(at/ap) 11 = ap/(p + dpc + oc). 7 Changing Pricing Policy in Education THE PREVIOUS CHAPTERS have argued that prices should be given a greater role in the mobilization ofresources to expand some types ofeducational and health services. Such a policy was recommended because of its potential for improving both efficiency and equity. In these last two chapters, the possible effects of such a policy in education and health are considered. An increase in user charges is warranted for educational services that do not exhibit "public goods" characteristics, such as externalities, and for which private demand is strong at current prices. Externalities lessen with the level of education. Thus, the case for greater cost recovery through pricing can be made for higher (postsecondary) and urban-based secondary educa­ tion. In primary education, externalities are important. But if budget re­ straints inhibit the increase of subsidies to that level and if private demand is strong enough (as it often is in urban areas), then a case can also be made for greater cost recovery for primary education. The applicability ofuser charges can be roughly categorized on the basis of some general characteristics of levels of education (see table 7-1). These conclusions may vary in specific cases because of country conditions. Behavioral Parameters Before estimating the magnitude of the impact of increased user charges, we need to review the literature on the parameters ofdemand and supply that will affect the calculations. Then the greater role of user charges at various levels of education, as well as accompanying policies, can be discussed in more quantitative terms. The most important parameters are those that govern preference and cost: how demand is likely to vary with respect to price changes and how cost is likely to vary with changes in scale. 77 78 CHANGING PRICING POLICY Thble 7-1. Stylized Market Characteristics of Educational Services Market for places at Level of education "Public good" characteristic at current prices Higher Lack of capital market Excess demand Secondary Urban Lack of capital market Excess demand Rural Lack of capital market; inadequate knowledge of benefits Excess supply Primary Urban Externalities Excess demand Rural Externalities; inadequate knowledge of benefits Excess supply The Responsiveness of Demand Families must make many decisions about education. Should their children go to school? If so, should all their children go, or should some remain at home? How much education should they have? And is academic schooling better than vocational training? The answers to these questions depend upon several variables, among them the cost of obtaining the education. The price responsiveness of educational demand is reflected in the price elasticity, which measures the percentage change in demand with respect to a 1 percent change in the price charged to the user of the educational service. Most demand studies are not able to measure price changes accurately. Some countries do not change prices at all, and those that do, vary them very slightly, which makes it difficult to observe behavioral responses to price variation. Since price is not the only cost-related variable that affects demand, however, most studies are able to use variations in directly incurred private costs or opportunity costs to infer behavioral parameters. When weighed by the share of price in total private costs, these elasticities can be interpreted as reasonable proxies for price elasticity (see table 7-2). The differences in the measures of demand (including household spending on education, the pro­ bability of having one child in school, and enrollment ratios) reflect the various choices that families make about schooling. The measures ofprivate costs used in table 7-2 also vary. Some are indirect measures, such as household size. A greater number of children enrolled im­ plies higher costs of educating an additional child and thus a lower enroll­ ment ratio among children in the household, if income is held constant (Birdsall 1980; Mingat, Tan, and Hoque 1984). Other measures that are used EDUCATION 79 include distance to school and mean wages (opportunity costs) for children of different age groups. Despite the variability in these measures, an increase in most components of private costs would result in a less than proportionate decrease in the de­ mand for education. 1 This conclusion can certainly be extended to fees since fees can be expected to be an even smaller proportion oftotal private costs. In the analysis that follows, the effects of alternative assumptions about fee elas­ ticity are simulated. Another way to estimate changes in demand would be to calculate the im­ pact of increased user charges on the perceived private rate of return to education. If this rate of return were to drop below that of investment in alternative activities, the family would be expected to stop investing in education. Table 7-3 presents the effects of price increases on primary and secondary school enrollment in Malawi. It is based upon the response of primary and secondary students who were asked whether they would drop out of school if certain fees were increased. The percentage declines in en­ rollment listed in column 3 are likely to be overstated since respondents. who would rather not pay a higher fee, have an incentive to understate their private rate ofreturn. Despite this potential bias. the responsiveness ofpoten­ tial enrollment is not very large in comparison with fee changes. At the secondary level, the percentage drop in enrollment is not substantially larger than the percentage fee increase. At higher fee increases, the former is smaller than the latter. At the primary level, the percentage drop in enroll­ ment is half or less than half the percentage increase in price. 2 In April 1982, the Malawian government did in fact increase school fees. The increase varied between urban and rural areas and by type of school. At the primary level an increase in fees of 45 to 50 percent produced a mean drop-out rate of 1.4 to 1.7 percent, respectively, while at the secondary level an increase of 50 to 150 percent produced a drop-out rate of 4.3 to 7.8 per­ cent, respectively (Tan, Lee, and Mingat 1984, table 8). The Responsiveness of Cost The level of public educational services depends upon the willingness and the ability of the public authorities to provide them. An inquiry into bureau­ 1. Of all the studies cited in table 7-2, only one (Taiwan), which uses economywide time series data and is likely fraught with aggregation bias, computes an elasticity that exceeds one. 2. It would, ofcourse, be misleading to infer the magnitude of the fee elasticity ofdemand from these data since elasticity is defined as percentage changes in demand with respect to very small changes in price. Thble 7-2. Price and Income Elasticities of Demand for Education Elasticitie Level of lAtest rear atIdIysis of data fkpendent varl4b1e IlICOme measure Price measure Income Colombia Household 1967-68 Spending on education Husband's income Household characteristics ~ (number of children of different age groups) 1.045 I Share of household budget on education 0.334 I Actual expenditures! predicted expenditures 1.035 I Household education achievement index 1.343 I EI Salvador Household 1980 Spending on education Household permanent Household characteristics income (proxied by expenditure) Santa Ana 0.967 I Sonsonate 0.023 I Malawi Household 1983 Household enrollment Father's income Total household cost - Malaysia Household 1976 Population of school- Household income Distance to secondary children school Age 6-11 0.097 -0 Age 12-18 0.318 -0 Mali Household 1982 Enrollment ratio Household income Fees to parent association -0 Distance to school - Quality (availability of books) Pakistan Individual 1978-79 Probability of going to Household income 0.01 to 0.15 school per capita Philippines Individual 1968 Years of completed Father's wage in 1968 Mean wage for children schooling Age 7-14 0.111 - - - OQ Age 15-19 0.111 Taiwan Countrywide 1950-69 Number taking college Per capita income Average tuition and fees 0.383 time series entrance exam Tanzania Individual 1981 Spending on education Father's income 0.03 ... Information not provided by the study. Sources: Birdsall (1980) for Colombia; World Bank data for EI Salvador; Tan, Lee, and Mingat (1984) fur Malawi; de Tray (1984) for Malaysia; Birdsall (1983a) for M (1984) for Pakistan; King and Lillard (1983) for Philippines; Kondrassis and Tseng (1976) for Uiwan; and Tan (1985a) for Unzania. 82 CHANGING PRICING POLICY Table 7-3. Percentage of Malawian Students Expecting to Leave School after Hypothetical Increase in Fees, 1983 Fee after Increase as percetl/- Perctntage decline increase (kwacha) age of total fee in enrollment (3) I (2) Level of education (1) (2) (3) (4) Primary 10 48 -22 -0.46 (cost recovery 20 196 -38 -0.19 K16 per student) 30 344 -18 -0.14 50 640 -57 -0.09 Secondary 40 33 -52 -1.58 (cost recovery 60 100 -75 -0.75 K.266 per student) 100 233 -85 -0.36 200 567 -91 -0.16 Note: GNP per capita in 1982 was approximately 219 kwacha (K) according to International Monetary Fund statistics. Source: Calrulated from Thn. Lee. and Mingat (1984). table 4. cratic behavior, however, is beyond the scope ofthis book. It is assumed here that government provision of education is consistent with a search for social efficiency-an economical and equitable allocation of resources. Education can be financed out ofgovernment subsidies and collected fees. Increased fees obviously strengthen the ability of public authorities to pro­ vide educational services. But collected fees are not always earmarked for education; they may be treated like general tax revenue. In this section, however, all user fees are assumed to be kept within the education sector ex­ cept when noted otherwise. The maximum amount of educational service that can be provided would therefore be the sum ofthe total subsidy and user charge revenue, divided by the unit cost. The education cost structure could thus influence the impact of user charges. Although not comprehensive, some evidence in developing countries sug­ gests that the average costs of providing education decline as enrollment increases-that is, that there are economies of scale. According to a study of average costs and enrollment ratios that was based on aggregate data from eighty-three countries, an increase of the higher education enrollment ratio from 1 percent to 2 percent would lower average costs by 15 percent (Psacharopoulos 1982). This magnitude is much less for greater enrollment ratios. An increase of the higher education enrollment ratio from 10 percent to 11 percent would lower average costs by only 6 percent. The same cross­ national study also compiled within-country estimates (see table 7-4). The qualitative results do not change. An annual increment in enrollment ofabout EDUCATION 83 Table 7-4. Changes in Enrollment in Higher Education and Average Cost per Student, Selected Periods (percent) Country Period Enrollment change Real cast change Egypt 1957-75 9.5 0.2 Ghana 1957-75 8.2 -9.2 Kenya 1968-70 20.4 -10.9 Mexico 1961-75 13.0 -0.8 Pakistan 1964-75 8.4 -6.3 Thailand 1954-64 6.4 0.1 Zambia 1969-73 21.5 -8.6 Average 12.5 -5.1 Source: Psacharopoulos (1982). 12 percent is associated with a fall of 5 percent in unit cost. This represents a total cost elasticity of 1.4.3 At the primary and secondary levels, there are only a few within-country studies that utilize microlevel data. The elasticity of cost with respect to a marginal primary student in Bolivia is 0.59 and in Paraguay, 0.5; for a marginal secondary student in Bolivia it is 0.54 (Jimenez 1986a). These find­ ings also suggest the presence of scale economies. Prices and Expansion As was shown in table 7-1, user charges are more appropriate for higher education and perhaps for the secondary level than for primary education, presuming that financial markets can be made to work better. The gains from user charges will be discussed in terms of expansion, efficiency. and equity for all three levels. Higher Education When education budgets are limited and the available subsidy is insuffi­ cient to fund the optimal amount of schooling, efficiency will be improved by increasing user fees and using the revenue to expand those levels ofeduca­ 3. In this context it should be kept in mind that there is only limited evidence on the elas­ ticity of demand for higher education in developing countries. and differences in elasticities across income groups and for different price levels could not be estimated. 84 CHANGING PRICING POLICY tion with the highest social rates of return. The feasibility of mobilizing suffi­ cient resources depends upon the responsiveness of demand and supply. If there is excess demand, prices can be raised to mobilize revenue until equili­ brium is restored. Beyond that point, the ability to mobilize resources depends upon the elasticity of demand. If demand is relatively price inelastic, the percentage amount by which consumption drops will be less than the per­ centage increase in prices, and there will be a revenue gain. Heavy unit subsidies for higher education, at a time when budgets are con­ strained, have led to excess demand in many countries. In Nigeria, for exam­ ple, the average acceptance rate for university education was only 16 percent in 1979-80, and in some specialties, such as business administration and law, it was even as low as 8 and 5 percent, respectively (Hinchliffe 1985). In In­ donesia, on average only 30 percent of all applicants to institutions of higher education were accommodated in the early 1980s (Hanovice 1984). In several Latin American countries, there is evidence the number of applications for university education exceeds the available places by a factor of at least two (Schiefelbein 1985). Further evidence of the insufficiency oflocal facilities to meet the private demand for higher education is the large and growing num­ ber of developing country nationals studying abroad (Lee and Tan 1984). For example, in Greece one out of every four postsecondary students attends a foreign institution, often with family financial support. In cases where there is excess demand, revenue will increase by the amount of the increase in user charges. Once the demand has been satisfied, the amount of revenue generated will depend upon demand elasticities, as explained in the previous chapter. Where should the resources be spent? They should be spent where the social rates of return are highest. In most cases, this means primary education. Table 7-5 shows the potential efficiency gains of expanding primary educa­ tion through increased private contributions to higher education in twelve African countries. By shifting the financial responsibility for living allow­ ances from governments to students or their families, enough public resour­ ces would be freed to allow, on average, an 18 percent expansion of the yearly primary education budget. These additional funds would allow greater coverage of the primary-school-aged population. An additional expansion of 23 percent in the primary school budget would be possible if fees recovered all operating costs in higher education. Thus, if both kinds of subsidies in higher education were fully withdrawn, the primary education budget would be expanded, on average, by about 40 percent. The expansionary effect differs from country to country.4 In Cote d'Ivoire, Malawi, Senegal, Sudan, Tanzania, and Togo, the potential expan­ 4. Unit costs are assumed to be constant in these calculations. EDUCATION 85 Table 7-5. Potential Increase in Primary Education Budget Financed by Raising Higher Education Fees in Selected African Countries, 1980 Hight:r education fees raised to recover Country Living expemes Operating costs Both Benin 18.9 5.0 23.9 Burkina Faso 18.6 8.0 26.6 Central African Republic 12.4 4.0 16.4 Congo. People's Republic ofche 17.6 5.8 23.4 Cote d'lvo ire 21.0 19.2 40.2 Malawi 8.6 45.8 54.4 Mali 21.6 8.6 30.2 Niger 9.6 2.4 12.0 Senegal 20.4 48.5 68.9 Sudan 2.9 40.2 43.1 Tanzania 24.2 31.0 55.2 Togo 40.4 51.6 92.0 Average 18.0 22.5 40.5 Source: Based on Mingat and Tan (1985a). sion in the primary school budget would be substantial, ranging from 40 to 90 percent. Since Tanzania and Togo have attained or are close to attaining uni­ versal primary education, the generated resources could be used to improve educational quality. In the remaining countries, if the additional resources were used for quantitative expansion, the enrollment ratio in primary educa­ tion would rise dramatically. Given present unit costs, this ratio would in­ crease from 76 to 100 percent in Cote d'Ivoire, 48 to 81 percent in Senegal, and 51 to 73 percent in Sudan (Mingat and Tan 1985a). In countries with a very low primary school enrollment rate, the returns to primary school expansion are likely to exceed the returns to secondary educa­ tion and to almost all subjects in higher education. But as coverage at the primary level is extended geographically and academically to a diverse pop­ ulation, the average costs will tend to rise and the marginal returns to fall. In some countries, such as those that are close to attaining universal primary education, it may be socially profitable to use the extra funds to expand secondary education. In countries where highly skilled personnel is relatively scarce, it may be profitable to expand fields within higher education that ex­ hibit high social rates of return, such as engineering, management, or the physical sciences (see appendix table D-3). 86 CHANGING PRICING POLICY Thble 7-6. Impact of a 10 Percent Increase in Fees on Enrollment in Public Higher Edllcation if Fee Revenlles Were Used to Expand Higher Edllcation Percentage increase inenrollment if cost elasticity were Unit Jee as Country percentage of unit cost 0 -0.5 Africa Burundi 14.8 1.7 4.2 Lesotho 5.0 0.5 1.1 Malavvi. 1.0 0.1 0.2 Nigeria 12.4 1.4 3.3 Asia India 29.1 4.1 13.9 Indonesia 13.0 1.5 3.5 Korea, Rep. of 23.4 3.1 8.8 Malaysia 5.8 0.6 1.3 Pakistan 2.1 0.2 0.4 Philippines 3.7 0.4 0.8 Thailand 6.9 0.7 1.6 Turkey 15.0 1.8 4.3 latin America Bolivia 1.0 0.1 0.2 Brazil 5.0 0.5 1.1 Chile 25.0 3.3 10.0 Colombia 3.4 0.4 0.7 Costa Rica 8.0 0.9 1.9 Dominican Rep. 1.0 0.1 0.2 Ecuador 2.0 0.2 0.4 Guatemala 10.0 1.1 2.5 Honduras 10.0 1.1 2.5 Paraguay 0.7 0.1 0.1 Uruguay 5.0 0.5 1.1 Average 8.8 1.1 2.8 Note: Excess demand for higher education is assumed. Source: Calculated from appendix table D-1, using equation C-18 in appendix C. EDUCATION 87 The extent to which the higher educational system can expand depends upon the impact ofschool size on unit costs. The extent to which fee increases can provide additional services depends upon the behavior of unit costs and the presence of scale economies. Since there is little information about the behavior of unit costs, only hypothetical cases based on reasonable assump­ tions about the unit cost elasticities at the margin are discussed here. Table 7-6 shows how a 10 percent increase in fees might affect the quantity of public higher education ifthe revenues were used to expand higher educa­ tion. On the assumption of constant costs (given that unit fees are, on average, 8.8 percent of unit costs for the sample of countries), this fee in­ crease would result in an increase in enrollment of 1 percent. For countries with a less than average cost-recovery percentage, such as Bolivia, the Dominican Republic, or Malawi (at 1 percent), the increase in enrollment would be 0.1 percent. For countries such as Korea, with a cost-recovery per­ centage of23, or Chile, with a cost-recovery percentage of25, the potential increase in enrollment that could be financed by a 10 percent increase in fees would exceed 3 percent. On the assumption that there are scale economies in the provision of higher education, the expansionary impact would be enhanced. If average costs declined by 0.5 percent for every 1 percent increase in enrollment (that is, a cost elasticity of-0.5), the potential expansionary impact ofa 10 percent increase in fees for the average country would almost triple from 1.1 percent to 2.8 percent. Macro-level evidence indicates that such an estimate of unit cost elasticity is plausible (Psacharopoulos 1982). Secondary Education There is less empirical evidence of excess demand for secondary educa­ tion, although it is known to exist in some places. In Malawi, for example, the supply of secondary school places accommodates only one-third of expressed demand (Tan, Lee, and Mingat 1984). In Kenya, there is evidence that many students retake the secondary entrance examination to enhance their chances of admission (Somerset 1974). There are other indications that private de­ mand is strong, particularly in urban areas. Recent restrictions on secondary school budgets in Pakistan have led to an unprecedented growth in private schools during the past five years Gimenez and Tan 1985). The same is true in Kenya, where the harambee system of locally financed schools has absorbed those unable to enter publicly provided secondary schools (Bertrand and Griffin 1983). A case study of the potential impact of fees at the secondary level in Malawi is particularly well documented (Mingat and Tan 1985c). Although 88 CHANGING PRICING POLICY only 4 percent of the secondary-school-age population attend school, the government has had to restrain increases in public spending in the sector because of adverse economic conditions. The supply of places for secondary education, as a result, has not kept up with demand. In 1982, the total private demand was estimated to be at least 50,000, but only 17,000 individuals could be accommodated. The cost per student place in Malawian secondary education was K266 (about US$280) a year in 1982. On average, students paid K.30 in tuition and K71 in boarding charges. In total, the revenue from tuition and boarding fees recovered about 38 percent of the entire cost of public secondary education. The actual annual cost to the government of enrolling each student was therefore K165 (0.62 X K266). Despite the already substantial level ofprivate costs, an increase in tuition fees is unlikely to result in lower enrollments. It probably would only reduce the extent of excess demand, not the actual overall level of enrollments, if the extra revenue generated was used solely to increase the supply of secondary education places. If the elasticity of demand were -0.03, excess demand for secondary education would persist as long as the new level of tuition fees remained below K95 (Mingat and Tan 1986c). If fees were raised to this level, the ad­ ditional revenue would allow secondary education to expand by 11,000 places, an increase of65 percent over the current provision. The potential ex­ pansion would, ofcourse, be smaller if demand were much more elastic than was assumed in this calculation. Even if elasticity was -1.0 instead of -0.5, however, fees as high as K68 would not eliminate excess demand. The revenue generated by raising fees to this level would permit a smaller expan­ sion of 5,100 extra places, an increase of 30 percent over the present supply. These simulations show that increased user charges for secondary education could generate efficiency gains since the additional resources would allow in­ creased investment in a highly profitable activity. For other countries where precise estimates of excess demand are not available, only calculations at the margin are possible. The results are pre­ sented in table 7-7. On average, a 10 percent increase in fees in twenty-seven developing countries would allow an expansion in enrollment of 2.2 percent if unit costs were unchanged. Ifunit costs dropped at a unit cost elasticity rate of approximately -0.5, a figure derived from a study ofParaguay and Bolivia by Jimenez (1986a), the expansion would be a substantial 11.1 percent. Primary Education It has been argued that mobilizing resources through user chargers for higher and perhaps even secondary education would produce efficiency gains EDUCATION 89 Table 7-7. Impact of a 10 Percent Increase in Fees on Enrollment in Public Secondary Education if Fee Revenues Were Used to Expand Secondary Education Percentage increase in enrollment if cost elasticity were Unitfoe as Country percentage of unit cost 0 -0.5 Africa Botswana 2.7 0.3 0.6 Burundi 6.3 0.7 1.4 Central African Rep. 2.7 0.3 0.6 Kenya 43.7 7.8 69.4 Lesotho 42.1 7.3 53.3 Malawi 38.0 6.1 31.7 Nigeria 39.0 6.4 35.5 Sierra Leone 20.3 2.5 6.8 Swaziland 26.3 3.6 11.1 Togo 5.0 0.5 1.1 Uganda 24.3 3.2 9.5 Zambia 11.6 1.3 3.0 Zimbabwe 5.0 0.5 1.1 Asia India 18.5 2.3 5.9 Indonesia 9.0 1.0 2.2 Korea, Rep. of 41.2 7.0 46.8 Malaysia 5.0 0.5 1.1 Pakistan 1.8 0.2 0.4 Solomon Islands 25.0 3.3 10.0 Thailand 12.5 1.4 3.3 Latin America Bolivia 0.4 0.0 0.1 Chile 0.9 0.1 0.2 Costa Rica 0.5 0.1 0.1 Haiti 3.4 0.4 0.7 Honduras 9.6 1.1 2.4 Paraguay 2.0 0.2 0.4 Uruguay 0.4 0.0 0.1 Average 14.7 2.2 11.1 Source: Calculated from appendix table D-l, using equation C-18 in appendix C. 90 CHANGING PRICING POLICY if the resources were used to expand the educational service with the highest rate of return. But for the primary level, other factors must be considered, and the conclusion may differ. First, because of externalities and other public goods characteristics, higher unit subsidies are needed to stimulate private de­ mand at the primary level than at other levels. Second, in many countries, there is not as much evidence ofexcess demand for primary education. There may be some excess demand in urban areas, and, if so, the analysis for higher and secondary education would hold. In rural areas, low enrollment ratios are due as much to lack ofdemand as to supply constraints. Thus, existing schools may not be able to generate resources without lowering enrollment. Under certain circumstances, however, it would also be efficient to charge higher prices at the primary level. If transfers of funds from other levels of education are administratively or politically too costly to implement or if these transfers do not generate sufficient resources (as in countries with relatively small enrollments at tertiary levels), then resources could be generated through higher user charges for primary education. If there is excess demand at the primary level, then the previous analysis holds. If not, the argument hinges on the demand for the quality of educa­ tion. It is assumed that the lack ofprivate demand for existing school places is due predominantly to the poor quality of the schooling. Thus, if parents and students knew that the additional resources would be used to improve school quality, they would be willing to pay the higher user charges. The structure of demand would then change. In fact, demand would increase as a result of the fee increase, if the fees were used to improve schools. One indication that primary school quality is low in developing countries is the shortage of textbooks and other school supplies. In Burkina Faso, for example, where books are provided free by the state at the primary level, "probably only a minority of pupils possess textbooks" (OrivelI983, 12). In rural Mali, there are about six textbooks per one hundred primary students, and seven out of twenty schools surveyed had no books at all (Birdsall 1983a, 10). In Kenya, despite an education commission's recommendation in 1976 that government expenditure on supplies and equipment be doubled, such resources appear to be declining (Bertrand and Griffin 1983,34). Other in­ dications of low quality include severe overcrowding and relatively high student-teacher ratios, such as in Malawi (Thobani 1983, 26), and a great reliance on untrained primary teachers (Bertrand and Griffin 1983, 32). Partly because of the difficulty in agreeing upon an appropriate definition of educational quality, direct evidence on the social returns to investment in quality is not extensive. The preceding evidence suggests, however, that the mix of quantity and quality in primary education may be inappropriate. Behrman and Birdsall (1983) conclude that the high social rate of return to primary schooling in Brazil can be largely ascribed to quality-enhancing EDUCATION 91 rather than quantity-enhancing investments. The relative neglect ofquality is partly explained by the recurrent cost problem outlined earlier. Moreover, most internationally set standards regarding schooling are based solely on quantity (such as universal education to a certain level). Providers who are in­ tent on meeting those standards with a given amount of resources may sac­ rifice quality to expand coverage. Ignoring the tradeoff between numbers and quality of learning is likely to result in a misallocation of resources. The imposition of user charges on quantity (the charge is levied per child) may induce a more efficient mix by decreasing utilization and allowing revenues to be used to increase quality. This is feasible because improved quality may stimulate demand and increase households' willingness to pay (Birdsall 1983b). Thus, any fees collected to expand quality may also allow society to expand coverage (financed by user fees), that is, to increase the number of primary students who are educated. Since fees are often ear­ marked for books and supplies, modest increases in fees may have a small im­ pact on the total government recurrent budget but a substantial effect on cer­ tain types of expenditures. For example, Birdsall (1983a) estimates that in Mali a 10 percent increase in user fees would be sufficient to double the sup­ ply of books available to students. Since the availability of books, in par­ ticular, and school quality, in general, affect learning throughout the world (see Heyneman and Loxley 1983), this would seem to be a great contribution not only to finance, but also to internal efficiency. The social gains depend upon the benefits of another unit of quality in relation to quantity. Birdsall's (1983a) model of this argument is depicted in figure 7-1, where the q axis denotes the quantity ofeducation being provided, given the quality. Initially, no prices are charged, and for a school of a given quality, provided at the cost to the government cp Co, the private demand structure is D~. There is no willingness to pay for schooling. If cp denotes the private cost of provid­ ing education, demand will be qo. If the goverrunent provides a subsidy of So, this demand will be met. So is the iso-subsidy line (see explanation of figure 6-1). At a higher quality level, it would be more costly to provide education, at Cl' But there would be greater willingness to pay, as well, at D;. If the sub­ sidy allocation remained the same at So, prices could be raised from zero to PICp and enrollment could still expand to ql' Because little is known about key behavioral parameters, such as the responsiveness of demand to changes in quality, the relative preference for quantity over quality, and the costs of in­ creasing quality, this model has yet to be tested. Some of the conceptual and empirical work, however, has begun (Mingat and Tan 1986b). If there is no excess demand for primary education, the effect of a fee in­ crease would depend upon the elasticity of demand. Consider, for example, the potential role of user charges in primary education as a response to a cut in government subsidies. The revenue from user charges could be used to 92 CHANGING PRICING POLICY Figure 7-1. Enrollment Expansion as the Result of Increased Fees to Finance Qualitative Improvements p CI~----~r---------~------------------ ~~--------~----------~----~~----~- ~~------------~----~--~------~--+-- minimize the contractionary impact of tighter budgetary restrictions. Figure 7-2 presents the financial impact of a marginal (10 percent) cut in government transfers to primary education and the increase in fee that would be necessary to maintain present enrollments. It is assumed that, at the current fee level, the system is at equilibrium with neither excess demand nor excess supply. A drop in the subsidy allocation from So to SI would result in a drop in enrollment from qo to ql' To mitigate this drop would require a price increase from zero to Cp P2' The welfare gain obtained would be ABED. The results of the analysis are presented in table 7-8. For the countries for which data are available, a drop of 10 percent in government subsidies would require a price increase of 9.4 percent to minimize social losses if demand is relatively inelastic, or 5.3 percent if demand is unit elastic. Prices and Efficiency within Schools Two common sources of internal inefficiency are the mix of purchased in­ puts, such as teachers' services and pedagogical materials, and the process of selecting students. EDUCATION 93 Figure 7-2. Impact of a Cut in Subsidies on Fees and Enrollment in Public Primary Education p P2 ~r-----------~--~~~~----------~----- q Buying the Right Inputs The mix of inputs is inefficient when more can be achieved if the same amount of funds is allocated differently among the various inputs. The pre­ sent financing and provision of education contributes to this type of inef­ ficiency. Most public school systems are highly centralized, both in the collection and distribution of revenue for education. Revenues are usually drawn from general tax sources, which are then distributed through budgetary allocations to the central ministry of education. In turn, the funds are allocated to schools and universities according to preset funding formulas. In many instances the central budgets pay teachers directly. In most cases the public school system has adjusted to new budgetary re­ strictions through a process of default, partly because managers have neither the incentive nor the authority to respond creatively. Rules regarding teachers' qualifications, employment, and salaries are normally inflexible, largely because teacher unions are a powerful force in defining and protect­ ing the status quo. In most cases teachers are employed centrally, and in­ dividual schools have no discretion over hiring. Consequently, much of the reduction in education funds has been absorbed by cuts in expenditure for Table 7-8. Impact of a 10 Percent Cut in Subsidies on Fees and Enrollment in Public Primary Education if Household Payments Were Used to Maintain Equilibrium Percentage increase in Percentage decrease in foe if demand elastidty enrollment if foes rise were and demand elasticity were Unit foe as Country percentage of unit cost -0.5 -1.0 -0.5 -1.0 Mrica Botswana 0.0 20.0 10.0 -10.0 -10.0 Burkina Faso 13.0 0.7 0.7 -0.4 -0.7 Burundi 0.0 20.0 10.0 -10.0 -10.0 Central African Rep. 2.5 3.3 2.9 -1.7 -2.9 Guinea 0.0 20.0 10.0 -10.0 -10.0 Kenya 4.0 2.2 2.0 -1.1 -2.0 Lesotho 9.0 1.1 1.0 -0.5 -1.0 Malawi 37.0 0.3 0.3 -0.1 -0.3 Mauritania 0.0 20.0 10.0 -10.0 -10.0 Nigeria 30.0 0.3 0.3 -0.2 -0.3 Sierra Leone 1.5 5.0 4.0 -2.5 -4.0 Swaziland 7.0 1.3 1.3 -0.7 -1.3 Togo 13.0 0.7 0.7 -0.4 -0.7 Uganda 27.0 0.4 0.4 -0.2 -0.4 Zambia 3.0 2.9 2.5 -0.4 -2.5 Zimbabwe 0.0 20.0 10.0 -10.0 -10.0 Asia India 2.0 4.0 3.3 -2.0 -3.3 Indonesia 0.0 20.0 10.0 -10.0 -10.0 Korea, Rep. of 3.7 2.4 2.1 -1.2 -2.1 Malaysia 5.0 1.8 1.7 -0.9 -1.7 Pakistan 1.2 5.9 4.5 -2.9 -4.5 Solomon Islands 0.0 20.0 10.0 -10.0 -10.0 Thailand 0.0 20.0 10.0 -10.0 -10.0 Turkey 0.0 20.0 10.0 -10.0 -10.0 Latin America Bolivia 0.8 7.7 5.6 -3.8 -5.6 Chile 1.6 4.8 3.8 -2.4 -3.8 Costa Rica 0.3 12.5 7.7 -6.3 -7.7 Dominican Rep. 0.0 20.0 10.0 -10.0 -10.0 Ecuador 0.0 20.0 10.0 -10.0 -10.0 Haiti 6.8 1.4 1.3 -0.7 -1.3 Honduras 0.0 20.0 10.0 -10.0 -10.0 Paraguay 4.1 2.2 2.0 -1.1 -2.0 Uruguay 0.5 10.0 6.7 -5.0 -6.7 Average 5.2 9.4 5.3 -4.7 -5.3 Source: Calculated from appendix table D-1, using equation C-ll in appendix C. 94 EDUCATION 95 other categories ofschool inputs. For example, in East African countries, such as Comoros, Ethiopia, Rwanda, and Tanzania, governments are currently fac­ ing difficulties in maintaining textbook programs. The result is severe text­ book shortages, especially in rural areas (Wolff 1984). Similarly, in Jamaica there is a general lack of instructional materials in primary schools and in some secondary schools, despite the fact that 20 percent of the state budget is allocated to the education sector. In fact, in most developing countries today expenditure on instructional materials is very low (1980 US$4.80 on average; for developed countries, the figure is 1980 US$106). On average in 1980, it accounts for 3.4 percent of the total public recurrent expenditure on educa­ tion (Unesco, Statistical Yearbook). Examples from Heller and Cheasty (1984) in Latin America have already been mentioned. A financing system that relied more heavily on user charges could improve efficiency since it would give schools more resources with which to respond to local conditions. The willingness to pay of households could be mobilized to buy the necessary inputs. School authorities would have to be held accoun­ table to parents and students to ensure that appropriate incentives existed for management. Choosing the Right Students A system of selecting students is likely to be inefficient ifit does not admit from the previous level the students with the most learning potential. If it is assumed that innate ability is randomly distributed in the population, then a system that grants access solely on the basis of willingness to pay would lead to inefficient selection. Even with free or heavily subsidized education, poor bright students are being left out. In Colombia, 54 percent offirst-year students in higher educa­ tion institutions scored lower on aptitude tests than did the corresponding cohort of secondary school graduates who did not continue their education. The students who did not continue their education came from families with incomes significantly lower than those of the families whose children en­ rolled in institutions of higher education (Jimenez 1985). Since there is a lack of credit markets from which students can borrow to finance their studies, the competition for places is limited to those who have the requisite private funds at the time of enrollment. Greater reliance on user fees might exacerbate this type ofinefficiency because it would force some of the poorer students to terminate their education. The extent to which the users would be affected depends, ofcourse, on the magnitude ofthe price in­ crease. Tables 7-9 to 7-11 show the distributional impact on household budgets caused by a price change that would recover all costs. Since there is 96 CHANGING PRICING POLICY Table 7-9. Household Expenditure for Public Education without Subsidies in Malaysia (1974 ringgit; all items per household) Average Current Total public cast Out-ofpocket cost annual public Quinti/e incallre (Y,l sub.riJy Amount Percentage of Y Amount Percentage of Y 1 1,552 471 583 38 221 18 2 2,681 419 527 20 215 10 3 3,786 483 585 15 295 10 4 5,230 419 511 9 288 8 5 12,440 422 513 4 291 6 Mean 5,662 444 544 13 265 10 Source: Meerman (1979), tables 3.1, 4.7, and 4.14. Table 7-10. Household Expenditure for Public Education without Subsidies in Colombia (1974 Colombian pesos; all items per household) Public cost of eaucation Primary Secondary University Total Average Perctnlans. One way of minimizing the negative equity and efficiency effects of increasing private financing is to improve individuals' access to financial markets through student loan schemes, particularly for higher education. In some countries, commercial credit markets work well but may not be available for funding education or training. Private banks may be unwilling to lend to students because they often lack acceptable forms of collateraL Also, banks may be afraid that unemployment will cause graduates to default on their loans. And in many developing countries, the legal and administra­ tive frameworks for enforcing financing contracts are weak. Finally, the ad­ ministrative costs of collection tend to be high because of the mobility of graduates. Many commercial banks in developing countries are simply too small to absorb the high risks and costs oflending to students without charg­ ing prohibitive interest rates. Governments can make it easier for students to obtain educational credit. Because their size allows them to absorb the risks that private lenders are un­ willing or unable to bear, governments can become lenders, or they can in­ sure commercial loans. In addition, governments can make collections less problematic. Since most graduates work in the formal wage sector, they can be traced through the government bureaucracy or through the income tax system. With governmental authority, employers may be willing to assist collection by making deductions from wages for loan repayment. In several devel0ped and developing countries, particularly in Latin America, the government is already active in providing educational credit (see table 7-12). Governments either have established state-owned banks to provide student loans, particularly for higher education, or have encouraged private banks to provide student loans with a government guarantee. The Colombian government has recently introduced a type of loan that is repaid partly by the students and partly by their parents. This is similar to the PLUS scheme (Parents' Loans for Undergraduate Study) in the United States that permits parents and students to borrow from the government to finance education. EDUCATION 99 Table 7-12. Student Loans in Latin America, 1978 Total loans awarded Country (student loan institution) (excluding loans already repaid) Argentina (INCE) 1,400 Bolivia (CIDEP) 476 Brazil APLUB 3,084 Caixa Economica Federal 354,588 Colombia (ICETEX) 53,865 Costa Rica (CONAPE) 1,286 Chile (Catholic University) 1,982 Dominican Republic (FCE) 10,097 Ecuador (IECE) 15,803 EI Salvador (Educredito) 2,350 Honduras (Educredito) 1,740 Jamaica (Students' Loan Bureau) 6,875 Nicaragua (Educredito) 630 Panama (IFARHIJ) 5,800 Peru (INABEC) 274 Venezuela Educredito 2,866 SACUEDO 2,770 Source: Woodhall (1983). Experience with student loan programs shows that high default rates and collection difficulties are sometimes exaggerated. In the United States, default rates on student loans have been falling in recent years. When allowance is made for funds collected on previously defaulted loans, the net annual default rate in 1981 was only 4 percent, which compares well with the default rate on other forms of credit (Hauptman 1983). In the developing countries, it is more difficult to evaluate the incidence of default since loan schemes are relatively recent and a high proportion of loans are not yet due. Latin America and parts of the Caribbean are probably the only developing regions with extensive experience with loan schemes. The incidence of default or late repayment has actually been quite low in several of these coun­ tries. In Costa Rica, only 0.5 percent of the debts due for repayment in 1978 were in default. In Brazil, the proportion was only 2 percent, and in Colom­ bia, Honduras, Jamaica, and Mexico, it was between 5 and 11 percent (Woodhall 1983). 100 CHANGING PRICING POLICY Evidence from a few countries suggests that administrative costs are not exorbitantly high. In Sweden, the Central Student Assistance Commitee, the state agency responsible for administering student loans and other forms of assistance, calculated that in 1980-81 the cost of administration represented only 1.8 percent of its total expenditure on student aid. And the U.S. Con­ gressional Budget Office estimated in 1980 that the annual cost of servicing student loans in the United States ranged from 1.5 to 2 percent of the loan principal, compared with the usual range of 0.25 to 3.75 percent with housing loans (Woodhall 1983). In several Latin American student loan institutions, from 12 to 23 percent of the total annual outlay was spent on administration. With good management, however, it would be possible to reduce this share to between 10 and 15 percent (Herrick, Shavlach, and Seville 1974). Although loan schemes tend to suffer from default and collection pro­ blems, these difficulties are neither unmanageable nor universal. The cost of administration can be kept reasonably low in most cases. In the absence of well-functioning commercial credit markets, individuals often must rely on funds borrowed from relatives, friends, or ad hoc moneylenders. In Malawi, more than half of the secondary school students in high and low socio­ economic groups obtained financing from relatives in 1983 and another 20 percent obtained external loans (Tan, Lee, and Mingat 1984, table 4). Not all students, however, know people who are willing and able to lend them money. A Further Efficiency Gain: Overcoming Rent Seeking. Quantity rationing schemes, such as entrance examinations or grades, may still rely on willing­ ness to pay. Many relatively wealthy students invest in private tutoring or repeat grades to improve their examination scores and thereby improve their chances of gaining access to the next level of education. This "over­ education" is unproductive and consumes resources. High education subsidies are a windfall to those who are able to obtain them. They impart economic rents; that is, they offer a private benefit that exceeds the social cost of provision. This gives an incentive for those with means to invest resources in order to receive the subsidies. In economic tenns, this is called rent seeking. Thus, students may deliberately repeat grades so as to compete more successfully for subsidies at higher levels of education, as in Malawi, and schools may overexpand or lobby the ministry of education, depending upon the rules regarding the provision of subsidies. Private providers can also engage in renk seeking if subsidies are distributed to them. In Kenya, many harambee schools are being established without much regard for financial viability because they expect to be subsidized in the future (Bertrand and Griffin 1983). EDUCATION 101 Prices and Equity To a large extent, inequities in the access to education would be alleviated by user charges that are accompauied by student loans and subsidies targeted to low-income groups. Moreover, the shift in subsidies from higher to primary education would, in itself, improve equity (Mingat and Tan 1985a). In developing countries as a group, 72 percent of a generation of schoolchildren (those with primary schooling or less) obtain only 25 percent of the total public expenditure on education (Mingat and Tan 1985a). Their share would increase to 64 percent if user charges could be introduced to recover all the costs of higher education and if the public resources that were generated could be used to finance additional primary school places for those who are currently deuied access. The redirection of subsidies toward primary education would also benefit those from low-income groups. At present, such groups receive very little of the education subsidies, whereas the high-income groups receive the most. In Chile, Colombia, Indonesia, and Malaysia, students from low-income families receive between 6 percent and 10 percent of the subsidies at the uui­ versity level, whereas those in the high-income group receive between 50 percent and 83 percent (see table 7-13). This is because the low-income groups are grossly underrepresented at this level of education. This inequit­ able outcome would be redressed by increasing subsidization at the primary level, where the low-income groups are most widely represented. As noted earlier in the discussion of student selection, any adverse equity effect within higher education might be neutralized through selective scholarships or exempting low-income students from paying fees. In some Latin American countries, uuiversity fees are related to family income; 1:able 7-13. The Share of Higher Education Subsidies by Income Group --­ ... Income group Counlly Low' MidJle High Chile 6 44 50 Colombia 6 35 60 Indonesia 7 10 83 Malaysia 10 38 51 a. The low-income group corresponds to the poorest 40 percent, except in the case of Chile, where it corresponds to the poorest 30 percent. Source: Appendix table D-4. 102 CHANGING PRICING POLICY students from poor families pay lower tuition fees than those from wealthier ones. A means-tested scholarship system would achieve the same result. Private Schools Another way to shift the financing burden to users is to allow private schools to provide more education. In some countries, however, private schools and universities are not allowed to operate. For example, the People's Republic ofthe Congo, Ethiopia, and, more recendy, Nigeria have abolished or have attempted to ban private schools through legislative action (Cowen and Mclean 1984). Lifting such a ban would lead to an expansion in educa­ tion, as in Pakistan, where the nationalization policy of the 1970s has recently been reversed. A more widely practiced form of restriction is the imposition of a set of norms regarding fees, the hiring of teachers, their qualifications and pay, curriculum development, and student selection. Although ostensibly less binding than outright prohibition, such regulations may have a stifling effect on private schools. The extent to which private education is discouraged depends upon the degree of regulation. In Cameroon, for example, the government determines the fee charged by private schools; similarly, in Colombia and Chile, private school fees are controlled by the government (Schiefelbein 1985). Other countries, such as Jordan and Zambia, have simply declared that all primary education must be free. Relaxing overly restrictive regulation, while monitoring to ensure against fraudulent institutions, would lead to educational expansion. This is not a recommendation for a completely private system, which would tend to provide less education than would be optimal. When public budgets are constrained, however, the coexistence of private and public sys­ tems would allow more flexibility in expanding educational resources, es­ pecially if there are constraints on raising fees for public schools. Figure 7-3 depicts the social gain from loosening restrictions in the private sector. The private and social demand curves are, respectively, Dp and Ds. To simplify the presentation, private nonfee costs are not shown in the diagram. Suppose that the long-run marginal cost of private and public schools is the same for both types of schools and constant at c. The effective supply curve for an underfunded public school system would be given by the iso-subsidy line S. If there was a cap on user charges that the public system could levy at p, only qo students could be accommodated at a certain quality level. With a restriction on private schools, this would lead to an effective excess demand of qo qt. But if the restrictions were lifted, the private sector would offer school places to cover costs. Initially, the first private school owners would be EDUCATION 103 Figure 7-3. The Social Gain from Loosening Restrictions in the Private Sector able to charge a price Eqo and might earn "economic profit"-profits in ex­ cess of a competitively determined rate of return. With free entry, however, prices would be bid down to c, where proprietors would recover a reasonable return on their investment. Economic profits would eventually drop to niL The "effective" supply curve in this mixed system would be line segment FC for quantities provided less than qo, and the straight line denoted by Al for quantities provided greater than qo. The social gain from allowing the private sector to operate is DCBA. Notice the following: • An increase in prices ofpublic schools from p to P1 would have the same social benefit as a policy ofeasing the restriction on private schools. The difference would be in the distribution of the benefits. The amount ABq2Qo accrues to the owners of schools rather than to the government or to parents and students in the latter policy. • This policy of allowing a mixed public-private system is dominated by one which raises prices in the public system until effective excess de­ mand is eliminated. A price increase in the public system to pz would result in an additional gain of CHIB relative to free entry in the private market. The reason is that the externalities to schooling are not taken into account by the private system. 104 CHANGING PRICING POLICY Greater reliance on private provision could have negative efficiency and equity effects in terms of access. This can be mitigated to some extent by in­ troducing a more selective entrance procedure in public institutions so that the meritorious needy are given priority. But to avoid a segregated system in which the poor go to public schools (which may be of high quality) and the rich to private schools, subsidies could be distributed to students for use in private schools. Governments subsidize private schools in many developing countries. The subsidized provider is not necessarily less efficient than one that is not sub­ sidized. It is the disposition of the proceeds that is important in determining efficiency. A sector that does not provide incentives for its owners and managers to minimize costs will be less efficient than another that does. A private system has such incentives imbedded in its structure because the returns flow to the owners, who can keep or sell their shares. A public system may not have such incentives since the owners-the general public-cannot transfer their shares in the enterprise and capitalize on any ofthe gains. Thus, subsidies to a private school system may still result in cost minimization. Education subsidies are, for the most part, uniformly distributed to all en­ rolled students because the school is usually the recipient. An alternative is for the government to channel the subsidy to the student rather than to the institution. For example, recipients ofgrants or scholarships would be chosen according to their academic performance and economic need and permitted to attend the schools oftheir choice, whether public or private. This arrange­ ment would broaden parental and student choices, increase institutional ac­ countability, and encourage schools to compete for pupils, thus stimulating diversity and experimentation and improving efficiency. The option of distributing centrally collected subsidies to individual students directly may be particularly beneficial at the levels of higher and even secondary education, where students and their families are likely to have more information about educational alternatives. At lower levels, the administrative costs of distributing scholarships and providing information about schools may be high enough to preclude the possibility of individual distribution of subsidies. An alternative might be to distribute subsidies to local authorities or neighborhood groups according to the economic needs of the groups. For example, communities in rural areas would be provided relatively more subsidies than would neighborhoods in urban areas. Local authorities could then use the centrally provided subsidies, local levies (both monetary and nonmonetary), and user fees to purchase the educational ser­ vices desired by their constituents. The central government's subsidy formula could be structured to encourage local authorities to be more fiscally inde­ pendent. The local group could be rewarded, for example, by a matching EDUCATION 105 grant, ill which case subsidies would be partly based on locally raised funds. Public subsidization of private education is not widespread. In developing countries, individually distributed public subsidies for private education are limited. In Pakistan, for example, the government has recently announced plans to establish scholarship programs at all levels of education for the children of families with limited income, but it did not specify the extent to which scholarship recipients would be allowed to attend private schools. In some Latin American countries, between 5 and 10 percent of the private secondary students receive scholarships provided by their school, usually in the form of exemptions from fees (Schiefelbein 1985). In other countries, such as Colombia, the government exerts pressure on private schools to pro­ vide education to poor students by requiring the schools to increase their scholarships in exchange for official permission to raise fees. Governments can also indirectly subsidize private schools by giving in­ come tax relief on private educational expenditures. In Brazil, for example, tax exemptions for families paying private school fees were equivalent in 1972 to 3 percent of total public educational expenditures (Brodersohn and Sanjuro 1978). The feasibility of Brazil's system for other countries depends on their methods of tax collection. 8 Changing Pricing Policy in Health THE EFFICIENCY AND EQUITY gains from increasing the role of pricing in the provision and allocation of health services, as in the case of education, vary with the type ofservice. The important criteria for determining this role are the service's public goods characteristics and initial market conditions, the responsiveness ofdemand and supply to price changes, and the administrative costs of implementing a pricing and collection scheme. Because it is so varied, health care is generally more difficult to categorize in terms of these criteria than is education. It encompasses individual health care, disease control, and drug provision, as well as programs indirectly related to health such as water supply, sanitation, education, the environ­ ment, and housing. Health care can be provided by delivery systems that range from rural clinics to modern hospitals administered by local, regional, or national authorities. (See appendix table D-5 for an example of these health care services and their providers.) Although the mix varies from coun­ try to country, a useful categorization scheme shown in table 8-1 has been devised by de Ferranti (1985). Curative health care includes "first-contact" services, mostly for out­ patients. The benefits from using these services tend to accrue almost ex­ clusively to individuals. Thus, there is no case for externalities. Moreover, users have adequate knowledge of these benefits. They do not generally need to know more than they learn from personal experience. Guided by the de­ gree of pain or other symptoms, most individuals know when they are ill or injured and when they should seek medical help. Another type of curative health care is "referred," and it usually involves the use of inpatient hospital services. Referral services have two types of public goods characteristics. First, the consumer has less knowledge ofwhat is needed to cure a malady, and thus, in a pay-for-service scheme, suppliers may have the incentive to offer more services than would be warranted. Second, for serious cases the cure, such as a prolonged hospital stay, may be unafford­ able if patients are not adequately covered by health insurance. 106 HEALTH 107 Thble 8-1. Stylized Market Characteristics of Different 1)rpes of Health Services Market Type of send.ee "Public good" characteristic characteristic Curative care (includes patient care Lack of access to insurance Excess demand by health facilities and independent market (inpatient case); in some providers and purchases by users of lack of knowledge countries medIcines) (referral services) "First-contact" services (mostly outpatient) Referral services (inpatient and some outpatient) Preventive care: patient-related Externalities (immuni­ No excess (includes well baby and child zation, perinatal care to demand health care, such as immunization, avoid permanent disability); growth monitoring, and child-care lack of knowledge of instruction) benefits Preventive care: non-patient-related Exclusivity; Cannot be (includes disease control, sanitation, externalities; determined promotion of health and hygiene, lack of knowledge control of pests and wonotic of benefits disease, and monitoring of disease patterns) Source: Adapted from de Ferranti (1985), p. 67. Some patient-related preventive services, such as immunization, also benefit nonusers. Because these services hinder the spread ofdisease to other people, externalities are an important consideration in pricing policy. Extern­ alities may also playa role in perinatal and infant care because without this care the chances of long-term disability may be large. and this may affect the welfare ofothers. For example, if medical services are subsidized, others will have to bear the possible cost of childhood infirmities because of lack of in­ fant care. Non-patient-related preventive care has almost all the characteristics of a public good, including nonexdusivity. These characteristics imply that a greater role for pricing policy is infeasible for non-patient-related preventive care, which accounts for roughly 3 to 10 percent of all health expenditures in developing countries. Although increased prices are feasible for patient­ 108 CHANGING PRICING POLICY related preventive care, which accounts for 10 to 20 percent of total health expenditures, the extent of the efficient increase would be constrained if ex­ ternalities were large. For curative health care, which represents 70 to 87 per­ cent of total health care, increased user charges can be feasible and efficient in the allocation and provision of health services (de Ferranti 1983). This chapter, an application and extension of the earlier analysis of educa­ tion pricing, will discuss how increasing user charges for curative health care influences efficiency and equity. The evidence is not based on ex post evalua­ tion of price increases but on simulations of possible effects. Estimated behavioral models are used. Behavioral Parameters Before the possible effects of pricing on the provision and allocation of health services can be detennined, it is necessary to consider the impact of pricing on demand and supply. Demand To make the best decisions about health care services, policymakers must be able to determine the demand for them. This is very difficult, however, because illnesses occur randomly and in different severities. It is also difficult for policymakers to assess the benefits of health services because placing an explicit value on human life is distasteful. Thus, health care policies are more difficult to analyze than are educational policies because health care informa­ tion is scarce for developing countries, and the analyst must frequently rely upon indirect evidence. Nonetheless, policymakers can work with several in­ dicators to formulate desirable health care policies. For example, most health care policies tend to affect morbidity rates, which can be measured through productivity rates. A healthy work force produces more than does one in which illness keeps workers at home or in the hospital. Policymakers may also look at consumption levels of health services to assess how people per­ ceive the benefits of seeking health care. Household decisions about the consumption ofcurative health services de­ pend upon the random incidence of illness, the cost of the services, and the income, size, and location of the consuming household. Consumption of these services also depends upon how much the household spends on prevention. The few micro economic estimates that have been done indicate that health demand, whether measured by expenditure or utilization, is relatively unre­ sponsive to price. Heller (1975) finds that the fee elasticity ofoutpatient visits HEALTH 109 Table 8-2. Income and Price Elasticltes of the Demand for Health Care in Developing Countries lAtest IncOtlU! Price year Dependent Country of d4Ja variable Measure Elasticity Measure Elastidty El Salvador' 1980 Medical Monthly 0.887 Distance -0.054 expenditure income Malaysia ca. 1975 Private Outpatient Monthly 11.32 Private fee -0.15 visits income Government fee -0.01 Private waiting time -0.25 Private travel time -0.14 Government travel time -0.18 Government Outpatient Monthly -11.43 Private fee 0.15 visits income Government fee -0.01 Private waiting time -0.07 Government travel time 0.10 Private travel time 0.26 Government travel time -0.05 Mali 1982 Willingness Monthly Distance to to pay for a income dispensary -0.0003 health worker Distance to drug outlet -0.0001 Quality of dispensary -0.18 Quality of output 0.04 (Tabk amtit!ues ot! the following page.) 110 CHANGING PRICING POLICY Table 8-2 (continued) lAtest Income Price year Dependent Country of data variable Measure Elasticity Measure Elasticity Philippines ca. 1980 Expenditures Monthly Distance to on drugs Income dispensary -0.002 Distance to drug outlet 0.0009 Quality of dispensary -0.45 Quality of outlet 0.17 - Not available. 3. For Santa Ana only. Sources: World Bank data for El Salvador; Heller (1975) for Malaysia; Birdsall and Chuhan (1983) for Mali; and Akin and others (1982) for Philippines. is significantly less than unity in Malaysia (see table 8-2). Akin and others (1982) conclude that price has no effect at all on outpatient visits in the Philippines. Relatively price-inelastic demand for health care is not surprising because health care is often viewed as a necessity. But this is not necessarily true for specific categories of health care (outpatient visits, for example) since sub­ stitutes, such as private sector services, may be available. The surprising find­ ing is that the cross-price elasticity of these substitutes is small. One possible explanation for the finding oflow elasticity is a measurement problem. Prices may be positively correlated with quality; thus the researcher concerned with measurement will find the dampening effect of prices on demand counter­ balanced as health care becomes relatively more expensive. This problem can be resolved only through careful control of quality in future studies. Finally, elasticity need not be constant. The estimated elasticities are measured in environments where consumers pay very low prices. Price re­ sponsiveness may be higher at higher price levels. It is apparent that con­ siderably more research is needed in this area. Table 8-3 presents only aggregate evidence of the price elasticity of de­ mand for health care. The table shows the demand for personal care, which includes medical expenditures, to be only one component of health care, albeit a major one (Lluch, Powell, and Williams 1977). The time series data for seventeen countries estimate a price elasticity that averages -0.6. For the poorer countries, the price elasticity tends to be somewhat lower-as low as HEALTH 111 Table 8-3. Personal Health Care Expenditures ~~~ .........- ­ Per capita subsistetlU Average share Total expenditure Price expenditure" afbudget elasticity elasticity (1970 U.S. dollars) Australia 0.057 2.34 -0.85 14.1 Germany, Fed. Rep. 0.037 1.21 -0.86 6.1 Greece 0.036 1.35 -0.80 3.3 Ireland 0.013 1.07 -0.30 6.3 Israel 0.066 0.99 -0.24 51.6 Italy 0.063 1.04 -0.62 19.8 Jamaica 0.026 2.35 -1.21 -2.4 Korea, Rep. of 0.042 1.76 -0.23 3.7 Panama 0.047 0.92 -0.52 11.2 Philippines 0.035 1.72 -0.22 3.5 Puerto Rico 0.069 1.70 -1.28 -20.6 South Mrica 0.048 1.02 -0.20 17.4 Sweden 0.037 1.43 -0.93 3.8 Taiwan 0.057 1.69 -0.46 4.9 Thailand 0.056 0.93 -0.46 3.4 United Kingdom 0.023 1.35 -0.78 6.3 United States 0.081 1.69 -0.92 14.9 Mean Valuesb 100-500 1.53 0.34 SOl-1,000 1.10 -0.51 1,001-1,500 1.20 -0.61 1,500+ 1.60 0.87 Overall 1.39 -0.60 a. The estimated parameters ofa Stone-Geary utility function. The numbers are interpreted as the cost of the minimum amount of personal care expenditure that will keep the utility function non­ negative. h. Class intervals refer to GNP per capita at sample midpoints in 1970 U.S. dollars as given in column 5 of table 3-2. Jamaica is excluded. Sour",: Lluch, Powell, and Williams (1977). -0.2. Fifteen ofthe countries exhibit relatively inelastic demand. The results from aggregate data are qualitatively consistent with those from micro-level data-the demand for health care is relatively price inelastic. Supply The total cost of the service will obviously be affected if unit costs are re­ sponsive to any changes in the quantity of the service supplied. The 112 CHANGING PRICING POLICY magnitude of the impact depends upon whether the mix of health services is altered and also upon the presence of possible scale economies. Information on the unit costs of providing health services is limited but usefuL Unit cost varies significantly depending upon the level of service that is offered. Table 8-4 summarizes some of these costs for five countries. In Colombia and Malaysia, an inpatient day costs four to seven times more than an outpatient visit. For Peru, the average inpatient admission costs thirty times more than the average outpatient consultation. The mix of resources allocated to inpatient and outpatient care greatly affects the extent to which government can provide adequate health services. There is no evidence that hospitals in developing countries experience a significant amount of scale economies. A comparison of unit costs of five Malaysian general hospitals of comparable quality indicates no perceptible downward trend in unit costs of inpatient days as the size of the facility in­ creases (see table 8-5). There is even an upward trend with respect to out­ patient visits. Although these comparisons are not strict because the controls for quality or differentials in input prices are rough, they still provide a pre- Thble 8-4. Public Unit Costs by 'JYpe of Medical Service Unit recurrent and capital wsts (1980 Country Tn'" of service U.S. Jolwrs) Botswana Outpatient visit 2-3 Colombia' Outpatient visits 1.91 Deliveries 2.20 Operations 56.23 Inpatient days 7.63 Malaysiah Hospital inpatient 26.71 Hospital outpatient 3.73 Rural clinic unit 4.06 Birth assistance by midwife 44.53 Peru Inpatient (per admission) Hospital 203.10 Health center 126.40 Thailand Outpatient visit 2-3 a. National health system hospitals. b. Figures for peninsula only. Sources: World Bank data for Botswana, Peru, and 'Thailand; Selowsky (1979) for Colombia; and Meer­ man (1979) for Malaysia. HEALTH 113 Thble 8-5. Unit Cost by Size of Hospital in Malaysia (1974 ringgit) Percentage of Outpatient seroices Inpatient services Ictal expenditures (cost per visit) (cost per inpatient Jay) Number of .;penton beds outpatients Recurrent Capital Total Recurrent Capital Total Satellite hospital 41 35.3 1.50 0.98 2.48 17.33 15.54 32.87 75 31.3 3.00 2.42 5.42 15.61 15.54 31.15 81 23.1 4.48 3.28 7.76 20.63 14.25 34.88 200 16.8 1.97 2.27 4.24 13.02 19.64 32.66 206 23.4 1.69 1.47 3.16 13.13 16.22 29.35 244 25.3 3.27 4.27 7.54 10.46 17.99 28.45 260 26.2 3.30 4.20 7.50 13.74 22.61 36.35 General hospital 368 16.1 2.98 4.94 21.96 41.03 789 16.3 3.52 6.29 17.33 36.54 903 25.6 3.25 5.92 17.92 35.03 1,067 16.7 3.75 6.69 2,016 16.8 5.69 8.63 23.70 37.28 Not available. Source: Heller (1975). liminary indication that hospitals may not be natural monopolies that operate on the downward sloping portion of their cost curves. Horton and Claquin (1983) report similar findings for rural health stations and hospitals in Bangladesh (see table 8-6). The long-run average cost per death averted is significantly lower for the smaller health service. Some con­ flicting evidence, however, prevents a firm conclusion regarding the shape of the average cost curve. For example, an analysis of a survey of seventy-five government hospitals in Kenya in 1975 concluded that economies of scale characterized the underlying cost structure ofthe sample facilities in the short run (Anderson 1980). Thus, hospitals are operating on the downward sloping portion of their short-run average cost curves. The study also found that for Kenya it would be more economical to increase output of existing facilities than to create new facilities. 1 1. In the analysis of the impact of marginal fee increases on health services, alternative assumptions about elasticities were used. 114 CHANGING PRICING POLICY 'table 8~. Unit Cost of Health Services in RuraJ Bangladesh (costs in U.S. dollars) Matlab Cost Type 1 Type 2 Ambulance Sotalid Number of users 10,618 10,618 4,359 891 Short-run average variable cost per patient 3.91 3.91 7.98 1.83 Short-run average cost per patient 15.56 16.68 11.47 3.10 Long-run average cost Per patient 15.65 16.77 12.80 3.36 Per severely ill patient 631.04 676.21 178.53 91.59 Cost per death averted 1,262.10 1,352.40 357.06 187.19 Note: Matlab and Sotalci are locations of treatment centers about ten miles apart. The Matlab type 2 dif­ fers from type 1 in having a microbiology laboratory and expatriate supervision and in using an existing puaa building rather than a rented kutcha building. Source: Horton and Claquin (1983). Reprinted with permission; copyright 1983 Pergamon Journals Ltd. Prices and Expansion The efficiency gains from user charges for curative health care services de­ pend, as in education, upon initial market characteristics. Although it is known that public health services are heavily subsidized in most developing countries, there is only anecdotal evidence of excess demand for them in some countries. Some facilities are underutilized (as in Malawi), and some provide very low quality service (as in Mali). Since these conditions could vary considerably from country to country, it is not possible to make general, concrete recommendations about user charges. This section will, instead, dis­ cuss their possible effects on expansion. Excess Demand The analytical and educational chapters have already outlined the con­ ditions-excess demand or low quality-under which it would be beneficial to raise prices. Suppose that the total subsidy level or the budget allocation for health is fixed. Ifa less than optimal amount offunding is being provided, an increase in prices would generate resources that could be used to improve HEALTH 115 the quality of the system or to offer services to more people. Many clinics in Africa simply have no drogs (Birdsall and Chuhan 1983). If user fees were implemented and the revenue were used to purchase drogs, more people would receive better service, and efficiency would be improved. In many developing countries, however, prevailing policies may require that all fees from public health facilities revert to general government accounts. If there is excess demand, moderate increases in user charges will increase rather than decrease the number of patients being served-even if quality is not enhanced to stimulate demand-because revenues can be used to hire more staff. The queue to obtain service will therefore shorten. (Estimates of the fee elasticity of demand indicate that the percentage decline in the queue would be less than the percentage rise in fee.) There are efficiency gains from shortening the queue: the first ones to leave will be those who are not willing to pay the moderate increase in fees. If these are people who are not very ill, then the fee hike will have increased the chance that the more seriously ill will be served. This allocation of resources is more efficient for society. If willingness to pay is a function of the severity of illness and income, then equity considerations arise. These are addressed below. As table 8-7 indicates, a 10 percent increase in fee revenue, if all of it is used to finance expansion of a health service, can be expected to increase the amount of the service by about 1 percent at constant unit cost. This is because the bulk ofthe financing ofpresent health services comes from subsidies. The ability of the service to expand is enhanced if there are scale economies. On average, a 10 percent fee increase could fmance a service expansion of2 per­ cent if the cost elasticity estimate of -0.5 holds. The policy prescription would, ofcourse, be different if there were excess demand and the current level of the service exceeded the socially optimal amount. An example of this would be an ultramodern hospital facility that is costly to ron and maintain per patient served. In this situation, if the total sub­ sidy is fixed at a high level and cannot be reduced in the short ron, fee revenue should not be used to expand the system. Instead it should be saved or unit fees should be lowered and the service rationed. No Excess Demand The effect of increased user charges on revenues depends upon whether the demand for the health service is a constraint. If there is excess demand, for example, then the entire percentage increase in fees could be used to ex­ pand the service. If not, an increase in fees would decrease demand. If de­ mand for the services were relatively price inelastic, then revenues would increase. 116 CHANGING PRICING POLICY Table 8-7. Impact of a 10 Percent Increase in Fees on Qllantity of Public Health Services if Revenues Were Used to Expand Service Percentage increase in health services if aJst elasticity were Ratio of unit fee Country to unit recu"ent cost 0.0 -0.5 -0.7 Mrica Botswana 2.8 0.3 0.6 1.0 Burundi 4.0 0.4 0.9 1.5 Ghana 3.0 0.3 0.6 1.1 Lesotho 6.0 0.6 1.4 2.5 Malawi 3.0 0.3 0.6 1.1 Rwanda 7.0 0.8 1.6 3.0 Sudan 1.4 0.1 0.3 0.5 Togo 6.0 0.6 1.4 2.5 Zimbabwe 2.2 0.2 0.5 0.8 Asia Indonesia 15.5 1.8 4.5 10.7 Malaysia 18.0 2.2 5.6 15.0 Pakistan 2.5 0.3 0.5 0.9 Philippines 6.8 0.7 1.6 2.9 Sri Lanka 0.7 0.1 0.1 0.2 Middle East Jordan 13.2 1.5 3.6 7.9 Tunisia 2.0 0.2 0.4 0.7 Latin America Colombia 28.4 4.0 13.1 177.5 Peru 8.0 0.9 1.9 3.6 Average 7.3 0.9 2.2 13.0 Note: The persistence of excess demand is assumed in these calculations. Source: Calculated from table 2-3, using equation C-IB from appendix C. Suppose that a country maintains the equilibrium level of services through a mixture of subsidies and fee revenues. If the country were committed to meeting all demand and subsidies were cut by 10 percent, what should the fee increase be? Table 8-8 shows that relatively modest increases in fees could compensate for the subsidy reduction. Even if demand had an elasticity of -0.5, a 2.8 percent increase in fees could, on average, finance the 10 percent loss in a subsidy allocation and still maintain equilibrium. The consequent decline in utilization would be about 1 percent. Such a fee increase would have very little effect on average household ex­ penditures. For example, fee payments in Malaysia account for 0.1 percent of HEALTH 117 Table 8-8. Impact of a 10 Percent Decline in Subsidies on Fees, Quantity of Public Health Care, and Household Payments to Minimize Social Losses Percentage decline in Percentage fee increase utiliziltion if fees rise and Unit fee as if demand elasticity were demand elasticity were percentage of Country unit cost -0.5 -1.0 -0.5 -1.0 Mrica Botswana 2.8 3.0 2.6 -1.5 -2.6 Burundi 4.0 2.2 2.0 -1.1 -2.0 Ghana 3.0 2.9 2.5 -1.4 -2.5 Lesotho 6.0 1.5 1.4 -0.8 -1.4 Malawi 3.0 2.9 2.5 -1.4 -2.5 Rwanda 7.0 1.3 1.3 -0.7 -1.3 Sudan 1.4 5.3 4.2 -2.6 -4.2 Togo 6.0 1.5 1.4 -0.8 -1.4 Zimbabwe 2.2 3.7 3.1 -1.9 -3.1 Asia Indonesia 15.5 0.6 0.6 -0.3 -0.6 Malaysia 18.0 0.5 0.5 -0.3 -0.5 Pakistan 2.5 3.3 2.9 -1.7 -2.9 Philippines 6.8 1.4 1.3 -0.7 -1.3 Sri Lanka 0.7 8.3 5.9 -4.2 -5.9 Middle East Jordan 13.2 0.7 0.7 -0.4 -0.7 Tunisia 2.0 4.0 3.3 -2.0 -3.3 Latin America Colombia 28.4 0.3 0.3 -0.2 -0.3 Peru 8.0 1.2 1.1 -0.6 -1.1 Average 7.2 2.5 2.1 -1.2 -2.1 Note: Although there is anecdotal evidence that many of the countries in tables 8-7 and 8-8 confront ex­ cess demand for health services, for illustrative purposes it is assumed in the calculations shown here that each system is in equilibrium. Source: Calculated from table 2-3. using equation C-ll from appendix c. household income. A 19 percent increase in household payments would mean that the proportion of household income for medical services would have to rise to 0.12 percent-an almost imperceptible increase. Even for the lowest­ income households, if demand elasticities were constant across income groups, the percentage of household income spent on publicly provided health care would rise from 0.36 percent to only 0.43 percent. Data for other countries are not available. 118 CHANGING PRICING POLICY The calculations made above assume constant unit costs. It is possible, however, that health services would exhibit increasing returns to scale. In this case lower utilization would result in higher unit costs. Since the evidence on the extent of scale economies in the health services is even less than what is available for education, further calculations of alternative assumptions about costs are not presented. Prices and Efficiency Aside from mobilizing resources to expand services, increased cost recovery could also improve efficiency within the health system. If health services are offered for free across the board, the demand for any service will be very heavy. The providers will be confronted by individuals with a range of maladies, some of which require immediate and sustained attention while others might require little or no care at all. It is difficult to determine who should be given priority access. Individuals would have an incentive to overutilize a free system. Ofcourse, not all who demand immediate attention would get it. Free ser­ vices are frequently rationed to match budgetary restraints. The most com­ monly used rationing device is the queue, which limits service to those who come first. To a certain extent, the cost of waiting is a private cost that may deter some individuals from seeking unnecessary health care. It is an ineffi­ cient allocative device, however, if some services have to be provided ex­ peditiously. Increased cost recovery for curative care would deter those with minor complaints from seeking unnecessary care. With uncertain and large health expenditures, such as hospitalization charges, it would not be efficient to rely solely on fees to finance services ren­ dered. Individuals cannot be expected to save large amounts ofmoney simply because some day they may need to finance a large hospital bill. Many could not mobilize these amounts, and if they could it would be better if they pooled their resources to minimize the risks through insurance schemes. 2 In insurance schemes, individuals pool their resources into a fund from which those who are sick can withdraw. A premium, paid regularly regardless of the occurrence of an illness, is charged to all the participants. Because not everyone will need assistance with medical expenses at the same time, the fund will be able to finance substantial costs, such as those for hospitalization, for some of the scheme's members. 2. The complex topic of insurance and its role in cost recovery is treated only superficially here. For a more comprehensive treatment and possible applications to developing countries, see Besley, Bevan, and Collier (1985). HEALTH 119 Insurance schemes suffer from two inherent drawbacks. First, there is a tendency in insurance markets for people to buy insurance when they know that they are likely to become ill. When these people do actually become ill, their health care is paid by the insurance. As more people become ill, the total cost of providing insurance increases. This causes insurers to increase the prices oftheir premiums. Expensive premiums deter healthy, low-risk people from purchasing insurance. As a result, there may not be enough healthy peo­ ple contributing to the premium pool to pay for the expenses of the sick. Under these conditions, insurers spend more than they receive in revenues from premiums. This problem, called adverse selection, might be resolved through compulsory public health insurance. Private insurance might also be effective if made compulsory. Second, insurance providers, whether public or private, must confront the problem of "moral hazard." The insured who falls ill consumes more health care than is needed. Once the premium is paid, there is nothing to hinder a patient from getting more services than are necessary. A fee for service together with a premium-based financing scheme might lessen this problem by discouraging overutilization. Another type of moral hazard arises when health service providers pre­ scribe more treatment than is required. For example, the private sector may use capital-intensive techniques when simpler ones would do, and prescribe more treatment than is necessary. Overtreatment occurs when uninformed patients fail to recognize that they are receiving unnecessary tests or medicines. It also occurs when insurance systems are heavily subsidized or charge no premiums. Patients then have no incentive to minimize costs, and physicians can prescribe extra services and gain financially, especially if they are paid directly for the services they render. A fee-for-service system would only partially resolve this problem because patients cannot shop for the most cost-effective treatment-they lack information about the performance of various medical services. In addition, physicians' practices should be regu­ larly monitored to control excessive treatments. One way to do this without placing an undue infonnational burden on the user would be through com­ petition among providers in private systems. In public systems, incomes of physicians might be prepaid or based upon how well they minimize costs, as they are in health maintenance organizations. Because the cost of treat­ ments would come out of their own budgets, they would have an incentive to keep costs down. The few reviews of the use of insurance in developing countries show that most medical insurance is provided as a component of social insurance. Higher-income countries in Latin America provide extensive social insurance coverage, while lower-income countries, which account for only 22 percent ofthe region's total population, provide only limited coverage (see table 8-9). Table 8-9. Care under Social Insurance in Sixteen Latin American Countries, ca. 1977 Per capita Equal cost of Percentage of social Health care costs Perce Country Percentage of Domirumt betl£fits medical care insurance financing from b as percentage of GDP tota (in order of population pottern of for social insurance insuran per capita income) covered medical cari' dependents (U.S. dollars) w.,rkers Employers Government Sodal insurance Public health to med Higher-income countries Argentina 80 Indirect Yes 38.0 51.8 0 Brazil 83 Indirect Yes 23 1.4 0.7 Costa Rica 82 Direct Yes 51 25.7 48.5 2.1 3.8 0.6 Mexico 56 Direct Yes Panama 47 Direct Yes 74 29.5 53.2 4.1 3.1 1.5 Uruguay 47 Indirect No 14 28.6 47.5 7.5 0.5 0.9 ..... Venezuela 30 Direct Yes 59 23.9 47.8 15.3 0.7 1.5 ~ Lower-income countries Bolivia 26 Direct Yes 52 32.3 53.9 0 1.3 1.0 Colombia 10 Direct No 49 24.3 58.2 5.4 0.9 0.8 Dominican Republic 4 Direct No 73 16.6 64.4 0 0.4 1.2 Ecuador 5 Direct No 89 0.7 1.3 El Salvador 5 Direct No 52 26.7 63.7 0 0.6 1.4 Guatemala 14 Direct No 25 30.3 55.6 12.6 0.5 0.8 Honduras 7 Direct No 48 0.8 1.6 Paraguay 13 Direct Yes Peru 12 Direct No 36 0.8 1.0 - Not available. a. Direct pattern implies a social insurance fund that operates medical facilities itself. Indirect pattern implies one that merely finances and regulates purchases fro or public providers. b. The sum of these three columns, when subtracted from 100, yields the percentage of social insurance financing from capital and other income. HEALTH 121 Coverage, primarily for urban workers, is financed through wage deductions and subsidies from employers and the government. A few forms of medical insurance are linked to public health facilities in developing countries. Donaldson (undated) evaluated such a scheme in Nepal and concluded that adverse selection was indeed a problem since sicker households tended to enroll in the plan with the highest premium. The two­ part tariff of premium plus fees, however, seemed to work well in controlling utilization. Insurance may be relevant only for middle- and high-income groups because of the high risks in insuring and collecting premiums from the lowest-income groups. The most visible formal insurance schemes are medi­ cal coverage under social security programs and employer plans serving urban workers, but other, more broadly defined varieties of insurance are employer-based schemes serving workers on agricultural estates, community­ financed schemes, and cooperatives. In these smaller bodies, the costs of monitoring risk and collection may be lower than for larger groups. The ability to pool risks, however, declines "lith size. It may be efficient to sub­ sidize insurance targeted to low-income groups. In fact, some analysts (see Culyer 1980) have argued that the public sector should provide the insurance, if, because of scale economies, it incurs lower costs than does the private sec­ tor. Public sector domination, however, would not be the only efficient solu­ tion even if scale economies did exist. Regulated private monopolies are also an alternative. Prices and Equity Present subsidy systems do not lead to equal consumption ofsome types of health services. The main problem is that subsidized health insurance schemes tend to cover primarily urban wage earners in the formal sector. In Colom­ bia, for example, making the system of hospitals in the social security system (which serves employees in the private sector) pay for itself would affect mostly the higher-income groups. Although differential charges among income groups could improve pro­ gressivity, this practice is still not widespread. There is more room for dif­ ferential charges between rural and urban areas. In Colombia and Malaysia, for example, urban areas receive a larger share of the health services. Thbles 8-10 and 8-11 present the distribution of costs and payments for Malaysian and Colombian households. Payments are the average amounts that households of each income group pay for services. These figures also de­ pend upon use of the services by the average household in each income group. The cost figures are the proportion of income that the average household would have to pay if it had to incur the full amount of recurrent 122 CHANGING PRICING POLlCY Thble 8-10. Affordability of Health Care in Malaysia: Payments and Social Costs as Percentage of Household Income by Income Quintile Total ¤t social cost [ruome Frequency Fee social per T:rpe of servia quintile of used payment cost b household' Inpatient days 2.4 0.27 3.48 6.18 2 3.5 0.23 2.89 5.18 3 1.3 0.06 0.78 1.38 4 3.0 0.09 1.22 2.17 5 2.5 0.03 0.45 0.80 Overall 2.5 0.08 1.00 1.78 Public hospital outpatient visits 5.3 0.03 1.20 1.85 2 4.9 0.02 0.64 0.98 3 5.5 0.02 0.50 0.78 4 4.5 0.01 0.28 0.43 5 4.0 0.003 0.11 0.18 Overall 4.8 om 0.29 0.45 Mean monthly household income (ringgit) Combined public medical servicesd 129.3 0.36 5.99 10.40 2 223.4 0.28 4.26 7.39 3 315.5 0.10 1.61 2.99 4 460.8 0.12 1.77 3.17 5 1,037.0 0.10 1.48 2.80 Overall 471.8 0.10 1.48 2.80 a. Number of inpatient days or number of public hospital outpatient visits. b. For inpatient days, the recurrent social cost equals unit current cost times the mean frequency ofuse (mean number of inpatient days per household, for all peninsular Malaysia households), divided by the mean annual household income for each income group. A similar method is used for outpatient visits. c. Mean current cost and mean capital service cost per household. d. Includes the preceding categories plus outpatient visits to rural clinics. Source: Meerman (1979). HEALTH 123 'Dtble 8-11. Affordability of Public Health Services in Colombia Cost of providing National Health Annual household National Health Service's Service rost per inrome hospital service and household as percentage of Income quintile (Colombian pesos) health center per household household income" 1 10,368 514 5.0 2 17,820 440 2.5 3 25,032 393 1.6 4 36,912 321 0.9 5 104,388 210 0.2 Overall 38,904 376 1.0 .. _--­ a. Extra private costs, such as those for transport to the facility, are not included. Source: Selowsky (1979). costs. Mean payments and social costs are lower for the higher-income quin­ tiles because they tend to use public services less frequendy than do lower­ income groups. For Malaysia, payments for combined health services take up an average one-tenth of 1 percent of household income. If the average household were to bear the full burden of the recurrent costs of all publicly provided health services, it would have to devote 1.6 percent of household income. If it were to bear the burden of all costs, it would have to devote almost 3 percent of income. The latter figure ranges from 1 percent for high­ income groups to 11 percent for low-income groups. Ifhouseholds in Colombia had to pay the full cost of the National Health Service. the healthier ones would pay 0.2 percent of their income, the average ones would pay 1.6 percent. and the poorer ones would pay 5 per­ cent. Clearly, fee payments comprise only a small portion of household in­ come, even among the poorest. 9 Feasibility of Policy Change COSTS ARE ASSOCIATED with any policy change. These costs, whether economic or political, can be an effective barrier to implementation. Even if overall gains from any changes outweigh the costs, including those of putting them in place, policymakers must be able to identify potential losers-and to know whether they can effectively block reform. The policy package must include measures to address these issues. The rdative importance of implementation issues will vary among coun­ tries. Thus, it is beyond the scope of this book to detail the steps toward im­ plementation. It is instructive, however, to discuss briefly some general principles. The efficiency gains from increased user charges hinge on the ability of governments to use the revenues thus generated within the social sectors. The ultimate disposition of the funds, however. depends upon those who are given control over them. Funds collected at the school or health unit may have to be turned over to central authorities. In some governments, the funds may have to be surrendered to the central fiscal authorities, which may allo­ cate them outside the sector. The issue is further complicated in federal sys­ tems where the central government may be responsible for one type of ser­ vice, such as higher education, and local authorities for other types, such as primary and secondary education. Thus, it is probably preferable that the funds from user fees be controlled by the authorities that control the social sectors so that they do not have to be first transferred to the fiscal authorities for distribution later. There is very little information on the costs of collecting fees in education or health. Nevertheless, it has been frequently argued that it would be pro­ hibitively expensive to collect charges directly. Since there is a paucity of hard evidence, it is not possible to compute the costs of collection per dollar of revenue. This would certainly be a fruitful area of future research. It is 124 FEASIBILITY OF POLICY CHANGE 125 possible, however, to sperulate on what such an investigation should encompass. Just as there are administrative costs in charging and collecting fees, so are there costs in levying and collecting taxes. If it can be argued that the marginal cost of an additional dollar of tax collected is zero in administrative terms (since taxes are collected anyway), the argument must also hold for an additional dollar of fees collected. More important, however, are the distor­ tionary effects of collecting revenue through taxes. Developing countries rely to a large extent on taxes that drive a wedge between the price that agents observe and the true scarcity value of a commodity. Income taxes are only a small proportion of total government revenue. Other indirect taxes and trade duties may cause efficiency losses and may be regressive. Moreover, because of the uncertainties in the budget-making process, the price ofobtaining a dollar of revenue to cover recurrent costs may be greater than the forgone resources (USAID 1982). These costs have to be weighed with those ofdirect collection from users, and it is not dear beforehand which will be greater. There is, of course, substantial room for improvement in the collection system for user fees, although the record of the education sector in enforcing existing rules is better than that of other social sectors, such as health (Ainsworth 1984). An alternative to central government control is to place more emphasis on localized collection efforts. In this way each community decides how much education or health it will provide and is primarily respon­ sible for paying for it. Since it may be diffirult for central authorities to collect from individuals, the distribution of subsidies will be left up to each community. But communities would, to a larger extent, be self-supporting. There are at least three reasons why this system may be appropriate. First, as already mentioned, it may be cheaper, since the money will have to go through fewer agencies before it is spent. Second, local authorities will have more of an incentive to collect charges if they are certain that the funds will be used locally. Third, users will be more inclined to pay a fee if they believe that it will be used to improve local facilities. These advantages would have to be weighed in each case and would depend upon the resources at the local level for handling financial transactions and the degree of inequality across localities that might result. A frequent argument against increasing user charges for potentially large investments, such as education, has been that household economies in the least developed countries (particularly in subsistence rural areas within them) are not monetized. As a result, it is virtually impossible to mobilize resources from users. The lack of a monetized economy should not, in itself, be a significant barrier to transferring greater responsibility for financing to users. Private 126 CHANGING PRICING POLICY resources may be mobilized in kind. In particular, communities frequently contribute labor to the capital costs and, in some instances, recurrent costs of building and running schools in some African countries. In education, the best-documented examples are the Kenyan harambee schools. As of 1981-82, approximately 20 percent (more than 82,000 students) of secondary school enrollment in Kenya was in unsubsidized harambee schools. Another 20 percent was in assisted harambee schools, where public subsidies accounted on average for only about 18 percent of unit costs per student (Bertrand and Griffm 1983, 42). Thus, some 40 percent of Kenyan secondary school students were enrolled in schools where private assistance was mobilized not only through cash contributions to cover operat­ ing costs, but also through local materials and voluntary labor in school­ building. Schools are organized by local committees that continue to guide their management. The committee determines the type of support that parents will give the school (for instance, by arranging work days for con­ struction tasks). Usually monetary "fines" are imposed in lieu oflabor. The committee also keeps the parents informed about school affairs. One exam­ ple is the Kenyatta High School in an isolated part of the Nyeri district, where 4,000 residents were mobilized mainly to contribute their own labor in September 1965. Material costs were minimized (to about half what the government would have incurred) by making the design simple and by using local materials (Roth 1987, chap. 2). Other innovative financing schemes are based on income-generating ac­ tivities by students and their families. In many elementary schools in Africa, students maintain income-earning farms. In one Rwandan school that has only one plot, eight- and nine-year-old students were able to grow $120 worth of potatoes-six times what the school received in government grants-and the profits were used to purchase equipment (Kulakow, Brace, and Morrill 1978, 15). There is little evidence, however, to determine whether such efforts at school-based production are generally successfuL In the health sector, a review of seventy community-based health schemes worldwide also describes various innovations (Stinson and American Public Health Association 1982). Aside from fees for service, voluntary labor and payments in output (as in China) are commonly used to contribute to capital costs as well as to compensate community health workers. The success of these schemes in maintaining the financial viability of the system depends upon the commitment of the community leaders and their resourcefulness in mobilizing support. The political feasibility of user charges obviously must be considered in adopting policies. Since this is likely to vary widely from country to country, it is up to the practitioner to decide how important each factor is. Great resis­ FEASIBILITY OF POLICY CHANGE 127 tance to the imposition of fees would argue for a gradual introduction. A more important consideration in cutting subsidies is the identity of the main recipients of the subsidies; they are sometimes in a position to decide or in­ fluence such policies. Political acceptability and administrative feasibility can be partially ac­ commodated by phasing in the policy change gradually. The precise sequence and timing would vary from country to country. For the health field, Birdsall (1985) suggests this sequence: (1) introduction of or increase in hospital charges for private patients; (2) greater efforts to recover costs from those with medical insurance, including user charges and premiums; (3) fees levied on patients who bypass an initial examination in countries where a referral system exists; (4) introduction of or increase in other inpatient charges (for meals, visitors, and special services); and (5) introduction of or increase in outpatient fees. If possible, governments should also encourage the growth of insurance markets to cover a wider range of people. Many countries have successfully implemented such schemes. In education, policies to diminish students' living allowance could precede actual increases in tuition. At the same time, measures could be taken to institute student credit schemes (Mingat, Tan, and Hoque 1984). Perhaps the most telling argument for the feasibility of pricing policies to mobilize resources in the social sectors is that nations have begun to imple­ ment them, according to recent studies conducted by the World Bank. For example, in education, university subsidies were reduced in Morocco in 1982 by half, with the exception of those to students from very poor homes. Similar policies are being considered in Ghana. In Malawi and Tanzania, fees were increased at the secondary level. (See World Bank 1986 for a more com­ prehensive list.) Birdsall (1985) documents some recent attempts to raise fee revenue in health systems, including some being financed by the World Bank. For ex­ ample, charges for medicines are being implemented in the Gambia and China; hospital fees are being charged in ten countries (Burundi, Cameroon, China, Ghana, Indonesia, Lesotho, Pakistan, Philippines, Rwanda, and Togo); and outpatients pay consultation fees in Botswana, Burundi, Indonesia, Lesotho, Pakistan, Rwanda, and Zimbabwe. Appendixes A A Review of Alternative Pricing Policies THE ARGUMENTS BELOW assume that prices elsewhere in the economy are given. They are based on partial equilibrium analysis. General Pricing Rules Figure A-I depicts marginal cost pricing if there are no externalities or other market imperfections. For the representative consumer the private marginal benefit (PMB) equals the social marginal benefit (SMB). As a person is satiated, PMB is assumed to decrease with each additional unit. The private marginal cost (PMC) equals the social marginal cost (SMC) and, for presen­ tational purposes, is assumed to be constant with each additional unit. At a price P = PIt 5MB = PMB >PMC = SMC, and consumption at q = ql would be suboptimal, since the value to society of an extra unit of consumption of ql exceeds the cost to society ofproviding it. At a price P = P2, 5MB = PMB < PMC = SMC, and consumption at q = q2 would be "overconsumption," since the value to society ofthe extra unit consumed at q2 is less than the cost ofprovid­ ing the extra unit. Efficiency is reached at P = po, where Po = 5MB = PMB = PMC = SMC for consumption at qo. The efficiency losses at p = Pl can be measured by the extent of the "underconsumption" (the difference between qo and ql) times the amount by which marginal social benefit exceeds marginal social cost at each q. This would be area x. The efficiency losses at p = P2 would, by a similar argument, be equal to area y. The analysis is easily extended if there is more than one individual. The in­ dividual PMB curves are first summed horizontally to form the aggregate PMB. In figure A-2 this aggregate curve is shown by PMBA + PMBB = PMB. Then the analysis proceeds in the same manner as above, where total consumption would be qo. 131 132 APPENDIXES Figure A-I. Marginal Cost Pricing without Externalities p PO~-------+----~----~-------------PMC=SMC I I I - - - - ..+----\-- I I I I I I I I PMB=SMB 'I Figure A-2. Marginal Cost Pricing without Externalities. with Multiple 'Ifpes of Demanders P PMBA + PMBa = PMB 5MB q ALTERNATIVE PRICING POLICIES 133 The pricing rule in the presence ofexternalities depends on their extent. In figure A-3, if externalities imply that the social marginal benefit is 5MBo, the price charged should be P = PI < po, the full cost-recovery price, to ensure a socially optimal level of consumption ql' If user charges were used to fully recover costs, there would be underconsumption by the amount qo ql' The difference between Po and PI would have to be financed by subsidies. If 5MBI applies, prices should be zero to obtain optimal consumption at qz. For any social marginal benefit schedule beyond 5MBl, public authorities would have to pay the consumer a certain amount to induce consumption at an optimal level. Basic Needs The basic needs level of consumption can be interpreted as a minimum amount that society or the policymaker considers desirable. In this case, the optimal level of subsidies can be analyzed as in the externalities case (although the "externalities" in this case are not as tangible and extend only up to a certain amount). A consumer whose private marginal benefit is less than the social marginal cost at the minimum amount of consumption would have to be subsidized by that difference. For other consumers whose optimal Figure A-3. Marginal Cost Pricing with Externalities P Po PMC= SMC I I PI ---I­ I I 5MB 1 I I qo q1 q2 q 134 APPENDIXES Figure A-4. Basic Needs p r-------~~--~~--------------PMC=SMC E q consumption level is beyond the minimum amount prescribed by society, the provision of a similar subsidy (which would occur in any uniform pricing scheme) would lead to overconsumption. Thus, the basic needs argument is one for differential pricing. The only individuals who should be charged a zero price are those for whom the private marginal beneht of another unit of consumption is exactly nil at the minimum amount of consumption that is prescribed for society. Figure A-4 illustrates this case. Let q* = the minimum amount of consumption considered desirable by the policymaker. In effect, this implies that the 5MB is line segment ACD for an individual whose private marginal beneht is PMB = line segment FCD. In­ dividual A would have to be subsidized an amount POPI to induce the minimum consumption q*. Individual B, whose PMB = line segment GBE, would not have to be subsidized at all. A subsidy of the same amount would imply an overconsumption by the amount qo ql for individual B. Distorted Markets There may be distortions in input markets. In figure A-5, if inputs are un­ derpriced, providers incur marginal costs that are less than the marginal cost to society of providing the service (SMC = SMC 1 > PMC). In this case, prices (p =PI) would be charged so that providers would run a financial profit. If in­ puts are overpriced, SMC = SMC2 < PMC and P - P2' There would have to be a subsidy Po P2' ALTERNATIVE PRICING POLICIES 135 Figure A-S. Distortions in Input Markets P PI r----~-----------SMCI ~~ ___ ~ __ ~ _ _ _ _ _ _ _ _ _ _ PMC P21----+---I---~~-----SMC2 q Alternatively, there could be distortions in output markets. If university graduates, for example, were paid a wage that exceeded the value of their marginal product, then the benefit to society from another graduate would be less than the private benefit. In this case, the provider should charge PI and run a financial profit to induce an optimal level ofconsumption ql (see figure A-6). Figure A-6. Distortions in Output Markets P ~I----~--~---------PMC SMC PMl! q 136 APPENDIXES Figure A-7. Pricing of Nonexclusive Goods p ~------~--~---T------------PMC=SMC q Pure Public Goods Pure public goods imply that the consumption by one individual does not exclude the consumption by another. Thus, the total willingness to pay for any unit of the commodity is the vertical sum of individuals' willingness to pay (such as PMBA and PMBB in figure A-7). Optimal price will differ by in­ dividual (PB and PA)' Economies of Scale A monopoly in the short run may be stuck with excess capacity. so that its operations are on the downward sloping portion of its average cost schedule. This is shown in figure A-B. The optimal policy in this case would be to en­ sure that 5MB = SMC at qo. That would imply that variable costs are subsidized by $1 and fixed costs are fully subsidized by the amount $2' The application of this model is described below. The total cost of providing social services is made up of two components: fixed cost, which does not vary with the amount of the service that is pro­ duced, and variable cost, which does. The fixed cost component might in­ clude the capital cost of building a facility. such as a school. Once the school is built, the fixed cost will not be altered no matter how many students are educated. The variable cost is the cost of running and maintaining the school ALTERNATIVE PRICING POLICIES 137 Figure A-S. Pricing When Capital Is Indivisible and Average Costs Are Decreasing p '10 'I and varies with the size of the school. To educate more students at a given quality, more labor and maintenance expenditures must be made. These con­ siderations imply that, in the short run, average costs are V-shaped. In other words, for low levels ofoutput, average costs are declining because of the ef­ fect ofdecreasing average fixed costs; at higher levels ofoutput, average costs are rising because more output is being squeezed out while at least some in­ puts (capital) are held fixed. For a school with low enrollment, average costs per student fall and then eventually rise as the level of enrollment increases. If the school were a publicly owned monopoly that attempted to provide the socially optimal number ofplaces and ifit faced a low enough demand for its places, the school might be operating on the downward sloping portion of its average cost curve in the short run. Such a situation is shown in figure A­ 8, where the hypothetical school faces a short-run average cost of SAC and an average variable cost of SAVC. Marginal cost is labeled SMC. If long-run average costs (LAC, not shown in the figure) were constant and tangent to SAC at B, the school would be suffering from overcapacity since it could prob­ ably do with a smaller capital stock. 1 Given that overcapacity because of, say, 1. The SAC of such a (smaller) school would be tangent to long-run average cost where LAC intersects PMB. 138 APPENDIXES indivisibilities, the optimal pricing solution would be to fully subsidize the average fixed cost (by 52) and charge po only to recover a portion of the variable cost (Saunders and Warford 1976; de Ferranti 1983). The oveull average subsidy would be 51 + 52 in figure A-8. These assumptions would be more likely to hold for facilities where fixed costs were relatively large, such as hospitals and urban schools. The above considerations apply ifproviders can minimize costs in the short run but not in the long run. But the same arguments hold even under long­ run cost minimization if there are economies of scale in the long run. As the system expands and more schools or health facilities are built, long-run average costs decline. Although such a scenario might not be true in the rural areas (where long-run average costs may increase because of the difficulty in serving a dispersed population), it probably is in urban areas. An expanding system may be able to obtain discounts in unit costs of inputs. In addition, once land was bought, newer multistory buildings would be established on the same site and thus lead to expansion without the costs of negotiating new land leases. The appropriate pricing strategy in this case would be to subsidize costs equal to the difference between long-run average costs and marginal costs. A full recovery policy would lead to underconsumption. An alternative that does not require a subsidy is a two-part pricing scheme. The first part is a lump sum periodic payment for the portion of costs above marginal costs that would have to be recovered. This lump sum payment is presumed to be insufficient to deter consumption, which is determined by the second part of the tariff, a user charge that is based on marginal cost. B Pricing Policies under Budgetary Restraints THIS APPENDIX DERIVES the optimal allocation when some parameters are fixed at levels that make the first-best infeasible. Fixed Subsidy Allocations Suppose that central governments set the total subsidy level (budget alloca­ tion) at S. To simplify the analysis, it is assumed that average social cost is constant at c (see figure B-1). This cost is the sum ofprivately incurred costs cp and the cost to the provider ofthe service. Dp andD, are the marginal willing­ ness to pay schedules for the user and the society, respectively. Thobani's (1983) iso-subsidy locus, which is the locus of combinations of unit subsidies and quantities that exhaust 5, is the curve So. Thus, Solc = qo is the amount of the service that would be provided if the service were free. The optimal sub­ sidy level is at 5*, which is the level ofsubsidy that would just finance the dif­ ference between the total cost of providing the optimal level of output, q*, and the amount that can be recovered through fee payments. As Thobani shows, for a subsidy that is 51. which is less than 5·, the price p*. which would have been optimal for the allocation 5·. would entail a loss of area ADE and excess demand qo q*. Under the tighter 51 budgetary con­ straint, the optimal policy would be to raise prices to P2 and use the additional revenue to expand the service until the market dears at ql' This would imply a smaller social loss at BFE. But if the initial price level is above Pl. at a sub­ sidy level 51> this will cause an oversupply of the service, even though the level of subsidy is "too small." The amount actually consumed would be less than the optimal amount. Thus, prices should be lowered and demand stimulated until equilibrium is reached. If the subsidy level is greater than the optimal level (say, at 52)' any price equal to or above the first-best price would result in excess supply. Any ex­ 139 140 APPENDIXES Figure B-1. Pricing Policies under Budgetary Constraints p Pl Po=P* ~~----~--~~~--r-----~------- qo q cess supply beyond the optimal level of consumption is wasteful, since at a lower price more benefits could be reaped with increased demand and with fewer resources. Equilibrium in this case would be reached beyond the first­ best consumption level, q*. Society could do better, however, by providing fewer services at a lower price even though excess demand will be caused. The optimal price level will be at the first-best consumption level but at a low enough price to use up all of the excess subsidy. (The alternative is to run a budget surplus.) Excess demand of q*q2 is optimal! This occurs precisely because the amount of the subsidy is too large. If there is to be rationing, it is crucial that the policy allocate places to the ones who value it the most­ those with the highest rate of return in education and those with the greatest need in health. Fixed Prices Sometimes political constraints require governments to set prices at a cer­ tain level. In this case, the appropriate subsidy policy to follow depends on the fixed level of prices. If p = p*, then subsidies should be adjusted until equilibrium is reached. If there is excess demand at S = SI> subsidies should be increased until the market clears at S = S1 and q*. If there is excess supply at S2' subsidies should be cut until S = S* and q = q*. The story differs if prices are fixed at p =1= p*. If P =P1 > p*, then subsidies should be adjusted until market clearing is reached. For example, at PI> a sub­ PRICING POLICIES UNDER BUDGETARY RESTRAINTS 141 sidy allocation S* would be too large and imply excess supply if all funds from fee payments and the subsidy were spent on q. Thus, S* should be lowered to S1' If P = P2 < p*, subsidies should be adjusted only until q* is provided (at S = Sf). This implies that, even if excess demand were ob­ served, subsidies should exceed S2' Excess demand of q*q2 would be optimal. If p were set at a ceiling p. rather than a level at which prices were fixed, < the analysis would not gready change. If p p*, the conclusion described above holds. If the maximum price were set above p*, then the appropriate subsidy policy would revert to that of obtaining the first-best solution, which is now feasible. c The Algebra of User Fees Impact of Changes in Subsidy Allocations on Fees TIus APPENDIX DERIVES a simple (partial equilibrium) system for calculating the impact of a cutback in government subsidies on private payments. It is assumed that the total amount of financial resources (R) to finance a certain consumption flow of social services is the sum of government subsidy allocations from general revenues (S) and private payments (H): (C-l) R = S + H. These revenues are spent on expenditures (E) that are equal to the unit cost (c) of providing the social services times the amount of social service (q) which is measured by the flow of homogeneous units (such as student years in the case of education) at a given quality level, z: E = cq, Unit costs are, in general, a function of q and z. The exact nature of this relationship depends upon the existence of scale economies. For costs to equal revenue, (C-2) R = E = qc(q, z). Government contributions, S, are assumed to be a control variable, the level of which is determined outside the system. Household payments are assumed to be equal to fee payments: (C-3) H = pq" where p is the fee needed to buy a unit of social service of a,'ven quality. The demand for the qualitatively homogeneous social service, , is a function of household income (.r), fee payments, and the quality (z) of the service that 142 THE ALGEBRA OF USER FEES 143 is provided. The latter variable is meant to capture some of the nonfee costs, such as transport, that need to be incurred in consumption: (C-4) To simplify the system, assume that the impact of quality on unit cost is zero and then substitute equations (C-2) and (C-3) into (C-l) to form (C-S) At equilibrium, supply equals demand, so that (C-6) and equations (C-4), (C-S), and (C-6) can be solved for three unknowns: q, C/' and p. Equilibrium to Equilibrium One interesting issue is to determine the impact of change in S on the amount of fee changes needed to maintain a certain level of quality, under the assumption that the system is in equilibrium before and after the subsidy change. In figure B-1 this is equivalent to estimating the change from po to PI ifthe subsidy provided by the government were to fall from So to S l' Equilib­ rium requires that the system move from points G to H. To measure responsiveness, equations C-4 and C-6 are substituted into equation C-S and the modified equation is then completely differentiated to form the system (C-7) ] dp = dS - [c - p + q(oc/oq)] (oq/oy) dy - [c - p + q (ac/Oq)] (ac/az) dz where] = [q(ac/aq) + c- p](aq/ap) - q = [(ec + s) 11- p](q/p); s c - p = the unit subsidy provided by the government; 11 == fJ1/q)(aq/ap) == the fee el­ asticity of demand for the social service; and e == (q/c)(ac/aq) == the elasticity of unit cost with respect to output. 1 As special cases, two assumptions are made about e, and these are discussed below. The results are summarized in table C-1. 1. The relationship between total cost elasticity ec and average cost elasticity e is straightforward. By definition, c(q) = C(q)/q. Thus, iJc/aq = q[ac/aq - CJ/tj. This implies that e = (¥c)(ac/aq) = (¥C) (aClaq) - 1 = ec 1. Most cost function studies estimate ec· 144 APPENDIXES Table C-1. The Impact of a Change in Government Transfers on Fees, Costs, and Household Payments under Equilibrium The impact of a change in S on percentage change in Household Assumptions Fees } 11P P +R z = z(P), i *0 (8 - ey) I {[(eels) + 1] (11 + ye) - } (11 + yelp p +R Constant unit costs (e = 0) z = y= 0 [1 I (11 <1»] 8 11P p +R z=z(P),i *0 [1 I (11 + ye <1»][8 - ei] (11 + yelp p +R Note: p = unit fees y = household income e unit cost s= e p C = total cost 4> pis R = total revenue 11 (PIq) (aq I a p) = fee elasticity of demand If = household payments Y (zlq) (aqI az) quahty elasticity of demand S = government payments e = (Ph) (az / a p) = fee elasticity of quahty ;; percentage change in x e = (qlc) (ac / aq) = unit cost elasticity q units of the service provided E = (y/q) (aq / ay) = income elasticity of demand z quality of q The first case assumes that there are no scale economies. Thus, long-run average costs are constant. If c were constant, then ocloq = 0 implies that e = O. With c, y, z held fixed, equation C-7 can be solved for the impact on fees of a cutback in S: dp Ipl[(sTJ p)q]} dS. With the appropriate substitutions and after the multiplication of both sides of the equation by (Sip), the changes can be written in percentage change terms as (C-8) p= [1/{11 - <1»]5 where the dot ( . ) signifies percentage changes and == pis == the price-to­ unit subsidy ratio, which is assumed to be stable. For 11 = -0.15 and a fee-to­ subsidy ratio of = 0.1, a 10 percent cut in government expenditures would be financed by a fee increase of [1/(-0.15 -0.1)] (-10) = 40 percent. The change in total revenue caused by the percent change is, from dif­ ferentiation of equation C-2, under the assumption that c is fixed: THE ALGEBRA OF USER FEES 145 (C-9) With the above numbers, a fee increase of 40 percent will cause a change in the total cost of providing the social service of (-0.15) (40) = -0.6 percent. The impact on household payments will be, from differentiation of equa­ tion C-3, . . (C-10) H = (p + R) = (1 + TJ)p. With the above numbers, household fee payments will change by (40 - 6) = 34 percent. The preceding formulas hold under the assumption that, when govern­ ment revenues are cut, other parameters that affect the demand, such as y and z, do not change. These conditions, of course, do not need to hold. In par­ ticular, Birdsall (1983a) argues that quality is affected by fees. The exact specification of this relationship z = z(P) cannot be determined without assumptions about government behavior. But if z were endogenous in the system (equation C-S), it can be shown that the determinant is ~ = s(iJcvop) + s(oqloz)(oz/op) - q (STJ + sy9 - p)(qlp), where'Y == (oqloz(zlq)== quality elasticity of demand and e (oz/op)(P/z) == the percentage increase in quality as a result of a percentage increase in fee payments. Assume for the moment that 9 1. Then, dp {pl[s(TJ + y) - p]q} dS. After appropriate substitutions, (C-ll) i = [S 1(11 + Y - <1»]. If income were to change, for reasons completely external to the system, simultaneously with the change in the availability of government funds, (C-12) p= (S - ey)/(ll + 'Y - <1» where e == (oqloy) (ylq) == the income elasticity of demand. The second case assumes variable costs. Schools and health services may face V-shaped long-run average cost curves because of scale economies and diseconomies. These services may operate on the downward sloping portion of these curves. If so (ocloq) < o. The amount of fee increase needed to finance a drop in government transfers of S is, in percentage terms, (C-13) p = S I {[(eels) + l]TJ }. 146 APPENDIXES This implies the change in revenue, k TJp. Thus, household payments are (C-14) Household Expenditures The effect of a change in p on the share of a social service in total household expenditures can be calculated by differentiatingHT = (H + Ho)/y where H = fee payments and Ho other expenditures on the social service. r This yields HT = [H/(H + Ho)]H - if iI 0 = O. The coefficient of iI is the share of fee payments in total household expenditures on the social service. Impact of a Fee Increase under a Fixed Subsidy Allocation with Excess Demand The general formula for a subsidy constraint is (C-1S) [c(q) p]q = s. This simply states that the difference between total provider-borne cost [qc(q)] and revenues received from user charges (pq) must be made by a sub­ sidy allocation (8). To determine the impact of a marginal fee (p) increase on the quantity provided (q), with 8 held fixed, the equation is totally differentiated: (C-16) [c(q) pJdq + q [c'dq - dp] = O. The resulting equation is then rearranged as follows: ( C-17) (c - p + qc')dq = qdp. Then, this can be solved for the following expression: dq/dp = q/(c(e + 1) - p] (C-18) q' = p' /[(c/p)(e + 1) - 1] when e = c'(q/c) = unit cost elasticity of quantity expansion. If e 0, the ex­ pression is equal to p/(c - p), or the ratio of the fee to unit subsidy, times the change in price. D Tables 'DIble D-l. Cost-Recovery Ratios in Education, 1980 Government Social" (user fees as percentage of unit (private cost as percentage of public cost) total social cost) Count'}' Primary Seamdary Higher Primary Secondary Higher East Africa Botswana 0.0 2.7 0.0 0.0 17.6 0.0 Burundi 0.0 6.3 14.8 0.0 18.8 21.6 Kenya 4.0 43.7 0.0 18.0 31.2 0.0 Lesotho 9.0 42.1 5.0 19.7 30.7 18.3 Malawi 37.0 38.0 1.0 29.0 29.3 17.0 Swaziland 7.0 26.3 19.0 25.4 Uganda 27.0 24.3 25.7 24.8 Zambia 3.0 11.6 17.7 20.5 Zimbabwe 0.0 5.0 0.0 18.3 West Mrica Burkina Faso 13.0 0.0 0.0 21.0 0 0.0 Central African Rep. 2.5 2.7 17.5 17.6 Guinea 0.0 0.0 0.0 0.0 0.0 0.0 Mauritania 0.0 0.0 0.0 0.0 0.0 0.0 Nigeria 30.0 39.0 12.4 26.7 29.7 20.8 Sierra Leone 1.5 20.3 17.2 23.4 Togo 13.0 5.0 21.0 18.3 Asia India 2.0 18.5 29.1 17.3 22.8 26.4 Indonesia 0.0 9.0 13.0 0 19.7 21.0 Korea 3.7 41.2 23.4 17.9 30.4 24.5 (Table continues on the following page.) 147 148 APPENDIXES Table D-1 (continued) Gwernment Social' (user fees as percentage of unit (private cost as percentage of public cost) total sodal cost) Country Primary Secomlary Higher Primary SeconJary Higher Malaysia 5.0 5.0 5.8 18.3 18.3 18.6 Pakistan 1.2 1.8 2.1 17.1 17.3 17.4 Philippines 3.7 17.9 Solomon Islands 0.0 25.0 0.0 0.0 25.0 0.0 Thailand 0.0 12.5 6.9 0.0 20.8 19.0 Turkey 0.0 0.0 15.0 0.0 21.7 Larin America Bolivia 0.8 0.4 1.0 16.9 16.8 17.0 Brazil 5.0 18.3 Chile 1.6 0.9 25.0 17.2 17.0 25.0 Colombia 3.4 17.8 Costa Rica 0.3 0.5 8.0 16.8 16.8 19.3 Dominican Republic 0.0 0.0 1.0 0.0 0.0 17.0 Ecuador 0.0 0.0 2.0 0.0 0.0 17.3 Guatemala 10.0 20.0 Haiti 6.8 3.4 18.9 17.8 Honduras 0.0 9.6 10.0 0.0 19.9 20.0 Paraguay 4.1 2.0 0.7 18.0 17.3 16.9 Uruguay 0.5 0.4 5.0 16.8 16.8 18.3 ~ Not available. a. Calculated from government columns using equation 2-6 in text. Sources; Calculated from Wolff (1984) for East Africa; Ainsworth (1984), World Bank data, Tilak and Varghese (1985), Benson and Harbison (1985) for West Africa; Schiefelbein (1985) for Asia and Latin America, except Gomez (1984) for Colombia and Ainsworth (1984) for Bolivia and Haiti. TABLES 149 Table D-2. The Components of Public Educational Recurrent Expenditure (percent) Country Year Salaries Materials Scholarship Other Bolivia 1979 87.4 0.3 12.3 Botswana 1979 57.6 • 14.2 28.3 Burkina Paso 1979 51.5 10.3 37.7 0.5 Bunmdi 1979 63.0 0.4 17.5 4.7 Central African Rep. 1978 73.0 4.8 17.5 4.7 Guinea 1979 62.8 • • 37.2 Haiti 1979 88.7 • 0.6 10.7 Indonesia Kenya Korea, Rep. of 1977 92.9 4.4 0.4 3.0 Lesotho Malawi 1979 72.2 4.1 23.6 Mauritania Mauritius 1979 73.8 0.2 • 26.0 Sierra Leone Swaziland 1977 73.7 27.3 Tanzania 1979 35.2 15.9 13.2 35.7 Togo 1979 57.8 5.0 10.6 26.6 Uganda 1975 60.8 26.0 4.8 8.5 - Not available. • Incorporated in salaries. Source: Unesco, 1982 Statistical YearVook. table 4.2. Table 0-3. Social Returns to Higher Education by Subject (percent) Sodnl Country Agriculture Engineering Sciences Medicine scietlCes Humanities Economics Arts Developing countries Brazil 5.2 17.3 11.9 16.1 India 16.6 12.7 Iran 13.8 18.2 14.2 15.3 18.5 Malaysia 9.8 13.4 12.4 Philippines 3.0 10.3 to.5 ..... :s Developed countries Belgium 8.0 11.5 9.5 Canada 2.0 9.0 Denmark 8.0 5.0 9.0 France 12.3 16.5 Norway 2.2 8.7 6.2 3.1 8.9 4.3 Sweden 7.5 13.0 9.0 United Kingdom 11.4 11.0 13.0 13.5 Developing countries average 8.0 15.2 14.2 12.2 n.a. 14.0 15.0 Developed countries average 2.2 7.5 9.4 8.2 13.0 n.a. 10.3 8.9 Not available. Source: Psacharopoulos (1982). table 9. Reprinted with permission. TABLES 151 Table D-4. Share of Educational Subsidies by Income Group (percent) Country Survey year Type of subsidy Income group Poorest 30% Middle 30% Upper 40% Chile 1983 Preprimary 50 35 15 Primary 53 29 18 Secondary 37 35 28 University 15 24 61 All levels 39 29 32 Poorest 40% Middle 40% Upper 20% Colombia 1979 Primary 59 36 6 Secondary 39 46 16 University 6 35 60 All levels 40 39 21 Poorest 40% Middle 30% Upper 30% Indonesia 1978 Primary 51 27 22 Junior secondary 45 21 33 Senior secondary 22 23 55 University 7 10 83 All levels 46 25 29 Malaysia 1974 Primary 50 40 9 Secondary 38 43 18 Postsecondary 10 38 51 All levels 41 41 18 Note: All rows total 100 percent except for rounding. Sources; Castaiieda (1984) for Chile; Selowsky (1979) for Colombia; Meesook (1984) for Indonesia; and Meerman (1979) for Malaysia. Thble D-S. The Stmcture of the Health Sector in Developing Countries: 1fpes of Services and Providers Central government Regional, Local government Industrial and Public works provincial, and agricultural Private Pr Sodal and other and district community organizations enterprises, volunl4ry pract Health St'curity Other support government parastatal organizations, mod Health services ministry agency agellCks' agencies agencies Municipalities Villages and private missions, and other trad Personal care ofpatients . Health facilities (hospitals• centers. physicians - offices) .~ .... Outpatient General .. . .. . . . .. Maternal and child health . . .. . .. .. .. Family planning . .. .. .. .. . .. Inpatient General (bed and nursing) .. . .. . .. .. .. Special services (deliveries, surgery. nutrition reha­ bilitation) .. . .. .. .. .. . Village health care Preventive .. .. . . Curative .. . . .. Disease control programs Vector control (spraying for malaria mosquitos) ,. ,. ,. ,. Population prophylaxis (mobile teams immunize or de para­ sitize whole villages) ,. • • .. Environmental inter­ vention (removing vegetation from stagnant waterways to control schistosomiasis) • . .. . • - ~ Drug salesb • • .. . .. .. .. Other programs Sanitation Human waste disposal . . . . • General sewerage . . • . Inspection (of food purveyors and processors) • . .. . Education and promotion of health and hygiene Institutions (schools) • • • • Media (radio, posters) • .. . i 'Dlble D-5 (continued) Central government l: Regicnal, LJcal government Industrial and I Public works provincial, and agricultural Private P Social and other and district community organizations enterprises, voluntary prac Health security Other support government parastatal organizations, mod i Health services ministry agency agencies· agencies agencies Municipalities Villages and private missions, and other tra I l ! Control of pests and zoonotic diseases } • • • Domesticated animals * I ! 1 l - ~ Wild animals Control of pollution Air • • • • • * Water (from industrial sources) * • • Monitoring of communicable diseases • • * * • • Service is offered by providers in this category. - Service not offered. Note: Examples in parentheses are merely illustrative and not a complete list of possibilities. a. For example, defense ministries that provide health services for military personnel; also police agencies, prisons, and mental institutions that provide services f ployees and inmates. b. Drug sales to individuals by private or public pharmacies, excluding drugs sold or provided free as part of services under "Personal care of patients" ab Source: de Ferranti (1985), table 1. References Ainsworth, Martha. 1983. "The Demand for Health and Schooling in Mali: Results of the Community and Service Provider Survey." World Bank, Country Policy Department, Dis­ cussion Paper 1983-7. Washington, D.C., March. _ _ _. 1984. "User Charges for Cost Recovery in the Social Sectors: Current Practices." World Bank, Country Policy Department, Discussion Paper 1984-6. Washington, D.C., March. - Ainsworth, Martha, Fran~ois Orivel, and Punarn Chuhan. 1983. "Cost Recovery for Health and Water Projects in Rural Mali: Household Ability to Pay and Organizational Capacity of Villages." In Three Studies orz Cost Recovery jrz Social Sector Projects. World Bank, Country Policy Department, Discussion Paper 1983-8. Washington, D.C., July. Akin, John, Charles Griffin, David Guilkey, and Barry Popkin. 1982. "The Demand for Primary Health Care in the Bieol Region of the Philippines." Paper presented at the Natlonal Council for International Health Conference, Washington, D.C., June 14-16. Anderson, D. L. 1980. "A Statistical Cost Function Study of Public General Hospitals in Kenya." Journal of Developirzg Areas 14:223-35. Armitage, Jane, and Richard Sabot. Forthcoming. "Efficiency and Equity Implications of Sub­ sidies of Secondary Education in Kenya." In David Newbery and Nicholas Stern, eds., The Theory of Taxatiorz for Developirzg Courztries. New York: Oxford University Press. Behrman, Jere, and Nancy Birdsall. 1983. "The Quality of Schooling: Quantity Alone is Mis­ leading." Americarz EcoMm;, Review 73:926-96. Benson, Charles, and Ralph Harbison. 1985. "Nigeria-Educatiorz Irzvestment Review. " World Bank, Education and Training Department. Washington, D.C. Preliminary draft. Berry, R. A. 1980. "Education, Income Productivity and Urban Poverty." In Timothy King. ed., Education arzd Irzcome. World Bank StafTWorking Paper 402. Washington, D.C. Bertrand, Trent, and Robert Griffin. 1983. "The Economics of Financing Education: A Case Study ofKenya ... World Bank. Country Policy Department, Discussion Paper. Washington. D.C.• December. Besley, T. J., D. L. Bevan, and Paul Collier. 1985. "Health Insurance and Cost Recovery in 155 156 REFERENCES Developing Countries." World Bank, Population, Health, and Nutrition Department. Washington, D.C., October. Bird, R. M. 1976. Chargingfor Public Services: A New Look at an Old Idea. Canadian Tax Papers 59. Toronto: Canadian Tax Foundation, December. Birdsall, Nancy. 1980. "The Cost of Siblings: Child Schooling in Urban Colombia." Research in Population Economics 2:115-50. - - - ' 1983a. "Demand for Primary Schooling in Rural Mali: Should User Fees Be In­ creased?" In Three Studies on Cost Recovery in Social Sector Projects. World Bank, Country Policy Department, Discussion Paper 1983-8. Washington, D.C., July. - - - ' 1983b. "Strategies for Analyzing Effects of User Charges in the Social Sectors." World Bank, Country Policy Department, Discussion Paper 1983-9. Washington, D.C., September. _ _ _. 1985. "Cost Recovery in Health and Education: Bank Policy and Operations." World Bank, Population, Health, and Nutrition Department. Washington, D.C. Processed. Birdsall, Nancy, and Punam Chuhan. 1983. "Willingness to Pay for Health and Water in Rural Mali: Do WTP Questions Work?" In Three Studies on Cost Recovery in Social Sector Projects. World Bank, Country Policy Department, Discussion Paper 1983-8. Washington, D.C., July. Blomqvist, A. G. 1979. The Health Care Business: International Evidence on Private versus Public Health Care Systems. Vancouver: Fraser Institute. - - - ' 1982. "Education, Unemployment and Government Job Creation for Graduates in IDCs." In T. E. Barker, A. S. Downes, and J. A. Sackey, eds., Perspectives on Economic Development: Essays in Honour of W. Arthur Lewis. Washington, D.C.: University Press of America. Briones, G. 1983. "La Distribucion de la Educacion in el Modelo de Economia Neo-Liberal: 1974-1982." Santiago, Chile: Programa Interdisciplinario de Investigaciones en Educacion (PilE), Academia de Humanismo Cristiano. July. Brooersohn, Mario S., and Maria Ester Sanjuro. 1978. "Seminario sobre Financiamiento de la Educacion en America Latina." In Mario S. Brodersohn and Maria Ester Sanjuro, eds., Financiamiento de la EJucacion en America Latina. Mexico City: Fondo de Cultura Economica and Banco Interamencano de Desarrollo (Inter-American Development Bank). Cameron, John, and Paul Hurst. 1983. International Handbook of Education Systems: Sub-Saharan Africa, North Africa, and the Middle East, vol. 2. New York: Wiley. Carnoy, Martin, Henry Levin, Reginald Nugent, Suleman Sumra, Carlos Torres, and Jeff Un­ sicker. 1982. "The Political Economy of Financing Education in Developing Countries." In Financing Educational Development, conference proceedings, IDRC-205e. Ottawa: In­ ternational Development Research Centre. May. Castaneda, Tarsicio. 1984. "La Evolucion del Gasto Social en Chile y su Impacto Redis­ tributivo." University of Chile, Department of Economics. December. Processed. Cochrane, Susan H. 1979. Fertility and Education: Mat Do We Really Know? Baltimore, Md.: Johns Hopkins University Press. Cochrane, Susan H., Donald J. O'Hara, and Joanne Leslie. 1980. The Effects of Education on Health. World Bank Staff Working Paper 405. Washington, D.C. REFERENCES 157 Cowen, Robert, and Martin Mclean. 1984. International Handbook of Education Systems: Asia, Australasia, and lAtin America, vol. 3. New York: Wiley. Culyer, A. J. 1980. The Political Economy of Social Policy. Oxford, Eng.: Martin Robertson. de Ferranti, David. 1983. "Health Sector Financing and Expenditure in Developing Coun­ tries: Current Issues." World Bank, Population, Health, and Nutrition Department. Washington, D.C., February 8. Draft. - __. 1985. Payingfor Health Services in Developing Countries: An Overview. World Bank Staff Working Paper 721. Washington, D.C. de Tray, Dennis. 1984. "Schooling in Malaysia: Historical Trends and Recent Enrollments." Rand Note N-2011-AID. Santa Monica, Calif.: Rand Corp. October. de Wulf, Luc. 1975. "Fiscal Incidence Studies in Developing Countries." IMF St4f Papers 22:61-131. Donaldson, D. S. n.d. "An Analysis of Health Post-Based Insurance Schemes in the Lalit per District, Nepal." USAID Paper. Washington, D.C.: U.S. Agency for International Develop­ ment. Processed. Dougherty, Christopher, and George Psacharopoulos. 1977. "Measuring the Cost of Mis­ allocation of Investment in Education." Journal of Human REsources 12(4):446-59. Dunlop, David W. 1982. "Health Care Financing: Recent Experience in Africa." Paper pre­ pared for the Conference on Health and Development in Africa, University of Bayreuth, Germany, June. Eicher, J. C. 1984. "L'Enseignement Superieur en Afrique de l'Ouest Francophone: Synthese de Cinq Etudes de Cas." World Bank, Education and Training Department. Washington, D.C., August. _ _ _. 1985. Educational Costing and Financing in Developing Countries with Special REference to Sub-Saharan Africa. World Bank Staff Working Paper 655. Washington. D.C. Feldstein. Martin, M. A. Piot, and T. K. Sunderesan. 1973. "Resource A1Iocation Model for Public Health Planning: A Case ofThberculosis ControL" In Bulletin of the World Health Organization, vol. 48, supplement. Fields, Gary S. 1974. "Private Returns and Social Equity in the Financing of Higher Educa­ tion." In David Court and D. P. Ghai, eds., Education, Society and Development: New Perspec­ tives from Kenya. Nairobi: Oxford University Press. _ _ _. 1980. "Education and Income Distribution in Developing Countries." In Timothy King, ed., Education and Income. World Bank Staff Working Paper 402. Washington, D.C. Foxley, Alejandro, Eduardo Animat, and J. P. Arellano. 1979. Redistributive Effects of Government Programs. Oxford, Eng.: Pergamon. Friedman, Milton. 1971. "Governrnent Revenue from Inflation." Journal of Political Economy Ouly/August). Gillis, Malcolm, and C. E. McClure. 1978. "Taxation and Income Distribution: The Colom­ bian Tax Reforms of 1974." Journal of Development Economics 5:233-48. Gomez, H. 1984. Finanzas Universitarias. Bogota: Fedesarollo. Hanovice, Karen. 1984. "The Private Higher Education Sector in Indonesia: Description and 158 REFERENCES Loan Feasibility." World Bank, East Asia and Pacific Projects Department. Washington, D.C. Processed. Hauptman, A. M. 1983. "Student Loan Default Rates in Perspective." American Council on Education. Washington, D.C., February. Haveman, Robert, and Barbara Wolfe. 1984. "Schooling and Economic Well-Being: The Role of Nonmarket Effects." Journal of Human Resources 19:377 -407 . Heller, Peter S. 1975. Issues in the Costing ofPublic Sector Outputs: The Public Medical Services of Malaysia. World Bank Staff Working Paper 207, Washington, D.C. _ _ _" 1979" "The Underfinancing of Recurrent Development Costs." Finance & Develop­ ment 16(1):38-41. _ _ _, 1981. "Testing the Impact of Value-Added and Global Income Tax Reforms in Korean Tax Incidence in 1976. IMF Staff Papers 28:375-410. _ _ _. 1982, "A Model of the Demand for Medical and Health Services in Peninsular Malaysia." Social Science and Medicine 16:267-84. Heller, Peter, and Joan Aghevli, 1985. "The Recurrent Cost Problem: An International Overview." In John Howell, ed., Recurrent Costs and Agricultural Development. London: Overseas Development Institute. Heller, Peter, and Adrienne Cheasty. 1984. "Sectoral Adjustment in Government Expenditure in the 1970s: The Education Sector in Latin America." World Develcpment 12(10): 1039-49. Published by Pergamon Journals Ltd., Oxford, U.K. Herrick, Allison B., Howard R. Shavlach, and Linda Seville. 1974. Intercountry Evaluation of Education Credit Institutions in Latin America. Washington, D.C.: U.S. Agency for Inter­ national Development. Heyneman, Stephen P. 1975. "Changes in Efficiency and in Equity Accruing from Govern­ ment Involvement in Ugandan Primary Education." African Studies Review 18(1):51­ 60. Heyneman, Stephen P., Dean T. Jamison, and Xavier Montenegro. 1984. "Textbooks in the Philippines: Evaluation of the Pedagogical Impact of a Nationwide Investment." Education Evaluation and Policy Analysis 6(2):139-50. Heyneman, Stephen P., and William A. Loxley. 1983. "The Effect ofPrimary-School Quality on Achievement across Twenty-Nine High- and Low-Income Countries." American Journal of Sociology 88(6}:1162-94. Hicks, Norman, and Anne Kubisch. 1984. "Cutting Government Expenditures in IDCs." Finance & Develcpment 21(3}:37-39. Hinchliffe, Keith. 1985. Issues Related to Higher Education in Sub-Saharan Africa. World Bank Staff Working Paper 780. Washington, D.C. Horton, Susan, and Pierre Claquin. 1983. "Cost Effectiveness and User Characteristics of Clinic-Based Services for Diarrhea." Social Science and Medicine 17:721-9. International Monetary Fund. 1981. 1982. Government Financial Statistics. Washington, D.C. Jallade, Jean-Pierre. 1973. The Financing ofEducation: An Examination ofBasiclssues. World Bank Staff Working Paper 157. Washington, D.C. REFERENCES 159 Jamison. Dean T., and Lawrence]. Lau. 1982. Farmer Education and Farm Efficiency. Baltimore. Md.: Johns Hopkins University Press. Jimenez. Emmanuel. 1985. "Selecting the Brightest: The Impact of a Hypothetical Policy Reform in Colombia." World Bank, Education and Training Department. Washington, D.C. _ _ _. 1986a. "The Structure of Educational Costs: Multiproduct Cost Functions for Primary and Secondary Schools in Latin America." Economics ofEducatiOfl Review 5(1):25­ 40. _ _ _. 1986b. "The Public Subsidization ofEducation and Health in Developing Countries: A Review of Equity and Efficiency." World Bank Research Observer 1(1):111-29. Jimenez, Emmanuel, and Jee-Peng Tan. 1985. "Educational Development in Pakistan: The Role of User Charges and Private Schools." World Bank, Education and Training Depart­ ment, Discussion Paper 16. Washington, D.C., December. King. E. M., and L. A. Lillard. 1983. "Determinants of Schooling Attairunent and Emollment Rates in the Philippines." Rand Note N-1962-AID. Santa Monica. Calif.: Rand Corp. April. Kolohe. Paki. and Tsie Pekeche. 1980. A Survey of the Financi